_ZN7roaring7RoaringC2EmPKj: 85| 12.3k| Roaring(size_t n, const uint32_t *data) : Roaring() { 86| 12.3k| api::roaring_bitmap_add_many(&roaring, n, data); 87| 12.3k| } _ZN7roaring7RoaringC2Ev: 76| 18.4k| Roaring() : roaring{} { 77| | // The empty constructor roaring{} silences warnings from pedantic 78| | // static analyzers. 79| 18.4k| api::roaring_bitmap_init_cleared(&roaring); 80| 18.4k| } _ZN7roaring7RoaringD2Ev: 907| 55.3k| ~Roaring() { 908| 55.3k| if (!(roaring.high_low_container.flags & ROARING_FLAG_FROZEN)) { ------------------ | | 47| 55.3k|#define ROARING_FLAG_FROZEN UINT8_C(0x2) ------------------ | Branch (908:13): [True: 55.3k, False: 0] ------------------ 909| 55.3k| api::roaring_bitmap_clear(&roaring); 910| 55.3k| } else { 911| | // The roaring member variable copies the `roaring_bitmap_t` and 912| | // nested `roaring_array_t` structures by value and is freed in the 913| | // constructor, however the underlying memory arena used for the 914| | // container data is not freed with it. Here we derive the arena 915| | // pointer from the second arena allocation in 916| | // `roaring_bitmap_frozen_view` and free it as well. 917| 0| roaring_bitmap_free( 918| 0| (roaring_bitmap_t *)((char *) 919| 0| roaring.high_low_container.containers - 920| 0| sizeof(roaring_bitmap_t))); 921| 0| } 922| 55.3k| } _ZN7roaring7Roaring11runOptimizeEv: 487| 12.3k| bool runOptimize() noexcept { 488| 12.3k| return api::roaring_bitmap_run_optimize(&roaring); 489| 12.3k| } _ZN7roaring7Roaring11shrinkToFitEv: 495| 6.15k| size_t shrinkToFit() noexcept { 496| 6.15k| return api::roaring_bitmap_shrink_to_fit(&roaring); 497| 6.15k| } _ZN7roaring7Roaring3addEj: 198| 6.15k| void add(uint32_t x) noexcept { api::roaring_bitmap_add(&roaring, x); } _ZN7roaring7Roaring10addCheckedEj: 205| 6.15k| bool addChecked(uint32_t x) noexcept { 206| 6.15k| return api::roaring_bitmap_add_checked(&roaring, x); 207| 6.15k| } _ZN7roaring7Roaring8addRangeEmm: 212| 6.15k| void addRange(const uint64_t min, const uint64_t max) noexcept { 213| 6.15k| return api::roaring_bitmap_add_range(&roaring, min, max); 214| 6.15k| } _ZN7roaring7Roaring7addManyEmPKj: 226| 6.15k| void addMany(size_t n_args, const uint32_t *vals) noexcept { 227| 6.15k| api::roaring_bitmap_add_many(&roaring, n_args, vals); 228| 6.15k| } _ZN7roaring7Roaring6removeEj: 258| 6.15k| void remove(uint32_t x) noexcept { 259| 6.15k| api::roaring_bitmap_remove(&roaring, x); 260| 6.15k| } _ZN7roaring7Roaring13removeCheckedEj: 267| 6.15k| bool removeChecked(uint32_t x) noexcept { 268| 6.15k| return api::roaring_bitmap_remove_checked(&roaring, x); 269| 6.15k| } _ZN7roaring7Roaring11removeRangeEmm: 274| 6.15k| void removeRange(uint64_t min, uint64_t max) noexcept { 275| 6.15k| return api::roaring_bitmap_remove_range(&roaring, min, max); 276| 6.15k| } _ZN7roaring7Roaring17removeRangeClosedEjj: 281| 6.15k| void removeRangeClosed(uint32_t min, uint32_t max) noexcept { 282| 6.15k| return api::roaring_bitmap_remove_range_closed(&roaring, min, max); 283| 6.15k| } _ZNK7roaring7Roaring7maximumEv: 315| 6.15k| uint32_t maximum() const noexcept { 316| 6.15k| return api::roaring_bitmap_maximum(&roaring); 317| 6.15k| } _ZNK7roaring7Roaring7minimumEv: 322| 6.15k| uint32_t minimum() const noexcept { 323| 6.15k| return api::roaring_bitmap_minimum(&roaring); 324| 6.15k| } _ZNK7roaring7Roaring8containsEj: 329| 6.15k| bool contains(uint32_t x) const noexcept { 330| 6.15k| return api::roaring_bitmap_contains(&roaring, x); 331| 6.15k| } _ZNK7roaring7Roaring13containsRangeEmm: 336| 6.15k| bool containsRange(const uint64_t x, const uint64_t y) const noexcept { 337| 6.15k| return api::roaring_bitmap_contains_range(&roaring, x, y); 338| 6.15k| } _ZNK7roaring7Roaring6selectEjPj: 520| 6.15k| bool select(uint32_t rnk, uint32_t *element) const noexcept { 521| 6.15k| return api::roaring_bitmap_select(&roaring, rnk, element); 522| 6.15k| } _ZNK7roaring7Roaring9intersectERKS0_: 534| 6.15k| bool intersect(const Roaring &r) const noexcept { 535| 6.15k| return api::roaring_bitmap_intersect(&roaring, &r.roaring); 536| 6.15k| } _ZNK7roaring7Roaring13jaccard_indexERKS0_: 545| 6.15k| double jaccard_index(const Roaring &r) const noexcept { 546| 6.15k| return api::roaring_bitmap_jaccard_index(&roaring, &r.roaring); 547| 6.15k| } _ZNK7roaring7Roaring14or_cardinalityERKS0_: 552| 6.15k| uint64_t or_cardinality(const Roaring &r) const noexcept { 553| 6.15k| return api::roaring_bitmap_or_cardinality(&roaring, &r.roaring); 554| 6.15k| } _ZNK7roaring7Roaring18andnot_cardinalityERKS0_: 559| 6.15k| uint64_t andnot_cardinality(const Roaring &r) const noexcept { 560| 6.15k| return api::roaring_bitmap_andnot_cardinality(&roaring, &r.roaring); 561| 6.15k| } _ZNK7roaring7Roaring15xor_cardinalityERKS0_: 567| 6.15k| uint64_t xor_cardinality(const Roaring &r) const noexcept { 568| 6.15k| return api::roaring_bitmap_xor_cardinality(&roaring, &r.roaring); 569| 6.15k| } _ZNK7roaring7Roaring4rankEj: 579| 6.15k| uint64_t rank(uint32_t x) const noexcept { 580| 6.15k| return api::roaring_bitmap_rank(&roaring, x); 581| 6.15k| } _ZNK7roaring7Roaring14getSizeInBytesEb: 726| 12.3k| size_t getSizeInBytes(bool portable = true) const noexcept { 727| 12.3k| if (portable) { ------------------ | Branch (727:13): [True: 12.3k, False: 0] ------------------ 728| 12.3k| return api::roaring_bitmap_portable_size_in_bytes(&roaring); 729| 12.3k| } else { 730| 0| return api::roaring_bitmap_size_in_bytes(&roaring); 731| 0| } 732| 12.3k| } _ZNK7roaring7RoaringanERKS0_: 788| 6.15k| Roaring operator&(const Roaring &o) const { 789| 6.15k| roaring_bitmap_t *r = api::roaring_bitmap_and(&roaring, &o.roaring); 790| 6.15k| if (r == NULL) { ------------------ | Branch (790:13): [True: 0, False: 6.15k] ------------------ 791| 0| ROARING_TERMINATE("failed materalization in and"); ------------------ | | 27| 0|#define ROARING_TERMINATE(_s) throw std::runtime_error(_s) ------------------ 792| 0| } 793| 6.15k| return Roaring(r); 794| 6.15k| } _ZN7roaring7RoaringC2EPNS_3api16roaring_bitmap_sE: 102| 30.7k| explicit Roaring(roaring_bitmap_t *s) noexcept : roaring(*s) { 103| 30.7k| roaring_free(s); // deallocate the passed-in pointer 104| 30.7k| } _ZNK7roaring7RoaringmiERKS0_: 801| 6.15k| Roaring operator-(const Roaring &o) const { 802| 6.15k| roaring_bitmap_t *r = api::roaring_bitmap_andnot(&roaring, &o.roaring); 803| 6.15k| if (r == NULL) { ------------------ | Branch (803:13): [True: 0, False: 6.15k] ------------------ 804| 0| ROARING_TERMINATE("failed materalization in andnot"); ------------------ | | 27| 0|#define ROARING_TERMINATE(_s) throw std::runtime_error(_s) ------------------ 805| 0| } 806| 6.15k| return Roaring(r); 807| 6.15k| } _ZNK7roaring7RoaringorERKS0_: 814| 6.15k| Roaring operator|(const Roaring &o) const { 815| 6.15k| roaring_bitmap_t *r = api::roaring_bitmap_or(&roaring, &o.roaring); 816| 6.15k| if (r == NULL) { ------------------ | Branch (816:13): [True: 0, False: 6.15k] ------------------ 817| 0| ROARING_TERMINATE("failed materalization in or"); ------------------ | | 27| 0|#define ROARING_TERMINATE(_s) throw std::runtime_error(_s) ------------------ 818| 0| } 819| 6.15k| return Roaring(r); 820| 6.15k| } _ZNK7roaring7RoaringeoERKS0_: 827| 6.15k| Roaring operator^(const Roaring &o) const { 828| 6.15k| roaring_bitmap_t *r = api::roaring_bitmap_xor(&roaring, &o.roaring); 829| 6.15k| if (r == NULL) { ------------------ | Branch (829:13): [True: 0, False: 6.15k] ------------------ 830| 0| ROARING_TERMINATE("failed materalization in xor"); ------------------ | | 27| 0|#define ROARING_TERMINATE(_s) throw std::runtime_error(_s) ------------------ 831| 0| } 832| 6.15k| return Roaring(r); 833| 6.15k| } _ZN7roaring7RoaringoRERKS0_: 375| 6.15k| Roaring &operator|=(const Roaring &r) noexcept { 376| 6.15k| api::roaring_bitmap_or_inplace(&roaring, &r.roaring); 377| 6.15k| return *this; 378| 6.15k| } _ZN7roaring7RoaringaNERKS0_: 353| 6.15k| Roaring &operator&=(const Roaring &r) noexcept { 354| 6.15k| api::roaring_bitmap_and_inplace(&roaring, &r.roaring); 355| 6.15k| return *this; 356| 6.15k| } _ZN7roaring7RoaringmIERKS0_: 363| 6.15k| Roaring &operator-=(const Roaring &r) noexcept { 364| 6.15k| api::roaring_bitmap_andnot_inplace(&roaring, &r.roaring); 365| 6.15k| return *this; 366| 6.15k| } _ZN7roaring7RoaringeOERKS0_: 385| 6.15k| Roaring &operator^=(const Roaring &r) noexcept { 386| 6.15k| api::roaring_bitmap_xor_inplace(&roaring, &r.roaring); 387| 6.15k| return *this; 388| 6.15k| } _ZNK7roaring7RoaringeqERKS0_: 452| 12.3k| bool operator==(const Roaring &r) const noexcept { 453| 12.3k| return api::roaring_bitmap_equals(&roaring, &r.roaring); 454| 12.3k| } _ZNK7roaring7Roaring11cardinalityEv: 398| 6.15k| uint64_t cardinality() const noexcept { 399| 6.15k| return api::roaring_bitmap_get_cardinality(&roaring); 400| 6.15k| } _ZNK7roaring7Roaring7isEmptyEv: 405| 12.3k| bool isEmpty() const noexcept { 406| 12.3k| return api::roaring_bitmap_is_empty(&roaring); 407| 12.3k| } _ZNK7roaring7Roaring13toUint32ArrayEPj: 438| 6.15k| void toUint32Array(uint32_t *ans) const noexcept { 439| 6.15k| api::roaring_bitmap_to_uint32_array(&roaring, ans); 440| 6.15k| } _ZNK7roaring7Roaring8isSubsetERKS0_: 422| 6.15k| bool isSubset(const Roaring &r) const noexcept { 423| 6.15k| return api::roaring_bitmap_is_subset(&roaring, &r.roaring); 424| 6.15k| } _ZNK7roaring7Roaring14isStrictSubsetERKS0_: 429| 6.15k| bool isStrictSubset(const Roaring &r) const noexcept { 430| 6.15k| return api::roaring_bitmap_is_strict_subset(&roaring, &r.roaring); 431| 6.15k| } _ZN7roaring7Roaring4flipEmm: 460| 6.15k| void flip(uint64_t range_start, uint64_t range_end) noexcept { 461| 6.15k| api::roaring_bitmap_flip_inplace(&roaring, range_start, range_end); 462| 6.15k| } _ZN7roaring7Roaring10flipClosedEjj: 468| 6.15k| void flipClosed(uint32_t range_start, uint32_t range_end) noexcept { 469| 6.15k| api::roaring_bitmap_flip_inplace_closed(&roaring, range_start, 470| 6.15k| range_end); 471| 6.15k| } _ZN7roaring7Roaring20removeRunCompressionEv: 477| 6.15k| bool removeRunCompression() noexcept { 478| 6.15k| return api::roaring_bitmap_remove_run_compression(&roaring); 479| 6.15k| } _ZN7roaring7RoaringC2ERKS0_: 110| 6.15k| Roaring(const Roaring &r) : Roaring() { 111| 6.15k| if (!api::roaring_bitmap_overwrite(&roaring, &r.roaring)) { ------------------ | Branch (111:13): [True: 0, False: 6.15k] ------------------ 112| 0| ROARING_TERMINATE("failed roaring_bitmap_overwrite in constructor"); ------------------ | | 27| 0|#define ROARING_TERMINATE(_s) throw std::runtime_error(_s) ------------------ 113| 0| } 114| 6.15k| api::roaring_bitmap_set_copy_on_write( 115| 6.15k| &roaring, api::roaring_bitmap_get_copy_on_write(&r.roaring)); 116| 6.15k| } _ZN7roaring7RoaringC2EOS0_: 122| 6.15k| Roaring(Roaring &&r) noexcept : roaring(r.roaring) { 123| | // 124| | // !!! This clones the bits of the roaring structure to a new location 125| | // and then overwrites the old bits...assuming that this will still 126| | // work. There are scenarios where this could break; e.g. if some of 127| | // those bits were pointers into the structure memory itself. If such 128| | // things were possible, a roaring_bitmap_move() API would be needed. 129| | // 130| 6.15k| api::roaring_bitmap_init_cleared(&r.roaring); 131| 6.15k| } _ZN7roaring7RoaringaSERKS0_: 152| 6.15k| Roaring &operator=(const Roaring &r) { 153| 6.15k| if (!api::roaring_bitmap_overwrite(&roaring, &r.roaring)) { ------------------ | Branch (153:13): [True: 0, False: 6.15k] ------------------ 154| 0| ROARING_TERMINATE("failed memory alloc in assignment"); ------------------ | | 27| 0|#define ROARING_TERMINATE(_s) throw std::runtime_error(_s) ------------------ 155| 0| } 156| 6.15k| api::roaring_bitmap_set_copy_on_write( 157| 6.15k| &roaring, api::roaring_bitmap_get_copy_on_write(&r.roaring)); 158| 6.15k| return *this; 159| 6.15k| } _ZN7roaring7RoaringaSEOS0_: 165| 6.15k| Roaring &operator=(Roaring &&r) noexcept { 166| 6.15k| api::roaring_bitmap_clear(&roaring); // free this class's allocations 167| | 168| | // !!! See notes in the Move Constructor regarding roaring_bitmap_move() 169| | // 170| 6.15k| roaring = r.roaring; 171| 6.15k| api::roaring_bitmap_init_cleared(&r.roaring); 172| | 173| 6.15k| return *this; 174| 6.15k| } _ZN7roaring7Roaring8readSafeEPKcm: 700| 12.3k| static Roaring readSafe(const char *buf, size_t maxbytes) { 701| 12.3k| roaring_bitmap_t *r = 702| 12.3k| api::roaring_bitmap_portable_deserialize_safe(buf, maxbytes); 703| 12.3k| if (r == NULL) { ------------------ | Branch (703:13): [True: 6.14k, False: 6.16k] ------------------ 704| 6.14k| ROARING_TERMINATE("failed alloc while reading"); ------------------ | | 27| 6.14k|#define ROARING_TERMINATE(_s) throw std::runtime_error(_s) ------------------ 705| 6.14k| } 706| 6.16k| return Roaring(r); 707| 12.3k| } _ZNK7roaring7Roaring5writeEPcb: 643| 6.15k| size_t write(char *buf, bool portable = true) const noexcept { 644| 6.15k| if (portable) { ------------------ | Branch (644:13): [True: 6.15k, False: 0] ------------------ 645| 6.15k| return api::roaring_bitmap_portable_serialize(&roaring, buf); 646| 6.15k| } else { 647| 0| return api::roaring_bitmap_serialize(&roaring, buf); 648| 0| } 649| 6.15k| } _ZNK7roaring7Roaring8toStringEv: 850| 6.15k| std::string toString() const noexcept { 851| 6.15k| struct iter_data { 852| 6.15k| std::string str{}; // The empty constructor silences warnings from 853| | // pedantic static analyzers. 854| 6.15k| char first_char = '{'; 855| 6.15k| } outer_iter_data; 856| 6.15k| if (!isEmpty()) { ------------------ | Branch (856:13): [True: 6.14k, False: 13] ------------------ 857| 6.14k| iterate( 858| 6.14k| [](uint32_t value, void *inner_iter_data) -> bool { 859| 6.14k| ((iter_data *)inner_iter_data)->str += 860| 6.14k| ((iter_data *)inner_iter_data)->first_char; 861| 6.14k| ((iter_data *)inner_iter_data)->str += 862| 6.14k| std::to_string(value); 863| 6.14k| ((iter_data *)inner_iter_data)->first_char = ','; 864| 6.14k| return true; 865| 6.14k| }, 866| 6.14k| (void *)&outer_iter_data); 867| 6.14k| } else 868| 13| outer_iter_data.str = '{'; 869| 6.15k| outer_iter_data.str += '}'; 870| 6.15k| return outer_iter_data.str; 871| 6.15k| } _ZNK7roaring7Roaring7iterateEPFbjPvES1_: 508| 6.14k| void iterate(api::roaring_iterator iterator, void *ptr) const { 509| 6.14k| api::roaring_iterate(&roaring, iterator, ptr); 510| 6.14k| } _ZZNK7roaring7Roaring8toStringEvENKUljPvE_clEjS1_: 858| 3.59G| [](uint32_t value, void *inner_iter_data) -> bool { 859| 3.59G| ((iter_data *)inner_iter_data)->str += 860| 3.59G| ((iter_data *)inner_iter_data)->first_char; 861| 3.59G| ((iter_data *)inner_iter_data)->str += 862| 3.59G| std::to_string(value); 863| 3.59G| ((iter_data *)inner_iter_data)->first_char = ','; 864| 3.59G| return true; 865| 3.59G| }, _ZNK7roaring7Roaring5beginEv: 1067| 12.3k|inline RoaringSetBitBiDirectionalIterator Roaring::begin() const { 1068| 12.3k| return RoaringSetBitBiDirectionalIterator(*this); 1069| 12.3k|} _ZN7roaring34RoaringSetBitBiDirectionalIteratorC2ERKNS_7RoaringEb: 961| 12.3k| bool exhausted = false) { 962| 12.3k| if (exhausted) { ------------------ | Branch (962:13): [True: 1, False: 12.3k] ------------------ 963| 1| i.parent = &parent.roaring; 964| 1| i.container_index = INT32_MAX; 965| 1| i.has_value = false; 966| 1| i.current_value = UINT32_MAX; 967| 12.3k| } else { 968| 12.3k| api::roaring_iterator_init(&parent.roaring, &i); 969| 12.3k| } 970| 12.3k| } _ZNK7roaring34RoaringSetBitBiDirectionalIteratorneERKS0_: 1058| 326k| bool operator!=(const RoaringSetBitBiDirectionalIterator &o) const { 1059| 326k| return i.current_value != *o || i.has_value != o.i.has_value; ------------------ | Branch (1059:16): [True: 320k, False: 6.15k] | Branch (1059:41): [True: 0, False: 6.15k] ------------------ 1060| 326k| } _ZNK7roaring34RoaringSetBitBiDirectionalIteratordeEv: 975| 326k| value_type operator*() const { return i.current_value; } _ZNK7roaring7Roaring3endEv: 1071| 326k|inline RoaringSetBitBiDirectionalIterator &Roaring::end() const { 1072| 326k| static RoaringSetBitBiDirectionalIterator e(*this, true); 1073| 326k| return e; 1074| 326k|} _ZN7roaring34RoaringSetBitBiDirectionalIteratorppEi: 1006| 320k| type_of_iterator operator++(int) { // i++, must return orig. value 1007| 320k| RoaringSetBitBiDirectionalIterator orig(*this); 1008| 320k| api::roaring_uint32_iterator_advance(&i); 1009| 320k| return orig; 1010| 320k| } _ZN7roaring34RoaringSetBitBiDirectionalIterator13equalorlargerEj: 1021| 6.15k| CROARING_DEPRECATED void equalorlarger(uint32_t val) { 1022| 6.15k| api::roaring_uint32_iterator_move_equalorlarger(&i, val); 1023| 6.15k| } binarySearch: 52| 7.04M| uint16_t ikey) { 53| 7.04M| const int32_t gap = 16; 54| 7.04M| if (lenarray < gap) { ------------------ | Branch (54:9): [True: 2.77M, False: 4.27M] ------------------ 55| 4.21M| for (int32_t j = 0; j < lenarray; j++) { ------------------ | Branch (55:29): [True: 4.20M, False: 7.25k] ------------------ 56| 4.20M| if (array[j] >= ikey) { ------------------ | Branch (56:17): [True: 2.76M, False: 1.44M] ------------------ 57| 2.76M| return (array[j] == ikey) ? j : -(j + 1); ------------------ | Branch (57:24): [True: 2.61M, False: 145k] ------------------ 58| 2.76M| } 59| 4.20M| } 60| 7.25k| return -(lenarray + 1); 61| 2.77M| } 62| 4.27M| const int32_t num_blocks = lenarray / gap; 63| 4.27M| int32_t base = 0; 64| 4.27M| int32_t n = num_blocks; 65| 8.56M| while (n > 3) { ------------------ | Branch (65:12): [True: 4.28M, False: 4.27M] ------------------ 66| 4.28M| int32_t quarter = n >> 2; 67| | 68| 4.28M| int32_t k1 = array[(base + quarter + 1) * gap - 1]; 69| 4.28M| int32_t k2 = array[(base + 2 * quarter + 1) * gap - 1]; 70| 4.28M| int32_t k3 = array[(base + 3 * quarter + 1) * gap - 1]; 71| | 72| 4.28M| int32_t c1 = (k1 < ikey); 73| 4.28M| int32_t c2 = (k2 < ikey); 74| 4.28M| int32_t c3 = (k3 < ikey); 75| | 76| 4.28M| base += (c1 + c2 + c3) * quarter; 77| 4.28M| n -= 3 * quarter; 78| 4.28M| } 79| 7.91M| while (n > 1) { ------------------ | Branch (79:12): [True: 3.63M, False: 4.27M] ------------------ 80| 3.63M| int32_t half = n >> 1; 81| 3.63M| base = (array[(base + half + 1) * gap - 1] < ikey) ? base + half : base; ------------------ | Branch (81:16): [True: 508k, False: 3.12M] ------------------ 82| 3.63M| n -= half; 83| 3.63M| } 84| 4.27M| int32_t lo = (array[(base + 1) * gap - 1] < ikey) ? base + 1 : base; ------------------ | Branch (84:18): [True: 1.98M, False: 2.28M] ------------------ 85| | 86| 4.27M| if (lo < num_blocks) { ------------------ | Branch (86:9): [True: 3.97M, False: 305k] ------------------ 87| 3.97M| const int32_t start = lo * gap; 88| 3.97M|#if defined(CROARING_IS_X64) 89| | // SSE2: subs_epu16 yields zero where lane >= ikey. movemask of an 90| | // epi16 compare gives 2 bits per lane; ctz>>1 = lane index. Scan 91| | // the first 8 lanes first and exit early when they contain the 92| | // answer; otherwise the block-narrowing invariant guarantees the 93| | // second-half mask is non-zero. 94| 3.97M| __m128i needle = _mm_set1_epi16((short)ikey); 95| 3.97M| __m128i zero = _mm_setzero_si128(); 96| 3.97M| __m128i v0 = _mm_loadu_si128((const __m128i *)(array + start)); 97| 3.97M| __m128i ge0 = _mm_cmpeq_epi16(_mm_subs_epu16(needle, v0), zero); 98| 3.97M| unsigned m0 = (unsigned)_mm_movemask_epi8(ge0); 99| 3.97M| if (m0 != 0) { ------------------ | Branch (99:13): [True: 2.82M, False: 1.14M] ------------------ 100| 2.82M| int32_t j = start + (int32_t)(roaring_trailing_zeroes(m0) >> 1); 101| 2.82M| return (array[j] == ikey) ? j : -(j + 1); ------------------ | Branch (101:20): [True: 2.09M, False: 731k] ------------------ 102| 2.82M| } 103| 1.14M| __m128i v1 = _mm_loadu_si128((const __m128i *)(array + start + 8)); 104| 1.14M| __m128i ge1 = _mm_cmpeq_epi16(_mm_subs_epu16(needle, v1), zero); 105| 1.14M| unsigned m1 = (unsigned)_mm_movemask_epi8(ge1); 106| 1.14M| int32_t j = start + 8 + (int32_t)(roaring_trailing_zeroes(m1) >> 1); 107| 1.14M| return (array[j] == ikey) ? j : -(j + 1); ------------------ | Branch (107:16): [True: 502k, False: 641k] ------------------ 108| |#else 109| | const int32_t end = start + gap; 110| | for (int32_t j = start; j < end; j++) { 111| | if (array[j] >= ikey) { 112| | return (array[j] == ikey) ? j : -(j + 1); 113| | } 114| | } 115| | // Unreachable: the narrowing guarantees the last element of the 116| | // selected block is >= ikey. 117| | return -(end + 1); 118| |#endif 119| 3.97M| } 120| | 121| 1.74M| for (int32_t j = num_blocks * gap; j < lenarray; j++) { ------------------ | Branch (121:40): [True: 1.73M, False: 10.6k] ------------------ 122| 1.73M| if (array[j] >= ikey) { ------------------ | Branch (122:13): [True: 294k, False: 1.43M] ------------------ 123| 294k| return (array[j] == ikey) ? j : -(j + 1); ------------------ | Branch (123:20): [True: 164k, False: 130k] ------------------ 124| 294k| } 125| 1.73M| } 126| 10.6k| return -(lenarray + 1); 127| 305k|} roaring.c:advanceUntil: 137| 518| int32_t length, uint16_t min) { 138| 518| int32_t lower = pos + 1; 139| | 140| 518| if ((lower >= length) || (array[lower] >= min)) { ------------------ | Branch (140:9): [True: 518, False: 0] | Branch (140:30): [True: 0, False: 0] ------------------ 141| 518| return lower; 142| 518| } 143| | 144| 0| int32_t spansize = 1; 145| | 146| 0| while ((lower + spansize < length) && (array[lower + spansize] < min)) { ------------------ | Branch (146:12): [True: 0, False: 0] | Branch (146:43): [True: 0, False: 0] ------------------ 147| 0| spansize <<= 1; 148| 0| } 149| 0| int32_t upper = (lower + spansize < length) ? lower + spansize : length - 1; ------------------ | Branch (149:21): [True: 0, False: 0] ------------------ 150| | 151| 0| if (array[upper] == min) { ------------------ | Branch (151:9): [True: 0, False: 0] ------------------ 152| 0| return upper; 153| 0| } 154| 0| if (array[upper] < min) { ------------------ | Branch (154:9): [True: 0, False: 0] ------------------ 155| | // means 156| | // array 157| | // has no 158| | // item 159| | // >= min 160| | // pos = array.length; 161| 0| return length; 162| 0| } 163| | 164| | // we know that the next-smallest span was too small 165| 0| lower += (spansize >> 1); 166| | 167| 0| int32_t mid = 0; 168| 0| while (lower + 1 != upper) { ------------------ | Branch (168:12): [True: 0, False: 0] ------------------ 169| 0| mid = (lower + upper) >> 1; 170| 0| if (array[mid] == min) { ------------------ | Branch (170:13): [True: 0, False: 0] ------------------ 171| 0| return mid; 172| 0| } else if (array[mid] < min) { ------------------ | Branch (172:20): [True: 0, False: 0] ------------------ 173| 0| lower = mid; 174| 0| } else { 175| 0| upper = mid; 176| 0| } 177| 0| } 178| 0| return upper; 179| 0|} roaring.c:count_greater: 197| 8.80k| uint16_t ikey) { 198| 8.80k| if (lenarray == 0) return 0; ------------------ | Branch (198:9): [True: 0, False: 8.80k] ------------------ 199| 8.80k| int32_t pos = binarySearch(array, lenarray, ikey); 200| 8.80k| if (pos >= 0) { ------------------ | Branch (200:9): [True: 1.92k, False: 6.88k] ------------------ 201| 1.92k| return lenarray - (pos + 1); 202| 6.88k| } else { 203| 6.88k| return lenarray - (-pos - 1); 204| 6.88k| } 205| 8.80k|} roaring.c:count_less: 186| 15.2k| uint16_t ikey) { 187| 15.2k| if (lenarray == 0) return 0; ------------------ | Branch (187:9): [True: 2.93k, False: 12.2k] ------------------ 188| 12.2k| int32_t pos = binarySearch(array, lenarray, ikey); 189| 12.2k| return pos >= 0 ? pos : -(pos + 1); ------------------ | Branch (189:12): [True: 9.23k, False: 3.02k] ------------------ 190| 15.2k|} array_util.c:advanceUntil: 137| 17.2k| int32_t length, uint16_t min) { 138| 17.2k| int32_t lower = pos + 1; 139| | 140| 17.2k| if ((lower >= length) || (array[lower] >= min)) { ------------------ | Branch (140:9): [True: 17, False: 17.2k] | Branch (140:30): [True: 6.19k, False: 11.0k] ------------------ 141| 6.21k| return lower; 142| 6.21k| } 143| | 144| 11.0k| int32_t spansize = 1; 145| | 146| 48.2k| while ((lower + spansize < length) && (array[lower + spansize] < min)) { ------------------ | Branch (146:12): [True: 47.0k, False: 1.19k] | Branch (146:43): [True: 37.1k, False: 9.89k] ------------------ 147| 37.1k| spansize <<= 1; 148| 37.1k| } 149| 11.0k| int32_t upper = (lower + spansize < length) ? lower + spansize : length - 1; ------------------ | Branch (149:21): [True: 9.89k, False: 1.19k] ------------------ 150| | 151| 11.0k| if (array[upper] == min) { ------------------ | Branch (151:9): [True: 2.56k, False: 8.51k] ------------------ 152| 2.56k| return upper; 153| 2.56k| } 154| 8.51k| if (array[upper] < min) { ------------------ | Branch (154:9): [True: 165, False: 8.34k] ------------------ 155| | // means 156| | // array 157| | // has no 158| | // item 159| | // >= min 160| | // pos = array.length; 161| 165| return length; 162| 165| } 163| | 164| | // we know that the next-smallest span was too small 165| 8.34k| lower += (spansize >> 1); 166| | 167| 8.34k| int32_t mid = 0; 168| 26.5k| while (lower + 1 != upper) { ------------------ | Branch (168:12): [True: 22.5k, False: 4.01k] ------------------ 169| 22.5k| mid = (lower + upper) >> 1; 170| 22.5k| if (array[mid] == min) { ------------------ | Branch (170:13): [True: 4.33k, False: 18.2k] ------------------ 171| 4.33k| return mid; 172| 18.2k| } else if (array[mid] < min) { ------------------ | Branch (172:20): [True: 9.19k, False: 9.03k] ------------------ 173| 9.19k| lower = mid; 174| 9.19k| } else { 175| 9.03k| upper = mid; 176| 9.03k| } 177| 22.5k| } 178| 4.01k| return upper; 179| 8.34k|} mixed_intersection.c:advanceUntil: 137| 128k| int32_t length, uint16_t min) { 138| 128k| int32_t lower = pos + 1; 139| | 140| 128k| if ((lower >= length) || (array[lower] >= min)) { ------------------ | Branch (140:9): [True: 494, False: 127k] | Branch (140:30): [True: 72.4k, False: 55.3k] ------------------ 141| 72.8k| return lower; 142| 72.8k| } 143| | 144| 55.3k| int32_t spansize = 1; 145| | 146| 131k| while ((lower + spansize < length) && (array[lower + spansize] < min)) { ------------------ | Branch (146:12): [True: 130k, False: 1.15k] | Branch (146:43): [True: 76.1k, False: 54.2k] ------------------ 147| 76.1k| spansize <<= 1; 148| 76.1k| } 149| 55.3k| int32_t upper = (lower + spansize < length) ? lower + spansize : length - 1; ------------------ | Branch (149:21): [True: 54.2k, False: 1.15k] ------------------ 150| | 151| 55.3k| if (array[upper] == min) { ------------------ | Branch (151:9): [True: 26.8k, False: 28.4k] ------------------ 152| 26.8k| return upper; 153| 26.8k| } 154| 28.4k| if (array[upper] < min) { ------------------ | Branch (154:9): [True: 538, False: 27.9k] ------------------ 155| | // means 156| | // array 157| | // has no 158| | // item 159| | // >= min 160| | // pos = array.length; 161| 538| return length; 162| 538| } 163| | 164| | // we know that the next-smallest span was too small 165| 27.9k| lower += (spansize >> 1); 166| | 167| 27.9k| int32_t mid = 0; 168| 49.4k| while (lower + 1 != upper) { ------------------ | Branch (168:12): [True: 33.6k, False: 15.8k] ------------------ 169| 33.6k| mid = (lower + upper) >> 1; 170| 33.6k| if (array[mid] == min) { ------------------ | Branch (170:13): [True: 12.1k, False: 21.5k] ------------------ 171| 12.1k| return mid; 172| 21.5k| } else if (array[mid] < min) { ------------------ | Branch (172:20): [True: 8.12k, False: 13.4k] ------------------ 173| 8.12k| lower = mid; 174| 13.4k| } else { 175| 13.4k| upper = mid; 176| 13.4k| } 177| 33.6k| } 178| 15.8k| return upper; 179| 27.9k|} mixed_subset.c:advanceUntil: 137| 380| int32_t length, uint16_t min) { 138| 380| int32_t lower = pos + 1; 139| | 140| 380| if ((lower >= length) || (array[lower] >= min)) { ------------------ | Branch (140:9): [True: 0, False: 380] | Branch (140:30): [True: 192, False: 188] ------------------ 141| 192| return lower; 142| 192| } 143| | 144| 188| int32_t spansize = 1; 145| | 146| 922| while ((lower + spansize < length) && (array[lower + spansize] < min)) { ------------------ | Branch (146:12): [True: 836, False: 86] | Branch (146:43): [True: 734, False: 102] ------------------ 147| 734| spansize <<= 1; 148| 734| } 149| 188| int32_t upper = (lower + spansize < length) ? lower + spansize : length - 1; ------------------ | Branch (149:21): [True: 102, False: 86] ------------------ 150| | 151| 188| if (array[upper] == min) { ------------------ | Branch (151:9): [True: 0, False: 188] ------------------ 152| 0| return upper; 153| 0| } 154| 188| if (array[upper] < min) { ------------------ | Branch (154:9): [True: 10, False: 178] ------------------ 155| | // means 156| | // array 157| | // has no 158| | // item 159| | // >= min 160| | // pos = array.length; 161| 10| return length; 162| 10| } 163| | 164| | // we know that the next-smallest span was too small 165| 178| lower += (spansize >> 1); 166| | 167| 178| int32_t mid = 0; 168| 638| while (lower + 1 != upper) { ------------------ | Branch (168:12): [True: 460, False: 178] ------------------ 169| 460| mid = (lower + upper) >> 1; 170| 460| if (array[mid] == min) { ------------------ | Branch (170:13): [True: 0, False: 460] ------------------ 171| 0| return mid; 172| 460| } else if (array[mid] < min) { ------------------ | Branch (172:20): [True: 270, False: 190] ------------------ 173| 270| lower = mid; 174| 270| } else { 175| 190| upper = mid; 176| 190| } 177| 460| } 178| 178| return upper; 179| 178|} mixed_andnot.c:advanceUntil: 137| 89.9k| int32_t length, uint16_t min) { 138| 89.9k| int32_t lower = pos + 1; 139| | 140| 89.9k| if ((lower >= length) || (array[lower] >= min)) { ------------------ | Branch (140:9): [True: 4.76k, False: 85.1k] | Branch (140:30): [True: 72.3k, False: 12.8k] ------------------ 141| 77.0k| return lower; 142| 77.0k| } 143| | 144| 12.8k| int32_t spansize = 1; 145| | 146| 25.9k| while ((lower + spansize < length) && (array[lower + spansize] < min)) { ------------------ | Branch (146:12): [True: 25.4k, False: 497] | Branch (146:43): [True: 13.0k, False: 12.3k] ------------------ 147| 13.0k| spansize <<= 1; 148| 13.0k| } 149| 12.8k| int32_t upper = (lower + spansize < length) ? lower + spansize : length - 1; ------------------ | Branch (149:21): [True: 12.3k, False: 497] ------------------ 150| | 151| 12.8k| if (array[upper] == min) { ------------------ | Branch (151:9): [True: 6.37k, False: 6.48k] ------------------ 152| 6.37k| return upper; 153| 6.37k| } 154| 6.48k| if (array[upper] < min) { ------------------ | Branch (154:9): [True: 446, False: 6.04k] ------------------ 155| | // means 156| | // array 157| | // has no 158| | // item 159| | // >= min 160| | // pos = array.length; 161| 446| return length; 162| 446| } 163| | 164| | // we know that the next-smallest span was too small 165| 6.04k| lower += (spansize >> 1); 166| | 167| 6.04k| int32_t mid = 0; 168| 9.31k| while (lower + 1 != upper) { ------------------ | Branch (168:12): [True: 5.41k, False: 3.90k] ------------------ 169| 5.41k| mid = (lower + upper) >> 1; 170| 5.41k| if (array[mid] == min) { ------------------ | Branch (170:13): [True: 2.14k, False: 3.27k] ------------------ 171| 2.14k| return mid; 172| 3.27k| } else if (array[mid] < min) { ------------------ | Branch (172:20): [True: 1.15k, False: 2.11k] ------------------ 173| 1.15k| lower = mid; 174| 2.11k| } else { 175| 2.11k| upper = mid; 176| 2.11k| } 177| 5.41k| } 178| 3.90k| return upper; 179| 6.04k|} roaring.c:bitset_lenrange_cardinality: 63| 1.27k| uint32_t lenminusone) { 64| 1.27k| uint32_t firstword = start / 64; 65| 1.27k| uint32_t endword = (start + lenminusone) / 64; 66| 1.27k| if (firstword == endword) { ------------------ | Branch (66:9): [True: 892, False: 386] ------------------ 67| 892| return roaring_hamming(words[firstword] & 68| 892| ((~UINT64_C(0)) >> ((63 - lenminusone) % 64)) 69| 892| << (start % 64)); 70| 892| } 71| 386| int answer = 72| 386| roaring_hamming(words[firstword] & ((~UINT64_C(0)) << (start % 64))); 73| 195k| for (uint32_t i = firstword + 1; i < endword; i++) { ------------------ | Branch (73:38): [True: 194k, False: 386] ------------------ 74| 194k| answer += roaring_hamming(words[i]); 75| 194k| } 76| 386| answer += roaring_hamming(words[endword] & 77| | (~UINT64_C(0)) >> 78| 386| (((~start + 1) - lenminusone - 1) % 64)); 79| 386| return answer; 80| 1.27k|} roaring.c:bitset_reset_range: 147| 1.23k| uint32_t end) { 148| 1.23k| if (start == end) return; ------------------ | Branch (148:9): [True: 0, False: 1.23k] ------------------ 149| 1.23k| uint32_t firstword = start / 64; 150| 1.23k| uint32_t endword = (end - 1) / 64; 151| 1.23k| if (firstword == endword) { ------------------ | Branch (151:9): [True: 892, False: 347] ------------------ 152| 892| words[firstword] &= ~(((~UINT64_C(0)) << (start % 64)) & 153| 892| ((~UINT64_C(0)) >> ((~end + 1) % 64))); 154| 892| return; 155| 892| } 156| 347| words[firstword] &= ~((~UINT64_C(0)) << (start % 64)); 157| 161k| for (uint32_t i = firstword + 1; i < endword; i++) { ------------------ | Branch (157:38): [True: 161k, False: 347] ------------------ 158| 161k| words[i] = UINT64_C(0); 159| 161k| } 160| | words[endword] &= ~((~UINT64_C(0)) >> ((~end + 1) % 64)); 161| 347|} roaring.c:bitset_set_lenrange: 112| 1.45k| uint32_t lenminusone) { 113| 1.45k| uint32_t firstword = start / 64; 114| 1.45k| uint32_t endword = (start + lenminusone) / 64; 115| 1.45k| if (firstword == endword) { ------------------ | Branch (115:9): [True: 0, False: 1.45k] ------------------ 116| 0| words[firstword] |= ((~UINT64_C(0)) >> ((63 - lenminusone) % 64)) 117| 0| << (start % 64); 118| 0| return; 119| 0| } 120| 1.45k| uint64_t temp = words[endword]; 121| 1.45k| words[firstword] |= (~UINT64_C(0)) << (start % 64); 122| 424k| for (uint32_t i = firstword + 1; i < endword; i += 2) ------------------ | Branch (122:38): [True: 422k, False: 1.45k] ------------------ 123| 422k| words[i] = words[i + 1] = ~UINT64_C(0); 124| 1.45k| words[endword] = 125| | temp | (~UINT64_C(0)) >> (((~start + 1) - lenminusone - 1) % 64); 126| 1.45k|} bitset.c:avx2_harley_seal_popcount256: 340| 3.19k| const uint64_t size) { 341| 3.19k| __m256i total = _mm256_setzero_si256(); 342| 3.19k| __m256i ones = _mm256_setzero_si256(); 343| 3.19k| __m256i twos = _mm256_setzero_si256(); 344| 3.19k| __m256i fours = _mm256_setzero_si256(); 345| 3.19k| __m256i eights = _mm256_setzero_si256(); 346| 3.19k| __m256i sixteens = _mm256_setzero_si256(); 347| 3.19k| __m256i twosA, twosB, foursA, foursB, eightsA, eightsB; 348| | 349| 3.19k| const uint64_t limit = size - size % 16; 350| 3.19k| uint64_t i = 0; 351| | 352| 54.3k| for (; i < limit; i += 16) { ------------------ | Branch (352:12): [True: 51.1k, False: 3.19k] ------------------ 353| 51.1k| CSA(&twosA, &ones, ones, _mm256_lddqu_si256(data + i), 354| 51.1k| _mm256_lddqu_si256(data + i + 1)); 355| 51.1k| CSA(&twosB, &ones, ones, _mm256_lddqu_si256(data + i + 2), 356| 51.1k| _mm256_lddqu_si256(data + i + 3)); 357| 51.1k| CSA(&foursA, &twos, twos, twosA, twosB); 358| 51.1k| CSA(&twosA, &ones, ones, _mm256_lddqu_si256(data + i + 4), 359| 51.1k| _mm256_lddqu_si256(data + i + 5)); 360| 51.1k| CSA(&twosB, &ones, ones, _mm256_lddqu_si256(data + i + 6), 361| 51.1k| _mm256_lddqu_si256(data + i + 7)); 362| 51.1k| CSA(&foursB, &twos, twos, twosA, twosB); 363| 51.1k| CSA(&eightsA, &fours, fours, foursA, foursB); 364| 51.1k| CSA(&twosA, &ones, ones, _mm256_lddqu_si256(data + i + 8), 365| 51.1k| _mm256_lddqu_si256(data + i + 9)); 366| 51.1k| CSA(&twosB, &ones, ones, _mm256_lddqu_si256(data + i + 10), 367| 51.1k| _mm256_lddqu_si256(data + i + 11)); 368| 51.1k| CSA(&foursA, &twos, twos, twosA, twosB); 369| 51.1k| CSA(&twosA, &ones, ones, _mm256_lddqu_si256(data + i + 12), 370| 51.1k| _mm256_lddqu_si256(data + i + 13)); 371| 51.1k| CSA(&twosB, &ones, ones, _mm256_lddqu_si256(data + i + 14), 372| 51.1k| _mm256_lddqu_si256(data + i + 15)); 373| 51.1k| CSA(&foursB, &twos, twos, twosA, twosB); 374| 51.1k| CSA(&eightsB, &fours, fours, foursA, foursB); 375| 51.1k| CSA(&sixteens, &eights, eights, eightsA, eightsB); 376| | 377| 51.1k| total = _mm256_add_epi64(total, popcount256(sixteens)); 378| 51.1k| } 379| | 380| 3.19k| total = _mm256_slli_epi64(total, 4); // * 16 381| 3.19k| total = _mm256_add_epi64( 382| 3.19k| total, _mm256_slli_epi64(popcount256(eights), 3)); // += 8 * ... 383| 3.19k| total = _mm256_add_epi64( 384| 3.19k| total, _mm256_slli_epi64(popcount256(fours), 2)); // += 4 * ... 385| 3.19k| total = _mm256_add_epi64( 386| 3.19k| total, _mm256_slli_epi64(popcount256(twos), 1)); // += 2 * ... 387| 3.19k| total = _mm256_add_epi64(total, popcount256(ones)); 388| 3.19k| for (; i < size; i++) ------------------ | Branch (388:12): [True: 0, False: 3.19k] ------------------ 389| 0| total = 390| 0| _mm256_add_epi64(total, popcount256(_mm256_lddqu_si256(data + i))); 391| | 392| 3.19k| return (uint64_t)(_mm256_extract_epi64(total, 0)) + 393| 3.19k| (uint64_t)(_mm256_extract_epi64(total, 1)) + 394| 3.19k| (uint64_t)(_mm256_extract_epi64(total, 2)) + 395| | (uint64_t)(_mm256_extract_epi64(total, 3)); 396| 3.19k|} bitset.c:CSA: 328| 1.44M| __m256i c) { 329| 1.44M| const __m256i u = _mm256_xor_si256(a, b); 330| 1.44M| *h = _mm256_or_si256(_mm256_and_si256(a, b), _mm256_and_si256(u, c)); 331| 1.44M| *l = _mm256_xor_si256(u, c); 332| 1.44M|} bitset.c:popcount256: 292| 120k|static inline __m256i popcount256(__m256i v) { 293| 120k| const __m256i lookuppos = _mm256_setr_epi8( 294| 120k| /* 0 */ 4 + 0, /* 1 */ 4 + 1, /* 2 */ 4 + 1, /* 3 */ 4 + 2, 295| 120k| /* 4 */ 4 + 1, /* 5 */ 4 + 2, /* 6 */ 4 + 2, /* 7 */ 4 + 3, 296| 120k| /* 8 */ 4 + 1, /* 9 */ 4 + 2, /* a */ 4 + 2, /* b */ 4 + 3, 297| 120k| /* c */ 4 + 2, /* d */ 4 + 3, /* e */ 4 + 3, /* f */ 4 + 4, 298| | 299| | /* 0 */ 4 + 0, /* 1 */ 4 + 1, /* 2 */ 4 + 1, /* 3 */ 4 + 2, 300| 120k| /* 4 */ 4 + 1, /* 5 */ 4 + 2, /* 6 */ 4 + 2, /* 7 */ 4 + 3, 301| 120k| /* 8 */ 4 + 1, /* 9 */ 4 + 2, /* a */ 4 + 2, /* b */ 4 + 3, 302| 120k| /* c */ 4 + 2, /* d */ 4 + 3, /* e */ 4 + 3, /* f */ 4 + 4); 303| 120k| const __m256i lookupneg = _mm256_setr_epi8( 304| 120k| /* 0 */ 4 - 0, /* 1 */ 4 - 1, /* 2 */ 4 - 1, /* 3 */ 4 - 2, 305| 120k| /* 4 */ 4 - 1, /* 5 */ 4 - 2, /* 6 */ 4 - 2, /* 7 */ 4 - 3, 306| 120k| /* 8 */ 4 - 1, /* 9 */ 4 - 2, /* a */ 4 - 2, /* b */ 4 - 3, 307| 120k| /* c */ 4 - 2, /* d */ 4 - 3, /* e */ 4 - 3, /* f */ 4 - 4, 308| | 309| | /* 0 */ 4 - 0, /* 1 */ 4 - 1, /* 2 */ 4 - 1, /* 3 */ 4 - 2, 310| 120k| /* 4 */ 4 - 1, /* 5 */ 4 - 2, /* 6 */ 4 - 2, /* 7 */ 4 - 3, 311| 120k| /* 8 */ 4 - 1, /* 9 */ 4 - 2, /* a */ 4 - 2, /* b */ 4 - 3, 312| 120k| /* c */ 4 - 2, /* d */ 4 - 3, /* e */ 4 - 3, /* f */ 4 - 4); 313| 120k| const __m256i low_mask = _mm256_set1_epi8(0x0f); 314| | 315| 120k| const __m256i lo = _mm256_and_si256(v, low_mask); 316| 120k| const __m256i hi = _mm256_and_si256(_mm256_srli_epi16(v, 4), low_mask); 317| 120k| const __m256i popcnt1 = _mm256_shuffle_epi8(lookuppos, lo); 318| 120k| const __m256i popcnt2 = _mm256_shuffle_epi8(lookupneg, hi); 319| 120k| return _mm256_sad_epu8(popcnt1, popcnt2); 320| 120k|} convert.c:bitset_set_lenrange: 112| 97.4k| uint32_t lenminusone) { 113| 97.4k| uint32_t firstword = start / 64; 114| 97.4k| uint32_t endword = (start + lenminusone) / 64; 115| 97.4k| if (firstword == endword) { ------------------ | Branch (115:9): [True: 30.5k, False: 66.9k] ------------------ 116| 30.5k| words[firstword] |= ((~UINT64_C(0)) >> ((63 - lenminusone) % 64)) 117| 30.5k| << (start % 64); 118| 30.5k| return; 119| 30.5k| } 120| 66.9k| uint64_t temp = words[endword]; 121| 66.9k| words[firstword] |= (~UINT64_C(0)) << (start % 64); 122| 33.1M| for (uint32_t i = firstword + 1; i < endword; i += 2) ------------------ | Branch (122:38): [True: 33.0M, False: 66.9k] ------------------ 123| 33.0M| words[i] = words[i + 1] = ~UINT64_C(0); 124| 66.9k| words[endword] = 125| | temp | (~UINT64_C(0)) >> (((~start + 1) - lenminusone - 1) % 64); 126| 66.9k|} mixed_intersection.c:bitset_lenrange_cardinality: 63| 203k| uint32_t lenminusone) { 64| 203k| uint32_t firstword = start / 64; 65| 203k| uint32_t endword = (start + lenminusone) / 64; 66| 203k| if (firstword == endword) { ------------------ | Branch (66:9): [True: 197k, False: 6.08k] ------------------ 67| 197k| return roaring_hamming(words[firstword] & 68| 197k| ((~UINT64_C(0)) >> ((63 - lenminusone) % 64)) 69| 197k| << (start % 64)); 70| 197k| } 71| 6.08k| int answer = 72| 6.08k| roaring_hamming(words[firstword] & ((~UINT64_C(0)) << (start % 64))); 73| 6.33k| for (uint32_t i = firstword + 1; i < endword; i++) { ------------------ | Branch (73:38): [True: 248, False: 6.08k] ------------------ 74| 248| answer += roaring_hamming(words[i]); 75| 248| } 76| 6.08k| answer += roaring_hamming(words[endword] & 77| | (~UINT64_C(0)) >> 78| 6.08k| (((~start + 1) - lenminusone - 1) % 64)); 79| 6.08k| return answer; 80| 203k|} mixed_intersection.c:bitset_lenrange_empty: 86| 3.58k| uint32_t lenminusone) { 87| 3.58k| uint32_t firstword = start / 64; 88| 3.58k| uint32_t endword = (start + lenminusone) / 64; 89| 3.58k| if (firstword == endword) { ------------------ | Branch (89:9): [True: 3.30k, False: 281] ------------------ 90| 3.30k| return (words[firstword] & ((~UINT64_C(0)) >> ((63 - lenminusone) % 64)) 91| 3.30k| << (start % 64)) == 0; 92| 3.30k| } 93| 281| if (((words[firstword] & ((~UINT64_C(0)) << (start % 64)))) != 0) { ------------------ | Branch (93:9): [True: 100, False: 181] ------------------ 94| 100| return false; 95| 100| } 96| 192| for (uint32_t i = firstword + 1; i < endword; i++) { ------------------ | Branch (96:38): [True: 13, False: 179] ------------------ 97| 13| if (words[i] != 0) { ------------------ | Branch (97:13): [True: 2, False: 11] ------------------ 98| 2| return false; 99| 2| } 100| 13| } 101| 179| if ((words[endword] & ------------------ | Branch (101:9): [True: 98, False: 81] ------------------ 102| 179| (~UINT64_C(0)) >> (((~start + 1) - lenminusone - 1) % 64)) != 0) { 103| 98| return false; 104| 98| } 105| 81| return true; 106| 179|} mixed_union.c:bitset_set_lenrange: 112| 101k| uint32_t lenminusone) { 113| 101k| uint32_t firstword = start / 64; 114| 101k| uint32_t endword = (start + lenminusone) / 64; 115| 101k| if (firstword == endword) { ------------------ | Branch (115:9): [True: 98.8k, False: 3.04k] ------------------ 116| 98.8k| words[firstword] |= ((~UINT64_C(0)) >> ((63 - lenminusone) % 64)) 117| 98.8k| << (start % 64); 118| 98.8k| return; 119| 98.8k| } 120| 3.04k| uint64_t temp = words[endword]; 121| 3.04k| words[firstword] |= (~UINT64_C(0)) << (start % 64); 122| 3.14k| for (uint32_t i = firstword + 1; i < endword; i += 2) ------------------ | Branch (122:38): [True: 96, False: 3.04k] ------------------ 123| 96| words[i] = words[i + 1] = ~UINT64_C(0); 124| 3.04k| words[endword] = 125| | temp | (~UINT64_C(0)) >> (((~start + 1) - lenminusone - 1) % 64); 126| 3.04k|} mixed_negation.c:bitset_flip_range: 132| 1.42k| uint32_t end) { 133| 1.42k| if (start == end) return; ------------------ | Branch (133:9): [True: 0, False: 1.42k] ------------------ 134| 1.42k| uint32_t firstword = start / 64; 135| 1.42k| uint32_t endword = (end - 1) / 64; 136| 1.42k| words[firstword] ^= ~((~UINT64_C(0)) << (start % 64)); 137| 181k| for (uint32_t i = firstword; i < endword; i++) { ------------------ | Branch (137:34): [True: 179k, False: 1.42k] ------------------ 138| 179k| words[i] = ~words[i]; 139| 179k| } 140| | words[endword] ^= ((~UINT64_C(0)) >> ((~end + 1) % 64)); 141| 1.42k|} mixed_xor.c:bitset_flip_range: 132| 54.6k| uint32_t end) { 133| 54.6k| if (start == end) return; ------------------ | Branch (133:9): [True: 0, False: 54.6k] ------------------ 134| 54.6k| uint32_t firstword = start / 64; 135| 54.6k| uint32_t endword = (end - 1) / 64; 136| 54.6k| words[firstword] ^= ~((~UINT64_C(0)) << (start % 64)); 137| 112k| for (uint32_t i = firstword; i < endword; i++) { ------------------ | Branch (137:34): [True: 57.4k, False: 54.6k] ------------------ 138| 57.4k| words[i] = ~words[i]; 139| 57.4k| } 140| | words[endword] ^= ((~UINT64_C(0)) >> ((~end + 1) % 64)); 141| 54.6k|} array_container_contains: 343| 3.23k| uint16_t pos) { 344| | /** 345| | * SIMD Quad algorithm 346| | * Daniel Lemire, "You can beat the binary search," in Daniel Lemire's blog, 347| | * April 27, 2026, 348| | * https://lemire.me/blog/2026/04/27/you-can-beat-the-binary-search/. 349| | */ 350| 3.23k| const int32_t gap = 16; 351| 3.23k| const uint16_t *carr = arr->array; 352| 3.23k| int32_t cardinality = arr->cardinality; 353| 3.23k| if (cardinality < gap) { ------------------ | Branch (353:9): [True: 945, False: 2.29k] ------------------ 354| 1.08k| for (int32_t j = 0; j < cardinality; j++) { ------------------ | Branch (354:29): [True: 1.05k, False: 34] ------------------ 355| 1.05k| if (carr[j] >= pos) return carr[j] == pos; ------------------ | Branch (355:17): [True: 911, False: 142] ------------------ 356| 1.05k| } 357| 34| return false; 358| 945| } 359| 2.29k| int32_t num_blocks = cardinality / gap; 360| 2.29k| int32_t base = 0; 361| 2.29k| int32_t n = num_blocks; 362| 5.26k| while (n > 3) { ------------------ | Branch (362:12): [True: 2.97k, False: 2.29k] ------------------ 363| 2.97k| int32_t quarter = n >> 2; 364| | 365| 2.97k| int32_t k1 = carr[(base + quarter + 1) * gap - 1]; 366| 2.97k| int32_t k2 = carr[(base + 2 * quarter + 1) * gap - 1]; 367| 2.97k| int32_t k3 = carr[(base + 3 * quarter + 1) * gap - 1]; 368| | 369| 2.97k| int32_t c1 = (k1 < pos); 370| 2.97k| int32_t c2 = (k2 < pos); 371| 2.97k| int32_t c3 = (k3 < pos); 372| | 373| 2.97k| base += (c1 + c2 + c3) * quarter; 374| 2.97k| n -= 3 * quarter; 375| 2.97k| } 376| 4.04k| while (n > 1) { ------------------ | Branch (376:12): [True: 1.75k, False: 2.29k] ------------------ 377| 1.75k| int32_t half = n >> 1; 378| 1.75k| base = (carr[(base + half + 1) * gap - 1] < pos) ? base + half : base; ------------------ | Branch (378:16): [True: 76, False: 1.67k] ------------------ 379| 1.75k| n -= half; 380| 1.75k| } 381| 2.29k| int32_t lo = (carr[(base + 1) * gap - 1] < pos) ? base + 1 : base; ------------------ | Branch (381:18): [True: 117, False: 2.17k] ------------------ 382| | 383| 2.29k| if (lo < num_blocks) { ------------------ | Branch (383:9): [True: 2.21k, False: 73] ------------------ 384| 2.21k| const uint16_t *blk = carr + lo * gap; 385| |#ifdef CROARING_USENEON 386| | uint16x8_t needle = vdupq_n_u16(pos); 387| | uint16x8_t v0 = vld1q_u16(blk); 388| | uint16x8_t v1 = vld1q_u16(blk + 8); 389| | uint16x8_t hit = 390| | vorrq_u16(vceqq_u16(v0, needle), vceqq_u16(v1, needle)); 391| | return vmaxvq_u16(hit) != 0; 392| |#elif defined(CROARING_IS_X64) 393| | __m128i needle = _mm_set1_epi16((short)pos); 394| 2.21k| __m128i v0 = _mm_loadu_si128((const __m128i *)blk); 395| 2.21k| __m128i v1 = _mm_loadu_si128((const __m128i *)(blk + 8)); 396| 2.21k| __m128i hit = _mm_or_si128(_mm_cmpeq_epi16(v0, needle), 397| 2.21k| _mm_cmpeq_epi16(v1, needle)); 398| 2.21k| return _mm_movemask_epi8(hit) != 0; 399| |#else 400| | for (int32_t j = 0; j < gap; j++) { 401| | if (blk[j] >= pos) return blk[j] == pos; 402| | } 403| | return false; 404| |#endif 405| 2.21k| } 406| | 407| 638| for (int32_t j = num_blocks * gap; j < cardinality; j++) { ------------------ | Branch (407:40): [True: 591, False: 47] ------------------ 408| 591| uint16_t v = carr[j]; 409| 591| if (v >= pos) return (v == pos); ------------------ | Branch (409:13): [True: 26, False: 565] ------------------ 410| 591| } 411| 47| return false; 412| 73|} array_container_minimum: 445| 3.51k|inline uint16_t array_container_minimum(const array_container_t *arr) { 446| 3.51k| if (arr->cardinality == 0) return 0; ------------------ | Branch (446:9): [True: 0, False: 3.51k] ------------------ 447| 3.51k| return arr->array[0]; 448| 3.51k|} array_container_maximum: 451| 12.8k|inline uint16_t array_container_maximum(const array_container_t *arr) { 452| 12.8k| if (arr->cardinality == 0) return 0; ------------------ | Branch (452:9): [True: 0, False: 12.8k] ------------------ 453| 12.8k| return arr->array[arr->cardinality - 1]; 454| 12.8k|} array_container_rank: 457| 3.63k|inline int array_container_rank(const array_container_t *arr, uint16_t x) { 458| 3.63k| const int32_t idx = binarySearch(arr->array, arr->cardinality, x); 459| 3.63k| const bool is_present = idx >= 0; 460| 3.63k| if (is_present) { ------------------ | Branch (460:9): [True: 3.06k, False: 566] ------------------ 461| 3.06k| return idx + 1; 462| 3.06k| } else { 463| 566| return -idx - 1; 464| 566| } 465| 3.63k|} array_container_index_equalorlarger: 507| 4.32k| uint16_t x) { 508| 4.32k| const int32_t idx = binarySearch(arr->array, arr->cardinality, x); 509| 4.32k| const bool is_present = idx >= 0; 510| 4.32k| if (is_present) { ------------------ | Branch (510:9): [True: 3.95k, False: 366] ------------------ 511| 3.95k| return idx; 512| 3.95k| } else { 513| 366| int32_t candidate = -idx - 1; 514| 366| if (candidate < arr->cardinality) return candidate; ------------------ | Branch (514:13): [True: 366, False: 0] ------------------ 515| 0| return -1; 516| 366| } 517| 4.32k|} roaring.c:array_container_add_range_nvals: 527| 539| int32_t nvals_greater) { 528| 539| int32_t union_cardinality = nvals_less + (max - min + 1) + nvals_greater; 529| 539| if (union_cardinality > array->capacity) { ------------------ | Branch (529:9): [True: 503, False: 36] ------------------ 530| 503| array_container_grow(array, union_cardinality, true); 531| 503| } 532| 539| memmove(&(array->array[union_cardinality - nvals_greater]), 533| 539| &(array->array[array->cardinality - nvals_greater]), 534| 539| nvals_greater * sizeof(uint16_t)); 535| 1.33M| for (uint32_t i = 0; i <= max - min; i++) { ------------------ | Branch (535:26): [True: 1.33M, False: 539] ------------------ 536| 1.33M| array->array[nvals_less + i] = (uint16_t)(min + i); 537| 1.33M| } 538| 539| array->cardinality = union_cardinality; 539| 539|} roaring.c:array_container_remove_range: 557| 3.13k| uint32_t pos, uint32_t count) { 558| 3.13k| if (count != 0) { ------------------ | Branch (558:9): [True: 324, False: 2.81k] ------------------ 559| 324| memmove(&(array->array[pos]), &(array->array[pos + count]), 560| 324| (array->cardinality - pos - count) * sizeof(uint16_t)); 561| 324| array->cardinality -= count; 562| 324| } 563| 3.13k|} roaring.c:array_container_cardinality: 82| 69.1k|static inline int array_container_cardinality(const array_container_t *array) { 83| 69.1k| return array->cardinality; 84| 69.1k|} roaring.c:array_container_try_add: 293| 7.02M| int32_t max_cardinality) { 294| 7.02M| const int32_t cardinality = arr->cardinality; 295| | 296| | // best case, we can append. 297| 7.02M| if ((array_container_empty(arr) || arr->array[cardinality - 1] < value) && ------------------ | Branch (297:10): [True: 15.5k, False: 7.01M] | Branch (297:40): [True: 49.9k, False: 6.96M] ------------------ 298| 65.4k| cardinality < max_cardinality) { ------------------ | Branch (298:9): [True: 65.4k, False: 0] ------------------ 299| 65.4k| array_container_append(arr, value); 300| 65.4k| return 1; 301| 65.4k| } 302| | 303| 6.96M| const int32_t loc = binarySearch(arr->array, cardinality, value); 304| | 305| 6.96M| if (loc >= 0) { ------------------ | Branch (305:9): [True: 5.33M, False: 1.62M] ------------------ 306| 5.33M| return 0; 307| 5.33M| } else if (cardinality < max_cardinality) { ------------------ | Branch (307:16): [True: 1.62M, False: 103] ------------------ 308| 1.62M| if (array_container_full(arr)) { ------------------ | Branch (308:13): [True: 73.3k, False: 1.55M] ------------------ 309| 73.3k| array_container_grow(arr, arr->capacity + 1, true); 310| 73.3k| } 311| 1.62M| const int32_t insert_idx = -loc - 1; 312| 1.62M| memmove(arr->array + insert_idx + 1, arr->array + insert_idx, 313| 1.62M| (cardinality - insert_idx) * sizeof(uint16_t)); 314| 1.62M| arr->array[insert_idx] = value; 315| 1.62M| arr->cardinality++; 316| 1.62M| return 1; 317| 1.62M| } else { 318| 103| return -1; 319| 103| } 320| 6.96M|} roaring.c:array_container_empty: 100| 7.02M|static inline bool array_container_empty(const array_container_t *array) { 101| 7.02M| return array->cardinality == 0; 102| 7.02M|} roaring.c:array_container_append: 274| 65.4k| uint16_t pos) { 275| 65.4k| const int32_t capacity = arr->capacity; 276| | 277| 65.4k| if (array_container_full(arr)) { ------------------ | Branch (277:9): [True: 31.3k, False: 34.1k] ------------------ 278| 31.3k| array_container_grow(arr, capacity + 1, true); 279| 31.3k| } 280| | 281| 65.4k| arr->array[arr->cardinality++] = pos; 282| 65.4k|} roaring.c:array_container_full: 106| 1.69M|static inline bool array_container_full(const array_container_t *array) { 107| 1.69M| return array->cardinality == array->capacity; 108| 1.69M|} roaring.c:array_container_remove: 329| 5.96k| uint16_t pos) { 330| 5.96k| const int32_t idx = binarySearch(arr->array, arr->cardinality, pos); 331| 5.96k| const bool is_present = idx >= 0; 332| 5.96k| if (is_present) { ------------------ | Branch (332:9): [True: 2.98k, False: 2.97k] ------------------ 333| 2.98k| memmove(arr->array + idx, arr->array + idx + 1, 334| 2.98k| (arr->cardinality - idx - 1) * sizeof(uint16_t)); 335| 2.98k| arr->cardinality--; 336| 2.98k| } 337| | 338| 5.96k| return is_present; 339| 5.96k|} roaring.c:array_container_nonzero_cardinality: 87| 98.4k| const array_container_t *array) { 88| 98.4k| return array->cardinality > 0; 89| 98.4k|} roaring.c:array_container_equals: 230| 9.26k| const array_container_t *container2) { 231| 9.26k| if (container1->cardinality != container2->cardinality) { ------------------ | Branch (231:9): [True: 2.20k, False: 7.06k] ------------------ 232| 2.20k| return false; 233| 2.20k| } 234| 7.06k| return memequals(container1->array, container2->array, 235| 7.06k| container1->cardinality * 2); 236| 9.26k|} roaring.c:array_container_select: 252| 3.63k| uint32_t *element) { 253| 3.63k| int card = array_container_cardinality(container); 254| 3.63k| if (*start_rank + card <= rank) { ------------------ | Branch (254:9): [True: 331, False: 3.30k] ------------------ 255| 331| *start_rank += card; 256| 331| return false; 257| 3.30k| } else { 258| 3.30k| *element = container->array[rank - *start_rank]; 259| | return true; 260| 3.30k| } 261| 3.63k|} roaring.c:array_container_contains_range: 421| 94| uint32_t range_end) { 422| 94| const int32_t range_count = range_end - range_start; 423| 94| const uint16_t rs_included = (uint16_t)range_start; 424| 94| const uint16_t re_included = (uint16_t)(range_end - 1); 425| | 426| | // Empty range is always included 427| 94| if (range_count <= 0) { ------------------ | Branch (427:9): [True: 0, False: 94] ------------------ 428| 0| return true; 429| 0| } 430| 94| if (range_count > arr->cardinality) { ------------------ | Branch (430:9): [True: 39, False: 55] ------------------ 431| 39| return false; 432| 39| } 433| | 434| 55| const int32_t start = 435| 55| binarySearch(arr->array, arr->cardinality, rs_included); 436| | // If this sorted array contains all items in the range: 437| | // * the start item must be found 438| | // * the last item in range range_count must exist, and be the expected end 439| | // value 440| 55| return (start >= 0) && (arr->cardinality >= start + range_count) && ------------------ | Branch (440:12): [True: 35, False: 20] | Branch (440:28): [True: 32, False: 3] ------------------ 441| 32| (arr->array[start + range_count - 1] == re_included); ------------------ | Branch (441:12): [True: 7, False: 25] ------------------ 442| 94|} roaring_array.c:array_container_size_in_bytes: 221| 17.4k| const array_container_t *container) { 222| 17.4k| return container->cardinality * sizeof(uint16_t); 223| 17.4k|} roaring_array.c:array_container_cardinality: 82| 6.91k|static inline int array_container_cardinality(const array_container_t *array) { 83| 6.91k| return array->cardinality; 84| 6.91k|} array.c:array_container_size_in_bytes: 221| 13.8k| const array_container_t *container) { 222| 13.8k| return container->cardinality * sizeof(uint16_t); 223| 13.8k|} convert.c:array_container_cardinality: 82| 14.4k|static inline int array_container_cardinality(const array_container_t *array) { 83| 14.4k| return array->cardinality; 84| 14.4k|} convert.c:array_container_serialized_size_in_bytes: 173| 140k|static inline int32_t array_container_serialized_size_in_bytes(int32_t card) { 174| 140k| return card * sizeof(uint16_t); 175| 140k|} mixed_intersection.c:array_container_empty: 100| 34|static inline bool array_container_empty(const array_container_t *array) { 101| 34| return array->cardinality == 0; 102| 34|} mixed_andnot.c:array_container_cardinality: 82| 350|static inline int array_container_cardinality(const array_container_t *array) { 83| 350| return array->cardinality; 84| 350|} bitset_container_get: 193| 1.03M| uint16_t pos) { 194| 1.03M| const uint64_t word = bitset->words[pos >> 6]; 195| 1.03M| return (word >> (pos & 63)) & 1; 196| 1.03M|} bitset_container_contains: 232| 896k| uint16_t pos) { 233| 896k| return bitset_container_get(bitset, pos); 234| 896k|} roaring.c:bitset_container_cardinality: 248| 10.3k| const bitset_container_t *bitset) { 249| 10.3k| return bitset->cardinality; 250| 10.3k|} roaring.c:bitset_container_set: 147| 303k| uint16_t pos) { 148| 303k| const uint64_t old_word = bitset->words[pos >> 6]; 149| 303k| const int index = pos & 63; 150| | const uint64_t new_word = old_word | (UINT64_C(1) << index); 151| 303k| bitset->cardinality += (uint32_t)((old_word ^ new_word) >> index); 152| 303k| bitset->words[pos >> 6] = new_word; 153| 303k|} roaring.c:bitset_container_add: 168| 103| uint16_t pos) { 169| 103| const uint64_t old_word = bitset->words[pos >> 6]; 170| 103| const int index = pos & 63; 171| | const uint64_t new_word = old_word | (UINT64_C(1) << index); 172| 103| const uint64_t increment = (old_word ^ new_word) >> index; 173| 103| bitset->cardinality += (uint32_t)increment; 174| 103| bitset->words[pos >> 6] = new_word; 175| 103| return increment > 0; 176| 103|} roaring.c:bitset_container_remove: 181| 2.09k| uint16_t pos) { 182| 2.09k| const uint64_t old_word = bitset->words[pos >> 6]; 183| 2.09k| const int index = pos & 63; 184| | const uint64_t new_word = old_word & (~(UINT64_C(1) << index)); 185| 2.09k| const uint64_t increment = (old_word ^ new_word) >> index; 186| 2.09k| bitset->cardinality -= (uint32_t)increment; 187| 2.09k| bitset->words[pos >> 6] = new_word; 188| 2.09k| return increment > 0; 189| 2.09k|} roaring.c:bitset_container_const_nonzero_cardinality: 281| 6.47k| const bitset_container_t *bitset) { 282| 6.47k| return !bitset_container_empty(bitset); 283| 6.47k|} roaring.c:bitset_container_empty: 268| 6.47k|static inline bool bitset_container_empty(const bitset_container_t *bitset) { 269| 6.47k| if (bitset->cardinality == BITSET_UNKNOWN_CARDINALITY) { ------------------ | Branch (269:9): [True: 0, False: 6.47k] ------------------ 270| 0| for (int i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; i++) { ------------------ | Branch (270:25): [True: 0, False: 0] ------------------ 271| 0| if ((bitset->words[i]) != 0) return false; ------------------ | Branch (271:17): [True: 0, False: 0] ------------------ 272| 0| } 273| 0| return true; 274| 0| } 275| 6.47k| return bitset->cardinality == 0; 276| 6.47k|} roaring.c:bitset_container_get_range: 206| 233| uint32_t pos_end) { 207| 233| const uint32_t start = pos_start >> 6; 208| 233| const uint32_t end = pos_end >> 6; 209| | 210| 233| const uint64_t first = ~((1ULL << (pos_start & 0x3F)) - 1); 211| 233| const uint64_t last = (1ULL << (pos_end & 0x3F)) - 1; 212| | 213| 233| if (start == end) ------------------ | Branch (213:9): [True: 50, False: 183] ------------------ 214| 50| return ((bitset->words[end] & first & last) == (first & last)); 215| 183| if ((bitset->words[start] & first) != first) return false; ------------------ | Branch (215:9): [True: 66, False: 117] ------------------ 216| | 217| 117| if ((end < BITSET_CONTAINER_SIZE_IN_WORDS) && ------------------ | Branch (217:9): [True: 103, False: 14] ------------------ 218| 103| ((bitset->words[end] & last) != last)) { ------------------ | Branch (218:9): [True: 89, False: 14] ------------------ 219| 89| return false; 220| 89| } 221| | 222| 28| for (uint32_t i = start + 1; 223| 10.6k| (i < BITSET_CONTAINER_SIZE_IN_WORDS) && (i < end); ++i) { ------------------ | Branch (223:10): [True: 10.6k, False: 12] | Branch (223:50): [True: 10.6k, False: 11] ------------------ 224| 10.6k| if (bitset->words[i] != UINT64_C(0xFFFFFFFFFFFFFFFF)) return false; ------------------ | Branch (224:13): [True: 5, False: 10.6k] ------------------ 225| 10.6k| } 226| | 227| 23| return true; 228| 28|} roaring_array.c:bitset_container_size_in_bytes: 468| 133k| const bitset_container_t *container) { 469| 133k| (void)container; 470| 133k| return BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t); 471| 133k|} roaring_array.c:bitset_container_cardinality: 248| 66.7k| const bitset_container_t *bitset) { 249| 66.7k| return bitset->cardinality; 250| 66.7k|} bitset.c:bitset_container_size_in_bytes: 468| 133k| const bitset_container_t *container) { 469| 133k| (void)container; 470| 133k| return BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t); 471| 133k|} convert.c:bitset_container_set: 147| 2.07M| uint16_t pos) { 148| 2.07M| const uint64_t old_word = bitset->words[pos >> 6]; 149| 2.07M| const int index = pos & 63; 150| | const uint64_t new_word = old_word | (UINT64_C(1) << index); 151| 2.07M| bitset->cardinality += (uint32_t)((old_word ^ new_word) >> index); 152| 2.07M| bitset->words[pos >> 6] = new_word; 153| 2.07M|} convert.c:bitset_container_serialized_size_in_bytes: 427| 128k|static inline int32_t bitset_container_serialized_size_in_bytes(void) { 428| 128k| return BITSET_CONTAINER_SIZE_IN_WORDS * 8; 429| 128k|} container_iterator_next: 2456| 320k| uint16_t *value) { 2457| 320k| switch (typecode) { 2458| 0| case BITSET_CONTAINER_TYPE: { ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (2458:9): [True: 0, False: 320k] ------------------ 2459| 0| const bitset_container_t *bc = const_CAST_bitset(c); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2460| 0| it->index++; 2461| | 2462| 0| uint32_t wordindex = it->index / 64; 2463| 0| if (wordindex >= BITSET_CONTAINER_SIZE_IN_WORDS) { ------------------ | Branch (2463:17): [True: 0, False: 0] ------------------ 2464| 0| return false; 2465| 0| } 2466| | 2467| 0| uint64_t word = 2468| 0| bc->words[wordindex] & (UINT64_MAX << (it->index % 64)); 2469| | // next part could be optimized/simplified 2470| 0| while (word == 0 && ------------------ | Branch (2470:20): [True: 0, False: 0] ------------------ 2471| 0| (wordindex + 1 < BITSET_CONTAINER_SIZE_IN_WORDS)) { ------------------ | Branch (2471:20): [True: 0, False: 0] ------------------ 2472| 0| wordindex++; 2473| 0| word = bc->words[wordindex]; 2474| 0| } 2475| 0| if (word != 0) { ------------------ | Branch (2475:17): [True: 0, False: 0] ------------------ 2476| 0| it->index = wordindex * 64 + roaring_trailing_zeroes(word); 2477| 0| *value = it->index; 2478| 0| return true; 2479| 0| } 2480| 0| return false; 2481| 0| } 2482| 278k| case ARRAY_CONTAINER_TYPE: { ------------------ | | 49| 278k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (2482:9): [True: 278k, False: 42.6k] ------------------ 2483| 278k| const array_container_t *ac = const_CAST_array(c); ------------------ | | 55| 278k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 278k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2484| 278k| it->index++; 2485| 278k| if (it->index < ac->cardinality) { ------------------ | Branch (2485:17): [True: 272k, False: 5.45k] ------------------ 2486| 272k| *value = ac->array[it->index]; 2487| 272k| return true; 2488| 272k| } 2489| 5.45k| return false; 2490| 278k| } 2491| 42.6k| case RUN_CONTAINER_TYPE: { ------------------ | | 50| 42.6k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (2491:9): [True: 42.6k, False: 278k] ------------------ 2492| 42.6k| if (*value == UINT16_MAX) { // Avoid overflow to zero ------------------ | Branch (2492:17): [True: 0, False: 42.6k] ------------------ 2493| 0| return false; 2494| 0| } 2495| | 2496| 42.6k| const run_container_t *rc = const_CAST_run(c); ------------------ | | 78| 42.6k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 42.6k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2497| 42.6k| uint32_t limit = 2498| 42.6k| rc->runs[it->index].value + rc->runs[it->index].length; 2499| 42.6k| if (*value < limit) { ------------------ | Branch (2499:17): [True: 27.2k, False: 15.4k] ------------------ 2500| 27.2k| (*value)++; 2501| 27.2k| return true; 2502| 27.2k| } 2503| | 2504| 15.4k| it->index++; 2505| 15.4k| if (it->index < rc->n_runs) { ------------------ | Branch (2505:17): [True: 14.9k, False: 483] ------------------ 2506| 14.9k| *value = rc->runs[it->index].value; 2507| 14.9k| return true; 2508| 14.9k| } 2509| 483| return false; 2510| 15.4k| } 2511| 0| default: ------------------ | Branch (2511:9): [True: 0, False: 320k] ------------------ 2512| 0| assert(false); 2513| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2514| 0| return false; 2515| 320k| } 2516| 320k|} roaring.c:container_from_range: 317| 68.7k| uint32_t max, uint16_t step) { 318| 68.7k| if (step == 0) return NULL; // being paranoid ------------------ | Branch (318:9): [True: 0, False: 68.7k] ------------------ 319| 68.7k| if (step == 1) { ------------------ | Branch (319:9): [True: 68.7k, False: 0] ------------------ 320| 68.7k| return container_range_of_ones(min, max, type); 321| | // Note: the result is not always a run (need to check the cardinality) 322| | //*type = RUN_CONTAINER_TYPE; 323| | // return run_container_create_range(min, max); 324| 68.7k| } 325| 0| int size = (max - min + step - 1) / step; 326| 0| if (size <= DEFAULT_MAX_SIZE) { // array container ------------------ | Branch (326:9): [True: 0, False: 0] ------------------ 327| 0| *type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 328| 0| array_container_t *array = array_container_create_given_capacity(size); 329| 0| array_container_add_from_range(array, min, max, step); 330| 0| assert(array->cardinality == size); 331| 0| return array; 332| 0| } else { // bitset container 333| 0| *type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 334| 0| bitset_container_t *bitset = bitset_container_create(); 335| 0| bitset_container_add_from_range(bitset, min, max, step); 336| | assert(bitset->cardinality == size); 337| 0| return bitset; 338| 0| } 339| 0|} roaring.c:container_range_of_ones: 302| 84.7k| uint8_t *result_type) { 303| 84.7k| assert(range_end >= range_start); 304| 84.7k| uint64_t cardinality = range_end - range_start + 1; 305| 84.7k| if (cardinality <= 2) { ------------------ | Branch (305:9): [True: 150, False: 84.6k] ------------------ 306| 150| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 150|#define ARRAY_CONTAINER_TYPE 2 ------------------ 307| 150| return array_container_create_range(range_start, range_end); 308| 84.6k| } else { 309| 84.6k| *result_type = RUN_CONTAINER_TYPE; ------------------ | | 50| 84.6k|#define RUN_CONTAINER_TYPE 3 ------------------ 310| 84.6k| return run_container_create_range(range_start, range_end); 311| 84.6k| } 312| 84.7k|} roaring.c:container_add_range: 2282| 3.06k| uint8_t *result_type) { 2283| | // NB: when selecting new container type, we perform only inexpensive checks 2284| 3.06k| switch (type) { 2285| 0| case BITSET_CONTAINER_TYPE: { ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (2285:9): [True: 0, False: 3.06k] ------------------ 2286| 0| bitset_container_t *bitset = CAST_bitset(c); ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2287| | 2288| 0| int32_t union_cardinality = 0; 2289| 0| union_cardinality += bitset->cardinality; 2290| 0| union_cardinality += max - min + 1; 2291| 0| union_cardinality -= 2292| 0| bitset_lenrange_cardinality(bitset->words, min, max - min); 2293| | 2294| 0| if (union_cardinality == INT32_C(0x10000)) { ------------------ | Branch (2294:17): [True: 0, False: 0] ------------------ 2295| 0| *result_type = RUN_CONTAINER_TYPE; ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 2296| 0| return run_container_create_range(0, INT32_C(0x10000)); 2297| 0| } else { 2298| 0| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 2299| 0| bitset_set_lenrange(bitset->words, min, max - min); 2300| 0| bitset->cardinality = union_cardinality; 2301| 0| return bitset; 2302| 0| } 2303| 0| } 2304| 2.65k| case ARRAY_CONTAINER_TYPE: { ------------------ | | 49| 2.65k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (2304:9): [True: 2.65k, False: 411] ------------------ 2305| 2.65k| array_container_t *array = CAST_array(c); ------------------ | | 54| 2.65k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 2.65k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2306| | 2307| 2.65k| int32_t nvals_greater = 2308| 2.65k| count_greater(array->array, array->cardinality, (uint16_t)max); 2309| 2.65k| int32_t nvals_less = 2310| 2.65k| count_less(array->array, array->cardinality - nvals_greater, 2311| 2.65k| (uint16_t)min); 2312| 2.65k| int32_t union_cardinality = 2313| 2.65k| nvals_less + (max - min + 1) + nvals_greater; 2314| | 2315| 2.65k| if (union_cardinality == INT32_C(0x10000)) { ------------------ | Branch (2315:17): [True: 657, False: 1.99k] ------------------ 2316| 657| *result_type = RUN_CONTAINER_TYPE; ------------------ | | 50| 657|#define RUN_CONTAINER_TYPE 3 ------------------ 2317| 657| return run_container_create_range(0, INT32_C(0x10000)); 2318| 1.99k| } else if (union_cardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (2318:24): [True: 539, False: 1.45k] ------------------ 2319| 539| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 539|#define ARRAY_CONTAINER_TYPE 2 ------------------ 2320| 539| array_container_add_range_nvals(array, min, max, nvals_less, 2321| 539| nvals_greater); 2322| 539| return array; 2323| 1.45k| } else { 2324| 1.45k| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 1.45k|#define BITSET_CONTAINER_TYPE 1 ------------------ 2325| 1.45k| bitset_container_t *bitset = bitset_container_from_array(array); 2326| 1.45k| bitset_set_lenrange(bitset->words, min, max - min); 2327| 1.45k| bitset->cardinality = union_cardinality; 2328| 1.45k| return bitset; 2329| 1.45k| } 2330| 2.65k| } 2331| 411| case RUN_CONTAINER_TYPE: { ------------------ | | 50| 411|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (2331:9): [True: 411, False: 2.65k] ------------------ 2332| 411| run_container_t *run = CAST_run(c); ------------------ | | 77| 411|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 411|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2333| | 2334| 411| int32_t nruns_greater = 2335| 411| rle16_count_greater(run->runs, run->n_runs, (uint16_t)max); 2336| 411| int32_t nruns_less = rle16_count_less( 2337| 411| run->runs, run->n_runs - nruns_greater, (uint16_t)min); 2338| | 2339| 411| int32_t run_size_bytes = 2340| 411| (nruns_less + 1 + nruns_greater) * sizeof(rle16_t); 2341| 411| int32_t bitset_size_bytes = 2342| 411| BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t); 2343| | 2344| 411| if (run_size_bytes <= bitset_size_bytes) { ------------------ | Branch (2344:17): [True: 411, False: 0] ------------------ 2345| 411| run_container_add_range_nruns(run, min, max, nruns_less, 2346| 411| nruns_greater); 2347| 411| *result_type = RUN_CONTAINER_TYPE; ------------------ | | 50| 411|#define RUN_CONTAINER_TYPE 3 ------------------ 2348| 411| return run; 2349| 411| } else { 2350| 0| return container_from_run_range(run, min, max, result_type); 2351| 0| } 2352| 411| } 2353| 0| default: ------------------ | Branch (2353:9): [True: 0, False: 3.06k] ------------------ 2354| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2355| 3.06k| } 2356| 3.06k|} roaring.c:container_remove_range: 2371| 21.2k| uint8_t *result_type) { 2372| 21.2k| switch (type) { 2373| 1.27k| case BITSET_CONTAINER_TYPE: { ------------------ | | 48| 1.27k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (2373:9): [True: 1.27k, False: 19.9k] ------------------ 2374| 1.27k| bitset_container_t *bitset = CAST_bitset(c); ------------------ | | 52| 1.27k|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2375| | 2376| 1.27k| int32_t result_cardinality = 2377| 1.27k| bitset->cardinality - 2378| 1.27k| bitset_lenrange_cardinality(bitset->words, min, max - min); 2379| | 2380| 1.27k| if (result_cardinality == 0) { ------------------ | Branch (2380:17): [True: 39, False: 1.23k] ------------------ 2381| 39| return NULL; 2382| 1.23k| } else if (result_cardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (2382:24): [True: 106, False: 1.13k] ------------------ 2383| 106| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 106|#define ARRAY_CONTAINER_TYPE 2 ------------------ 2384| 106| bitset_reset_range(bitset->words, min, max + 1); 2385| 106| bitset->cardinality = result_cardinality; 2386| 106| return array_container_from_bitset(bitset); 2387| 1.13k| } else { 2388| 1.13k| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 1.13k|#define BITSET_CONTAINER_TYPE 1 ------------------ 2389| 1.13k| bitset_reset_range(bitset->words, min, max + 1); 2390| 1.13k| bitset->cardinality = result_cardinality; 2391| 1.13k| return bitset; 2392| 1.13k| } 2393| 1.27k| } 2394| 3.49k| case ARRAY_CONTAINER_TYPE: { ------------------ | | 49| 3.49k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (2394:9): [True: 3.49k, False: 17.7k] ------------------ 2395| 3.49k| array_container_t *array = CAST_array(c); ------------------ | | 54| 3.49k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 3.49k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2396| | 2397| 3.49k| int32_t nvals_greater = 2398| 3.49k| count_greater(array->array, array->cardinality, (uint16_t)max); 2399| 3.49k| int32_t nvals_less = 2400| 3.49k| count_less(array->array, array->cardinality - nvals_greater, 2401| 3.49k| (uint16_t)min); 2402| 3.49k| int32_t result_cardinality = nvals_less + nvals_greater; 2403| | 2404| 3.49k| if (result_cardinality == 0) { ------------------ | Branch (2404:17): [True: 357, False: 3.13k] ------------------ 2405| 357| return NULL; 2406| 3.13k| } else { 2407| 3.13k| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 3.13k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 2408| 3.13k| array_container_remove_range( 2409| 3.13k| array, nvals_less, array->cardinality - result_cardinality); 2410| 3.13k| return array; 2411| 3.13k| } 2412| 3.49k| } 2413| 16.4k| case RUN_CONTAINER_TYPE: { ------------------ | | 50| 16.4k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (2413:9): [True: 16.4k, False: 4.76k] ------------------ 2414| 16.4k| run_container_t *run = CAST_run(c); ------------------ | | 77| 16.4k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 16.4k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2415| | 2416| 16.4k| if (run->n_runs == 0) { ------------------ | Branch (2416:17): [True: 0, False: 16.4k] ------------------ 2417| 0| return NULL; 2418| 0| } 2419| 16.4k| if (min <= run_container_minimum(run) && ------------------ | Branch (2419:17): [True: 16.1k, False: 369] ------------------ 2420| 16.1k| max >= run_container_maximum(run)) { ------------------ | Branch (2420:17): [True: 14.6k, False: 1.43k] ------------------ 2421| 14.6k| return NULL; 2422| 14.6k| } 2423| | 2424| 1.80k| run_container_remove_range(run, min, max); 2425| 1.80k| return convert_run_to_efficient_container(run, result_type); 2426| 16.4k| } 2427| 0| default: ------------------ | Branch (2427:9): [True: 0, False: 21.2k] ------------------ 2428| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2429| 21.2k| } 2430| 21.2k|} roaring.c:container_get_cardinality: 244| 365k| uint8_t typecode) { 245| 365k| c = container_unwrap_shared(c, &typecode); 246| 365k| switch (typecode) { ------------------ | Branch (246:13): [True: 365k, False: 0] ------------------ 247| 9.07k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 9.07k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (247:9): [True: 9.07k, False: 356k] ------------------ 248| 9.07k| return bitset_container_cardinality(const_CAST_bitset(c)); ------------------ | | 53| 9.07k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 9.07k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 249| 61.8k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 61.8k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (249:9): [True: 61.8k, False: 303k] ------------------ 250| 61.8k| return array_container_cardinality(const_CAST_array(c)); ------------------ | | 55| 61.8k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 61.8k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 251| 294k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 294k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (251:9): [True: 294k, False: 70.9k] ------------------ 252| 294k| return run_container_cardinality(const_CAST_run(c)); ------------------ | | 78| 294k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 294k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 253| 365k| } 254| 365k| assert(false); 255| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 256| 0| return 0; // unreached 257| 365k|} roaring.c:get_container_type: 129| 73.6k|static inline uint8_t get_container_type(const container_t *c, uint8_t type) { 130| 73.6k| if (type == SHARED_CONTAINER_TYPE) { ------------------ | | 51| 73.6k|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (130:9): [True: 0, False: 73.6k] ------------------ 131| 0| return const_CAST_shared(c)->typecode; ------------------ | | 80| 0|#define const_CAST_shared(c) CAST(const shared_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 132| 73.6k| } else { 133| 73.6k| return type; 134| 73.6k| } 135| 73.6k|} roaring.c:get_writable_copy_if_shared: 146| 7.94M| uint8_t *type) { 147| 7.94M| if (*type == SHARED_CONTAINER_TYPE) { // shared, return enclosed container ------------------ | | 51| 7.94M|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (147:9): [True: 0, False: 7.94M] ------------------ 148| 0| return shared_container_extract_copy(CAST_shared(c), type); ------------------ | | 79| 0|#define CAST_shared(c) CAST(shared_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 149| 7.94M| } else { 150| 7.94M| return c; // not shared, so return as-is 151| 7.94M| } 152| 7.94M|} roaring.c:container_add: 492| 7.72M| uint8_t *new_typecode) { 493| 7.72M| c = get_writable_copy_if_shared(c, &typecode); 494| 7.72M| switch (typecode) { 495| 303k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 303k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (495:9): [True: 303k, False: 7.41M] ------------------ 496| 303k| bitset_container_set(CAST_bitset(c), val); ------------------ | | 52| 303k|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 303k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 497| 303k| *new_typecode = BITSET_CONTAINER_TYPE; ------------------ | | 48| 303k|#define BITSET_CONTAINER_TYPE 1 ------------------ 498| 303k| return c; 499| 7.02M| case ARRAY_CONTAINER_TYPE: { ------------------ | | 49| 7.02M|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (499:9): [True: 7.02M, False: 694k] ------------------ 500| 7.02M| array_container_t *ac = CAST_array(c); ------------------ | | 54| 7.02M|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 7.02M|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 501| 7.02M| if (array_container_try_add(ac, val, DEFAULT_MAX_SIZE) != -1) { ------------------ | Branch (501:17): [True: 7.02M, False: 103] ------------------ 502| 7.02M| *new_typecode = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 7.02M|#define ARRAY_CONTAINER_TYPE 2 ------------------ 503| 7.02M| return ac; 504| 7.02M| } else { 505| 103| bitset_container_t *bitset = bitset_container_from_array(ac); 506| 103| bitset_container_add(bitset, val); 507| 103| *new_typecode = BITSET_CONTAINER_TYPE; ------------------ | | 48| 103|#define BITSET_CONTAINER_TYPE 1 ------------------ 508| 103| return bitset; 509| 103| } 510| 7.02M| } break; 511| 390k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 390k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (511:9): [True: 390k, False: 7.33M] ------------------ 512| | // per Java, no container type adjustments are done (revisit?) 513| 390k| run_container_add(CAST_run(c), val); ------------------ | | 77| 390k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 390k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 514| 390k| *new_typecode = RUN_CONTAINER_TYPE; ------------------ | | 50| 390k|#define RUN_CONTAINER_TYPE 3 ------------------ 515| 390k| return c; 516| 0| default: ------------------ | Branch (516:9): [True: 0, False: 7.72M] ------------------ 517| 0| assert(false); 518| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 519| 0| return NULL; 520| 7.72M| } 521| 7.72M|} roaring.c:container_remove: 533| 11.0k| uint8_t *new_typecode) { 534| 11.0k| c = get_writable_copy_if_shared(c, &typecode); 535| 11.0k| switch (typecode) { 536| 2.09k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 2.09k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (536:9): [True: 2.09k, False: 8.91k] ------------------ 537| 2.09k| if (bitset_container_remove(CAST_bitset(c), val)) { ------------------ | | 52| 2.09k|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 2.09k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (537:17): [True: 1.25k, False: 846] ------------------ 538| 1.25k| int card = bitset_container_cardinality(CAST_bitset(c)); ------------------ | | 52| 1.25k|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.25k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 539| 1.25k| if (card <= DEFAULT_MAX_SIZE) { ------------------ | Branch (539:21): [True: 17, False: 1.23k] ------------------ 540| 17| *new_typecode = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 17|#define ARRAY_CONTAINER_TYPE 2 ------------------ 541| 17| return array_container_from_bitset(CAST_bitset(c)); ------------------ | | 52| 17|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 17|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 542| 17| } 543| 1.25k| } 544| 2.07k| *new_typecode = typecode; 545| 2.07k| return c; 546| 5.96k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 5.96k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (546:9): [True: 5.96k, False: 5.04k] ------------------ 547| 5.96k| *new_typecode = typecode; 548| 5.96k| array_container_remove(CAST_array(c), val); ------------------ | | 54| 5.96k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 5.96k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 549| 5.96k| return c; 550| 2.95k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 2.95k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (550:9): [True: 2.95k, False: 8.05k] ------------------ 551| | // per Java, no container type adjustments are done (revisit?) 552| 2.95k| run_container_remove(CAST_run(c), val); ------------------ | | 77| 2.95k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 2.95k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 553| 2.95k| *new_typecode = RUN_CONTAINER_TYPE; ------------------ | | 50| 2.95k|#define RUN_CONTAINER_TYPE 3 ------------------ 554| 2.95k| return c; 555| 0| default: ------------------ | Branch (555:9): [True: 0, False: 11.0k] ------------------ 556| 0| assert(false); 557| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 558| 0| return NULL; 559| 11.0k| } 560| 11.0k|} roaring.c:container_nonzero_cardinality: 437| 165k| uint8_t typecode) { 438| 165k| c = container_unwrap_shared(c, &typecode); 439| 165k| switch (typecode) { ------------------ | Branch (439:13): [True: 165k, False: 0] ------------------ 440| 6.47k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 6.47k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (440:9): [True: 6.47k, False: 159k] ------------------ 441| 6.47k| return bitset_container_const_nonzero_cardinality( 442| 6.47k| const_CAST_bitset(c)); ------------------ | | 53| 6.47k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 6.47k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 443| 98.4k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 98.4k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (443:9): [True: 98.4k, False: 67.3k] ------------------ 444| 98.4k| return array_container_nonzero_cardinality(const_CAST_array(c)); ------------------ | | 55| 98.4k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 98.4k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 445| 60.9k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 60.9k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (445:9): [True: 60.9k, False: 104k] ------------------ 446| 60.9k| return run_container_nonzero_cardinality(const_CAST_run(c)); ------------------ | | 78| 60.9k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 60.9k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 447| 165k| } 448| 165k| assert(false); 449| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 450| 0| return 0; // unreached 451| 165k|} roaring.c:container_and: 728| 6.03k| uint8_t *result_type) { 729| 6.03k| c1 = container_unwrap_shared(c1, &type1); 730| 6.03k| c2 = container_unwrap_shared(c2, &type2); 731| 6.03k| container_t *result = NULL; 732| 6.03k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 6.03k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 733| 0| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (733:9): [True: 0, False: 6.03k] ------------------ 734| 0| *result_type = 735| 0| bitset_bitset_container_intersection( ------------------ | Branch (735:17): [True: 0, False: 0] ------------------ 736| 0| const_CAST_bitset(c1), const_CAST_bitset(c2), &result) ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_bitset(c1), const_CAST_bitset(c2), &result) ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 737| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 738| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 739| 0| return result; 740| | 741| 2.27k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 2.27k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (741:9): [True: 2.27k, False: 3.76k] ------------------ 742| 2.27k| result = array_container_create(); 743| 2.27k| array_container_intersection( 744| 2.27k| const_CAST_array(c1), const_CAST_array(c2), CAST_array(result)); ------------------ | | 55| 2.27k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_array(c2), CAST_array(result)); ------------------ | | 55| 2.27k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_array(c2), CAST_array(result)); ------------------ | | 54| 2.27k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 745| 2.27k| *result_type = ARRAY_CONTAINER_TYPE; // never bitset ------------------ | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 746| 2.27k| return result; 747| | 748| 677| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 677| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 677|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 677|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (748:9): [True: 677, False: 5.35k] ------------------ 749| 677| result = run_container_create(); 750| 677| run_container_intersection(const_CAST_run(c1), const_CAST_run(c2), ------------------ | | 78| 677|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ run_container_intersection(const_CAST_run(c1), const_CAST_run(c2), ------------------ | | 78| 677|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 751| 677| CAST_run(result)); ------------------ | | 77| 677|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 752| 677| return convert_run_to_efficient_container_and_free(CAST_run(result), ------------------ | | 77| 677|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 753| 677| result_type); 754| | 755| 0| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (755:9): [True: 0, False: 6.03k] ------------------ 756| 0| result = array_container_create(); 757| 0| array_bitset_container_intersection(const_CAST_array(c2), ------------------ | | 55| 0|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 758| 0| const_CAST_bitset(c1), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 759| 0| CAST_array(result)); ------------------ | | 54| 0|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 760| 0| *result_type = ARRAY_CONTAINER_TYPE; // never bitset ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 761| 0| return result; 762| | 763| 553| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 553| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 553|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 553|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (763:9): [True: 553, False: 5.48k] ------------------ 764| 553| result = array_container_create(); 765| 553| *result_type = ARRAY_CONTAINER_TYPE; // never bitset ------------------ | | 49| 553|#define ARRAY_CONTAINER_TYPE 2 ------------------ 766| 553| array_bitset_container_intersection(const_CAST_array(c1), ------------------ | | 55| 553|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 767| 553| const_CAST_bitset(c2), ------------------ | | 53| 553|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 768| 553| CAST_array(result)); ------------------ | | 54| 553|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 769| 553| return result; 770| | 771| 0| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (771:9): [True: 0, False: 6.03k] ------------------ 772| 0| *result_type = 773| 0| run_bitset_container_intersection( ------------------ | Branch (773:17): [True: 0, False: 0] ------------------ 774| 0| const_CAST_run(c2), const_CAST_bitset(c1), &result) ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c2), const_CAST_bitset(c1), &result) ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 775| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 776| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 777| 0| return result; 778| | 779| 606| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 606| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 606|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 606|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (779:9): [True: 606, False: 5.42k] ------------------ 780| 606| *result_type = 781| 606| run_bitset_container_intersection( ------------------ | Branch (781:17): [True: 0, False: 606] ------------------ 782| 606| const_CAST_run(c1), const_CAST_bitset(c2), &result) ------------------ | | 78| 606|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c1), const_CAST_bitset(c2), &result) ------------------ | | 53| 606|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 783| 606| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 784| 606| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 1.21k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 785| 606| return result; 786| | 787| 723| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 723| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 723|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 723|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (787:9): [True: 723, False: 5.31k] ------------------ 788| 723| result = array_container_create(); 789| 723| *result_type = ARRAY_CONTAINER_TYPE; // never bitset ------------------ | | 49| 723|#define ARRAY_CONTAINER_TYPE 2 ------------------ 790| 723| array_run_container_intersection( 791| 723| const_CAST_array(c1), const_CAST_run(c2), CAST_array(result)); ------------------ | | 55| 723|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_run(c2), CAST_array(result)); ------------------ | | 78| 723|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_run(c2), CAST_array(result)); ------------------ | | 54| 723|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 792| 723| return result; 793| | 794| 1.20k| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 1.20k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 1.20k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 1.20k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (794:9): [True: 1.20k, False: 4.83k] ------------------ 795| 1.20k| result = array_container_create(); 796| 1.20k| *result_type = ARRAY_CONTAINER_TYPE; // never bitset ------------------ | | 49| 1.20k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 797| 1.20k| array_run_container_intersection( 798| 1.20k| const_CAST_array(c2), const_CAST_run(c1), CAST_array(result)); ------------------ | | 55| 1.20k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c2), const_CAST_run(c1), CAST_array(result)); ------------------ | | 78| 1.20k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c2), const_CAST_run(c1), CAST_array(result)); ------------------ | | 54| 1.20k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 799| 1.20k| return result; 800| | 801| 0| default: ------------------ | Branch (801:9): [True: 0, False: 6.03k] ------------------ 802| 0| assert(false); 803| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 804| 0| return NULL; 805| 6.03k| } 806| 6.03k|} roaring.c:container_iand: 923| 62.9k| uint8_t *result_type) { 924| 62.9k| c1 = get_writable_copy_if_shared(c1, &type1); 925| 62.9k| c2 = container_unwrap_shared(c2, &type2); 926| 62.9k| container_t *result = NULL; 927| 62.9k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 62.9k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 928| 1.39k| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 1.39k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 1.39k|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 1.39k|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (928:9): [True: 1.39k, False: 61.5k] ------------------ 929| 1.39k| *result_type = bitset_bitset_container_intersection_inplace( ------------------ | Branch (929:28): [True: 1.39k, False: 0] ------------------ 930| 1.39k| CAST_bitset(c1), const_CAST_bitset(c2), &result) ------------------ | | 52| 1.39k|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.39k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_bitset(c1), const_CAST_bitset(c2), &result) ------------------ | | 53| 1.39k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 1.39k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 931| 1.39k| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 1.39k|#define BITSET_CONTAINER_TYPE 1 ------------------ 932| 1.39k| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 1.39k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 933| 1.39k| return result; 934| | 935| 4.64k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 4.64k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 4.64k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 4.64k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (935:9): [True: 4.64k, False: 58.3k] ------------------ 936| 4.64k| array_container_intersection_inplace(CAST_array(c1), ------------------ | | 54| 4.64k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 4.64k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 937| 4.64k| const_CAST_array(c2)); ------------------ | | 55| 4.64k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 4.64k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 938| 4.64k| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 4.64k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 939| 4.64k| return c1; 940| | 941| 55.6k| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 55.6k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 55.6k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 55.6k|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (941:9): [True: 55.6k, False: 7.29k] ------------------ 942| 55.6k| result = run_container_create(); 943| 55.6k| run_container_intersection(const_CAST_run(c1), const_CAST_run(c2), ------------------ | | 78| 55.6k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 55.6k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ run_container_intersection(const_CAST_run(c1), const_CAST_run(c2), ------------------ | | 78| 55.6k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 55.6k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 944| 55.6k| CAST_run(result)); ------------------ | | 77| 55.6k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 55.6k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 945| | // as of January 2016, Java code used non-in-place intersection for 946| | // two runcontainers 947| 55.6k| return convert_run_to_efficient_container_and_free(CAST_run(result), ------------------ | | 77| 55.6k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 55.6k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 948| 55.6k| result_type); 949| | 950| 49| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 49| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 49|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 49|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (950:9): [True: 49, False: 62.9k] ------------------ 951| | // c1 is a bitmap so no inplace possible 952| 49| result = array_container_create(); 953| 49| array_bitset_container_intersection(const_CAST_array(c2), ------------------ | | 55| 49|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 49|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 954| 49| const_CAST_bitset(c1), ------------------ | | 53| 49|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 49|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 955| 49| CAST_array(result)); ------------------ | | 54| 49|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 49|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 956| 49| *result_type = ARRAY_CONTAINER_TYPE; // never bitset ------------------ | | 49| 49|#define ARRAY_CONTAINER_TYPE 2 ------------------ 957| 49| return result; 958| | 959| 0| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (959:9): [True: 0, False: 62.9k] ------------------ 960| 0| *result_type = ARRAY_CONTAINER_TYPE; // never bitset ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 961| 0| array_bitset_container_intersection( 962| 0| const_CAST_array(c1), const_CAST_bitset(c2), ------------------ | | 55| 0|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_bitset(c2), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 963| 0| CAST_array(c1)); // result is allowed to be same as c1 ------------------ | | 54| 0|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 964| 0| return c1; 965| | 966| 0| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (966:9): [True: 0, False: 62.9k] ------------------ 967| | // will attempt in-place computation 968| 0| *result_type = run_bitset_container_intersection( ------------------ | Branch (968:28): [True: 0, False: 0] ------------------ 969| 0| const_CAST_run(c2), const_CAST_bitset(c1), &c1) ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c2), const_CAST_bitset(c1), &c1) ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 970| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 971| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 972| 0| return c1; 973| | 974| 0| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (974:9): [True: 0, False: 62.9k] ------------------ 975| 0| *result_type = 976| 0| run_bitset_container_intersection( ------------------ | Branch (976:17): [True: 0, False: 0] ------------------ 977| 0| const_CAST_run(c1), const_CAST_bitset(c2), &result) ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c1), const_CAST_bitset(c2), &result) ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 978| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 979| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 980| 0| return result; 981| | 982| 189| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 189| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 189|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 189|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (982:9): [True: 189, False: 62.7k] ------------------ 983| 189| result = array_container_create(); 984| 189| *result_type = ARRAY_CONTAINER_TYPE; // never bitset ------------------ | | 49| 189|#define ARRAY_CONTAINER_TYPE 2 ------------------ 985| 189| array_run_container_intersection( 986| 189| const_CAST_array(c1), const_CAST_run(c2), CAST_array(result)); ------------------ | | 55| 189|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 189|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_run(c2), CAST_array(result)); ------------------ | | 78| 189|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 189|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_run(c2), CAST_array(result)); ------------------ | | 54| 189|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 189|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 987| 189| return result; 988| | 989| 1.01k| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 1.01k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 1.01k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 1.01k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (989:9): [True: 1.01k, False: 61.9k] ------------------ 990| 1.01k| result = array_container_create(); 991| 1.01k| *result_type = ARRAY_CONTAINER_TYPE; // never bitset ------------------ | | 49| 1.01k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 992| 1.01k| array_run_container_intersection( 993| 1.01k| const_CAST_array(c2), const_CAST_run(c1), CAST_array(result)); ------------------ | | 55| 1.01k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 1.01k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c2), const_CAST_run(c1), CAST_array(result)); ------------------ | | 78| 1.01k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 1.01k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c2), const_CAST_run(c1), CAST_array(result)); ------------------ | | 54| 1.01k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.01k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 994| 1.01k| return result; 995| | 996| 0| default: ------------------ | Branch (996:9): [True: 0, False: 62.9k] ------------------ 997| 0| assert(false); 998| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 999| 0| return NULL; 1000| 62.9k| } 1001| 62.9k|} roaring.c:container_or: 1010| 6.03k| uint8_t *result_type) { 1011| 6.03k| c1 = container_unwrap_shared(c1, &type1); 1012| 6.03k| c2 = container_unwrap_shared(c2, &type2); 1013| 6.03k| container_t *result = NULL; 1014| 6.03k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 6.03k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 1015| 0| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1015:9): [True: 0, False: 6.03k] ------------------ 1016| 0| result = bitset_container_create(); 1017| 0| bitset_container_or(const_CAST_bitset(c1), const_CAST_bitset(c2), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ bitset_container_or(const_CAST_bitset(c1), const_CAST_bitset(c2), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1018| 0| CAST_bitset(result)); ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1019| 0| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1020| 0| return result; 1021| | 1022| 2.27k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 2.27k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1022:9): [True: 2.27k, False: 3.76k] ------------------ 1023| 2.27k| *result_type = 1024| 2.27k| array_array_container_union(const_CAST_array(c1), ------------------ | | 55| 2.27k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1024:17): [True: 49, False: 2.22k] ------------------ 1025| 2.27k| const_CAST_array(c2), &result) ------------------ | | 55| 2.27k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1026| 2.27k| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 49|#define BITSET_CONTAINER_TYPE 1 ------------------ 1027| 2.27k| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 4.49k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1028| 2.27k| return result; 1029| | 1030| 677| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 677| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 677|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 677|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1030:9): [True: 677, False: 5.35k] ------------------ 1031| 677| result = run_container_create(); 1032| 677| run_container_union(const_CAST_run(c1), const_CAST_run(c2), ------------------ | | 78| 677|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ run_container_union(const_CAST_run(c1), const_CAST_run(c2), ------------------ | | 78| 677|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1033| 677| CAST_run(result)); ------------------ | | 77| 677|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1034| 677| *result_type = RUN_CONTAINER_TYPE; ------------------ | | 50| 677|#define RUN_CONTAINER_TYPE 3 ------------------ 1035| | // todo: could be optimized since will never convert to array 1036| 677| result = convert_run_to_efficient_container_and_free( 1037| 677| CAST_run(result), result_type); ------------------ | | 77| 677|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1038| 677| return result; 1039| | 1040| 0| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1040:9): [True: 0, False: 6.03k] ------------------ 1041| 0| result = bitset_container_create(); 1042| 0| array_bitset_container_union(const_CAST_array(c2), ------------------ | | 55| 0|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1043| 0| const_CAST_bitset(c1), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1044| 0| CAST_bitset(result)); ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1045| 0| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1046| 0| return result; 1047| | 1048| 553| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 553| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 553|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 553|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1048:9): [True: 553, False: 5.48k] ------------------ 1049| 553| result = bitset_container_create(); 1050| 553| array_bitset_container_union(const_CAST_array(c1), ------------------ | | 55| 553|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1051| 553| const_CAST_bitset(c2), ------------------ | | 53| 553|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1052| 553| CAST_bitset(result)); ------------------ | | 52| 553|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1053| 553| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 553|#define BITSET_CONTAINER_TYPE 1 ------------------ 1054| 553| return result; 1055| | 1056| 0| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1056:9): [True: 0, False: 6.03k] ------------------ 1057| 0| if (run_container_is_full(const_CAST_run(c2))) { ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1057:17): [True: 0, False: 0] ------------------ 1058| 0| result = run_container_create(); 1059| 0| *result_type = RUN_CONTAINER_TYPE; ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 1060| 0| run_container_copy(const_CAST_run(c2), CAST_run(result)); ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ run_container_copy(const_CAST_run(c2), CAST_run(result)); ------------------ | | 77| 0|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1061| 0| return result; 1062| 0| } 1063| 0| result = bitset_container_create(); 1064| 0| run_bitset_container_union( 1065| 0| const_CAST_run(c2), const_CAST_bitset(c1), CAST_bitset(result)); ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c2), const_CAST_bitset(c1), CAST_bitset(result)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c2), const_CAST_bitset(c1), CAST_bitset(result)); ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1066| 0| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1067| 0| return result; 1068| | 1069| 606| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 606| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 606|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 606|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1069:9): [True: 606, False: 5.42k] ------------------ 1070| 606| if (run_container_is_full(const_CAST_run(c1))) { ------------------ | | 78| 606|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1070:17): [True: 0, False: 606] ------------------ 1071| 0| result = run_container_create(); 1072| 0| *result_type = RUN_CONTAINER_TYPE; ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 1073| 0| run_container_copy(const_CAST_run(c1), CAST_run(result)); ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ run_container_copy(const_CAST_run(c1), CAST_run(result)); ------------------ | | 77| 0|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1074| 0| return result; 1075| 0| } 1076| 606| result = bitset_container_create(); 1077| 606| run_bitset_container_union( 1078| 606| const_CAST_run(c1), const_CAST_bitset(c2), CAST_bitset(result)); ------------------ | | 78| 606|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c1), const_CAST_bitset(c2), CAST_bitset(result)); ------------------ | | 53| 606|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c1), const_CAST_bitset(c2), CAST_bitset(result)); ------------------ | | 52| 606|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1079| 606| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 606|#define BITSET_CONTAINER_TYPE 1 ------------------ 1080| 606| return result; 1081| | 1082| 723| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 723| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 723|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 723|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1082:9): [True: 723, False: 5.31k] ------------------ 1083| 723| result = run_container_create(); 1084| 723| array_run_container_union(const_CAST_array(c1), const_CAST_run(c2), ------------------ | | 55| 723|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ array_run_container_union(const_CAST_array(c1), const_CAST_run(c2), ------------------ | | 78| 723|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1085| 723| CAST_run(result)); ------------------ | | 77| 723|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1086| 723| result = convert_run_to_efficient_container_and_free( 1087| 723| CAST_run(result), result_type); ------------------ | | 77| 723|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1088| 723| return result; 1089| | 1090| 1.20k| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 1.20k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 1.20k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 1.20k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1090:9): [True: 1.20k, False: 4.83k] ------------------ 1091| 1.20k| result = run_container_create(); 1092| 1.20k| array_run_container_union(const_CAST_array(c2), const_CAST_run(c1), ------------------ | | 55| 1.20k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ array_run_container_union(const_CAST_array(c2), const_CAST_run(c1), ------------------ | | 78| 1.20k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1093| 1.20k| CAST_run(result)); ------------------ | | 77| 1.20k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1094| 1.20k| result = convert_run_to_efficient_container_and_free( 1095| 1.20k| CAST_run(result), result_type); ------------------ | | 77| 1.20k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1096| 1.20k| return result; 1097| | 1098| 0| default: ------------------ | Branch (1098:9): [True: 0, False: 6.03k] ------------------ 1099| 0| assert(false); 1100| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 1101| 0| return NULL; // unreached 1102| 6.03k| } 1103| 6.03k|} roaring.c:container_is_full: 262| 7.23k|static inline bool container_is_full(const container_t *c, uint8_t typecode) { 263| 7.23k| c = container_unwrap_shared(c, &typecode); 264| 7.23k| switch (typecode) { ------------------ | Branch (264:13): [True: 7.23k, False: 0] ------------------ 265| 0| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (265:9): [True: 0, False: 7.23k] ------------------ 266| 0| return bitset_container_cardinality(const_CAST_bitset(c)) == ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 267| 0| (1 << 16); 268| 3.63k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 3.63k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (268:9): [True: 3.63k, False: 3.59k] ------------------ 269| 3.63k| return array_container_cardinality(const_CAST_array(c)) == ------------------ | | 55| 3.63k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 3.63k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 270| 3.63k| (1 << 16); 271| 3.59k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 3.59k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (271:9): [True: 3.59k, False: 3.63k] ------------------ 272| 3.59k| return run_container_is_full(const_CAST_run(c)); ------------------ | | 78| 3.59k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 3.59k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 273| 7.23k| } 274| 7.23k| assert(false); 275| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 276| 0| return 0; // unreached 277| 7.23k|} roaring.c:container_ior: 1228| 6.15k| uint8_t *result_type) { 1229| 6.15k| c1 = get_writable_copy_if_shared(c1, &type1); 1230| 6.15k| c2 = container_unwrap_shared(c2, &type2); 1231| 6.15k| container_t *result = NULL; 1232| 6.15k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 6.15k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 1233| 0| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1233:9): [True: 0, False: 6.15k] ------------------ 1234| 0| bitset_container_or(const_CAST_bitset(c1), const_CAST_bitset(c2), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ bitset_container_or(const_CAST_bitset(c1), const_CAST_bitset(c2), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1235| 0| CAST_bitset(c1)); ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1236| 0|#ifdef OR_BITSET_CONVERSION_TO_FULL 1237| 0| if (CAST_bitset(c1)->cardinality == (1 << 16)) { // we convert ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1237:17): [True: 0, False: 0] ------------------ 1238| 0| result = run_container_create_range(0, (1 << 16)); 1239| 0| *result_type = RUN_CONTAINER_TYPE; ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 1240| 0| return result; 1241| 0| } 1242| 0|#endif 1243| 0| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1244| 0| return c1; 1245| | 1246| 2.45k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 2.45k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.45k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.45k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1246:9): [True: 2.45k, False: 3.70k] ------------------ 1247| 2.45k| *result_type = array_array_container_inplace_union( ------------------ | Branch (1247:28): [True: 0, False: 2.45k] ------------------ 1248| 2.45k| CAST_array(c1), const_CAST_array(c2), &result) ------------------ | | 54| 2.45k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 2.45k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_array(c1), const_CAST_array(c2), &result) ------------------ | | 55| 2.45k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.45k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1249| 2.45k| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1250| 2.45k| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 4.90k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1251| 2.45k| if ((result == NULL) && (*result_type == ARRAY_CONTAINER_TYPE)) { ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (1251:17): [True: 0, False: 2.45k] | Branch (1251:37): [True: 0, False: 0] ------------------ 1252| 0| return c1; // the computation was done in-place! 1253| 0| } 1254| 2.45k| return result; 1255| | 1256| 1.68k| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 1.68k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 1.68k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 1.68k|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1256:9): [True: 1.68k, False: 4.47k] ------------------ 1257| 1.68k| run_container_union_inplace(CAST_run(c1), const_CAST_run(c2)); ------------------ | | 77| 1.68k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.68k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ run_container_union_inplace(CAST_run(c1), const_CAST_run(c2)); ------------------ | | 78| 1.68k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 1.68k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1258| 1.68k| return convert_run_to_efficient_container(CAST_run(c1), ------------------ | | 77| 1.68k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.68k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1259| 1.68k| result_type); 1260| | 1261| 0| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1261:9): [True: 0, False: 6.15k] ------------------ 1262| 0| array_bitset_container_union( 1263| 0| const_CAST_array(c2), const_CAST_bitset(c1), CAST_bitset(c1)); ------------------ | | 55| 0|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c2), const_CAST_bitset(c1), CAST_bitset(c1)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c2), const_CAST_bitset(c1), CAST_bitset(c1)); ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1264| 0| *result_type = BITSET_CONTAINER_TYPE; // never array ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1265| 0| return c1; 1266| | 1267| 602| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 602| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 602|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 602|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1267:9): [True: 602, False: 5.55k] ------------------ 1268| | // c1 is an array, so no in-place possible 1269| 602| result = bitset_container_create(); 1270| 602| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 602|#define BITSET_CONTAINER_TYPE 1 ------------------ 1271| 602| array_bitset_container_union(const_CAST_array(c1), ------------------ | | 55| 602|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 602|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1272| 602| const_CAST_bitset(c2), ------------------ | | 53| 602|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 602|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1273| 602| CAST_bitset(result)); ------------------ | | 52| 602|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 602|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1274| 602| return result; 1275| | 1276| 0| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1276:9): [True: 0, False: 6.15k] ------------------ 1277| 0| if (run_container_is_full(const_CAST_run(c2))) { ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1277:17): [True: 0, False: 0] ------------------ 1278| 0| result = run_container_create(); 1279| 0| *result_type = RUN_CONTAINER_TYPE; ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 1280| 0| run_container_copy(const_CAST_run(c2), CAST_run(result)); ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ run_container_copy(const_CAST_run(c2), CAST_run(result)); ------------------ | | 77| 0|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1281| 0| return result; 1282| 0| } 1283| 0| run_bitset_container_union(const_CAST_run(c2), ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1284| 0| const_CAST_bitset(c1), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1285| 0| CAST_bitset(c1)); // allowed ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1286| 0| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1287| 0| return c1; 1288| | 1289| 606| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 606| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 606|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 606|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1289:9): [True: 606, False: 5.54k] ------------------ 1290| 606| if (run_container_is_full(const_CAST_run(c1))) { ------------------ | | 78| 606|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1290:17): [True: 0, False: 606] ------------------ 1291| 0| *result_type = RUN_CONTAINER_TYPE; ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 1292| 0| return c1; 1293| 0| } 1294| 606| result = bitset_container_create(); 1295| 606| run_bitset_container_union( 1296| 606| const_CAST_run(c1), const_CAST_bitset(c2), CAST_bitset(result)); ------------------ | | 78| 606|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c1), const_CAST_bitset(c2), CAST_bitset(result)); ------------------ | | 53| 606|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c1), const_CAST_bitset(c2), CAST_bitset(result)); ------------------ | | 52| 606|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1297| 606| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 606|#define BITSET_CONTAINER_TYPE 1 ------------------ 1298| 606| return result; 1299| | 1300| 580| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 580| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 580|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 580|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1300:9): [True: 580, False: 5.57k] ------------------ 1301| 580| result = run_container_create(); 1302| 580| array_run_container_union(const_CAST_array(c1), const_CAST_run(c2), ------------------ | | 55| 580|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 580|#define CAST(type, value) ((type)value) | | ------------------ ------------------ array_run_container_union(const_CAST_array(c1), const_CAST_run(c2), ------------------ | | 78| 580|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 580|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1303| 580| CAST_run(result)); ------------------ | | 77| 580|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 580|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1304| 580| result = convert_run_to_efficient_container_and_free( 1305| 580| CAST_run(result), result_type); ------------------ | | 77| 580|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 580|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1306| 580| return result; 1307| | 1308| 233| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 233| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 233|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 233|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1308:9): [True: 233, False: 5.92k] ------------------ 1309| 233| array_run_container_inplace_union(const_CAST_array(c2), ------------------ | | 55| 233|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 233|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1310| 233| CAST_run(c1)); ------------------ | | 77| 233|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 233|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1311| 233| c1 = convert_run_to_efficient_container(CAST_run(c1), result_type); ------------------ | | 77| 233|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 233|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1312| 233| return c1; 1313| | 1314| 0| default: ------------------ | Branch (1314:9): [True: 0, False: 6.15k] ------------------ 1315| 0| assert(false); 1316| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 1317| 0| return NULL; 1318| 6.15k| } 1319| 6.15k|} roaring.c:container_xor: 1451| 6.03k| uint8_t *result_type) { 1452| 6.03k| c1 = container_unwrap_shared(c1, &type1); 1453| 6.03k| c2 = container_unwrap_shared(c2, &type2); 1454| 6.03k| container_t *result = NULL; 1455| 6.03k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 6.03k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 1456| 0| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1456:9): [True: 0, False: 6.03k] ------------------ 1457| 0| *result_type = 1458| 0| bitset_bitset_container_xor(const_CAST_bitset(c1), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1458:17): [True: 0, False: 0] ------------------ 1459| 0| const_CAST_bitset(c2), &result) ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1460| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1461| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1462| 0| return result; 1463| | 1464| 2.27k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 2.27k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1464:9): [True: 2.27k, False: 3.76k] ------------------ 1465| 2.27k| *result_type = 1466| 2.27k| array_array_container_xor(const_CAST_array(c1), ------------------ | | 55| 2.27k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1466:17): [True: 9, False: 2.26k] ------------------ 1467| 2.27k| const_CAST_array(c2), &result) ------------------ | | 55| 2.27k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1468| 2.27k| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 9|#define BITSET_CONTAINER_TYPE 1 ------------------ 1469| 2.27k| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 4.53k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1470| 2.27k| return result; 1471| | 1472| 677| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 677| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 677|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 677|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1472:9): [True: 677, False: 5.35k] ------------------ 1473| 677| *result_type = (uint8_t)run_run_container_xor( 1474| 677| const_CAST_run(c1), const_CAST_run(c2), &result); ------------------ | | 78| 677|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c1), const_CAST_run(c2), &result); ------------------ | | 78| 677|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1475| 677| return result; 1476| | 1477| 0| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1477:9): [True: 0, False: 6.03k] ------------------ 1478| 0| *result_type = 1479| 0| array_bitset_container_xor(const_CAST_array(c2), ------------------ | | 55| 0|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1479:17): [True: 0, False: 0] ------------------ 1480| 0| const_CAST_bitset(c1), &result) ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1481| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1482| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1483| 0| return result; 1484| | 1485| 553| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 553| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 553|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 553|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1485:9): [True: 553, False: 5.48k] ------------------ 1486| 553| *result_type = 1487| 553| array_bitset_container_xor(const_CAST_array(c1), ------------------ | | 55| 553|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1487:17): [True: 518, False: 35] ------------------ 1488| 553| const_CAST_bitset(c2), &result) ------------------ | | 53| 553|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1489| 553| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 518|#define BITSET_CONTAINER_TYPE 1 ------------------ 1490| 553| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 588|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1491| 553| return result; 1492| | 1493| 0| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1493:9): [True: 0, False: 6.03k] ------------------ 1494| 0| *result_type = 1495| 0| run_bitset_container_xor(const_CAST_run(c2), ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1495:17): [True: 0, False: 0] ------------------ 1496| 0| const_CAST_bitset(c1), &result) ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1497| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1498| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1499| 0| return result; 1500| | 1501| 606| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 606| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 606|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 606|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1501:9): [True: 606, False: 5.42k] ------------------ 1502| 606| *result_type = 1503| 606| run_bitset_container_xor(const_CAST_run(c1), ------------------ | | 78| 606|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1503:17): [True: 559, False: 47] ------------------ 1504| 606| const_CAST_bitset(c2), &result) ------------------ | | 53| 606|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1505| 606| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 559|#define BITSET_CONTAINER_TYPE 1 ------------------ 1506| 606| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 653|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1507| 606| return result; 1508| | 1509| 723| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 723| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 723|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 723|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1509:9): [True: 723, False: 5.31k] ------------------ 1510| 723| *result_type = (uint8_t)array_run_container_xor( 1511| 723| const_CAST_array(c1), const_CAST_run(c2), &result); ------------------ | | 55| 723|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_run(c2), &result); ------------------ | | 78| 723|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1512| 723| return result; 1513| | 1514| 1.20k| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 1.20k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 1.20k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 1.20k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1514:9): [True: 1.20k, False: 4.83k] ------------------ 1515| 1.20k| *result_type = (uint8_t)array_run_container_xor( 1516| 1.20k| const_CAST_array(c2), const_CAST_run(c1), &result); ------------------ | | 55| 1.20k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c2), const_CAST_run(c1), &result); ------------------ | | 78| 1.20k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1517| 1.20k| return result; 1518| | 1519| 0| default: ------------------ | Branch (1519:9): [True: 0, False: 6.03k] ------------------ 1520| 0| assert(false); 1521| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 1522| 0| return NULL; // unreached 1523| 6.03k| } 1524| 6.03k|} roaring.c:container_ixor: 1667| 61.4k| uint8_t *result_type) { 1668| 61.4k| c1 = get_writable_copy_if_shared(c1, &type1); 1669| 61.4k| c2 = container_unwrap_shared(c2, &type2); 1670| 61.4k| container_t *result = NULL; 1671| 61.4k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 61.4k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 1672| 1.43k| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 1.43k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 1.43k|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 1.43k|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1672:9): [True: 1.43k, False: 60.0k] ------------------ 1673| 1.43k| *result_type = bitset_bitset_container_ixor( ------------------ | Branch (1673:28): [True: 0, False: 1.43k] ------------------ 1674| 1.43k| CAST_bitset(c1), const_CAST_bitset(c2), &result) ------------------ | | 52| 1.43k|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.43k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_bitset(c1), const_CAST_bitset(c2), &result) ------------------ | | 53| 1.43k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 1.43k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1675| 1.43k| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1676| 1.43k| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 2.87k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1677| 1.43k| return result; 1678| | 1679| 4.82k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 4.82k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 4.82k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 4.82k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1679:9): [True: 4.82k, False: 56.6k] ------------------ 1680| 4.82k| *result_type = array_array_container_ixor( ------------------ | Branch (1680:28): [True: 0, False: 4.82k] ------------------ 1681| 4.82k| CAST_array(c1), const_CAST_array(c2), &result) ------------------ | | 54| 4.82k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 4.82k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_array(c1), const_CAST_array(c2), &result) ------------------ | | 55| 4.82k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 4.82k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1682| 4.82k| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1683| 4.82k| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 9.65k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1684| 4.82k| return result; 1685| | 1686| 54.7k| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 54.7k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 54.7k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 54.7k|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1686:9): [True: 54.7k, False: 6.73k] ------------------ 1687| 54.7k| *result_type = (uint8_t)run_run_container_ixor( 1688| 54.7k| CAST_run(c1), const_CAST_run(c2), &result); ------------------ | | 77| 54.7k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 54.7k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_run(c1), const_CAST_run(c2), &result); ------------------ | | 78| 54.7k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 54.7k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1689| 54.7k| return result; 1690| | 1691| 0| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1691:9): [True: 0, False: 61.4k] ------------------ 1692| 0| *result_type = bitset_array_container_ixor( ------------------ | Branch (1692:28): [True: 0, False: 0] ------------------ 1693| 0| CAST_bitset(c1), const_CAST_array(c2), &result) ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_bitset(c1), const_CAST_array(c2), &result) ------------------ | | 55| 0|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1694| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1695| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1696| 0| return result; 1697| | 1698| 89| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 89| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 89|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 89|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1698:9): [True: 89, False: 61.3k] ------------------ 1699| 89| *result_type = array_bitset_container_ixor( ------------------ | Branch (1699:28): [True: 0, False: 89] ------------------ 1700| 89| CAST_array(c1), const_CAST_bitset(c2), &result) ------------------ | | 54| 89|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 89|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_array(c1), const_CAST_bitset(c2), &result) ------------------ | | 53| 89|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 89|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1701| 89| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1702| 89| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 178|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1703| 89| return result; 1704| | 1705| 68| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 68| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 68|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 68|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1705:9): [True: 68, False: 61.3k] ------------------ 1706| 68| *result_type = bitset_run_container_ixor( ------------------ | Branch (1706:28): [True: 0, False: 68] ------------------ 1707| 68| CAST_bitset(c1), const_CAST_run(c2), &result) ------------------ | | 52| 68|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 68|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_bitset(c1), const_CAST_run(c2), &result) ------------------ | | 78| 68|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 68|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1708| 68| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1709| 68| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 136|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1710| | 1711| 68| return result; 1712| | 1713| 10| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 10| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 10|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 10|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1713:9): [True: 10, False: 61.4k] ------------------ 1714| 10| *result_type = run_bitset_container_ixor( ------------------ | Branch (1714:28): [True: 0, False: 10] ------------------ 1715| 10| CAST_run(c1), const_CAST_bitset(c2), &result) ------------------ | | 77| 10|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 10|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_run(c1), const_CAST_bitset(c2), &result) ------------------ | | 53| 10|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 10|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1716| 10| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1717| 10| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 20|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1718| 10| return result; 1719| | 1720| 253| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 253| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 253|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 253|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1720:9): [True: 253, False: 61.2k] ------------------ 1721| 253| *result_type = (uint8_t)array_run_container_ixor( 1722| 253| CAST_array(c1), const_CAST_run(c2), &result); ------------------ | | 54| 253|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 253|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_array(c1), const_CAST_run(c2), &result); ------------------ | | 78| 253|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 253|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1723| 253| return result; 1724| | 1725| 53| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 53| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 53|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 53|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1725:9): [True: 53, False: 61.4k] ------------------ 1726| 53| *result_type = (uint8_t)run_array_container_ixor( 1727| 53| CAST_run(c1), const_CAST_array(c2), &result); ------------------ | | 77| 53|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 53|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_run(c1), const_CAST_array(c2), &result); ------------------ | | 55| 53|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 53|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1728| 53| return result; 1729| | 1730| 0| default: ------------------ | Branch (1730:9): [True: 0, False: 61.4k] ------------------ 1731| 0| assert(false); 1732| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 1733| 0| return NULL; 1734| 61.4k| } 1735| 61.4k|} roaring.c:container_andnot: 1787| 6.03k| uint8_t *result_type) { 1788| 6.03k| c1 = container_unwrap_shared(c1, &type1); 1789| 6.03k| c2 = container_unwrap_shared(c2, &type2); 1790| 6.03k| container_t *result = NULL; 1791| 6.03k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 6.03k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 1792| 0| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1792:9): [True: 0, False: 6.03k] ------------------ 1793| 0| *result_type = 1794| 0| bitset_bitset_container_andnot(const_CAST_bitset(c1), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1794:17): [True: 0, False: 0] ------------------ 1795| 0| const_CAST_bitset(c2), &result) ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1796| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1797| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1798| 0| return result; 1799| | 1800| 2.27k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 2.27k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1800:9): [True: 2.27k, False: 3.76k] ------------------ 1801| 2.27k| result = array_container_create(); 1802| 2.27k| array_array_container_andnot( 1803| 2.27k| const_CAST_array(c1), const_CAST_array(c2), CAST_array(result)); ------------------ | | 55| 2.27k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_array(c2), CAST_array(result)); ------------------ | | 55| 2.27k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_array(c2), CAST_array(result)); ------------------ | | 54| 2.27k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1804| 2.27k| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1805| 2.27k| return result; 1806| | 1807| 677| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 677| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 677|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 677|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1807:9): [True: 677, False: 5.35k] ------------------ 1808| 677| if (run_container_is_full(const_CAST_run(c2))) { ------------------ | | 78| 677|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1808:17): [True: 14, False: 663] ------------------ 1809| 14| result = array_container_create(); 1810| 14| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 14|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1811| 14| return result; 1812| 14| } 1813| 663| *result_type = (uint8_t)run_run_container_andnot( 1814| 663| const_CAST_run(c1), const_CAST_run(c2), &result); ------------------ | | 78| 663|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 663|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c1), const_CAST_run(c2), &result); ------------------ | | 78| 663|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 663|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1815| 663| return result; 1816| | 1817| 0| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1817:9): [True: 0, False: 6.03k] ------------------ 1818| 0| *result_type = 1819| 0| bitset_array_container_andnot(const_CAST_bitset(c1), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1819:17): [True: 0, False: 0] ------------------ 1820| 0| const_CAST_array(c2), &result) ------------------ | | 55| 0|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1821| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1822| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1823| 0| return result; 1824| | 1825| 553| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 553| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 553|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 553|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1825:9): [True: 553, False: 5.48k] ------------------ 1826| 553| result = array_container_create(); 1827| 553| array_bitset_container_andnot(const_CAST_array(c1), ------------------ | | 55| 553|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1828| 553| const_CAST_bitset(c2), ------------------ | | 53| 553|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1829| 553| CAST_array(result)); ------------------ | | 54| 553|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1830| 553| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 553|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1831| 553| return result; 1832| | 1833| 0| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1833:9): [True: 0, False: 6.03k] ------------------ 1834| 0| if (run_container_is_full(const_CAST_run(c2))) { ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1834:17): [True: 0, False: 0] ------------------ 1835| 0| result = array_container_create(); 1836| 0| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1837| 0| return result; 1838| 0| } 1839| 0| *result_type = 1840| 0| bitset_run_container_andnot(const_CAST_bitset(c1), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1840:17): [True: 0, False: 0] ------------------ 1841| 0| const_CAST_run(c2), &result) ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1842| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1843| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1844| 0| return result; 1845| | 1846| 606| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 606| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 606|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 606|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1846:9): [True: 606, False: 5.42k] ------------------ 1847| 606| *result_type = 1848| 606| run_bitset_container_andnot(const_CAST_run(c1), ------------------ | | 78| 606|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1848:17): [True: 0, False: 606] ------------------ 1849| 606| const_CAST_bitset(c2), &result) ------------------ | | 53| 606|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1850| 606| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1851| 606| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 1.21k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1852| 606| return result; 1853| | 1854| 723| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 723| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 723|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 723|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1854:9): [True: 723, False: 5.31k] ------------------ 1855| 723| if (run_container_is_full(const_CAST_run(c2))) { ------------------ | | 78| 723|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (1855:17): [True: 34, False: 689] ------------------ 1856| 34| result = array_container_create(); 1857| 34| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 34|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1858| 34| return result; 1859| 34| } 1860| 689| result = array_container_create(); 1861| 689| array_run_container_andnot(const_CAST_array(c1), const_CAST_run(c2), ------------------ | | 55| 689|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 689|#define CAST(type, value) ((type)value) | | ------------------ ------------------ array_run_container_andnot(const_CAST_array(c1), const_CAST_run(c2), ------------------ | | 78| 689|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 689|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1862| 689| CAST_array(result)); ------------------ | | 54| 689|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 689|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1863| 689| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 689|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1864| 689| return result; 1865| | 1866| 1.20k| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 1.20k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 1.20k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 1.20k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1866:9): [True: 1.20k, False: 4.83k] ------------------ 1867| 1.20k| *result_type = (uint8_t)run_array_container_andnot( 1868| 1.20k| const_CAST_run(c1), const_CAST_array(c2), &result); ------------------ | | 78| 1.20k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c1), const_CAST_array(c2), &result); ------------------ | | 55| 1.20k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1869| 1.20k| return result; 1870| | 1871| 0| default: ------------------ | Branch (1871:9): [True: 0, False: 6.03k] ------------------ 1872| 0| assert(false); 1873| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 1874| 0| return NULL; // unreached 1875| 6.03k| } 1876| 6.03k|} roaring.c:container_iandnot: 1891| 5.73k| uint8_t *result_type) { 1892| 5.73k| c1 = get_writable_copy_if_shared(c1, &type1); 1893| 5.73k| c2 = container_unwrap_shared(c2, &type2); 1894| 5.73k| container_t *result = NULL; 1895| 5.73k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 5.73k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 1896| 0| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1896:9): [True: 0, False: 5.73k] ------------------ 1897| 0| *result_type = bitset_bitset_container_iandnot( ------------------ | Branch (1897:28): [True: 0, False: 0] ------------------ 1898| 0| CAST_bitset(c1), const_CAST_bitset(c2), &result) ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_bitset(c1), const_CAST_bitset(c2), &result) ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1899| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1900| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1901| 0| return result; 1902| | 1903| 3.43k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 3.43k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 3.43k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 3.43k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1903:9): [True: 3.43k, False: 2.29k] ------------------ 1904| 3.43k| array_array_container_iandnot(CAST_array(c1), const_CAST_array(c2)); ------------------ | | 54| 3.43k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 3.43k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ array_array_container_iandnot(CAST_array(c1), const_CAST_array(c2)); ------------------ | | 55| 3.43k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 3.43k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1905| 3.43k| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 3.43k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1906| 3.43k| return c1; 1907| | 1908| 439| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 439| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 439|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 439|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1908:9): [True: 439, False: 5.29k] ------------------ 1909| 439| *result_type = (uint8_t)run_run_container_iandnot( 1910| 439| CAST_run(c1), const_CAST_run(c2), &result); ------------------ | | 77| 439|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 439|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_run(c1), const_CAST_run(c2), &result); ------------------ | | 78| 439|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 439|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1911| 439| return result; 1912| | 1913| 1.11k| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 1.11k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 1.11k|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 1.11k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1913:9): [True: 1.11k, False: 4.61k] ------------------ 1914| 1.11k| *result_type = bitset_array_container_iandnot( ------------------ | Branch (1914:28): [True: 966, False: 153] ------------------ 1915| 1.11k| CAST_bitset(c1), const_CAST_array(c2), &result) ------------------ | | 52| 1.11k|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.11k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_bitset(c1), const_CAST_array(c2), &result) ------------------ | | 55| 1.11k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 1.11k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1916| 1.11k| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 966|#define BITSET_CONTAINER_TYPE 1 ------------------ 1917| 1.11k| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 1.27k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1918| 1.11k| return result; 1919| | 1920| 0| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1920:9): [True: 0, False: 5.73k] ------------------ 1921| 0| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1922| 0| array_bitset_container_iandnot(CAST_array(c1), ------------------ | | 54| 0|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1923| 0| const_CAST_bitset(c2)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1924| 0| return c1; 1925| | 1926| 0| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1926:9): [True: 0, False: 5.73k] ------------------ 1927| 0| *result_type = bitset_run_container_iandnot( ------------------ | Branch (1927:28): [True: 0, False: 0] ------------------ 1928| 0| CAST_bitset(c1), const_CAST_run(c2), &result) ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_bitset(c1), const_CAST_run(c2), &result) ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1929| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1930| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1931| 0| return result; 1932| | 1933| 0| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (1933:9): [True: 0, False: 5.73k] ------------------ 1934| 0| *result_type = run_bitset_container_iandnot( ------------------ | Branch (1934:28): [True: 0, False: 0] ------------------ 1935| 0| CAST_run(c1), const_CAST_bitset(c2), &result) ------------------ | | 77| 0|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_run(c1), const_CAST_bitset(c2), &result) ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1936| 0| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 1937| 0| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1938| 0| return result; 1939| | 1940| 46| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 46| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 46|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 46|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (1940:9): [True: 46, False: 5.68k] ------------------ 1941| 46| *result_type = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 46|#define ARRAY_CONTAINER_TYPE 2 ------------------ 1942| 46| array_run_container_iandnot(CAST_array(c1), const_CAST_run(c2)); ------------------ | | 54| 46|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 46|#define CAST(type, value) ((type)value) | | ------------------ ------------------ array_run_container_iandnot(CAST_array(c1), const_CAST_run(c2)); ------------------ | | 78| 46|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 46|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1943| 46| return c1; 1944| | 1945| 692| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 692| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 692|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 692|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (1945:9): [True: 692, False: 5.04k] ------------------ 1946| 692| *result_type = (uint8_t)run_array_container_iandnot( 1947| 692| CAST_run(c1), const_CAST_array(c2), &result); ------------------ | | 77| 692|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 692|#define CAST(type, value) ((type)value) | | ------------------ ------------------ CAST_run(c1), const_CAST_array(c2), &result); ------------------ | | 55| 692|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 692|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1948| 692| return result; 1949| | 1950| 0| default: ------------------ | Branch (1950:9): [True: 0, False: 5.73k] ------------------ 1951| 0| assert(false); 1952| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 1953| 0| return NULL; 1954| 5.73k| } 1955| 5.73k|} roaring.c:container_rank: 2209| 6.15k| uint16_t x) { 2210| 6.15k| c = container_unwrap_shared(c, &type); 2211| 6.15k| switch (type) { 2212| 0| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (2212:9): [True: 0, False: 6.15k] ------------------ 2213| 0| return bitset_container_rank(const_CAST_bitset(c), x); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2214| 3.63k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 3.63k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (2214:9): [True: 3.63k, False: 2.51k] ------------------ 2215| 3.63k| return array_container_rank(const_CAST_array(c), x); ------------------ | | 55| 3.63k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 3.63k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2216| 2.51k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 2.51k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (2216:9): [True: 2.51k, False: 3.63k] ------------------ 2217| 2.51k| return run_container_rank(const_CAST_run(c), x); ------------------ | | 78| 2.51k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 2.51k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2218| 0| default: ------------------ | Branch (2218:9): [True: 0, False: 6.15k] ------------------ 2219| 0| assert(false); 2220| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2221| 6.15k| } 2222| 6.15k| assert(false); 2223| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2224| 0| return false; 2225| 6.15k|} roaring.c:container_shrink_to_fit: 279| 6.15k|static inline int container_shrink_to_fit(container_t *c, uint8_t type) { 280| 6.15k| c = container_mutable_unwrap_shared(c, &type); 281| 6.15k| switch (type) { ------------------ | Branch (281:13): [True: 6.15k, False: 0] ------------------ 282| 0| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (282:9): [True: 0, False: 6.15k] ------------------ 283| 0| return 0; // no shrinking possible 284| 3.63k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 3.63k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (284:9): [True: 3.63k, False: 2.51k] ------------------ 285| 3.63k| return array_container_shrink_to_fit(CAST_array(c)); ------------------ | | 54| 3.63k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 3.63k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 286| 2.51k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 2.51k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (286:9): [True: 2.51k, False: 3.63k] ------------------ 287| 2.51k| return run_container_shrink_to_fit(CAST_run(c)); ------------------ | | 77| 2.51k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 2.51k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 288| 6.15k| } 289| 6.15k| assert(false); 290| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 291| 0| return 0; // unreached 292| 6.15k|} roaring.c:container_iterate: 1964| 63.0k| void *ptr) { 1965| 63.0k| c = container_unwrap_shared(c, &type); 1966| 63.0k| switch (type) { 1967| 1.53k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 1.53k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (1967:9): [True: 1.53k, False: 61.5k] ------------------ 1968| 1.53k| return bitset_container_iterate(const_CAST_bitset(c), base, ------------------ | | 53| 1.53k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 1.53k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1969| 1.53k| iterator, ptr); 1970| 6.25k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 6.25k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (1970:9): [True: 6.25k, False: 56.8k] ------------------ 1971| 6.25k| return array_container_iterate(const_CAST_array(c), base, iterator, ------------------ | | 55| 6.25k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 6.25k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1972| 6.25k| ptr); 1973| 55.2k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 55.2k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (1973:9): [True: 55.2k, False: 7.79k] ------------------ 1974| 55.2k| return run_container_iterate(const_CAST_run(c), base, iterator, ------------------ | | 78| 55.2k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 55.2k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1975| 55.2k| ptr); 1976| 0| default: ------------------ | Branch (1976:9): [True: 0, False: 63.0k] ------------------ 1977| 0| assert(false); 1978| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 1979| 63.0k| } 1980| 63.0k| assert(false); 1981| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 1982| 0| return false; 1983| 63.0k|} roaring.c:container_maximum: 2171| 12.1k|static inline uint16_t container_maximum(const container_t *c, uint8_t type) { 2172| 12.1k| c = container_unwrap_shared(c, &type); 2173| 12.1k| switch (type) { 2174| 2.29k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 2.29k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (2174:9): [True: 2.29k, False: 9.87k] ------------------ 2175| 2.29k| return bitset_container_maximum(const_CAST_bitset(c)); ------------------ | | 53| 2.29k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 2.29k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2176| 8.54k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 8.54k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (2176:9): [True: 8.54k, False: 3.63k] ------------------ 2177| 8.54k| return array_container_maximum(const_CAST_array(c)); ------------------ | | 55| 8.54k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 8.54k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2178| 1.33k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 1.33k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (2178:9): [True: 1.33k, False: 10.8k] ------------------ 2179| 1.33k| return run_container_maximum(const_CAST_run(c)); ------------------ | | 78| 1.33k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 1.33k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2180| 0| default: ------------------ | Branch (2180:9): [True: 0, False: 12.1k] ------------------ 2181| 0| assert(false); 2182| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2183| 12.1k| } 2184| 12.1k| assert(false); 2185| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2186| 0| return false; 2187| 12.1k|} roaring.c:container_equals: 620| 77.3k| const container_t *c2, uint8_t type2) { 621| 77.3k| c1 = container_unwrap_shared(c1, &type1); 622| 77.3k| c2 = container_unwrap_shared(c2, &type2); 623| 77.3k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 77.3k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 624| 66.7k| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 66.7k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 66.7k|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 66.7k|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (624:9): [True: 66.7k, False: 10.6k] ------------------ 625| 66.7k| return bitset_container_equals(const_CAST_bitset(c1), ------------------ | | 53| 66.7k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 66.7k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 626| 66.7k| const_CAST_bitset(c2)); ------------------ | | 53| 66.7k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 66.7k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 627| | 628| 0| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (628:9): [True: 0, False: 77.3k] ------------------ 629| 0| return run_container_equals_bitset(const_CAST_run(c2), ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 630| 0| const_CAST_bitset(c1)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 631| | 632| 0| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (632:9): [True: 0, False: 77.3k] ------------------ 633| 0| return run_container_equals_bitset(const_CAST_run(c1), ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 634| 0| const_CAST_bitset(c2)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 635| | 636| 0| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (636:9): [True: 0, False: 77.3k] ------------------ 637| | // java would always return false? 638| 0| return array_container_equal_bitset(const_CAST_array(c2), ------------------ | | 55| 0|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 639| 0| const_CAST_bitset(c1)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 640| | 641| 403| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 403| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 403|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 403|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (641:9): [True: 403, False: 76.9k] ------------------ 642| | // java would always return false? 643| 403| return array_container_equal_bitset(const_CAST_array(c1), ------------------ | | 55| 403|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 403|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 644| 403| const_CAST_bitset(c2)); ------------------ | | 53| 403|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 403|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 645| | 646| 584| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 584| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 584|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 584|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (646:9): [True: 584, False: 76.7k] ------------------ 647| 584| return run_container_equals_array(const_CAST_run(c2), ------------------ | | 78| 584|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 584|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 648| 584| const_CAST_array(c1)); ------------------ | | 55| 584|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 584|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 649| | 650| 255| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 255| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 255|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 255|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (650:9): [True: 255, False: 77.1k] ------------------ 651| 255| return run_container_equals_array(const_CAST_run(c1), ------------------ | | 78| 255|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 255|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 652| 255| const_CAST_array(c2)); ------------------ | | 55| 255|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 255|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 653| | 654| 9.26k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 9.26k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 9.26k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 9.26k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (654:9): [True: 9.26k, False: 68.0k] ------------------ 655| 9.26k| return array_container_equals(const_CAST_array(c1), ------------------ | | 55| 9.26k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 9.26k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 656| 9.26k| const_CAST_array(c2)); ------------------ | | 55| 9.26k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 9.26k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 657| | 658| 135| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 135| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 135|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 135|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (658:9): [True: 135, False: 77.2k] ------------------ 659| 135| return run_container_equals(const_CAST_run(c1), const_CAST_run(c2)); ------------------ | | 78| 135|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 135|#define CAST(type, value) ((type)value) | | ------------------ ------------------ return run_container_equals(const_CAST_run(c1), const_CAST_run(c2)); ------------------ | | 78| 135|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 135|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 660| | 661| 0| default: ------------------ | Branch (661:9): [True: 0, False: 77.3k] ------------------ 662| 0| assert(false); 663| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 664| 0| return false; 665| 77.3k| } 666| 77.3k|} roaring.c:container_is_subset: 673| 10.5k| const container_t *c2, uint8_t type2) { 674| 10.5k| c1 = container_unwrap_shared(c1, &type1); 675| 10.5k| c2 = container_unwrap_shared(c2, &type2); 676| 10.5k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 10.5k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 677| 0| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (677:9): [True: 0, False: 10.5k] ------------------ 678| 0| return bitset_container_is_subset(const_CAST_bitset(c1), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 679| 0| const_CAST_bitset(c2)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 680| | 681| 0| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (681:9): [True: 0, False: 10.5k] ------------------ 682| 0| return bitset_container_is_subset_run(const_CAST_bitset(c1), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 683| 0| const_CAST_run(c2)); ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 684| | 685| 0| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (685:9): [True: 0, False: 10.5k] ------------------ 686| 0| return run_container_is_subset_bitset(const_CAST_run(c1), ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 687| 0| const_CAST_bitset(c2)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 688| | 689| 0| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (689:9): [True: 0, False: 10.5k] ------------------ 690| 0| return false; // by construction, size(c1) > size(c2) 691| | 692| 2.24k| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 2.24k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.24k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 2.24k|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (692:9): [True: 2.24k, False: 8.28k] ------------------ 693| 2.24k| return array_container_is_subset_bitset(const_CAST_array(c1), ------------------ | | 55| 2.24k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.24k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 694| 2.24k| const_CAST_bitset(c2)); ------------------ | | 53| 2.24k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 2.24k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 695| | 696| 2.12k| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 2.12k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.12k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 2.12k|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (696:9): [True: 2.12k, False: 8.40k] ------------------ 697| 2.12k| return array_container_is_subset_run(const_CAST_array(c1), ------------------ | | 55| 2.12k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.12k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 698| 2.12k| const_CAST_run(c2)); ------------------ | | 78| 2.12k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 2.12k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 699| | 700| 374| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 374| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 374|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 374|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (700:9): [True: 374, False: 10.1k] ------------------ 701| 374| return run_container_is_subset_array(const_CAST_run(c1), ------------------ | | 78| 374|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 374|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 702| 374| const_CAST_array(c2)); ------------------ | | 55| 374|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 374|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 703| | 704| 5.54k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 5.54k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 5.54k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 5.54k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (704:9): [True: 5.54k, False: 4.98k] ------------------ 705| 5.54k| return array_container_is_subset(const_CAST_array(c1), ------------------ | | 55| 5.54k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 5.54k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 706| 5.54k| const_CAST_array(c2)); ------------------ | | 55| 5.54k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 5.54k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 707| | 708| 242| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 242| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 242|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 242|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (708:9): [True: 242, False: 10.2k] ------------------ 709| 242| return run_container_is_subset(const_CAST_run(c1), ------------------ | | 78| 242|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 242|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 710| 242| const_CAST_run(c2)); ------------------ | | 78| 242|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 242|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 711| | 712| 0| default: ------------------ | Branch (712:9): [True: 0, False: 10.5k] ------------------ 713| 0| assert(false); 714| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 715| 0| return false; 716| 10.5k| } 717| 10.5k|} roaring.c:container_inot_range: 2110| 6.28k| uint8_t *result_type) { 2111| 6.28k| c = get_writable_copy_if_shared(c, &type); 2112| 6.28k| container_t *result = NULL; 2113| 6.28k| switch (type) { 2114| 1.24k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 1.24k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (2114:9): [True: 1.24k, False: 5.03k] ------------------ 2115| 1.24k| *result_type = bitset_container_negation_range_inplace( ------------------ | Branch (2115:28): [True: 1.21k, False: 25] ------------------ 2116| 1.24k| CAST_bitset(c), range_start, range_end, &result) ------------------ | | 52| 1.24k|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.24k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2117| 1.24k| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 1.21k|#define BITSET_CONTAINER_TYPE 1 ------------------ 2118| 1.24k| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 1.26k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 2119| 1.24k| return result; 2120| 3.41k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 3.41k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (2120:9): [True: 3.41k, False: 2.86k] ------------------ 2121| 3.41k| *result_type = array_container_negation_range_inplace( ------------------ | Branch (2121:28): [True: 121, False: 3.29k] ------------------ 2122| 3.41k| CAST_array(c), range_start, range_end, &result) ------------------ | | 54| 3.41k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 3.41k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2123| 3.41k| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 121|#define BITSET_CONTAINER_TYPE 1 ------------------ 2124| 3.41k| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 6.70k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 2125| 3.41k| return result; 2126| 1.62k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 1.62k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (2126:9): [True: 1.62k, False: 4.65k] ------------------ 2127| 1.62k| *result_type = (uint8_t)run_container_negation_range_inplace( 2128| 1.62k| CAST_run(c), range_start, range_end, &result); ------------------ | | 77| 1.62k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.62k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2129| 1.62k| return result; 2130| | 2131| 0| default: ------------------ | Branch (2131:9): [True: 0, False: 6.28k] ------------------ 2132| 0| assert(false); 2133| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2134| 6.28k| } 2135| 6.28k| assert(false); 2136| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2137| 0| return NULL; 2138| 6.28k|} roaring.c:container_inot: 2076| 5.80k| uint8_t *result_type) { 2077| 5.80k| c = get_writable_copy_if_shared(c, &type); 2078| 5.80k| container_t *result = NULL; 2079| 5.80k| switch (type) { 2080| 57| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 57|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (2080:9): [True: 57, False: 5.74k] ------------------ 2081| 57| *result_type = 2082| 57| bitset_container_negation_inplace(CAST_bitset(c), &result) ------------------ | | 52| 57|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 57|#define CAST(type, value) ((type)value) | | ------------------ ------------------ | Branch (2082:17): [True: 26, False: 31] ------------------ 2083| 57| ? BITSET_CONTAINER_TYPE ------------------ | | 48| 26|#define BITSET_CONTAINER_TYPE 1 ------------------ 2084| 57| : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 88|#define ARRAY_CONTAINER_TYPE 2 ------------------ 2085| 57| return result; 2086| 151| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 151|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (2086:9): [True: 151, False: 5.65k] ------------------ 2087| | // will never be inplace 2088| 151| result = bitset_container_create(); 2089| 151| *result_type = BITSET_CONTAINER_TYPE; ------------------ | | 48| 151|#define BITSET_CONTAINER_TYPE 1 ------------------ 2090| 151| array_container_negation(CAST_array(c), CAST_bitset(result)); ------------------ | | 54| 151|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 151|#define CAST(type, value) ((type)value) | | ------------------ ------------------ array_container_negation(CAST_array(c), CAST_bitset(result)); ------------------ | | 52| 151|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 151|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2091| 151| array_container_free(CAST_array(c)); ------------------ | | 54| 151|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 151|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2092| 151| return result; 2093| 5.59k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 5.59k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (2093:9): [True: 5.59k, False: 208] ------------------ 2094| 5.59k| *result_type = 2095| 5.59k| (uint8_t)run_container_negation_inplace(CAST_run(c), &result); ------------------ | | 77| 5.59k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 5.59k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2096| 5.59k| return result; 2097| | 2098| 0| default: ------------------ | Branch (2098:9): [True: 0, False: 5.80k] ------------------ 2099| 0| assert(false); 2100| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2101| 5.80k| } 2102| 5.80k| assert(false); 2103| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2104| 0| return NULL; 2105| 5.80k|} roaring.c:container_unwrap_shared: 106| 1.07M| const container_t *candidate_shared_container, uint8_t *type) { 107| 1.07M| if (*type == SHARED_CONTAINER_TYPE) { ------------------ | | 51| 1.07M|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (107:9): [True: 0, False: 1.07M] ------------------ 108| 0| *type = const_CAST_shared(candidate_shared_container)->typecode; ------------------ | | 80| 0|#define const_CAST_shared(c) CAST(const shared_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 109| 0| assert(*type != SHARED_CONTAINER_TYPE); 110| 0| return const_CAST_shared(candidate_shared_container)->container; ------------------ | | 80| 0|#define const_CAST_shared(c) CAST(const shared_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 111| 1.07M| } else { 112| 1.07M| return candidate_shared_container; 113| 1.07M| } 114| 1.07M|} roaring.c:container_mutable_unwrap_shared: 118| 6.15k| uint8_t *type) { 119| 6.15k| if (*type == SHARED_CONTAINER_TYPE) { // the passed in container is shared ------------------ | | 51| 6.15k|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (119:9): [True: 0, False: 6.15k] ------------------ 120| 0| *type = CAST_shared(c)->typecode; ------------------ | | 79| 0|#define CAST_shared(c) CAST(shared_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 121| 0| assert(*type != SHARED_CONTAINER_TYPE); 122| 0| return CAST_shared(c)->container; // return the enclosed container ------------------ | | 79| 0|#define CAST_shared(c) CAST(shared_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 123| 6.15k| } else { 124| 6.15k| return c; // wasn't shared, so return as-is 125| 6.15k| } 126| 6.15k|} roaring.c:container_minimum: 2189| 6.10k|static inline uint16_t container_minimum(const container_t *c, uint8_t type) { 2190| 6.10k| c = container_unwrap_shared(c, &type); 2191| 6.10k| switch (type) { 2192| 1.16k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 1.16k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (2192:9): [True: 1.16k, False: 4.93k] ------------------ 2193| 1.16k| return bitset_container_minimum(const_CAST_bitset(c)); ------------------ | | 53| 1.16k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 1.16k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2194| 3.51k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 3.51k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (2194:9): [True: 3.51k, False: 2.59k] ------------------ 2195| 3.51k| return array_container_minimum(const_CAST_array(c)); ------------------ | | 55| 3.51k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 3.51k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2196| 1.42k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 1.42k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (2196:9): [True: 1.42k, False: 4.68k] ------------------ 2197| 1.42k| return run_container_minimum(const_CAST_run(c)); ------------------ | | 78| 1.42k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 1.42k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2198| 0| default: ------------------ | Branch (2198:9): [True: 0, False: 6.10k] ------------------ 2199| 0| assert(false); 2200| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2201| 6.10k| } 2202| 6.10k| assert(false); 2203| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2204| 0| return false; 2205| 6.10k|} roaring.c:container_select: 2150| 6.15k| uint32_t *element) { 2151| 6.15k| c = container_unwrap_shared(c, &type); 2152| 6.15k| switch (type) { 2153| 0| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (2153:9): [True: 0, False: 6.15k] ------------------ 2154| 0| return bitset_container_select(const_CAST_bitset(c), start_rank, ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2155| 0| rank, element); 2156| 3.63k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 3.63k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (2156:9): [True: 3.63k, False: 2.51k] ------------------ 2157| 3.63k| return array_container_select(const_CAST_array(c), start_rank, rank, ------------------ | | 55| 3.63k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 3.63k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2158| 3.63k| element); 2159| 2.51k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 2.51k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (2159:9): [True: 2.51k, False: 3.63k] ------------------ 2160| 2.51k| return run_container_select(const_CAST_run(c), start_rank, rank, ------------------ | | 78| 2.51k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 2.51k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2161| 2.51k| element); 2162| 0| default: ------------------ | Branch (2162:9): [True: 0, False: 6.15k] ------------------ 2163| 0| assert(false); 2164| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2165| 6.15k| } 2166| 6.15k| assert(false); 2167| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2168| 0| return false; 2169| 6.15k|} roaring.c:container_intersect: 865| 6.03k| const container_t *c2, uint8_t type2) { 866| 6.03k| c1 = container_unwrap_shared(c1, &type1); 867| 6.03k| c2 = container_unwrap_shared(c2, &type2); 868| 6.03k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 6.03k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 869| 0| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (869:9): [True: 0, False: 6.03k] ------------------ 870| 0| return bitset_container_intersect(const_CAST_bitset(c1), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 871| 0| const_CAST_bitset(c2)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 872| | 873| 2.27k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 2.27k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.27k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (873:9): [True: 2.27k, False: 3.76k] ------------------ 874| 2.27k| return array_container_intersect(const_CAST_array(c1), ------------------ | | 55| 2.27k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 875| 2.27k| const_CAST_array(c2)); ------------------ | | 55| 2.27k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.27k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 876| | 877| 677| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 677| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 677|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 677|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (877:9): [True: 677, False: 5.35k] ------------------ 878| 677| return run_container_intersect(const_CAST_run(c1), ------------------ | | 78| 677|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 879| 677| const_CAST_run(c2)); ------------------ | | 78| 677|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 677|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 880| | 881| 0| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (881:9): [True: 0, False: 6.03k] ------------------ 882| 0| return array_bitset_container_intersect(const_CAST_array(c2), ------------------ | | 55| 0|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 883| 0| const_CAST_bitset(c1)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 884| | 885| 553| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 553| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 553|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 553|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (885:9): [True: 553, False: 5.48k] ------------------ 886| 553| return array_bitset_container_intersect(const_CAST_array(c1), ------------------ | | 55| 553|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 887| 553| const_CAST_bitset(c2)); ------------------ | | 53| 553|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 553|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 888| | 889| 0| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (889:9): [True: 0, False: 6.03k] ------------------ 890| 0| return run_bitset_container_intersect(const_CAST_run(c2), ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 891| 0| const_CAST_bitset(c1)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 892| | 893| 606| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 606| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 606|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 606|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (893:9): [True: 606, False: 5.42k] ------------------ 894| 606| return run_bitset_container_intersect(const_CAST_run(c1), ------------------ | | 78| 606|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 895| 606| const_CAST_bitset(c2)); ------------------ | | 53| 606|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 606|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 896| | 897| 723| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 723| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 723|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 723|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (897:9): [True: 723, False: 5.31k] ------------------ 898| 723| return array_run_container_intersect(const_CAST_array(c1), ------------------ | | 55| 723|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 899| 723| const_CAST_run(c2)); ------------------ | | 78| 723|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 723|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 900| | 901| 1.20k| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 1.20k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 1.20k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 1.20k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (901:9): [True: 1.20k, False: 4.83k] ------------------ 902| 1.20k| return array_run_container_intersect(const_CAST_array(c2), ------------------ | | 55| 1.20k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 903| 1.20k| const_CAST_run(c1)); ------------------ | | 78| 1.20k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 1.20k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 904| | 905| 0| default: ------------------ | Branch (905:9): [True: 0, False: 6.03k] ------------------ 906| 0| assert(false); 907| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 908| 0| return 0; 909| 6.03k| } 910| 6.03k|} roaring.c:container_and_cardinality: 814| 24.1k| uint8_t type2) { 815| 24.1k| c1 = container_unwrap_shared(c1, &type1); 816| 24.1k| c2 = container_unwrap_shared(c2, &type2); 817| 24.1k| switch (PAIR_CONTAINER_TYPES(type1, type2)) { ------------------ | | 62| 24.1k|#define PAIR_CONTAINER_TYPES(type1, type2) (4 * (type1) + (type2)) ------------------ 818| 0| case CONTAINER_PAIR(BITSET, BITSET): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (818:9): [True: 0, False: 24.1k] ------------------ 819| 0| return bitset_container_and_justcard(const_CAST_bitset(c1), ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 820| 0| const_CAST_bitset(c2)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 821| | 822| 9.08k| case CONTAINER_PAIR(ARRAY, ARRAY): ------------------ | | 65| 9.08k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 9.08k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 9.08k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (822:9): [True: 9.08k, False: 15.0k] ------------------ 823| 9.08k| return array_container_intersection_cardinality( 824| 9.08k| const_CAST_array(c1), const_CAST_array(c2)); ------------------ | | 55| 9.08k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 9.08k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_array(c2)); ------------------ | | 55| 9.08k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 9.08k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 825| | 826| 2.70k| case CONTAINER_PAIR(RUN, RUN): ------------------ | | 65| 2.70k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 2.70k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 2.70k|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (826:9): [True: 2.70k, False: 21.4k] ------------------ 827| 2.70k| return run_container_intersection_cardinality(const_CAST_run(c1), ------------------ | | 78| 2.70k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 2.70k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 828| 2.70k| const_CAST_run(c2)); ------------------ | | 78| 2.70k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 2.70k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 829| | 830| 0| case CONTAINER_PAIR(BITSET, ARRAY): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 0|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (830:9): [True: 0, False: 24.1k] ------------------ 831| 0| return array_bitset_container_intersection_cardinality( 832| 0| const_CAST_array(c2), const_CAST_bitset(c1)); ------------------ | | 55| 0|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c2), const_CAST_bitset(c1)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 833| | 834| 2.21k| case CONTAINER_PAIR(ARRAY, BITSET): ------------------ | | 65| 2.21k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.21k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 2.21k|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (834:9): [True: 2.21k, False: 21.9k] ------------------ 835| 2.21k| return array_bitset_container_intersection_cardinality( 836| 2.21k| const_CAST_array(c1), const_CAST_bitset(c2)); ------------------ | | 55| 2.21k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.21k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_bitset(c2)); ------------------ | | 53| 2.21k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 2.21k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 837| | 838| 0| case CONTAINER_PAIR(BITSET, RUN): ------------------ | | 65| 0| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 0|#define BITSET_CONTAINER_TYPE 1 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 0|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (838:9): [True: 0, False: 24.1k] ------------------ 839| 0| return run_bitset_container_intersection_cardinality( 840| 0| const_CAST_run(c2), const_CAST_bitset(c1)); ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c2), const_CAST_bitset(c1)); ------------------ | | 53| 0|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 841| | 842| 2.42k| case CONTAINER_PAIR(RUN, BITSET): ------------------ | | 65| 2.42k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 2.42k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 48| 2.42k|#define BITSET_CONTAINER_TYPE 1 | | ------------------ ------------------ | Branch (842:9): [True: 2.42k, False: 21.7k] ------------------ 843| 2.42k| return run_bitset_container_intersection_cardinality( 844| 2.42k| const_CAST_run(c1), const_CAST_bitset(c2)); ------------------ | | 78| 2.42k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 2.42k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_run(c1), const_CAST_bitset(c2)); ------------------ | | 53| 2.42k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 2.42k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 845| | 846| 2.89k| case CONTAINER_PAIR(ARRAY, RUN): ------------------ | | 65| 2.89k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 2.89k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 2.89k|#define RUN_CONTAINER_TYPE 3 | | ------------------ ------------------ | Branch (846:9): [True: 2.89k, False: 21.2k] ------------------ 847| 2.89k| return array_run_container_intersection_cardinality( 848| 2.89k| const_CAST_array(c1), const_CAST_run(c2)); ------------------ | | 55| 2.89k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 2.89k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c1), const_CAST_run(c2)); ------------------ | | 78| 2.89k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 2.89k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 849| | 850| 4.81k| case CONTAINER_PAIR(RUN, ARRAY): ------------------ | | 65| 4.81k| (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 50| 4.81k|#define RUN_CONTAINER_TYPE 3 | | ------------------ | | (4 * (name1##_CONTAINER_TYPE) + (name2##_CONTAINER_TYPE)) | | ------------------ | | | | 49| 4.81k|#define ARRAY_CONTAINER_TYPE 2 | | ------------------ ------------------ | Branch (850:9): [True: 4.81k, False: 19.3k] ------------------ 851| 4.81k| return array_run_container_intersection_cardinality( 852| 4.81k| const_CAST_array(c2), const_CAST_run(c1)); ------------------ | | 55| 4.81k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 4.81k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ const_CAST_array(c2), const_CAST_run(c1)); ------------------ | | 78| 4.81k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 4.81k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 853| | 854| 0| default: ------------------ | Branch (854:9): [True: 0, False: 24.1k] ------------------ 855| 0| assert(false); 856| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 857| 0| return 0; 858| 24.1k| } 859| 24.1k|} roaring.c:container_contains_range: 596| 507|) { 597| 507| c = container_unwrap_shared(c, &typecode); 598| 507| switch (typecode) { 599| 233| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 233|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (599:9): [True: 233, False: 274] ------------------ 600| 233| return bitset_container_get_range(const_CAST_bitset(c), range_start, ------------------ | | 53| 233|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 233|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 601| 233| range_end); 602| 94| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 94|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (602:9): [True: 94, False: 413] ------------------ 603| 94| return array_container_contains_range(const_CAST_array(c), ------------------ | | 55| 94|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 94|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 604| 94| range_start, range_end); 605| 180| case RUN_CONTAINER_TYPE: ------------------ | | 50| 180|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (605:9): [True: 180, False: 327] ------------------ 606| 180| return run_container_contains_range(const_CAST_run(c), range_start, ------------------ | | 78| 180|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 180|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 607| 180| range_end); 608| 0| default: ------------------ | Branch (608:9): [True: 0, False: 507] ------------------ 609| 0| assert(false); 610| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 611| 0| return false; 612| 507| } 613| 507|} roaring_array.c:container_to_uint32_array: 465| 62.9k| uint8_t typecode, uint32_t base) { 466| 62.9k| c = container_unwrap_shared(c, &typecode); 467| 62.9k| switch (typecode) { ------------------ | Branch (467:13): [True: 62.9k, False: 0] ------------------ 468| 1.39k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 1.39k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (468:9): [True: 1.39k, False: 61.5k] ------------------ 469| 1.39k| return bitset_container_to_uint32_array(output, 470| 1.39k| const_CAST_bitset(c), base); ------------------ | | 53| 1.39k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 1.39k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 471| 5.70k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 5.70k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (471:9): [True: 5.70k, False: 57.2k] ------------------ 472| 5.70k| return array_container_to_uint32_array(output, const_CAST_array(c), ------------------ | | 55| 5.70k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 5.70k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 473| 5.70k| base); 474| 55.8k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 55.8k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (474:9): [True: 55.8k, False: 7.10k] ------------------ 475| 55.8k| return run_container_to_uint32_array(output, const_CAST_run(c), ------------------ | | 78| 55.8k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 55.8k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 476| 55.8k| base); 477| 62.9k| } 478| 62.9k| assert(false); 479| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 480| 0| return 0; // unreached 481| 62.9k|} roaring_array.c:container_unwrap_shared: 106| 363k| const container_t *candidate_shared_container, uint8_t *type) { 107| 363k| if (*type == SHARED_CONTAINER_TYPE) { ------------------ | | 51| 363k|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (107:9): [True: 0, False: 363k] ------------------ 108| 0| *type = const_CAST_shared(candidate_shared_container)->typecode; ------------------ | | 80| 0|#define const_CAST_shared(c) CAST(const shared_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 109| 0| assert(*type != SHARED_CONTAINER_TYPE); 110| 0| return const_CAST_shared(candidate_shared_container)->container; ------------------ | | 80| 0|#define const_CAST_shared(c) CAST(const shared_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 111| 363k| } else { 112| 363k| return candidate_shared_container; 113| 363k| } 114| 363k|} roaring_array.c:get_container_type: 129| 153k|static inline uint8_t get_container_type(const container_t *c, uint8_t type) { 130| 153k| if (type == SHARED_CONTAINER_TYPE) { ------------------ | | 51| 153k|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (130:9): [True: 0, False: 153k] ------------------ 131| 0| return const_CAST_shared(c)->typecode; ------------------ | | 80| 0|#define const_CAST_shared(c) CAST(const shared_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 132| 153k| } else { 133| 153k| return type; 134| 153k| } 135| 153k|} roaring_array.c:container_size_in_bytes: 403| 153k| uint8_t typecode) { 404| 153k| c = container_unwrap_shared(c, &typecode); 405| 153k| switch (typecode) { ------------------ | Branch (405:13): [True: 153k, False: 0] ------------------ 406| 133k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 133k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (406:9): [True: 133k, False: 19.9k] ------------------ 407| 133k| return bitset_container_size_in_bytes(const_CAST_bitset(c)); ------------------ | | 53| 133k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 133k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 408| 17.4k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 17.4k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (408:9): [True: 17.4k, False: 135k] ------------------ 409| 17.4k| return array_container_size_in_bytes(const_CAST_array(c)); ------------------ | | 55| 17.4k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 17.4k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 410| 2.51k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 2.51k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (410:9): [True: 2.51k, False: 150k] ------------------ 411| 2.51k| return run_container_size_in_bytes(const_CAST_run(c)); ------------------ | | 78| 2.51k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 2.51k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 412| 153k| } 413| 153k| assert(false); 414| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 415| 0| return 0; // unreached 416| 153k|} roaring_array.c:container_get_cardinality: 244| 73.6k| uint8_t typecode) { 245| 73.6k| c = container_unwrap_shared(c, &typecode); 246| 73.6k| switch (typecode) { ------------------ | Branch (246:13): [True: 73.6k, False: 0] ------------------ 247| 66.7k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 66.7k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (247:9): [True: 66.7k, False: 6.91k] ------------------ 248| 66.7k| return bitset_container_cardinality(const_CAST_bitset(c)); ------------------ | | 53| 66.7k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 66.7k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 249| 6.91k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 6.91k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (249:9): [True: 6.91k, False: 66.7k] ------------------ 250| 6.91k| return array_container_cardinality(const_CAST_array(c)); ------------------ | | 55| 6.91k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 6.91k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 251| 0| case RUN_CONTAINER_TYPE: ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (251:9): [True: 0, False: 73.6k] ------------------ 252| 0| return run_container_cardinality(const_CAST_run(c)); ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 253| 73.6k| } 254| 73.6k| assert(false); 255| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 256| 0| return 0; // unreached 257| 73.6k|} roaring_array.c:container_write: 382| 73.6k| char *buf) { 383| 73.6k| c = container_unwrap_shared(c, &typecode); 384| 73.6k| switch (typecode) { ------------------ | Branch (384:13): [True: 73.6k, False: 0] ------------------ 385| 66.7k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 66.7k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (385:9): [True: 66.7k, False: 6.91k] ------------------ 386| 66.7k| return bitset_container_write(const_CAST_bitset(c), buf); ------------------ | | 53| 66.7k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 66.7k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 387| 6.91k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 6.91k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (387:9): [True: 6.91k, False: 66.7k] ------------------ 388| 6.91k| return array_container_write(const_CAST_array(c), buf); ------------------ | | 55| 6.91k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 6.91k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 389| 0| case RUN_CONTAINER_TYPE: ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (389:9): [True: 0, False: 73.6k] ------------------ 390| 0| return run_container_write(const_CAST_run(c), buf); ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 391| 73.6k| } 392| 73.6k| assert(false); 393| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 394| 0| return 0; // unreached 395| 73.6k|} containers.c:container_unwrap_shared: 106| 6.30k| const container_t *candidate_shared_container, uint8_t *type) { 107| 6.30k| if (*type == SHARED_CONTAINER_TYPE) { ------------------ | | 51| 6.30k|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (107:9): [True: 0, False: 6.30k] ------------------ 108| 0| *type = const_CAST_shared(candidate_shared_container)->typecode; ------------------ | | 80| 0|#define const_CAST_shared(c) CAST(const shared_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 109| 0| assert(*type != SHARED_CONTAINER_TYPE); 110| 0| return const_CAST_shared(candidate_shared_container)->container; ------------------ | | 80| 0|#define const_CAST_shared(c) CAST(const shared_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 111| 6.30k| } else { 112| 6.30k| return candidate_shared_container; 113| 6.30k| } 114| 6.30k|} containers.c:container_maximum: 2171| 6.05k|static inline uint16_t container_maximum(const container_t *c, uint8_t type) { 2172| 6.05k| c = container_unwrap_shared(c, &type); 2173| 6.05k| switch (type) { 2174| 1.73k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 1.73k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (2174:9): [True: 1.73k, False: 4.32k] ------------------ 2175| 1.73k| return bitset_container_maximum(const_CAST_bitset(c)); ------------------ | | 53| 1.73k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 1.73k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2176| 4.32k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 4.32k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (2176:9): [True: 4.32k, False: 1.73k] ------------------ 2177| 4.32k| return array_container_maximum(const_CAST_array(c)); ------------------ | | 55| 4.32k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 4.32k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2178| 0| case RUN_CONTAINER_TYPE: ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (2178:9): [True: 0, False: 6.05k] ------------------ 2179| 0| return run_container_maximum(const_CAST_run(c)); ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 2180| 0| default: ------------------ | Branch (2180:9): [True: 0, False: 6.05k] ------------------ 2181| 0| assert(false); 2182| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2183| 6.05k| } 2184| 6.05k| assert(false); 2185| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 2186| 0| return false; 2187| 6.05k|} run_container_contains: 260| 8.48k|inline bool run_container_contains(const run_container_t *run, uint16_t pos) { 261| 8.48k| int32_t index = interleavedBinarySearch(run->runs, run->n_runs, pos); 262| 8.48k| if (index >= 0) return true; ------------------ | Branch (262:9): [True: 5.95k, False: 2.52k] ------------------ 263| 2.52k| index = -index - 2; // points to preceding value, possibly -1 264| 2.52k| if (index != -1) { // possible match ------------------ | Branch (264:9): [True: 896, False: 1.63k] ------------------ 265| 896| int32_t offset = pos - run->runs[index].value; 266| 896| int32_t le = run->runs[index].length; 267| 896| if (offset <= le) return true; ------------------ | Branch (267:13): [True: 568, False: 328] ------------------ 268| 896| } 269| 1.95k| return false; 270| 2.52k|} interleavedBinarySearch: 112| 402k| uint16_t ikey) { 113| 402k| int32_t low = 0; 114| 402k| int32_t high = lenarray - 1; 115| 1.85M| while (low <= high) { ------------------ | Branch (115:12): [True: 1.63M, False: 220k] ------------------ 116| 1.63M| int32_t middleIndex = (low + high) >> 1; 117| 1.63M| uint16_t middleValue = array[middleIndex].value; 118| 1.63M| if (middleValue < ikey) { ------------------ | Branch (118:13): [True: 636k, False: 997k] ------------------ 119| 636k| low = middleIndex + 1; 120| 997k| } else if (middleValue > ikey) { ------------------ | Branch (120:20): [True: 815k, False: 181k] ------------------ 121| 815k| high = middleIndex - 1; 122| 815k| } else { 123| 181k| return middleIndex; 124| 181k| } 125| 1.63M| } 126| 220k| return -(low + 1); 127| 402k|} run_container_minimum: 563| 17.9k|inline uint16_t run_container_minimum(const run_container_t *run) { 564| 17.9k| if (run->n_runs == 0) return 0; ------------------ | Branch (564:9): [True: 0, False: 17.9k] ------------------ 565| 17.9k| return run->runs[0].value; 566| 17.9k|} run_container_maximum: 569| 17.4k|inline uint16_t run_container_maximum(const run_container_t *run) { 570| 17.4k| if (run->n_runs == 0) return 0; ------------------ | Branch (570:9): [True: 0, False: 17.4k] ------------------ 571| 17.4k| return run->runs[run->n_runs - 1].value + run->runs[run->n_runs - 1].length; 572| 17.4k|} roaring.c:run_container_create_range: 532| 85.2k| uint32_t stop) { 533| 85.2k| run_container_t *rc = run_container_create_given_capacity(1); 534| 85.2k| if (rc) { ------------------ | Branch (534:9): [True: 85.2k, False: 0] ------------------ 535| 85.2k| rle16_t r; 536| 85.2k| r.value = (uint16_t)start; 537| 85.2k| r.length = (uint16_t)(stop - start - 1); 538| 85.2k| run_container_append_first(rc, r); 539| 85.2k| } 540| 85.2k| return rc; 541| 85.2k|} roaring.c:run_container_append_first: 351| 85.2k| rle16_t vl) { 352| 85.2k| run->runs[run->n_runs] = vl; 353| 85.2k| run->n_runs++; 354| 85.2k| return vl; 355| 85.2k|} roaring.c:rle16_count_greater: 178| 411| int32_t lenarray, uint16_t key) { 179| 411| if (lenarray == 0) return 0; ------------------ | Branch (179:9): [True: 0, False: 411] ------------------ 180| 411| int32_t low = 0; 181| 411| int32_t high = lenarray - 1; 182| 3.11k| while (low <= high) { ------------------ | Branch (182:12): [True: 2.75k, False: 369] ------------------ 183| 2.75k| int32_t middleIndex = (low + high) >> 1; 184| 2.75k| uint16_t min_value = array[middleIndex].value; 185| 2.75k| uint16_t max_value = 186| 2.75k| array[middleIndex].value + array[middleIndex].length; 187| 2.75k| if (max_value < key) { ------------------ | Branch (187:13): [True: 2.50k, False: 244] ------------------ 188| 2.50k| low = middleIndex + 1; 189| 2.50k| } else if (key + UINT32_C(1) < min_value) { // uint32 arithmetic ------------------ | Branch (189:20): [True: 202, False: 42] ------------------ 190| 202| high = middleIndex - 1; 191| 202| } else { 192| 42| return lenarray - (middleIndex + 1); 193| 42| } 194| 2.75k| } 195| 369| return lenarray - low; 196| 411|} roaring.c:rle16_count_less: 157| 411| uint16_t key) { 158| 411| if (lenarray == 0) return 0; ------------------ | Branch (158:9): [True: 1, False: 410] ------------------ 159| 410| int32_t low = 0; 160| 410| int32_t high = lenarray - 1; 161| 2.62k| while (low <= high) { ------------------ | Branch (161:12): [True: 2.43k, False: 184] ------------------ 162| 2.43k| int32_t middleIndex = (low + high) >> 1; 163| 2.43k| uint16_t min_value = array[middleIndex].value; 164| 2.43k| uint16_t max_value = 165| 2.43k| array[middleIndex].value + array[middleIndex].length; 166| 2.43k| if (max_value + UINT32_C(1) < key) { // uint32 arithmetic ------------------ | Branch (166:13): [True: 1.05k, False: 1.38k] ------------------ 167| 1.05k| low = middleIndex + 1; 168| 1.38k| } else if (key < min_value) { ------------------ | Branch (168:20): [True: 1.15k, False: 226] ------------------ 169| 1.15k| high = middleIndex - 1; 170| 1.15k| } else { 171| 226| return middleIndex; 172| 226| } 173| 2.43k| } 174| 184| return low; 175| 410|} roaring.c:run_container_add_range_nruns: 610| 411| int32_t nruns_greater) { 611| 411| int32_t nruns_common = run->n_runs - nruns_less - nruns_greater; 612| 411| if (nruns_common == 0) { ------------------ | Branch (612:9): [True: 102, False: 309] ------------------ 613| 102| makeRoomAtIndex(run, (uint16_t)nruns_less); 614| 102| run->runs[nruns_less].value = (uint16_t)min; 615| 102| run->runs[nruns_less].length = (uint16_t)(max - min); 616| 309| } else { 617| 309| uint32_t common_min = run->runs[nruns_less].value; 618| 309| uint32_t common_max = run->runs[nruns_less + nruns_common - 1].value + 619| 309| run->runs[nruns_less + nruns_common - 1].length; 620| 309| uint32_t result_min = (common_min < min) ? common_min : min; ------------------ | Branch (620:31): [True: 64, False: 245] ------------------ 621| 309| uint32_t result_max = (common_max > max) ? common_max : max; ------------------ | Branch (621:31): [True: 35, False: 274] ------------------ 622| | 623| 309| run->runs[nruns_less].value = (uint16_t)result_min; 624| 309| run->runs[nruns_less].length = (uint16_t)(result_max - result_min); 625| | 626| 309| memmove(&(run->runs[nruns_less + 1]), 627| 309| &(run->runs[run->n_runs - nruns_greater]), 628| 309| nruns_greater * sizeof(rle16_t)); 629| 309| run->n_runs = nruns_less + 1 + nruns_greater; 630| 309| } 631| 411|} roaring.c:makeRoomAtIndex: 208| 755|static inline void makeRoomAtIndex(run_container_t *run, uint16_t index) { 209| | /* This function calls realloc + memmove sequentially to move by one index. 210| | * Potentially copying twice the array. 211| | */ 212| 755| if (run->n_runs + 1 > run->capacity) ------------------ | Branch (212:9): [True: 630, False: 125] ------------------ 213| 630| run_container_grow(run, run->n_runs + 1, true); 214| 755| memmove(run->runs + 1 + index, run->runs + index, 215| 755| (run->n_runs - index) * sizeof(rle16_t)); 216| 755| run->n_runs++; 217| 755|} roaring.c:run_container_remove_range: 667| 1.80k| uint32_t min, uint32_t max) { 668| 1.80k| int32_t first = rle16_find_run(run->runs, run->n_runs, (uint16_t)min); 669| 1.80k| int32_t last = rle16_find_run(run->runs, run->n_runs, (uint16_t)max); 670| | 671| 1.80k| if (first >= 0 && min > run->runs[first].value && ------------------ | Branch (671:9): [True: 694, False: 1.11k] | Branch (671:23): [True: 292, False: 402] ------------------ 672| 292| max < ((uint32_t)run->runs[first].value + ------------------ | Branch (672:9): [True: 40, False: 252] ------------------ 673| 292| (uint32_t)run->runs[first].length)) { 674| | // split this run into two adjacent runs 675| | 676| | // right subinterval 677| 40| makeRoomAtIndex(run, (uint16_t)(first + 1)); 678| 40| run->runs[first + 1].value = (uint16_t)(max + 1); 679| 40| run->runs[first + 1].length = 680| 40| (uint16_t)((run->runs[first].value + run->runs[first].length) - 681| 40| (max + 1)); 682| | 683| | // left subinterval 684| 40| run->runs[first].length = 685| 40| (uint16_t)((min - 1) - run->runs[first].value); 686| | 687| 40| return; 688| 40| } 689| | 690| | // update left-most partial run 691| 1.76k| if (first >= 0) { ------------------ | Branch (691:9): [True: 654, False: 1.11k] ------------------ 692| 654| if (min > run->runs[first].value) { ------------------ | Branch (692:13): [True: 252, False: 402] ------------------ 693| 252| run->runs[first].length = 694| 252| (uint16_t)((min - 1) - run->runs[first].value); 695| 252| first++; 696| 252| } 697| 1.11k| } else { 698| 1.11k| first = -first - 1; 699| 1.11k| } 700| | 701| | // update right-most run 702| 1.76k| if (last >= 0) { ------------------ | Branch (702:9): [True: 631, False: 1.13k] ------------------ 703| 631| uint16_t run_max = run->runs[last].value + run->runs[last].length; 704| 631| if (run_max > max) { ------------------ | Branch (704:13): [True: 409, False: 222] ------------------ 705| 409| run->runs[last].value = (uint16_t)(max + 1); 706| 409| run->runs[last].length = (uint16_t)(run_max - (max + 1)); 707| 409| last--; 708| 409| } 709| 1.13k| } else { 710| 1.13k| last = (-last - 1) - 1; 711| 1.13k| } 712| | 713| | // remove intermediate runs 714| 1.76k| if (first <= last) { ------------------ | Branch (714:9): [True: 146, False: 1.62k] ------------------ 715| 146| run_container_shift_tail(run, run->n_runs - (last + 1), 716| 146| -(last - first + 1)); 717| 146| } 718| 1.76k|} roaring.c:rle16_find_run: 133| 3.61k| uint16_t ikey) { 134| 3.61k| int32_t low = 0; 135| 3.61k| int32_t high = lenarray - 1; 136| 14.5k| while (low <= high) { ------------------ | Branch (136:12): [True: 12.3k, False: 2.24k] ------------------ 137| 12.3k| int32_t middleIndex = (low + high) >> 1; 138| 12.3k| uint16_t min = array[middleIndex].value; 139| 12.3k| uint16_t max = array[middleIndex].value + array[middleIndex].length; 140| 12.3k| if (ikey > max) { ------------------ | Branch (140:13): [True: 1.42k, False: 10.9k] ------------------ 141| 1.42k| low = middleIndex + 1; 142| 10.9k| } else if (ikey < min) { ------------------ | Branch (142:20): [True: 9.54k, False: 1.36k] ------------------ 143| 9.54k| high = middleIndex - 1; 144| 9.54k| } else { 145| 1.36k| return middleIndex; 146| 1.36k| } 147| 12.3k| } 148| 2.24k| return -(low + 1); 149| 3.61k|} roaring.c:run_container_shift_tail: 650| 146| int32_t distance) { 651| 146| if (distance > 0) { ------------------ | Branch (651:9): [True: 0, False: 146] ------------------ 652| 0| if (run->capacity < count + distance) { ------------------ | Branch (652:13): [True: 0, False: 0] ------------------ 653| | run_container_grow(run, count + distance, true); 654| 0| } 655| 0| } 656| 146| int32_t srcpos = run->n_runs - count; 657| 146| int32_t dstpos = srcpos + distance; 658| 146| memmove(&(run->runs[dstpos]), &(run->runs[srcpos]), 659| 146| sizeof(rle16_t) * count); 660| 146| run->n_runs += distance; 661| 146|} roaring.c:run_container_remove: 223| 2.95k|static inline bool run_container_remove(run_container_t *run, uint16_t pos) { 224| 2.95k| int32_t index = interleavedBinarySearch(run->runs, run->n_runs, pos); 225| 2.95k| if (index >= 0) { ------------------ | Branch (225:9): [True: 1.16k, False: 1.78k] ------------------ 226| 1.16k| int32_t le = run->runs[index].length; 227| 1.16k| if (le == 0) { ------------------ | Branch (227:13): [True: 263, False: 905] ------------------ 228| 263| recoverRoomAtIndex(run, (uint16_t)index); 229| 905| } else { 230| 905| run->runs[index].value++; 231| 905| run->runs[index].length--; 232| 905| } 233| 1.16k| return true; 234| 1.16k| } 235| 1.78k| index = -index - 2; // points to preceding value, possibly -1 236| 1.78k| if (index >= 0) { // possible match ------------------ | Branch (236:9): [True: 817, False: 968] ------------------ 237| 817| int32_t offset = pos - run->runs[index].value; 238| 817| int32_t le = run->runs[index].length; 239| 817| if (offset < le) { ------------------ | Branch (239:13): [True: 613, False: 204] ------------------ 240| | // need to break in two 241| 613| run->runs[index].length = (uint16_t)(offset - 1); 242| | // need to insert 243| 613| uint16_t newvalue = pos + 1; 244| 613| int32_t newlength = le - offset - 1; 245| 613| makeRoomAtIndex(run, (uint16_t)(index + 1)); 246| 613| run->runs[index + 1].value = newvalue; 247| 613| run->runs[index + 1].length = (uint16_t)newlength; 248| 613| return true; 249| | 250| 613| } else if (offset == le) { ------------------ | Branch (250:20): [True: 24, False: 180] ------------------ 251| 24| run->runs[index].length--; 252| 24| return true; 253| 24| } 254| 817| } 255| | // no match 256| 1.14k| return false; 257| 1.78k|} roaring.c:recoverRoomAtIndex: 102| 263|static inline void recoverRoomAtIndex(run_container_t *run, uint16_t index) { 103| 263| memmove(run->runs + index, run->runs + (1 + index), 104| 263| (run->n_runs - index - 1) * sizeof(rle16_t)); 105| 263| run->n_runs--; 106| 263|} roaring.c:run_container_nonzero_cardinality: 309| 60.9k| const run_container_t *run) { 310| 60.9k| return run->n_runs > 0; // runs never empty 311| 60.9k|} roaring.c:run_container_is_full: 394| 6.20k|static inline bool run_container_is_full(const run_container_t *run) { 395| 6.20k| rle16_t vl = run->runs[0]; 396| 6.20k| return (run->n_runs == 1) && (vl.value == 0) && (vl.length == 0xFFFF); ------------------ | Branch (396:12): [True: 1.47k, False: 4.73k] | Branch (396:34): [True: 1.25k, False: 224] | Branch (396:53): [True: 1.10k, False: 142] ------------------ 397| 6.20k|} roaring.c:run_container_equals: 502| 135| const run_container_t *container2) { 503| 135| if (container1->n_runs != container2->n_runs) { ------------------ | Branch (503:9): [True: 103, False: 32] ------------------ 504| 103| return false; 505| 103| } 506| 32| return memequals(container1->runs, container2->runs, 507| 32| container1->n_runs * sizeof(rle16_t)); 508| 135|} roaring.c:run_container_contains_range: 278| 180| uint32_t pos_end) { 279| 180| uint32_t count = 0; 280| 180| int32_t index = 281| 180| interleavedBinarySearch(run->runs, run->n_runs, (uint16_t)pos_start); 282| 180| if (index < 0) { ------------------ | Branch (282:9): [True: 100, False: 80] ------------------ 283| 100| index = -index - 2; 284| 100| if ((index == -1) || ------------------ | Branch (284:13): [True: 16, False: 84] ------------------ 285| 84| ((pos_start - run->runs[index].value) > run->runs[index].length)) { ------------------ | Branch (285:13): [True: 12, False: 72] ------------------ 286| 28| return false; 287| 28| } 288| 100| } 289| 2.38k| for (int32_t i = index; i < run->n_runs; ++i) { ------------------ | Branch (289:29): [True: 2.31k, False: 71] ------------------ 290| 2.31k| const uint32_t stop = run->runs[i].value + run->runs[i].length; 291| 2.31k| if (run->runs[i].value >= pos_end) break; ------------------ | Branch (291:13): [True: 12, False: 2.29k] ------------------ 292| 2.29k| if (stop >= pos_end) { ------------------ | Branch (292:13): [True: 69, False: 2.22k] ------------------ 293| 69| count += (((pos_end - run->runs[i].value) > 0) ------------------ | Branch (293:23): [True: 69, False: 0] ------------------ 294| 69| ? (pos_end - run->runs[i].value) 295| 69| : 0); 296| 69| break; 297| 69| } 298| 2.22k| const uint32_t min = (stop - pos_start) > 0 ? (stop - pos_start) : 0; ------------------ | Branch (298:30): [True: 2.21k, False: 15] ------------------ 299| 2.22k| count += (min < run->runs[i].length) ? min : run->runs[i].length; ------------------ | Branch (299:18): [True: 57, False: 2.17k] ------------------ 300| 2.22k| } 301| 152| return count >= (pos_end - pos_start - 1); 302| 180|} roaring_array.c:run_container_size_in_bytes: 493| 2.51k| const run_container_t *container) { 494| 2.51k| return run_container_serialized_size_in_bytes(container->n_runs); 495| 2.51k|} roaring_array.c:run_container_serialized_size_in_bytes: 456| 2.51k|static inline int32_t run_container_serialized_size_in_bytes(int32_t num_runs) { 457| 2.51k| return sizeof(uint16_t) + 458| 2.51k| sizeof(rle16_t) * num_runs; // each run requires 2 2-byte entries. 459| 2.51k|} convert.c:run_container_serialized_size_in_bytes: 456| 140k|static inline int32_t run_container_serialized_size_in_bytes(int32_t num_runs) { 457| 140k| return sizeof(uint16_t) + 458| 140k| sizeof(rle16_t) * num_runs; // each run requires 2 2-byte entries. 459| 140k|} mixed_intersection.c:run_container_is_full: 394| 16.3k|static inline bool run_container_is_full(const run_container_t *run) { 395| 16.3k| rle16_t vl = run->runs[0]; 396| 16.3k| return (run->n_runs == 1) && (vl.value == 0) && (vl.length == 0xFFFF); ------------------ | Branch (396:12): [True: 1.46k, False: 14.9k] | Branch (396:34): [True: 700, False: 760] | Branch (396:53): [True: 204, False: 496] ------------------ 397| 16.3k|} mixed_union.c:run_container_is_full: 394| 2.73k|static inline bool run_container_is_full(const run_container_t *run) { 395| 2.73k| rle16_t vl = run->runs[0]; 396| 2.73k| return (run->n_runs == 1) && (vl.value == 0) && (vl.length == 0xFFFF); ------------------ | Branch (396:12): [True: 354, False: 2.38k] | Branch (396:34): [True: 229, False: 125] | Branch (396:53): [True: 106, False: 123] ------------------ 397| 2.73k|} mixed_union.c:run_container_append_first: 351| 2.03k| rle16_t vl) { 352| 2.03k| run->runs[run->n_runs] = vl; 353| 2.03k| run->n_runs++; 354| 2.03k| return vl; 355| 2.03k|} mixed_union.c:run_container_append_value_first: 385| 599| uint16_t val) { 386| 599| rle16_t newrle = CROARING_MAKE_RLE16(val, 0); ------------------ | | 60| 599| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 387| 599| run->runs[run->n_runs] = newrle; 388| 599| run->n_runs++; 389| 599| return newrle; 390| 599|} mixed_union.c:run_container_append: 332| 154k| rle16_t *previousrl) { 333| 154k| const uint32_t previousend = previousrl->value + previousrl->length; 334| 154k| if (vl.value > previousend + 1) { // we add a new one ------------------ | Branch (334:9): [True: 143k, False: 11.3k] ------------------ 335| 143k| run->runs[run->n_runs] = vl; 336| 143k| run->n_runs++; 337| 143k| *previousrl = vl; 338| 143k| } else { 339| 11.3k| uint32_t newend = vl.value + vl.length + UINT32_C(1); 340| 11.3k| if (newend > previousend) { // we merge ------------------ | Branch (340:13): [True: 11.3k, False: 0] ------------------ 341| 11.3k| previousrl->length = (uint16_t)(newend - 1 - previousrl->value); 342| 11.3k| run->runs[run->n_runs - 1] = *previousrl; 343| 11.3k| } 344| 11.3k| } 345| 154k|} mixed_union.c:run_container_append_value: 368| 467k| rle16_t *previousrl) { 369| 467k| const uint32_t previousend = previousrl->value + previousrl->length; 370| 467k| if (val > previousend + 1) { // we add a new one ------------------ | Branch (370:9): [True: 51.9k, False: 415k] ------------------ 371| 51.9k| *previousrl = CROARING_MAKE_RLE16(val, 0); ------------------ | | 60| 51.9k| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 372| 51.9k| run->runs[run->n_runs] = *previousrl; 373| 51.9k| run->n_runs++; 374| 415k| } else if (val == previousend + 1) { // we merge ------------------ | Branch (374:16): [True: 142k, False: 273k] ------------------ 375| 142k| previousrl->length++; 376| 142k| run->runs[run->n_runs - 1] = *previousrl; 377| 142k| } 378| 467k|} run.c:run_container_is_full: 394| 124k|static inline bool run_container_is_full(const run_container_t *run) { 395| 124k| rle16_t vl = run->runs[0]; 396| 124k| return (run->n_runs == 1) && (vl.value == 0) && (vl.length == 0xFFFF); ------------------ | Branch (396:12): [True: 109k, False: 14.3k] | Branch (396:34): [True: 108k, False: 1.79k] | Branch (396:53): [True: 105k, False: 2.19k] ------------------ 397| 124k|} run.c:run_container_serialized_size_in_bytes: 456| 59|static inline int32_t run_container_serialized_size_in_bytes(int32_t num_runs) { 457| 59| return sizeof(uint16_t) + 458| 59| sizeof(rle16_t) * num_runs; // each run requires 2 2-byte entries. 459| 59|} run.c:recoverRoomAtIndex: 102| 6.31k|static inline void recoverRoomAtIndex(run_container_t *run, uint16_t index) { 103| 6.31k| memmove(run->runs + index, run->runs + (1 + index), 104| 6.31k| (run->n_runs - index - 1) * sizeof(rle16_t)); 105| 6.31k| run->n_runs--; 106| 6.31k|} run.c:makeRoomAtIndex: 208| 42.5k|static inline void makeRoomAtIndex(run_container_t *run, uint16_t index) { 209| | /* This function calls realloc + memmove sequentially to move by one index. 210| | * Potentially copying twice the array. 211| | */ 212| 42.5k| if (run->n_runs + 1 > run->capacity) ------------------ | Branch (212:9): [True: 2.41k, False: 40.1k] ------------------ 213| 2.41k| run_container_grow(run, run->n_runs + 1, true); 214| 42.5k| memmove(run->runs + 1 + index, run->runs + index, 215| 42.5k| (run->n_runs - index) * sizeof(rle16_t)); 216| 42.5k| run->n_runs++; 217| 42.5k|} run.c:run_container_append_first: 351| 2.31k| rle16_t vl) { 352| 2.31k| run->runs[run->n_runs] = vl; 353| 2.31k| run->n_runs++; 354| 2.31k| return vl; 355| 2.31k|} run.c:run_container_append: 332| 292k| rle16_t *previousrl) { 333| 292k| const uint32_t previousend = previousrl->value + previousrl->length; 334| 292k| if (vl.value > previousend + 1) { // we add a new one ------------------ | Branch (334:9): [True: 151k, False: 141k] ------------------ 335| 151k| run->runs[run->n_runs] = vl; 336| 151k| run->n_runs++; 337| 151k| *previousrl = vl; 338| 151k| } else { 339| 141k| uint32_t newend = vl.value + vl.length + UINT32_C(1); 340| 141k| if (newend > previousend) { // we merge ------------------ | Branch (340:13): [True: 108k, False: 32.7k] ------------------ 341| 108k| previousrl->length = (uint16_t)(newend - 1 - previousrl->value); 342| 108k| run->runs[run->n_runs - 1] = *previousrl; 343| 108k| } 344| 141k| } 345| 292k|} run.c:run_container_empty: 314| 14|static inline bool run_container_empty(const run_container_t *run) { 315| 14| return run->n_runs == 0; // runs never empty 316| 14|} run.c:run_container_size_in_bytes: 493| 59| const run_container_t *container) { 494| 59| return run_container_serialized_size_in_bytes(container->n_runs); 495| 59|} roaring_trailing_zeroes: 235| 63.9M|inline int roaring_trailing_zeroes(unsigned long long input_num) { 236| 63.9M| return __builtin_ctzll(input_num); 237| 63.9M|} roaring_leading_zeroes: 239| 4.03k|inline int roaring_leading_zeroes(unsigned long long input_num) { 240| 4.03k| return __builtin_clzll(input_num); 241| 4.03k|} roaring.c:roaring_hamming: 280| 196k|static inline int roaring_hamming(uint64_t x) { 281| |#if defined(_WIN64) && defined(CROARING_REGULAR_VISUAL_STUDIO) && \ 282| | CROARING_REGULAR_VISUAL_STUDIO 283| |#ifdef CROARING_USENEON 284| | return vaddv_u8(vcnt_u8(vcreate_u8(input_num))); 285| |#elif defined(_M_ARM64) 286| | return roaring_hamming_backup(x); 287| | // (int) _CountOneBits64(x); is unavailable 288| |#else // _M_ARM64 289| | return (int)__popcnt64(x); 290| |#endif // _M_ARM64 291| |#elif defined(_WIN32) && defined(CROARING_REGULAR_VISUAL_STUDIO) && \ 292| | CROARING_REGULAR_VISUAL_STUDIO 293| |#ifdef _M_ARM 294| | return roaring_hamming_backup(x); 295| | // _CountOneBits is unavailable 296| |#else // _M_ARM 297| | return (int)__popcnt((unsigned int)x) + 298| | (int)__popcnt((unsigned int)(x >> 32)); 299| |#endif // _M_ARM 300| |#else 301| 196k| return __builtin_popcountll(x); 302| 196k|#endif 303| 196k|} mixed_intersection.c:roaring_hamming: 280| 210k|static inline int roaring_hamming(uint64_t x) { 281| |#if defined(_WIN64) && defined(CROARING_REGULAR_VISUAL_STUDIO) && \ 282| | CROARING_REGULAR_VISUAL_STUDIO 283| |#ifdef CROARING_USENEON 284| | return vaddv_u8(vcnt_u8(vcreate_u8(input_num))); 285| |#elif defined(_M_ARM64) 286| | return roaring_hamming_backup(x); 287| | // (int) _CountOneBits64(x); is unavailable 288| |#else // _M_ARM64 289| | return (int)__popcnt64(x); 290| |#endif // _M_ARM64 291| |#elif defined(_WIN32) && defined(CROARING_REGULAR_VISUAL_STUDIO) && \ 292| | CROARING_REGULAR_VISUAL_STUDIO 293| |#ifdef _M_ARM 294| | return roaring_hamming_backup(x); 295| | // _CountOneBits is unavailable 296| |#else // _M_ARM 297| | return (int)__popcnt((unsigned int)x) + 298| | (int)__popcnt((unsigned int)(x >> 32)); 299| |#endif // _M_ARM 300| |#else 301| 210k| return __builtin_popcountll(x); 302| 210k|#endif 303| 210k|} roaring_bitmap_init_cleared: 72| 30.7k|inline void roaring_bitmap_init_cleared(roaring_bitmap_t *r) { 73| 30.7k| roaring_bitmap_init_with_capacity(r, 0); 74| 30.7k|} roaring_bitmap_add_range: 434| 6.15k| uint64_t max) { 435| 6.15k| if (max <= min || min > (uint64_t)UINT32_MAX + 1) { ------------------ | Branch (435:9): [True: 3.49k, False: 2.66k] | Branch (435:23): [True: 0, False: 2.66k] ------------------ 436| 3.49k| return; 437| 3.49k| } 438| 2.66k| roaring_bitmap_add_range_closed(r, (uint32_t)min, (uint32_t)(max - 1)); 439| 2.66k|} roaring_bitmap_remove_range: 456| 6.15k| uint64_t max) { 457| 6.15k| if (max <= min || min > (uint64_t)UINT32_MAX + 1) { ------------------ | Branch (457:9): [True: 5.18k, False: 970] | Branch (457:23): [True: 0, False: 970] ------------------ 458| 5.18k| return; 459| 5.18k| } 460| 970| roaring_bitmap_remove_range_closed(r, (uint32_t)min, (uint32_t)(max - 1)); 461| 970|} roaring_bitmap_contains: 478| 6.22k|inline bool roaring_bitmap_contains(const roaring_bitmap_t *r, uint32_t val) { 479| | // For performance reasons, this function is inline and uses internal 480| | // functions directly. 481| 6.22k|#ifdef __cplusplus 482| 6.22k| using namespace ::roaring::internal; 483| 6.22k|#endif 484| 6.22k| const uint16_t hb = val >> 16; 485| | /* 486| | * the next function call involves a binary search and lots of branching. 487| | */ 488| 6.22k| int32_t i = ra_get_index(&r->high_low_container, hb); 489| 6.22k| if (i < 0) return false; ------------------ | Branch (489:9): [True: 651, False: 5.57k] ------------------ 490| | 491| 5.57k| uint8_t typecode; 492| | // next call ought to be cheap 493| 5.57k| container_t *container = ra_get_container_at_index(&r->high_low_container, 494| 5.57k| (uint16_t)i, &typecode); 495| | // rest might be a tad expensive, possibly involving another round of binary 496| | // search 497| 5.57k| return container_contains(container, val & 0xFFFF, typecode); 498| 6.22k|} roaring_bitmap_set_copy_on_write: 117| 56.8k|inline void roaring_bitmap_set_copy_on_write(roaring_bitmap_t *r, bool cow) { 118| 56.8k| if (cow) { ------------------ | Branch (118:9): [True: 0, False: 56.8k] ------------------ 119| 0| r->high_low_container.flags |= ROARING_FLAG_COW; ------------------ | | 46| 0|#define ROARING_FLAG_COW UINT8_C(0x1) ------------------ 120| 56.8k| } else { 121| 56.8k| if (roaring_bitmap_get_copy_on_write(r)) { ------------------ | Branch (121:13): [True: 0, False: 56.8k] ------------------ 122| 0| roaring_unshare_all(r); 123| 0| } 124| | r->high_low_container.flags &= ~ROARING_FLAG_COW; ------------------ | | 46| 56.8k|#define ROARING_FLAG_COW UINT8_C(0x1) ------------------ 125| 56.8k| } 126| 56.8k|} roaring_bitmap_get_copy_on_write: 114| 69.2k|inline bool roaring_bitmap_get_copy_on_write(const roaring_bitmap_t *r) { 115| | return r->high_low_container.flags & ROARING_FLAG_COW; ------------------ | | 46| 69.2k|#define ROARING_FLAG_COW UINT8_C(0x1) ------------------ 116| 69.2k|} ra_get_index: 93| 84.6k|inline int32_t ra_get_index(const roaring_array_t *ra, uint16_t x) { 94| 84.6k| if ((ra->size == 0) || ra->keys[ra->size - 1] == x) return ra->size - 1; ------------------ | Branch (94:9): [True: 12.4k, False: 72.1k] | Branch (94:28): [True: 31.3k, False: 40.8k] ------------------ 95| 40.8k| return binarySearch(ra->keys, (int32_t)ra->size, x); 96| 84.6k|} ra_get_container_at_index: 102| 540k| uint16_t i, uint8_t *typecode) { 103| 540k| *typecode = ra->typecodes[i]; 104| 540k| return ra->containers[i]; 105| 540k|} ra_get_key_at_index: 110| 404k|inline uint16_t ra_get_key_at_index(const roaring_array_t *ra, uint16_t i) { 111| 404k| return ra->keys[i]; 112| 404k|} ra_set_container_at_index: 175| 106k| container_t *c, uint8_t typecode) { 176| | assert(i < ra->size); 177| 106k| ra->containers[i] = c; 178| 106k| ra->typecodes[i] = typecode; 179| 106k|} ra_get_size: 191| 13.0k|inline int32_t ra_get_size(const roaring_array_t *ra) { return ra->size; } ra_replace_key_and_container_at_index: 205| 145k| uint8_t typecode) { 206| 145k| assert(i < ra->size); 207| | 208| 145k| ra->keys[i] = key; 209| 145k| ra->containers[i] = c; 210| 145k| ra->typecodes[i] = typecode; 211| 145k|} roaring.c:ra_advance_until: 194| 518| int32_t pos) { 195| 518| return advanceUntil(ra->keys, pos, ra->size, x); 196| 518|} roaring.c:ra_unshare_container_at_index: 268| 62.8k| uint16_t i) { 269| | assert(i < ra->size); 270| 62.8k| ra->containers[i] = 271| 62.8k| get_writable_copy_if_shared(ra->containers[i], &ra->typecodes[i]); 272| 62.8k|} intersect_vector16: 386| 1.74k| size_t s_b, uint16_t *C) { 387| 1.74k| size_t count = 0; 388| 1.74k| size_t i_a = 0, i_b = 0; 389| 1.74k| const int vectorlength = sizeof(__m128i) / sizeof(uint16_t); 390| 1.74k| const size_t st_a = (s_a / vectorlength) * vectorlength; 391| 1.74k| const size_t st_b = (s_b / vectorlength) * vectorlength; 392| 1.74k| __m128i v_a, v_b; 393| 1.74k| if ((i_a < st_a) && (i_b < st_b)) { ------------------ | Branch (393:9): [True: 1.61k, False: 127] | Branch (393:25): [True: 1.47k, False: 146] ------------------ 394| 1.47k| v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); 395| 1.47k| v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); 396| 3.94k| while ((A[i_a] == 0) || (B[i_b] == 0)) { ------------------ | Branch (396:16): [True: 2.42k, False: 1.51k] | Branch (396:33): [True: 184, False: 1.33k] ------------------ 397| 2.60k| const __m128i res_v = _mm_cmpestrm( 398| 2.60k| v_b, vectorlength, v_a, vectorlength, 399| 2.60k| _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); 400| 2.60k| const int r = _mm_extract_epi32(res_v, 0); 401| 2.60k| __m128i sm16 = _mm_loadu_si128((const __m128i *)shuffle_mask16 + r); 402| 2.60k| __m128i p = _mm_shuffle_epi8(v_a, sm16); 403| 2.60k| _mm_storeu_si128((__m128i *)&C[count], p); // can overflow 404| 2.60k| count += _mm_popcnt_u32(r); 405| 2.60k| const uint16_t a_max = A[i_a + vectorlength - 1]; 406| 2.60k| const uint16_t b_max = B[i_b + vectorlength - 1]; 407| 2.60k| if (a_max <= b_max) { ------------------ | Branch (407:17): [True: 1.50k, False: 1.10k] ------------------ 408| 1.50k| i_a += vectorlength; 409| 1.50k| if (i_a == st_a) break; ------------------ | Branch (409:21): [True: 101, False: 1.40k] ------------------ 410| 1.40k| v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); 411| 1.40k| } 412| 2.50k| if (b_max <= a_max) { ------------------ | Branch (412:17): [True: 1.21k, False: 1.29k] ------------------ 413| 1.21k| i_b += vectorlength; 414| 1.21k| if (i_b == st_b) break; ------------------ | Branch (414:21): [True: 36, False: 1.17k] ------------------ 415| 1.17k| v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); 416| 1.17k| } 417| 2.50k| } 418| 1.47k| if ((i_a < st_a) && (i_b < st_b)) ------------------ | Branch (418:13): [True: 1.37k, False: 101] | Branch (418:29): [True: 1.33k, False: 36] ------------------ 419| 42.7k| while (true) { ------------------ | Branch (419:20): [True: 42.7k, Folded] ------------------ 420| 42.7k| const __m128i res_v = _mm_cmpistrm( 421| 42.7k| v_b, v_a, 422| 42.7k| _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); 423| 42.7k| const int r = _mm_extract_epi32(res_v, 0); 424| 42.7k| __m128i sm16 = 425| 42.7k| _mm_loadu_si128((const __m128i *)shuffle_mask16 + r); 426| 42.7k| __m128i p = _mm_shuffle_epi8(v_a, sm16); 427| 42.7k| _mm_storeu_si128((__m128i *)&C[count], p); // can overflow 428| 42.7k| count += _mm_popcnt_u32(r); 429| 42.7k| const uint16_t a_max = A[i_a + vectorlength - 1]; 430| 42.7k| const uint16_t b_max = B[i_b + vectorlength - 1]; 431| 42.7k| if (a_max <= b_max) { ------------------ | Branch (431:21): [True: 18.8k, False: 23.8k] ------------------ 432| 18.8k| i_a += vectorlength; 433| 18.8k| if (i_a == st_a) break; ------------------ | Branch (433:25): [True: 756, False: 18.0k] ------------------ 434| 18.0k| v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); 435| 18.0k| } 436| 41.9k| if (b_max <= a_max) { ------------------ | Branch (436:21): [True: 25.2k, False: 16.6k] ------------------ 437| 25.2k| i_b += vectorlength; 438| 25.2k| if (i_b == st_b) break; ------------------ | Branch (438:25): [True: 579, False: 24.7k] ------------------ 439| 24.7k| v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); 440| 24.7k| } 441| 41.9k| } 442| 1.47k| } 443| | // intersect the tail using scalar intersection 444| 27.3k| while (i_a < s_a && i_b < s_b) { ------------------ | Branch (444:12): [True: 25.9k, False: 1.37k] | Branch (444:25): [True: 25.6k, False: 369] ------------------ 445| 25.6k| uint16_t a = A[i_a]; 446| 25.6k| uint16_t b = B[i_b]; 447| 25.6k| if (a < b) { ------------------ | Branch (447:13): [True: 10.0k, False: 15.5k] ------------------ 448| 10.0k| i_a++; 449| 15.5k| } else if (b < a) { ------------------ | Branch (449:20): [True: 10.0k, False: 5.50k] ------------------ 450| 10.0k| i_b++; 451| 10.0k| } else { 452| 5.50k| C[count] = a; //==b; 453| 5.50k| count++; 454| 5.50k| i_a++; 455| 5.50k| i_b++; 456| 5.50k| } 457| 25.6k| } 458| 1.74k| return (int32_t)count; 459| 1.74k|} array_container_to_uint32_array_vector16: 464| 5.70k| uint32_t base) { 465| 5.70k| int outpos = 0; 466| 5.70k| uint32_t *out = (uint32_t *)vout; 467| 5.70k| size_t i = 0; 468| 181k| for (; i + sizeof(__m128i) / sizeof(uint16_t) <= cardinality; ------------------ | Branch (468:12): [True: 176k, False: 5.70k] ------------------ 469| 176k| i += sizeof(__m128i) / sizeof(uint16_t)) { 470| 176k| __m128i vinput = _mm_loadu_si128((const __m128i *)(array + i)); 471| 176k| __m256i voutput = _mm256_add_epi32(_mm256_cvtepu16_epi32(vinput), 472| 176k| _mm256_set1_epi32(base)); 473| 176k| _mm256_storeu_si256((__m256i *)(out + outpos), voutput); 474| 176k| outpos += sizeof(__m256i) / sizeof(uint32_t); 475| 176k| } 476| 19.6k| for (; i < cardinality; ++i) { ------------------ | Branch (476:12): [True: 13.9k, False: 5.70k] ------------------ 477| 13.9k| const uint32_t val = base + array[i]; 478| 13.9k| memcpy(out + outpos, &val, 479| 13.9k| sizeof(uint32_t)); // should be compiled as a MOV on x64 480| 13.9k| outpos++; 481| 13.9k| } 482| 5.70k| return outpos; 483| 5.70k|} intersect_vector16_inplace: 486| 4.43k| size_t s_b) { 487| 4.43k| size_t count = 0; 488| 4.43k| size_t i_a = 0, i_b = 0; 489| 4.43k| const int vectorlength = sizeof(__m128i) / sizeof(uint16_t); 490| 4.43k| const size_t st_a = (s_a / vectorlength) * vectorlength; 491| 4.43k| const size_t st_b = (s_b / vectorlength) * vectorlength; 492| 4.43k| __m128i v_a, v_b; 493| 4.43k| if ((i_a < st_a) && (i_b < st_b)) { ------------------ | Branch (493:9): [True: 2.21k, False: 2.22k] | Branch (493:25): [True: 2.03k, False: 180] ------------------ 494| 2.03k| v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); 495| 2.03k| v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); 496| 2.03k| __m128i tmp[2] = {_mm_setzero_si128()}; 497| 2.03k| size_t tmp_count = 0; 498| 4.03k| while ((A[i_a] == 0) || (B[i_b] == 0)) { ------------------ | Branch (498:16): [True: 1.79k, False: 2.23k] | Branch (498:33): [True: 291, False: 1.94k] ------------------ 499| 2.08k| const __m128i res_v = _mm_cmpestrm( 500| 2.08k| v_b, vectorlength, v_a, vectorlength, 501| 2.08k| _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); 502| 2.08k| const int r = _mm_extract_epi32(res_v, 0); 503| 2.08k| __m128i sm16 = _mm_loadu_si128((const __m128i *)shuffle_mask16 + r); 504| 2.08k| __m128i p = _mm_shuffle_epi8(v_a, sm16); 505| 2.08k| _mm_storeu_si128((__m128i *)&((uint16_t *)tmp)[tmp_count], p); 506| 2.08k| tmp_count += _mm_popcnt_u32(r); 507| 2.08k| const uint16_t a_max = A[i_a + vectorlength - 1]; 508| 2.08k| const uint16_t b_max = B[i_b + vectorlength - 1]; 509| 2.08k| if (a_max <= b_max) { ------------------ | Branch (509:17): [True: 1.94k, False: 145] ------------------ 510| 1.94k| _mm_storeu_si128((__m128i *)&A[count], tmp[0]); 511| 1.94k| _mm_storeu_si128(tmp, _mm_setzero_si128()); 512| 1.94k| count += tmp_count; 513| 1.94k| tmp_count = 0; 514| 1.94k| i_a += vectorlength; 515| 1.94k| if (i_a == st_a) break; ------------------ | Branch (515:21): [True: 72, False: 1.87k] ------------------ 516| 1.87k| v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); 517| 1.87k| } 518| 2.01k| if (b_max <= a_max) { ------------------ | Branch (518:17): [True: 332, False: 1.68k] ------------------ 519| 332| i_b += vectorlength; 520| 332| if (i_b == st_b) break; ------------------ | Branch (520:21): [True: 13, False: 319] ------------------ 521| 319| v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); 522| 319| } 523| 2.01k| } 524| 2.03k| if ((i_a < st_a) && (i_b < st_b)) { ------------------ | Branch (524:13): [True: 1.96k, False: 72] | Branch (524:29): [True: 1.94k, False: 13] ------------------ 525| 220k| while (true) { ------------------ | Branch (525:20): [True: 220k, Folded] ------------------ 526| 220k| const __m128i res_v = _mm_cmpistrm( 527| 220k| v_b, v_a, 528| 220k| _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); 529| 220k| const int r = _mm_extract_epi32(res_v, 0); 530| 220k| __m128i sm16 = 531| 220k| _mm_loadu_si128((const __m128i *)shuffle_mask16 + r); 532| 220k| __m128i p = _mm_shuffle_epi8(v_a, sm16); 533| 220k| _mm_storeu_si128((__m128i *)&((uint16_t *)tmp)[tmp_count], p); 534| 220k| tmp_count += _mm_popcnt_u32(r); 535| 220k| const uint16_t a_max = A[i_a + vectorlength - 1]; 536| 220k| const uint16_t b_max = B[i_b + vectorlength - 1]; 537| 220k| if (a_max <= b_max) { ------------------ | Branch (537:21): [True: 130k, False: 90.1k] ------------------ 538| 130k| _mm_storeu_si128((__m128i *)&A[count], tmp[0]); 539| 130k| _mm_storeu_si128(tmp, _mm_setzero_si128()); 540| 130k| count += tmp_count; 541| 130k| tmp_count = 0; 542| 130k| i_a += vectorlength; 543| 130k| if (i_a == st_a) break; ------------------ | Branch (543:25): [True: 1.07k, False: 129k] ------------------ 544| 129k| v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); 545| 129k| } 546| 219k| if (b_max <= a_max) { ------------------ | Branch (546:21): [True: 124k, False: 95.0k] ------------------ 547| 124k| i_b += vectorlength; 548| 124k| if (i_b == st_b) break; ------------------ | Branch (548:25): [True: 874, False: 123k] ------------------ 549| 123k| v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); 550| 123k| } 551| 219k| } 552| 1.94k| } 553| | // tmp_count <= 8, so this does not affect efficiency so much 554| 4.44k| for (size_t i = 0; i < tmp_count; i++) { ------------------ | Branch (554:28): [True: 2.41k, False: 2.03k] ------------------ 555| 2.41k| A[count] = ((uint16_t *)tmp)[i]; 556| 2.41k| count++; 557| 2.41k| } 558| 2.03k| i_a += tmp_count; // We can at least jump pass $tmp_count elements in A 559| 2.03k| } 560| | // intersect the tail using scalar intersection 561| 31.2k| while (i_a < s_a && i_b < s_b) { ------------------ | Branch (561:12): [True: 27.2k, False: 4.00k] | Branch (561:25): [True: 26.8k, False: 426] ------------------ 562| 26.8k| uint16_t a = A[i_a]; 563| 26.8k| uint16_t b = B[i_b]; 564| 26.8k| if (a < b) { ------------------ | Branch (564:13): [True: 8.54k, False: 18.2k] ------------------ 565| 8.54k| i_a++; 566| 18.2k| } else if (b < a) { ------------------ | Branch (566:20): [True: 6.77k, False: 11.5k] ------------------ 567| 6.77k| i_b++; 568| 11.5k| } else { 569| 11.5k| A[count] = a; //==b; 570| 11.5k| count++; 571| 11.5k| i_a++; 572| 11.5k| i_b++; 573| 11.5k| } 574| 26.8k| } 575| 4.43k| return (int32_t)count; 576| 4.43k|} intersect_vector16_cardinality: 581| 6.98k| const uint16_t *B, size_t s_b) { 582| 6.98k| size_t count = 0; 583| 6.98k| size_t i_a = 0, i_b = 0; 584| 6.98k| const int vectorlength = sizeof(__m128i) / sizeof(uint16_t); 585| 6.98k| const size_t st_a = (s_a / vectorlength) * vectorlength; 586| 6.98k| const size_t st_b = (s_b / vectorlength) * vectorlength; 587| 6.98k| __m128i v_a, v_b; 588| 6.98k| if ((i_a < st_a) && (i_b < st_b)) { ------------------ | Branch (588:9): [True: 6.47k, False: 508] | Branch (588:25): [True: 5.88k, False: 584] ------------------ 589| 5.88k| v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); 590| 5.88k| v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); 591| 15.7k| while ((A[i_a] == 0) || (B[i_b] == 0)) { ------------------ | Branch (591:16): [True: 9.69k, False: 6.07k] | Branch (591:33): [True: 736, False: 5.34k] ------------------ 592| 10.4k| const __m128i res_v = _mm_cmpestrm( 593| 10.4k| v_b, vectorlength, v_a, vectorlength, 594| 10.4k| _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); 595| 10.4k| const int r = _mm_extract_epi32(res_v, 0); 596| 10.4k| count += _mm_popcnt_u32(r); 597| 10.4k| const uint16_t a_max = A[i_a + vectorlength - 1]; 598| 10.4k| const uint16_t b_max = B[i_b + vectorlength - 1]; 599| 10.4k| if (a_max <= b_max) { ------------------ | Branch (599:17): [True: 6.01k, False: 4.41k] ------------------ 600| 6.01k| i_a += vectorlength; 601| 6.01k| if (i_a == st_a) break; ------------------ | Branch (601:21): [True: 404, False: 5.61k] ------------------ 602| 5.61k| v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); 603| 5.61k| } 604| 10.0k| if (b_max <= a_max) { ------------------ | Branch (604:17): [True: 4.86k, False: 5.16k] ------------------ 605| 4.86k| i_b += vectorlength; 606| 4.86k| if (i_b == st_b) break; ------------------ | Branch (606:21): [True: 144, False: 4.71k] ------------------ 607| 4.71k| v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); 608| 4.71k| } 609| 10.0k| } 610| 5.88k| if ((i_a < st_a) && (i_b < st_b)) ------------------ | Branch (610:13): [True: 5.48k, False: 404] | Branch (610:29): [True: 5.34k, False: 144] ------------------ 611| 170k| while (true) { ------------------ | Branch (611:20): [True: 170k, Folded] ------------------ 612| 170k| const __m128i res_v = _mm_cmpistrm( 613| 170k| v_b, v_a, 614| 170k| _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); 615| 170k| const int r = _mm_extract_epi32(res_v, 0); 616| 170k| count += _mm_popcnt_u32(r); 617| 170k| const uint16_t a_max = A[i_a + vectorlength - 1]; 618| 170k| const uint16_t b_max = B[i_b + vectorlength - 1]; 619| 170k| if (a_max <= b_max) { ------------------ | Branch (619:21): [True: 75.3k, False: 95.4k] ------------------ 620| 75.3k| i_a += vectorlength; 621| 75.3k| if (i_a == st_a) break; ------------------ | Branch (621:25): [True: 3.02k, False: 72.3k] ------------------ 622| 72.3k| v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); 623| 72.3k| } 624| 167k| if (b_max <= a_max) { ------------------ | Branch (624:21): [True: 101k, False: 66.5k] ------------------ 625| 101k| i_b += vectorlength; 626| 101k| if (i_b == st_b) break; ------------------ | Branch (626:25): [True: 2.31k, False: 98.8k] ------------------ 627| 98.8k| v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); 628| 98.8k| } 629| 167k| } 630| 5.88k| } 631| | // intersect the tail using scalar intersection 632| 109k| while (i_a < s_a && i_b < s_b) { ------------------ | Branch (632:12): [True: 103k, False: 5.50k] | Branch (632:25): [True: 102k, False: 1.47k] ------------------ 633| 102k| uint16_t a = A[i_a]; 634| 102k| uint16_t b = B[i_b]; 635| 102k| if (a < b) { ------------------ | Branch (635:13): [True: 40.3k, False: 62.1k] ------------------ 636| 40.3k| i_a++; 637| 62.1k| } else if (b < a) { ------------------ | Branch (637:20): [True: 40.1k, False: 22.0k] ------------------ 638| 40.1k| i_b++; 639| 40.1k| } else { 640| 22.0k| count++; 641| 22.0k| i_a++; 642| 22.0k| i_b++; 643| 22.0k| } 644| 102k| } 645| 6.98k| return (int32_t)count; 646| 6.98k|} difference_vector16: 655| 2.27k| size_t s_b, uint16_t *C) { 656| | // we handle the degenerate case 657| 2.27k| if (s_a == 0) return 0; ------------------ | Branch (657:9): [True: 0, False: 2.27k] ------------------ 658| 2.27k| if (s_b == 0) { ------------------ | Branch (658:9): [True: 0, False: 2.27k] ------------------ 659| 0| if (A != C) memcpy(C, A, sizeof(uint16_t) * s_a); ------------------ | Branch (659:13): [True: 0, False: 0] ------------------ 660| 0| return (int32_t)s_a; 661| 0| } 662| | // handle the leading zeroes, it is messy but it allows us to use the fast 663| | // _mm_cmpistrm instrinsic safely 664| 2.27k| int32_t count = 0; 665| 2.27k| if ((A[0] == 0) || (B[0] == 0)) { ------------------ | Branch (665:9): [True: 2.08k, False: 187] | Branch (665:24): [True: 3, False: 184] ------------------ 666| 2.08k| if ((A[0] == 0) && (B[0] == 0)) { ------------------ | Branch (666:13): [True: 2.08k, False: 3] | Branch (666:28): [True: 318, False: 1.76k] ------------------ 667| 318| A++; 668| 318| s_a--; 669| 318| B++; 670| 318| s_b--; 671| 1.76k| } else if (A[0] == 0) { ------------------ | Branch (671:20): [True: 1.76k, False: 3] ------------------ 672| 1.76k| C[count++] = 0; 673| 1.76k| A++; 674| 1.76k| s_a--; 675| 1.76k| } else { 676| 3| B++; 677| 3| s_b--; 678| 3| } 679| 2.08k| } 680| | // at this point, we have two non-empty arrays, made of non-zero 681| | // increasing values. 682| 2.27k| size_t i_a = 0, i_b = 0; 683| 2.27k| const size_t vectorlength = sizeof(__m128i) / sizeof(uint16_t); 684| 2.27k| const size_t st_a = (s_a / vectorlength) * vectorlength; 685| 2.27k| const size_t st_b = (s_b / vectorlength) * vectorlength; 686| 2.27k| if ((i_a < st_a) && (i_b < st_b)) { // this is the vectorized code path ------------------ | Branch (686:9): [True: 1.87k, False: 393] | Branch (686:25): [True: 1.59k, False: 288] ------------------ 687| 1.59k| __m128i v_a, v_b; //, v_bmax; 688| | // we load a vector from A and a vector from B 689| 1.59k| v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); 690| 1.59k| v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); 691| | // we have a runningmask which indicates which values from A have been 692| | // spotted in B, these don't get written out. 693| 1.59k| __m128i runningmask_a_found_in_b = _mm_setzero_si128(); 694| | /**** 695| | * start of the main vectorized loop 696| | *****/ 697| 45.9k| while (true) { ------------------ | Branch (697:16): [True: 45.9k, Folded] ------------------ 698| | // afoundinb will contain a mask indicate for each entry in A 699| | // whether it is seen 700| | // in B 701| 45.9k| const __m128i a_found_in_b = _mm_cmpistrm( 702| 45.9k| v_b, v_a, 703| 45.9k| _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); 704| 45.9k| runningmask_a_found_in_b = 705| 45.9k| _mm_or_si128(runningmask_a_found_in_b, a_found_in_b); 706| | // we always compare the last values of A and B 707| 45.9k| const uint16_t a_max = A[i_a + vectorlength - 1]; 708| 45.9k| const uint16_t b_max = B[i_b + vectorlength - 1]; 709| 45.9k| if (a_max <= b_max) { ------------------ | Branch (709:17): [True: 20.6k, False: 25.3k] ------------------ 710| | // Ok. In this code path, we are ready to write our v_a 711| | // because there is no need to read more from B, they will 712| | // all be large values. 713| 20.6k| const int bitmask_belongs_to_difference = 714| 20.6k| _mm_extract_epi32(runningmask_a_found_in_b, 0) ^ 0xFF; 715| | /*** next few lines are probably expensive *****/ 716| 20.6k| __m128i sm16 = _mm_loadu_si128((const __m128i *)shuffle_mask16 + 717| 20.6k| bitmask_belongs_to_difference); 718| 20.6k| __m128i p = _mm_shuffle_epi8(v_a, sm16); 719| 20.6k| _mm_storeu_si128((__m128i *)&C[count], p); // can overflow 720| 20.6k| count += _mm_popcnt_u32(bitmask_belongs_to_difference); 721| | // we advance a 722| 20.6k| i_a += vectorlength; 723| 20.6k| if (i_a == st_a) // no more ------------------ | Branch (723:21): [True: 1.10k, False: 19.5k] ------------------ 724| 1.10k| break; 725| 19.5k| runningmask_a_found_in_b = _mm_setzero_si128(); 726| 19.5k| v_a = _mm_lddqu_si128((__m128i *)&A[i_a]); 727| 19.5k| } 728| 44.8k| if (b_max <= a_max) { ------------------ | Branch (728:17): [True: 28.3k, False: 16.4k] ------------------ 729| | // in this code path, the current v_b has become useless 730| 28.3k| i_b += vectorlength; 731| 28.3k| if (i_b == st_b) break; ------------------ | Branch (731:21): [True: 481, False: 27.9k] ------------------ 732| 27.9k| v_b = _mm_lddqu_si128((__m128i *)&B[i_b]); 733| 27.9k| } 734| 44.8k| } 735| | // at this point, either we have i_a == st_a, which is the end of the 736| | // vectorized processing, 737| | // or we have i_b == st_b, and we are not done processing the vector... 738| | // so we need to finish it off. 739| 1.59k| if (i_a < st_a) { // we have unfinished business... ------------------ | Branch (739:13): [True: 481, False: 1.10k] ------------------ 740| 481| uint16_t buffer[8]; // buffer to do a masked load 741| 481| memset(buffer, 0, 8 * sizeof(uint16_t)); 742| 481| memcpy(buffer, B + i_b, (s_b - i_b) * sizeof(uint16_t)); 743| 481| v_b = _mm_lddqu_si128((__m128i *)buffer); 744| 481| const __m128i a_found_in_b = _mm_cmpistrm( 745| 481| v_b, v_a, 746| 481| _SIDD_UWORD_OPS | _SIDD_CMP_EQUAL_ANY | _SIDD_BIT_MASK); 747| 481| runningmask_a_found_in_b = 748| 481| _mm_or_si128(runningmask_a_found_in_b, a_found_in_b); 749| 481| const int bitmask_belongs_to_difference = 750| 481| _mm_extract_epi32(runningmask_a_found_in_b, 0) ^ 0xFF; 751| 481| __m128i sm16 = _mm_loadu_si128((const __m128i *)shuffle_mask16 + 752| 481| bitmask_belongs_to_difference); 753| 481| __m128i p = _mm_shuffle_epi8(v_a, sm16); 754| 481| _mm_storeu_si128((__m128i *)&C[count], p); // can overflow 755| 481| count += _mm_popcnt_u32(bitmask_belongs_to_difference); 756| 481| i_a += vectorlength; 757| 481| } 758| | // at this point we should have i_a == st_a and i_b == st_b 759| 1.59k| } 760| | // do the tail using scalar code 761| 98.1k| while (i_a < s_a && i_b < s_b) { ------------------ | Branch (761:12): [True: 96.3k, False: 1.73k] | Branch (761:25): [True: 95.8k, False: 534] ------------------ 762| 95.8k| uint16_t a = A[i_a]; 763| 95.8k| uint16_t b = B[i_b]; 764| 95.8k| if (b < a) { ------------------ | Branch (764:13): [True: 67.5k, False: 28.3k] ------------------ 765| 67.5k| i_b++; 766| 67.5k| } else if (a < b) { ------------------ | Branch (766:20): [True: 23.0k, False: 5.30k] ------------------ 767| 23.0k| C[count] = a; 768| 23.0k| count++; 769| 23.0k| i_a++; 770| 23.0k| } else { //== 771| 5.30k| i_a++; 772| 5.30k| i_b++; 773| 5.30k| } 774| 95.8k| } 775| 2.27k| if (i_a < s_a) { ------------------ | Branch (775:9): [True: 534, False: 1.73k] ------------------ 776| 534| if (C == A) { ------------------ | Branch (776:13): [True: 0, False: 534] ------------------ 777| 0| assert((size_t)count <= i_a); 778| 0| if ((size_t)count < i_a) { ------------------ | Branch (778:17): [True: 0, False: 0] ------------------ 779| 0| memmove(C + count, A + i_a, sizeof(uint16_t) * (s_a - i_a)); 780| 0| } 781| 534| } else { 782| 26.7k| for (size_t i = 0; i < (s_a - i_a); i++) { ------------------ | Branch (782:32): [True: 26.1k, False: 534] ------------------ 783| 26.1k| C[count + i] = A[i + i_a]; 784| 26.1k| } 785| 534| } 786| 534| count += (int32_t)(s_a - i_a); 787| 534| } 788| 2.27k| return count; 789| 2.27k|} intersect_skewed_uint16: 855| 735| uint16_t *buffer) { 856| 735| size_t pos = 0, idx_l = 0, idx_s = 0; 857| | 858| 735| if (0 == size_s) { ------------------ | Branch (858:9): [True: 0, False: 735] ------------------ 859| 0| return 0; 860| 0| } 861| 735| int32_t index1 = 0, index2 = 0, index3 = 0, index4 = 0; 862| 1.94k| while ((idx_s + 4 <= size_s) && (idx_l < size_l)) { ------------------ | Branch (862:12): [True: 1.21k, False: 734] | Branch (862:37): [True: 1.21k, False: 1] ------------------ 863| 1.21k| uint16_t target1 = small[idx_s]; 864| 1.21k| uint16_t target2 = small[idx_s + 1]; 865| 1.21k| uint16_t target3 = small[idx_s + 2]; 866| 1.21k| uint16_t target4 = small[idx_s + 3]; 867| 1.21k| binarySearch4(large + idx_l, (int32_t)(size_l - idx_l), target1, 868| 1.21k| target2, target3, target4, &index1, &index2, &index3, 869| 1.21k| &index4); 870| 1.21k| if ((index1 + idx_l < size_l) && (large[idx_l + index1] == target1)) { ------------------ | Branch (870:13): [True: 1.20k, False: 2] | Branch (870:42): [True: 582, False: 627] ------------------ 871| 582| buffer[pos++] = target1; 872| 582| } 873| 1.21k| if ((index2 + idx_l < size_l) && (large[idx_l + index2] == target2)) { ------------------ | Branch (873:13): [True: 1.20k, False: 4] | Branch (873:42): [True: 703, False: 504] ------------------ 874| 703| buffer[pos++] = target2; 875| 703| } 876| 1.21k| if ((index3 + idx_l < size_l) && (large[idx_l + index3] == target3)) { ------------------ | Branch (876:13): [True: 1.20k, False: 5] | Branch (876:42): [True: 676, False: 530] ------------------ 877| 676| buffer[pos++] = target3; 878| 676| } 879| 1.21k| if ((index4 + idx_l < size_l) && (large[idx_l + index4] == target4)) { ------------------ | Branch (879:13): [True: 1.20k, False: 8] | Branch (879:42): [True: 647, False: 556] ------------------ 880| 647| buffer[pos++] = target4; 881| 647| } 882| 1.21k| idx_s += 4; 883| 1.21k| idx_l += index4; 884| 1.21k| } 885| 735| if ((idx_s + 2 <= size_s) && (idx_l < size_l)) { ------------------ | Branch (885:9): [True: 370, False: 365] | Branch (885:34): [True: 368, False: 2] ------------------ 886| 368| uint16_t target1 = small[idx_s]; 887| 368| uint16_t target2 = small[idx_s + 1]; 888| 368| binarySearch2(large + idx_l, (int32_t)(size_l - idx_l), target1, 889| 368| target2, &index1, &index2); 890| 368| if ((index1 + idx_l < size_l) && (large[idx_l + index1] == target1)) { ------------------ | Branch (890:13): [True: 363, False: 5] | Branch (890:42): [True: 241, False: 122] ------------------ 891| 241| buffer[pos++] = target1; 892| 241| } 893| 368| if ((index2 + idx_l < size_l) && (large[idx_l + index2] == target2)) { ------------------ | Branch (893:13): [True: 352, False: 16] | Branch (893:42): [True: 245, False: 107] ------------------ 894| 245| buffer[pos++] = target2; 895| 245| } 896| 368| idx_s += 2; 897| 368| idx_l += index2; 898| 368| } 899| 735| if ((idx_s < size_s) && (idx_l < size_l)) { ------------------ | Branch (899:9): [True: 343, False: 392] | Branch (899:29): [True: 340, False: 3] ------------------ 900| 340| uint16_t val_s = small[idx_s]; 901| 340| int32_t index = 902| 340| binarySearch(large + idx_l, (int32_t)(size_l - idx_l), val_s); 903| 340| if (index >= 0) buffer[pos++] = val_s; ------------------ | Branch (903:13): [True: 242, False: 98] ------------------ 904| 340| } 905| 735| return (int32_t)pos; 906| 735|} intersect_skewed_uint16_cardinality: 912| 2.10k| size_t size_l) { 913| 2.10k| size_t pos = 0, idx_l = 0, idx_s = 0; 914| | 915| 2.10k| if (0 == size_s) { ------------------ | Branch (915:9): [True: 0, False: 2.10k] ------------------ 916| 0| return 0; 917| 0| } 918| | 919| 2.10k| uint16_t val_l = large[idx_l], val_s = small[idx_s]; 920| | 921| 27.6k| while (true) { ------------------ | Branch (921:12): [True: 27.6k, Folded] ------------------ 922| 27.6k| if (val_l < val_s) { ------------------ | Branch (922:13): [True: 5.80k, False: 21.8k] ------------------ 923| 5.80k| idx_l = advanceUntil(large, (int32_t)idx_l, (int32_t)size_l, val_s); 924| 5.80k| if (idx_l == size_l) break; ------------------ | Branch (924:17): [True: 92, False: 5.70k] ------------------ 925| 5.70k| val_l = large[idx_l]; 926| 21.8k| } else if (val_s < val_l) { ------------------ | Branch (926:20): [True: 10.1k, False: 11.7k] ------------------ 927| 10.1k| idx_s++; 928| 10.1k| if (idx_s == size_s) break; ------------------ | Branch (928:17): [True: 892, False: 9.22k] ------------------ 929| 9.22k| val_s = small[idx_s]; 930| 11.7k| } else { 931| 11.7k| pos++; 932| 11.7k| idx_s++; 933| 11.7k| if (idx_s == size_s) break; ------------------ | Branch (933:17): [True: 1.05k, False: 10.7k] ------------------ 934| 10.7k| val_s = small[idx_s]; 935| 10.7k| idx_l = advanceUntil(large, (int32_t)idx_l, (int32_t)size_l, val_s); 936| 10.7k| if (idx_l == size_l) break; ------------------ | Branch (936:17): [True: 68, False: 10.6k] ------------------ 937| 10.6k| val_l = large[idx_l]; 938| 10.6k| } 939| 27.6k| } 940| | 941| 2.10k| return (int32_t)pos; 942| 2.10k|} intersect_skewed_uint16_nonempty: 945| 526| const uint16_t *large, size_t size_l) { 946| 526| size_t idx_l = 0, idx_s = 0; 947| | 948| 526| if (0 == size_s) { ------------------ | Branch (948:9): [True: 0, False: 526] ------------------ 949| 0| return false; 950| 0| } 951| | 952| 526| uint16_t val_l = large[idx_l], val_s = small[idx_s]; 953| | 954| 2.05k| while (true) { ------------------ | Branch (954:12): [True: 2.05k, Folded] ------------------ 955| 2.05k| if (val_l < val_s) { ------------------ | Branch (955:13): [True: 786, False: 1.26k] ------------------ 956| 786| idx_l = advanceUntil(large, (int32_t)idx_l, (int32_t)size_l, val_s); 957| 786| if (idx_l == size_l) break; ------------------ | Branch (957:17): [True: 22, False: 764] ------------------ 958| 764| val_l = large[idx_l]; 959| 1.26k| } else if (val_s < val_l) { ------------------ | Branch (959:20): [True: 840, False: 425] ------------------ 960| 840| idx_s++; 961| 840| if (idx_s == size_s) break; ------------------ | Branch (961:17): [True: 79, False: 761] ------------------ 962| 761| val_s = small[idx_s]; 963| 761| } else { 964| 425| return true; 965| 425| } 966| 2.05k| } 967| | 968| 101| return false; 969| 526|} intersect_uint16_nonempty: 1025| 1.74k| const uint16_t *B, const size_t lenB) { 1026| 1.74k| if (lenA == 0 || lenB == 0) return 0; ------------------ | Branch (1026:9): [True: 0, False: 1.74k] | Branch (1026:22): [True: 0, False: 1.74k] ------------------ 1027| 1.74k| const uint16_t *endA = A + lenA; 1028| 1.74k| const uint16_t *endB = B + lenB; 1029| | 1030| 1.74k| while (1) { ------------------ | Branch (1030:12): [True: 1.74k, Folded] ------------------ 1031| 9.38k| while (*A < *B) { ------------------ | Branch (1031:16): [True: 7.16k, False: 2.22k] ------------------ 1032| 7.68k| SKIP_FIRST_COMPARE: 1033| 7.68k| if (++A == endA) return false; ------------------ | Branch (1033:17): [True: 39, False: 7.64k] ------------------ 1034| 7.68k| } 1035| 4.83k| while (*A > *B) { ------------------ | Branch (1035:16): [True: 2.61k, False: 2.22k] ------------------ 1036| 2.61k| if (++B == endB) return false; ------------------ | Branch (1036:17): [True: 8, False: 2.60k] ------------------ 1037| 2.61k| } 1038| 2.22k| if (*A == *B) { ------------------ | Branch (1038:13): [True: 1.69k, False: 522] ------------------ 1039| 1.69k| return true; 1040| 1.69k| } else { 1041| 522| goto SKIP_FIRST_COMPARE; 1042| 522| } 1043| 2.22k| } 1044| 0| return false; // NOTREACHED 1045| 1.74k|} union_uint16: 1105| 4.52k| size_t size_2, uint16_t *buffer) { 1106| 4.52k| size_t pos = 0, idx_1 = 0, idx_2 = 0; 1107| | 1108| 4.52k| if (0 == size_2) { ------------------ | Branch (1108:9): [True: 64, False: 4.46k] ------------------ 1109| 64| memmove(buffer, set_1, size_1 * sizeof(uint16_t)); 1110| 64| return size_1; 1111| 64| } 1112| 4.46k| if (0 == size_1) { ------------------ | Branch (1112:9): [True: 0, False: 4.46k] ------------------ 1113| 0| memmove(buffer, set_2, size_2 * sizeof(uint16_t)); 1114| 0| return size_2; 1115| 0| } 1116| | 1117| 4.46k| uint16_t val_1 = set_1[idx_1], val_2 = set_2[idx_2]; 1118| | 1119| 192k| while (true) { ------------------ | Branch (1119:12): [True: 192k, Folded] ------------------ 1120| 192k| if (val_1 < val_2) { ------------------ | Branch (1120:13): [True: 7.67k, False: 184k] ------------------ 1121| 7.67k| buffer[pos++] = val_1; 1122| 7.67k| ++idx_1; 1123| 7.67k| if (idx_1 >= size_1) break; ------------------ | Branch (1123:17): [True: 383, False: 7.28k] ------------------ 1124| 7.28k| val_1 = set_1[idx_1]; 1125| 184k| } else if (val_2 < val_1) { ------------------ | Branch (1125:20): [True: 153k, False: 31.3k] ------------------ 1126| 153k| buffer[pos++] = val_2; 1127| 153k| ++idx_2; 1128| 153k| if (idx_2 >= size_2) break; ------------------ | Branch (1128:17): [True: 176, False: 153k] ------------------ 1129| 153k| val_2 = set_2[idx_2]; 1130| 153k| } else { 1131| 31.3k| buffer[pos++] = val_1; 1132| 31.3k| ++idx_1; 1133| 31.3k| ++idx_2; 1134| 31.3k| if (idx_1 >= size_1 || idx_2 >= size_2) break; ------------------ | Branch (1134:17): [True: 3.77k, False: 27.5k] | Branch (1134:36): [True: 128, False: 27.4k] ------------------ 1135| 27.4k| val_1 = set_1[idx_1]; 1136| 27.4k| val_2 = set_2[idx_2]; 1137| 27.4k| } 1138| 192k| } 1139| | 1140| 4.46k| if (idx_1 < size_1) { ------------------ | Branch (1140:9): [True: 304, False: 4.16k] ------------------ 1141| 304| const size_t n_elems = size_1 - idx_1; 1142| 304| memmove(buffer + pos, set_1 + idx_1, n_elems * sizeof(uint16_t)); 1143| 304| pos += n_elems; 1144| 4.16k| } else if (idx_2 < size_2) { ------------------ | Branch (1144:16): [True: 1.75k, False: 2.40k] ------------------ 1145| 1.75k| const size_t n_elems = size_2 - idx_2; 1146| 1.75k| memmove(buffer + pos, set_2 + idx_2, n_elems * sizeof(uint16_t)); 1147| 1.75k| pos += n_elems; 1148| 1.75k| } 1149| | 1150| 4.46k| return pos; 1151| 4.46k|} difference_uint16: 1154| 3.43k| int length2, uint16_t *a_out) { 1155| 3.43k| int out_card = 0; 1156| 3.43k| int k1 = 0, k2 = 0; 1157| 3.43k| if (length1 == 0) return 0; ------------------ | Branch (1157:9): [True: 0, False: 3.43k] ------------------ 1158| 3.43k| if (length2 == 0) { ------------------ | Branch (1158:9): [True: 0, False: 3.43k] ------------------ 1159| 0| if (a1 != a_out) memcpy(a_out, a1, sizeof(uint16_t) * length1); ------------------ | Branch (1159:13): [True: 0, False: 0] ------------------ 1160| 0| return length1; 1161| 0| } 1162| 3.43k| uint16_t s1 = a1[k1]; 1163| 3.43k| uint16_t s2 = a2[k2]; 1164| 423k| while (true) { ------------------ | Branch (1164:12): [True: 423k, Folded] ------------------ 1165| 423k| if (s1 < s2) { ------------------ | Branch (1165:13): [True: 207k, False: 216k] ------------------ 1166| 207k| a_out[out_card++] = s1; 1167| 207k| ++k1; 1168| 207k| if (k1 >= length1) { ------------------ | Branch (1168:17): [True: 0, False: 207k] ------------------ 1169| 0| break; 1170| 0| } 1171| 207k| s1 = a1[k1]; 1172| 216k| } else if (s1 == s2) { ------------------ | Branch (1172:20): [True: 216k, False: 0] ------------------ 1173| 216k| ++k1; 1174| 216k| ++k2; 1175| 216k| if (k1 >= length1) { ------------------ | Branch (1175:17): [True: 2.23k, False: 214k] ------------------ 1176| 2.23k| break; 1177| 2.23k| } 1178| 214k| if (k2 >= length2) { ------------------ | Branch (1178:17): [True: 1.20k, False: 213k] ------------------ 1179| 1.20k| memmove(a_out + out_card, a1 + k1, 1180| 1.20k| sizeof(uint16_t) * (length1 - k1)); 1181| 1.20k| return out_card + length1 - k1; 1182| 1.20k| } 1183| 213k| s1 = a1[k1]; 1184| 213k| s2 = a2[k2]; 1185| 213k| } else { // if (val1>val2) 1186| 0| ++k2; 1187| 0| if (k2 >= length2) { ------------------ | Branch (1187:17): [True: 0, False: 0] ------------------ 1188| 0| memmove(a_out + out_card, a1 + k1, 1189| 0| sizeof(uint16_t) * (length1 - k1)); 1190| 0| return out_card + length1 - k1; 1191| 0| } 1192| 0| s2 = a2[k2]; 1193| 0| } 1194| 423k| } 1195| 2.23k| return out_card; 1196| 3.43k|} xor_uint16: 1199| 7.70k| const uint16_t *array_2, int32_t card_2, uint16_t *out) { 1200| 7.70k| int32_t pos1 = 0, pos2 = 0, pos_out = 0; 1201| 206k| while (pos1 < card_1 && pos2 < card_2) { ------------------ | Branch (1201:12): [True: 199k, False: 7.12k] | Branch (1201:29): [True: 198k, False: 580] ------------------ 1202| 198k| const uint16_t v1 = array_1[pos1]; 1203| 198k| const uint16_t v2 = array_2[pos2]; 1204| 198k| if (v1 == v2) { ------------------ | Branch (1204:13): [True: 38.2k, False: 160k] ------------------ 1205| 38.2k| ++pos1; 1206| 38.2k| ++pos2; 1207| 38.2k| continue; 1208| 38.2k| } 1209| 160k| if (v1 < v2) { ------------------ | Branch (1209:13): [True: 28.1k, False: 132k] ------------------ 1210| 28.1k| out[pos_out++] = v1; 1211| 28.1k| ++pos1; 1212| 132k| } else { 1213| 132k| out[pos_out++] = v2; 1214| 132k| ++pos2; 1215| 132k| } 1216| 160k| } 1217| 7.70k| if (pos1 < card_1) { ------------------ | Branch (1217:9): [True: 580, False: 7.12k] ------------------ 1218| 580| const size_t n_elems = card_1 - pos1; 1219| 580| memcpy(out + pos_out, array_1 + pos1, n_elems * sizeof(uint16_t)); 1220| 580| pos_out += (int32_t)n_elems; 1221| 7.12k| } else if (pos2 < card_2) { ------------------ | Branch (1221:16): [True: 2.23k, False: 4.88k] ------------------ 1222| 2.23k| const size_t n_elems = card_2 - pos2; 1223| 2.23k| memcpy(out + pos_out, array_2 + pos2, n_elems * sizeof(uint16_t)); 1224| 2.23k| pos_out += (int32_t)n_elems; 1225| 2.23k| } 1226| 7.70k| return pos_out; 1227| 7.70k|} union_vector16: 1654| 4.52k| uint16_t *output) { 1655| 4.52k| if ((length1 < 8) || (length2 < 8)) { ------------------ | Branch (1655:9): [True: 990, False: 3.53k] | Branch (1655:26): [True: 0, False: 3.53k] ------------------ 1656| 990| return (uint32_t)union_uint16(array1, length1, array2, length2, output); 1657| 990| } 1658| 3.53k| __m128i vA, vB, V, vecMin, vecMax; 1659| 3.53k| __m128i laststore; 1660| 3.53k| uint16_t *initoutput = output; 1661| 3.53k| uint32_t len1 = length1 / 8; 1662| 3.53k| uint32_t len2 = length2 / 8; 1663| 3.53k| uint32_t pos1 = 0; 1664| 3.53k| uint32_t pos2 = 0; 1665| | // we start the machine 1666| 3.53k| vA = _mm_lddqu_si128((const __m128i *)array1 + pos1); 1667| 3.53k| pos1++; 1668| 3.53k| vB = _mm_lddqu_si128((const __m128i *)array2 + pos2); 1669| 3.53k| pos2++; 1670| 3.53k| sse_merge(&vA, &vB, &vecMin, &vecMax); 1671| 3.53k| laststore = _mm_set1_epi16(-1); 1672| 3.53k| output += store_unique(laststore, vecMin, output); 1673| 3.53k| laststore = vecMin; 1674| 3.53k| if ((pos1 < len1) && (pos2 < len2)) { ------------------ | Branch (1674:9): [True: 3.03k, False: 509] | Branch (1674:26): [True: 3.03k, False: 0] ------------------ 1675| 3.03k| uint16_t curA, curB; 1676| 3.03k| curA = array1[8 * pos1]; 1677| 3.03k| curB = array2[8 * pos2]; 1678| 102k| while (true) { ------------------ | Branch (1678:16): [True: 102k, Folded] ------------------ 1679| 102k| if (curA <= curB) { ------------------ | Branch (1679:17): [True: 43.2k, False: 59.2k] ------------------ 1680| 43.2k| V = _mm_lddqu_si128((const __m128i *)array1 + pos1); 1681| 43.2k| pos1++; 1682| 43.2k| if (pos1 < len1) { ------------------ | Branch (1682:21): [True: 41.0k, False: 2.27k] ------------------ 1683| 41.0k| curA = array1[8 * pos1]; 1684| 41.0k| } else { 1685| 2.27k| break; 1686| 2.27k| } 1687| 59.2k| } else { 1688| 59.2k| V = _mm_lddqu_si128((const __m128i *)array2 + pos2); 1689| 59.2k| pos2++; 1690| 59.2k| if (pos2 < len2) { ------------------ | Branch (1690:21): [True: 58.5k, False: 751] ------------------ 1691| 58.5k| curB = array2[8 * pos2]; 1692| 58.5k| } else { 1693| 751| break; 1694| 751| } 1695| 59.2k| } 1696| 99.5k| sse_merge(&V, &vecMax, &vecMin, &vecMax); 1697| 99.5k| output += store_unique(laststore, vecMin, output); 1698| 99.5k| laststore = vecMin; 1699| 99.5k| } 1700| 3.03k| sse_merge(&V, &vecMax, &vecMin, &vecMax); 1701| 3.03k| output += store_unique(laststore, vecMin, output); 1702| 3.03k| laststore = vecMin; 1703| 3.03k| } 1704| | // we finish the rest off using a scalar algorithm 1705| | // could be improved? 1706| | // 1707| | // copy the small end on a tmp buffer 1708| 3.53k| uint32_t len = (uint32_t)(output - initoutput); 1709| 3.53k| uint16_t buffer[16]; 1710| 3.53k| uint32_t leftoversize = store_unique(laststore, vecMax, buffer); 1711| 3.53k| if (pos1 == len1) { ------------------ | Branch (1711:9): [True: 2.78k, False: 751] ------------------ 1712| 2.78k| memcpy(buffer + leftoversize, array1 + 8 * pos1, 1713| 2.78k| (length1 - 8 * len1) * sizeof(uint16_t)); 1714| 2.78k| leftoversize += length1 - 8 * len1; 1715| 2.78k| qsort(buffer, leftoversize, sizeof(uint16_t), uint16_compare); 1716| | 1717| 2.78k| leftoversize = unique(buffer, leftoversize); 1718| 2.78k| len += (uint32_t)union_uint16(buffer, leftoversize, array2 + 8 * pos2, 1719| 2.78k| length2 - 8 * pos2, output); 1720| 2.78k| } else { 1721| 751| memcpy(buffer + leftoversize, array2 + 8 * pos2, 1722| 751| (length2 - 8 * len2) * sizeof(uint16_t)); 1723| 751| leftoversize += length2 - 8 * len2; 1724| 751| qsort(buffer, leftoversize, sizeof(uint16_t), uint16_compare); 1725| 751| leftoversize = unique(buffer, leftoversize); 1726| 751| len += (uint32_t)union_uint16(buffer, leftoversize, array1 + 8 * pos1, 1727| 751| length1 - 8 * pos1, output); 1728| 751| } 1729| 3.53k| return len; 1730| 4.52k|} xor_vector16: 1778| 7.70k| uint16_t *output) { 1779| 7.70k| if ((length1 < 8) || (length2 < 8)) { ------------------ | Branch (1779:9): [True: 3.10k, False: 4.60k] | Branch (1779:26): [True: 348, False: 4.25k] ------------------ 1780| 3.45k| return xor_uint16(array1, length1, array2, length2, output); 1781| 3.45k| } 1782| 4.25k| __m128i vA, vB, V, vecMin, vecMax; 1783| 4.25k| __m128i laststore; 1784| 4.25k| uint16_t *initoutput = output; 1785| 4.25k| uint32_t len1 = length1 / 8; 1786| 4.25k| uint32_t len2 = length2 / 8; 1787| 4.25k| uint32_t pos1 = 0; 1788| 4.25k| uint32_t pos2 = 0; 1789| | // we start the machine 1790| 4.25k| vA = _mm_lddqu_si128((const __m128i *)array1 + pos1); 1791| 4.25k| pos1++; 1792| 4.25k| vB = _mm_lddqu_si128((const __m128i *)array2 + pos2); 1793| 4.25k| pos2++; 1794| 4.25k| sse_merge(&vA, &vB, &vecMin, &vecMax); 1795| 4.25k| laststore = _mm_set1_epi16(-1); 1796| 4.25k| uint16_t buffer[17]; 1797| 4.25k| output += store_unique_xor(laststore, vecMin, output); 1798| | 1799| 4.25k| laststore = vecMin; 1800| 4.25k| if ((pos1 < len1) && (pos2 < len2)) { ------------------ | Branch (1800:9): [True: 3.91k, False: 338] | Branch (1800:26): [True: 3.79k, False: 118] ------------------ 1801| 3.79k| uint16_t curA, curB; 1802| 3.79k| curA = array1[8 * pos1]; 1803| 3.79k| curB = array2[8 * pos2]; 1804| 179k| while (true) { ------------------ | Branch (1804:16): [True: 179k, Folded] ------------------ 1805| 179k| if (curA <= curB) { ------------------ | Branch (1805:17): [True: 85.6k, False: 94.1k] ------------------ 1806| 85.6k| V = _mm_lddqu_si128((const __m128i *)array1 + pos1); 1807| 85.6k| pos1++; 1808| 85.6k| if (pos1 < len1) { ------------------ | Branch (1808:21): [True: 83.1k, False: 2.50k] ------------------ 1809| 83.1k| curA = array1[8 * pos1]; 1810| 83.1k| } else { 1811| 2.50k| break; 1812| 2.50k| } 1813| 94.1k| } else { 1814| 94.1k| V = _mm_lddqu_si128((const __m128i *)array2 + pos2); 1815| 94.1k| pos2++; 1816| 94.1k| if (pos2 < len2) { ------------------ | Branch (1816:21): [True: 92.8k, False: 1.28k] ------------------ 1817| 92.8k| curB = array2[8 * pos2]; 1818| 92.8k| } else { 1819| 1.28k| break; 1820| 1.28k| } 1821| 94.1k| } 1822| 176k| sse_merge(&V, &vecMax, &vecMin, &vecMax); 1823| | // conditionally stores the last value of laststore as well as all 1824| | // but the 1825| | // last value of vecMin 1826| 176k| output += store_unique_xor(laststore, vecMin, output); 1827| 176k| laststore = vecMin; 1828| 176k| } 1829| 3.79k| sse_merge(&V, &vecMax, &vecMin, &vecMax); 1830| | // conditionally stores the last value of laststore as well as all but 1831| | // the 1832| | // last value of vecMin 1833| 3.79k| output += store_unique_xor(laststore, vecMin, output); 1834| 3.79k| laststore = vecMin; 1835| 3.79k| } 1836| 4.25k| uint32_t len = (uint32_t)(output - initoutput); 1837| | 1838| | // we finish the rest off using a scalar algorithm 1839| | // could be improved? 1840| | // conditionally stores the last value of laststore as well as all but the 1841| | // last value of vecMax, 1842| | // we store to "buffer" 1843| 4.25k| int leftoversize = store_unique_xor(laststore, vecMax, buffer); 1844| 4.25k| uint16_t vec7 = (uint16_t)_mm_extract_epi16(vecMax, 7); 1845| 4.25k| uint16_t vec6 = (uint16_t)_mm_extract_epi16(vecMax, 6); 1846| 4.25k| if (vec7 != vec6) buffer[leftoversize++] = vec7; ------------------ | Branch (1846:9): [True: 4.20k, False: 49] ------------------ 1847| 4.25k| if (pos1 == len1) { ------------------ | Branch (1847:9): [True: 2.84k, False: 1.40k] ------------------ 1848| 2.84k| memcpy(buffer + leftoversize, array1 + 8 * pos1, 1849| 2.84k| (length1 - 8 * len1) * sizeof(uint16_t)); 1850| 2.84k| leftoversize += length1 - 8 * len1; 1851| 2.84k| if (leftoversize == 0) { // trivial case ------------------ | Branch (1851:13): [True: 7, False: 2.84k] ------------------ 1852| 7| memcpy(output, array2 + 8 * pos2, 1853| 7| (length2 - 8 * pos2) * sizeof(uint16_t)); 1854| 7| len += (length2 - 8 * pos2); 1855| 2.84k| } else { 1856| 2.84k| qsort(buffer, leftoversize, sizeof(uint16_t), uint16_compare); 1857| 2.84k| leftoversize = unique_xor(buffer, leftoversize); 1858| 2.84k| len += xor_uint16(buffer, leftoversize, array2 + 8 * pos2, 1859| 2.84k| length2 - 8 * pos2, output); 1860| 2.84k| } 1861| 2.84k| } else { 1862| 1.40k| memcpy(buffer + leftoversize, array2 + 8 * pos2, 1863| 1.40k| (length2 - 8 * len2) * sizeof(uint16_t)); 1864| 1.40k| leftoversize += length2 - 8 * len2; 1865| 1.40k| if (leftoversize == 0) { // trivial case ------------------ | Branch (1865:13): [True: 2, False: 1.40k] ------------------ 1866| 2| memcpy(output, array1 + 8 * pos1, 1867| 2| (length1 - 8 * pos1) * sizeof(uint16_t)); 1868| 2| len += (length1 - 8 * pos1); 1869| 1.40k| } else { 1870| 1.40k| qsort(buffer, leftoversize, sizeof(uint16_t), uint16_compare); 1871| 1.40k| leftoversize = unique_xor(buffer, leftoversize); 1872| 1.40k| len += xor_uint16(buffer, leftoversize, array1 + 8 * pos1, 1873| 1.40k| length1 - 8 * pos1, output); 1874| 1.40k| } 1875| 1.40k| } 1876| 4.25k| return len; 1877| 7.70k|} fast_union_uint16: 1980| 4.52k| uint16_t *buffer) { 1981| 4.52k|#if CROARING_IS_X64 1982| 4.52k| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (1982:9): [True: 4.52k, False: 0] ------------------ 1983| | // compute union with smallest array first 1984| 4.52k| if (size_1 < size_2) { ------------------ | Branch (1984:13): [True: 2.28k, False: 2.24k] ------------------ 1985| 2.28k| return union_vector16(set_1, (uint32_t)size_1, set_2, 1986| 2.28k| (uint32_t)size_2, buffer); 1987| 2.28k| } else { 1988| 2.24k| return union_vector16(set_2, (uint32_t)size_2, set_1, 1989| 2.24k| (uint32_t)size_1, buffer); 1990| 2.24k| } 1991| 4.52k| } else { 1992| | // compute union with smallest array first 1993| 0| if (size_1 < size_2) { ------------------ | Branch (1993:13): [True: 0, False: 0] ------------------ 1994| 0| return union_uint16(set_1, size_1, set_2, size_2, buffer); 1995| 0| } else { 1996| 0| return union_uint16(set_2, size_2, set_1, size_1, buffer); 1997| 0| } 1998| 0| } 1999| |#else 2000| | // compute union with smallest array first 2001| | if (size_1 < size_2) { 2002| | return union_uint16(set_1, size_1, set_2, size_2, buffer); 2003| | } else { 2004| | return union_uint16(set_2, size_2, set_1, size_1, buffer); 2005| | } 2006| |#endif 2007| 4.52k|} memequals: 2111| 7.09k|bool memequals(const void *s1, const void *s2, size_t n) { 2112| 7.09k| if (n == 0) { ------------------ | Branch (2112:9): [True: 0, False: 7.09k] ------------------ 2113| 0| return true; 2114| 0| } 2115| 7.09k|#if CROARING_IS_X64 2116| 7.09k| int support = croaring_hardware_support(); 2117| 7.09k|#if CROARING_COMPILER_SUPPORTS_AVX512 2118| 7.09k| if (support & ROARING_SUPPORTS_AVX512) { ------------------ | Branch (2118:9): [True: 0, False: 7.09k] ------------------ 2119| 0| return _avx512_memequals(s1, s2, n); 2120| 0| } else 2121| 7.09k|#endif // CROARING_COMPILER_SUPPORTS_AVX512 2122| 7.09k| if (support & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (2122:13): [True: 7.09k, False: 0] ------------------ 2123| 7.09k| return _avx2_memequals(s1, s2, n); 2124| 7.09k| } else { 2125| 0| return memcmp(s1, s2, n) == 0; 2126| 0| } 2127| |#else 2128| | return memcmp(s1, s2, n) == 0; 2129| |#endif 2130| 7.09k|} array_util.c:binarySearch4: 804| 1.21k| int32_t *index4) { 805| 1.21k| const uint16_t *base1 = array; 806| 1.21k| const uint16_t *base2 = array; 807| 1.21k| const uint16_t *base3 = array; 808| 1.21k| const uint16_t *base4 = array; 809| 1.21k| if (n == 0) return; ------------------ | Branch (809:9): [True: 0, False: 1.21k] ------------------ 810| 15.1k| while (n > 1) { ------------------ | Branch (810:12): [True: 13.9k, False: 1.21k] ------------------ 811| 13.9k| int32_t half = n >> 1; 812| 13.9k| base1 = (base1[half] < target1) ? &base1[half] : base1; ------------------ | Branch (812:17): [True: 1.04k, False: 12.8k] ------------------ 813| 13.9k| base2 = (base2[half] < target2) ? &base2[half] : base2; ------------------ | Branch (813:17): [True: 1.94k, False: 11.9k] ------------------ 814| 13.9k| base3 = (base3[half] < target3) ? &base3[half] : base3; ------------------ | Branch (814:17): [True: 2.52k, False: 11.3k] ------------------ 815| 13.9k| base4 = (base4[half] < target4) ? &base4[half] : base4; ------------------ | Branch (815:17): [True: 2.96k, False: 10.9k] ------------------ 816| 13.9k| n -= half; 817| 13.9k| } 818| 1.21k| *index1 = (int32_t)((*base1 < target1) + base1 - array); 819| 1.21k| *index2 = (int32_t)((*base2 < target2) + base2 - array); 820| 1.21k| *index3 = (int32_t)((*base3 < target3) + base3 - array); 821| 1.21k| *index4 = (int32_t)((*base4 < target4) + base4 - array); 822| 1.21k|} array_util.c:binarySearch2: 833| 368| uint16_t target2, int32_t *index1, int32_t *index2) { 834| 368| const uint16_t *base1 = array; 835| 368| const uint16_t *base2 = array; 836| 368| if (n == 0) return; ------------------ | Branch (836:9): [True: 0, False: 368] ------------------ 837| 3.50k| while (n > 1) { ------------------ | Branch (837:12): [True: 3.14k, False: 368] ------------------ 838| 3.14k| int32_t half = n >> 1; 839| 3.14k| base1 = (base1[half] < target1) ? &base1[half] : base1; ------------------ | Branch (839:17): [True: 781, False: 2.36k] ------------------ 840| 3.14k| base2 = (base2[half] < target2) ? &base2[half] : base2; ------------------ | Branch (840:17): [True: 1.30k, False: 1.83k] ------------------ 841| 3.14k| n -= half; 842| 3.14k| } 843| 368| *index1 = (int32_t)((*base1 < target1) + base1 - array); 844| 368| *index2 = (int32_t)((*base2 < target2) + base2 - array); 845| 368|} array_util.c:sse_merge: 1244| 290k| __m128i *vecMin, __m128i *vecMax) { // output 1245| 290k| __m128i vecTmp; 1246| 290k| vecTmp = _mm_min_epu16(*vInput1, *vInput2); 1247| 290k| *vecMax = _mm_max_epu16(*vInput1, *vInput2); 1248| 290k| vecTmp = _mm_alignr_epi8(vecTmp, vecTmp, 2); 1249| 290k| *vecMin = _mm_min_epu16(vecTmp, *vecMax); 1250| 290k| *vecMax = _mm_max_epu16(vecTmp, *vecMax); 1251| 290k| vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); 1252| 290k| *vecMin = _mm_min_epu16(vecTmp, *vecMax); 1253| 290k| *vecMax = _mm_max_epu16(vecTmp, *vecMax); 1254| 290k| vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); 1255| 290k| *vecMin = _mm_min_epu16(vecTmp, *vecMax); 1256| 290k| *vecMax = _mm_max_epu16(vecTmp, *vecMax); 1257| 290k| vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); 1258| 290k| *vecMin = _mm_min_epu16(vecTmp, *vecMax); 1259| 290k| *vecMax = _mm_max_epu16(vecTmp, *vecMax); 1260| 290k| vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); 1261| 290k| *vecMin = _mm_min_epu16(vecTmp, *vecMax); 1262| 290k| *vecMax = _mm_max_epu16(vecTmp, *vecMax); 1263| 290k| vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); 1264| 290k| *vecMin = _mm_min_epu16(vecTmp, *vecMax); 1265| 290k| *vecMax = _mm_max_epu16(vecTmp, *vecMax); 1266| 290k| vecTmp = _mm_alignr_epi8(*vecMin, *vecMin, 2); 1267| 290k| *vecMin = _mm_min_epu16(vecTmp, *vecMax); 1268| 290k| *vecMax = _mm_max_epu16(vecTmp, *vecMax); 1269| | *vecMin = _mm_alignr_epi8(*vecMin, *vecMin, 2); 1270| 290k|} array_util.c:store_unique: 1619| 109k|static inline int store_unique(__m128i old, __m128i newval, uint16_t *output) { 1620| 109k| __m128i vecTmp = _mm_alignr_epi8(newval, old, 16 - 2); 1621| | // lots of high latency instructions follow (optimize?) 1622| 109k| int M = _mm_movemask_epi8( 1623| 109k| _mm_packs_epi16(_mm_cmpeq_epi16(vecTmp, newval), _mm_setzero_si128())); 1624| 109k| int numberofnewvalues = 8 - _mm_popcnt_u32(M); 1625| 109k| __m128i key = _mm_lddqu_si128((const __m128i *)uniqshuf + M); 1626| 109k| __m128i val = _mm_shuffle_epi8(newval, key); 1627| 109k| _mm_storeu_si128((__m128i *)output, val); 1628| 109k| return numberofnewvalues; 1629| 109k|} array_util.c:uint16_compare: 1645| 124k|static int uint16_compare(const void *a, const void *b) { 1646| 124k| return (*(uint16_t *)a - *(uint16_t *)b); 1647| 124k|} array_util.c:unique: 1634| 3.53k|static inline uint32_t unique(uint16_t *out, uint32_t len) { 1635| 3.53k| uint32_t pos = 1; 1636| 36.3k| for (uint32_t i = 1; i < len; ++i) { ------------------ | Branch (1636:26): [True: 32.8k, False: 3.53k] ------------------ 1637| 32.8k| if (out[i] != out[i - 1]) { ------------------ | Branch (1637:13): [True: 32.7k, False: 109] ------------------ 1638| 32.7k| out[pos++] = out[i]; 1639| 32.7k| } 1640| 32.8k| } 1641| 3.53k| return pos; 1642| 3.53k|} array_util.c:store_unique_xor: 1746| 188k| uint16_t *output) { 1747| 188k| __m128i vecTmp1 = _mm_alignr_epi8(newval, old, 16 - 4); 1748| | __m128i vecTmp2 = _mm_alignr_epi8(newval, old, 16 - 2); 1749| 188k| __m128i equalleft = _mm_cmpeq_epi16(vecTmp2, vecTmp1); 1750| 188k| __m128i equalright = _mm_cmpeq_epi16(vecTmp2, newval); 1751| 188k| __m128i equalleftoright = _mm_or_si128(equalleft, equalright); 1752| 188k| int M = _mm_movemask_epi8( 1753| 188k| _mm_packs_epi16(equalleftoright, _mm_setzero_si128())); 1754| 188k| int numberofnewvalues = 8 - _mm_popcnt_u32(M); 1755| 188k| __m128i key = _mm_lddqu_si128((const __m128i *)uniqshuf + M); 1756| 188k| __m128i val = _mm_shuffle_epi8(vecTmp2, key); 1757| 188k| _mm_storeu_si128((__m128i *)output, val); 1758| 188k| return numberofnewvalues; 1759| 188k|} array_util.c:unique_xor: 1764| 4.24k|static inline uint32_t unique_xor(uint16_t *out, uint32_t len) { 1765| 4.24k| uint32_t pos = 1; 1766| 42.0k| for (uint32_t i = 1; i < len; ++i) { ------------------ | Branch (1766:26): [True: 37.7k, False: 4.24k] ------------------ 1767| 37.7k| if (out[i] != out[i - 1]) { ------------------ | Branch (1767:13): [True: 37.5k, False: 223] ------------------ 1768| 37.5k| out[pos++] = out[i]; 1769| 37.5k| } else 1770| 223| pos--; // if it is identical to previous, delete it 1771| 37.7k| } 1772| 4.24k| return pos; 1773| 4.24k|} array_util.c:_avx2_memequals: 2069| 7.09k|static inline bool _avx2_memequals(const void *s1, const void *s2, size_t n) { 2070| 7.09k| const uint8_t *ptr1 = (const uint8_t *)s1; 2071| 7.09k| const uint8_t *ptr2 = (const uint8_t *)s2; 2072| 7.09k| const uint8_t *end1 = ptr1 + n; 2073| 7.09k| const uint8_t *end8 = ptr1 + n / 8 * 8; 2074| 7.09k| const uint8_t *end32 = ptr1 + n / 32 * 32; 2075| | 2076| 132k| while (ptr1 < end32) { ------------------ | Branch (2076:12): [True: 125k, False: 7.03k] ------------------ 2077| 125k| __m256i r1 = _mm256_loadu_si256((const __m256i *)ptr1); 2078| 125k| __m256i r2 = _mm256_loadu_si256((const __m256i *)ptr2); 2079| 125k| int mask = _mm256_movemask_epi8(_mm256_cmpeq_epi8(r1, r2)); 2080| 125k| if ((uint32_t)mask != UINT32_MAX) { ------------------ | Branch (2080:13): [True: 62, False: 125k] ------------------ 2081| 62| return false; 2082| 62| } 2083| 125k| ptr1 += 32; 2084| 125k| ptr2 += 32; 2085| 125k| } 2086| | 2087| 12.6k| while (ptr1 < end8) { ------------------ | Branch (2087:12): [True: 5.66k, False: 6.94k] ------------------ 2088| 5.66k| uint64_t v1, v2; 2089| 5.66k| memcpy(&v1, ptr1, sizeof(uint64_t)); 2090| 5.66k| memcpy(&v2, ptr2, sizeof(uint64_t)); 2091| 5.66k| if (v1 != v2) { ------------------ | Branch (2091:13): [True: 88, False: 5.58k] ------------------ 2092| 88| return false; 2093| 88| } 2094| 5.58k| ptr1 += 8; 2095| 5.58k| ptr2 += 8; 2096| 5.58k| } 2097| | 2098| 25.1k| while (ptr1 < end1) { ------------------ | Branch (2098:12): [True: 18.2k, False: 6.91k] ------------------ 2099| 18.2k| if (*ptr1 != *ptr2) { ------------------ | Branch (2099:13): [True: 34, False: 18.2k] ------------------ 2100| 34| return false; 2101| 34| } 2102| 18.2k| ptr1++; 2103| 18.2k| ptr2++; 2104| 18.2k| } 2105| | 2106| 6.91k| return true; 2107| 6.94k|} bitset_extract_setbits_avx2: 685| 932| uint32_t base) { 686| 932| uint32_t *initout = out; 687| 932| __m256i baseVec = _mm256_set1_epi32(base - 1); 688| 932| __m256i incVec = _mm256_set1_epi32(64); 689| 932| __m256i add8 = _mm256_set1_epi32(8); 690| 932| uint32_t *safeout = out + outcapacity; 691| 932| size_t i = 0; 692| 777k| for (; (i < length) && (out + 64 <= safeout); ++i) { ------------------ | Branch (692:12): [True: 777k, False: 267] | Branch (692:28): [True: 776k, False: 665] ------------------ 693| 776k| uint64_t w = words[i]; 694| 776k| if (w == 0) { ------------------ | Branch (694:13): [True: 111k, False: 664k] ------------------ 695| 111k| baseVec = _mm256_add_epi32(baseVec, incVec); 696| 664k| } else { 697| 3.32M| for (int k = 0; k < 4; ++k) { ------------------ | Branch (697:29): [True: 2.65M, False: 664k] ------------------ 698| 2.65M| uint8_t byteA = (uint8_t)w; 699| 2.65M| uint8_t byteB = (uint8_t)(w >> 8); 700| 2.65M| w >>= 16; 701| 2.65M| __m256i vecA = 702| 2.65M| _mm256_loadu_si256((const __m256i *)vecDecodeTable[byteA]); 703| 2.65M| __m256i vecB = 704| 2.65M| _mm256_loadu_si256((const __m256i *)vecDecodeTable[byteB]); 705| 2.65M| uint8_t advanceA = lengthTable[byteA]; 706| 2.65M| uint8_t advanceB = lengthTable[byteB]; 707| 2.65M| vecA = _mm256_add_epi32(baseVec, vecA); 708| 2.65M| baseVec = _mm256_add_epi32(baseVec, add8); 709| 2.65M| vecB = _mm256_add_epi32(baseVec, vecB); 710| 2.65M| baseVec = _mm256_add_epi32(baseVec, add8); 711| 2.65M| _mm256_storeu_si256((__m256i *)out, vecA); 712| 2.65M| out += advanceA; 713| 2.65M| _mm256_storeu_si256((__m256i *)out, vecB); 714| 2.65M| out += advanceB; 715| 2.65M| } 716| 664k| } 717| 776k| } 718| 932| base += i * 64; 719| 13.6k| for (; (i < length) && (out < safeout); ++i) { ------------------ | Branch (719:12): [True: 13.2k, False: 364] | Branch (719:28): [True: 12.7k, False: 568] ------------------ 720| 12.7k| uint64_t w = words[i]; 721| 31.6k| while ((w != 0) && (out < safeout)) { ------------------ | Branch (721:16): [True: 18.9k, False: 12.7k] | Branch (721:28): [True: 18.9k, False: 0] ------------------ 722| 18.9k| int r = 723| 18.9k| roaring_trailing_zeroes(w); // on x64, should compile to TZCNT 724| 18.9k| uint32_t val = r + base; 725| 18.9k| memcpy(out, &val, 726| 18.9k| sizeof(uint32_t)); // should be compiled as a MOV on x64 727| 18.9k| out++; 728| 18.9k| w &= (w - 1); 729| 18.9k| } 730| 12.7k| base += 64; 731| 12.7k| } 732| 932| return out - initout; 733| 932|} bitset_extract_setbits: 738| 465| uint32_t *out, uint32_t base) { 739| 465| int outpos = 0; 740| 476k| for (size_t i = 0; i < length; ++i) { ------------------ | Branch (740:24): [True: 476k, False: 465] ------------------ 741| 476k| uint64_t w = words[i]; 742| 2.78M| while (w != 0) { ------------------ | Branch (742:16): [True: 2.30M, False: 476k] ------------------ 743| 2.30M| int r = 744| 2.30M| roaring_trailing_zeroes(w); // on x64, should compile to TZCNT 745| 2.30M| uint32_t val = r + base; 746| 2.30M| memcpy(out + outpos, &val, 747| 2.30M| sizeof(uint32_t)); // should be compiled as a MOV on x64 748| 2.30M| outpos++; 749| 2.30M| w &= (w - 1); 750| 2.30M| } 751| 476k| base += 64; 752| 476k| } 753| 465| return outpos; 754| 465|} bitset_extract_setbits_uint16: 848| 2.66k| uint16_t *out, uint16_t base) { 849| 2.66k| int outpos = 0; 850| 2.72M| for (size_t i = 0; i < length; ++i) { ------------------ | Branch (850:24): [True: 2.72M, False: 2.66k] ------------------ 851| 2.72M| uint64_t w = words[i]; 852| 5.17M| while (w != 0) { ------------------ | Branch (852:16): [True: 2.45M, False: 2.72M] ------------------ 853| 2.45M| int r = roaring_trailing_zeroes(w); 854| 2.45M| out[outpos++] = (uint16_t)(r + base); 855| 2.45M| w &= (w - 1); 856| 2.45M| } 857| 2.72M| base += 64; 858| 2.72M| } 859| 2.66k| return outpos; 860| 2.66k|} bitset_clear_list: 1028| 1.54k| uint64_t length) { 1029| 1.54k| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (1029:9): [True: 1.54k, False: 0] ------------------ 1030| 1.54k| return _asm_bitset_clear_list(words, card, list, length); 1031| 1.54k| } else { 1032| 0| return _scalar_bitset_clear_list(words, card, list, length); 1033| 0| } 1034| 1.54k|} bitset_set_list_withcard: 1037| 1.35k| const uint16_t *list, uint64_t length) { 1038| 1.35k| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (1038:9): [True: 1.35k, False: 0] ------------------ 1039| 1.35k| return _asm_bitset_set_list_withcard(words, card, list, length); 1040| 1.35k| } else { 1041| 0| return _scalar_bitset_set_list_withcard(words, card, list, length); 1042| 0| } 1043| 1.35k|} bitset_set_list: 1045| 195|void bitset_set_list(uint64_t *words, const uint16_t *list, uint64_t length) { 1046| 195| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (1046:9): [True: 195, False: 0] ------------------ 1047| 195| _asm_bitset_set_list(words, list, length); 1048| 195| } else { 1049| 0| _scalar_bitset_set_list(words, list, length); 1050| 0| } 1051| 195|} bitset_flip_list_withcard: 1107| 1.34k| const uint16_t *list, uint64_t length) { 1108| 1.34k| uint64_t offset, load, newload, pos, index; 1109| 1.34k| const uint16_t *end = list + length; 1110| 2.33M| while (list != end) { ------------------ | Branch (1110:12): [True: 2.33M, False: 1.34k] ------------------ 1111| 2.33M| pos = *list; 1112| 2.33M| offset = pos >> 6; 1113| 2.33M| index = pos % 64; 1114| 2.33M| load = words[offset]; 1115| 2.33M| newload = load ^ (UINT64_C(1) << index); 1116| | // todo: is a branch here all that bad? 1117| 2.33M| card += 1118| | (1 - 2 * (((UINT64_C(1) << index) & load) >> index)); // +1 or -1 1119| 2.33M| words[offset] = newload; 1120| 2.33M| list++; 1121| 2.33M| } 1122| 1.34k| return card; 1123| 1.34k|} bitset_util.c:_asm_bitset_clear_list: 949| 1.54k| uint64_t length) { 950| 1.54k| uint64_t offset, load, pos; 951| 1.54k| uint64_t shift = 6; 952| 1.54k| const uint16_t *end = list + length; 953| 1.54k| if (!length) return card; ------------------ | Branch (953:9): [True: 0, False: 1.54k] ------------------ 954| | // btr is not available as an intrinsic in GCC 955| 1.54k| __asm volatile( 956| 1.54k| "1:\n" 957| 1.54k| "movzwq (%[list]), %[pos]\n" 958| 1.54k| "shrx %[shift], %[pos], %[offset]\n" 959| 1.54k| "mov (%[words],%[offset],8), %[load]\n" 960| 1.54k| "btr %[pos], %[load]\n" 961| 1.54k| "mov %[load], (%[words],%[offset],8)\n" 962| 1.54k| "sbb $0, %[card]\n" 963| 1.54k| "add $2, %[list]\n" 964| 1.54k| "cmp %[list], %[end]\n" 965| 1.54k| "jnz 1b" 966| 1.54k| : [card] "+&r"(card), [list] "+&r"(list), [load] "=&r"(load), 967| 1.54k| [pos] "=&r"(pos), [offset] "=&r"(offset) 968| 1.54k| : [end] "r"(end), [words] "r"(words), [shift] "r"(shift) 969| 1.54k| : 970| 1.54k| /* clobbers */ "memory"); 971| 1.54k| return card; 972| 1.54k|} bitset_util.c:_asm_bitset_set_list_withcard: 867| 1.35k| uint64_t length) { 868| 1.35k| uint64_t offset, load, pos; 869| 1.35k| uint64_t shift = 6; 870| 1.35k| const uint16_t *end = list + length; 871| 1.35k| if (!length) return card; ------------------ | Branch (871:9): [True: 0, False: 1.35k] ------------------ 872| | // TODO: could unroll for performance, see bitset_set_list 873| | // bts is not available as an intrinsic in GCC 874| 1.35k| __asm volatile( 875| 1.35k| "1:\n" 876| 1.35k| "movzwq (%[list]), %[pos]\n" 877| 1.35k| "shrx %[shift], %[pos], %[offset]\n" 878| 1.35k| "mov (%[words],%[offset],8), %[load]\n" 879| 1.35k| "bts %[pos], %[load]\n" 880| 1.35k| "mov %[load], (%[words],%[offset],8)\n" 881| 1.35k| "sbb $-1, %[card]\n" 882| 1.35k| "add $2, %[list]\n" 883| 1.35k| "cmp %[list], %[end]\n" 884| 1.35k| "jnz 1b" 885| 1.35k| : [card] "+&r"(card), [list] "+&r"(list), [load] "=&r"(load), 886| 1.35k| [pos] "=&r"(pos), [offset] "=&r"(offset) 887| 1.35k| : [end] "r"(end), [words] "r"(words), [shift] "r"(shift)); 888| 1.35k| return card; 889| 1.35k|} bitset_util.c:_asm_bitset_set_list: 892| 195| uint64_t length) { 893| 195| uint64_t pos; 894| 195| const uint16_t *end = list + length; 895| | 896| 195| uint64_t shift = 6; 897| 195| uint64_t offset; 898| 195| uint64_t load; 899| 5.58k| for (; list + 3 < end; list += 4) { ------------------ | Branch (899:12): [True: 5.38k, False: 195] ------------------ 900| 5.38k| pos = list[0]; 901| 5.38k| __asm volatile( 902| 5.38k| "shrx %[shift], %[pos], %[offset]\n" 903| 5.38k| "mov (%[words],%[offset],8), %[load]\n" 904| 5.38k| "bts %[pos], %[load]\n" 905| 5.38k| "mov %[load], (%[words],%[offset],8)" 906| 5.38k| : [load] "=&r"(load), [offset] "=&r"(offset) 907| 5.38k| : [words] "r"(words), [shift] "r"(shift), [pos] "r"(pos)); 908| 5.38k| pos = list[1]; 909| 5.38k| __asm volatile( 910| 5.38k| "shrx %[shift], %[pos], %[offset]\n" 911| 5.38k| "mov (%[words],%[offset],8), %[load]\n" 912| 5.38k| "bts %[pos], %[load]\n" 913| 5.38k| "mov %[load], (%[words],%[offset],8)" 914| 5.38k| : [load] "=&r"(load), [offset] "=&r"(offset) 915| 5.38k| : [words] "r"(words), [shift] "r"(shift), [pos] "r"(pos)); 916| 5.38k| pos = list[2]; 917| 5.38k| __asm volatile( 918| 5.38k| "shrx %[shift], %[pos], %[offset]\n" 919| 5.38k| "mov (%[words],%[offset],8), %[load]\n" 920| 5.38k| "bts %[pos], %[load]\n" 921| 5.38k| "mov %[load], (%[words],%[offset],8)" 922| 5.38k| : [load] "=&r"(load), [offset] "=&r"(offset) 923| 5.38k| : [words] "r"(words), [shift] "r"(shift), [pos] "r"(pos)); 924| 5.38k| pos = list[3]; 925| 5.38k| __asm volatile( 926| 5.38k| "shrx %[shift], %[pos], %[offset]\n" 927| 5.38k| "mov (%[words],%[offset],8), %[load]\n" 928| 5.38k| "bts %[pos], %[load]\n" 929| 5.38k| "mov %[load], (%[words],%[offset],8)" 930| 5.38k| : [load] "=&r"(load), [offset] "=&r"(offset) 931| 5.38k| : [words] "r"(words), [shift] "r"(shift), [pos] "r"(pos)); 932| 5.38k| } 933| | 934| 493| while (list != end) { ------------------ | Branch (934:12): [True: 298, False: 195] ------------------ 935| 298| pos = list[0]; 936| 298| __asm volatile( 937| 298| "shrx %[shift], %[pos], %[offset]\n" 938| 298| "mov (%[words],%[offset],8), %[load]\n" 939| 298| "bts %[pos], %[load]\n" 940| 298| "mov %[load], (%[words],%[offset],8)" 941| 298| : [load] "=&r"(load), [offset] "=&r"(offset) 942| 298| : [words] "r"(words), [shift] "r"(shift), [pos] "r"(pos)); 943| 298| list++; 944| 298| } 945| 195|} array_container_create_given_capacity: 49| 139k|array_container_t *array_container_create_given_capacity(int32_t size) { 50| 139k| array_container_t *container; 51| | 52| 139k| if ((container = (array_container_t *)roaring_malloc( ------------------ | Branch (52:9): [True: 0, False: 139k] ------------------ 53| 139k| sizeof(array_container_t))) == NULL) { 54| 0| return NULL; 55| 0| } 56| | 57| 139k| if (size <= 0) { // we don't want to rely on malloc(0) ------------------ | Branch (57:9): [True: 85.4k, False: 54.4k] ------------------ 58| 85.4k| container->array = NULL; 59| 85.4k| } else if ((container->array = (uint16_t *)roaring_malloc(sizeof(uint16_t) * ------------------ | Branch (59:16): [True: 0, False: 54.4k] ------------------ 60| 54.4k| size)) == NULL) { 61| 0| roaring_free(container); 62| 0| return NULL; 63| 0| } 64| | 65| 139k| container->capacity = size; 66| 139k| container->cardinality = 0; 67| | 68| 139k| return container; 69| 139k|} array_container_create: 72| 25.0k|array_container_t *array_container_create(void) { 73| 25.0k| return array_container_create_given_capacity(ARRAY_DEFAULT_INIT_SIZE); 74| 25.0k|} array_container_create_range: 77| 150|array_container_t *array_container_create_range(uint32_t min, uint32_t max) { 78| 150| array_container_t *answer = 79| 150| array_container_create_given_capacity(max - min + 1); 80| 150| if (answer == NULL) return answer; ------------------ | Branch (80:9): [True: 0, False: 150] ------------------ 81| 150| answer->cardinality = 0; 82| 300| for (uint32_t k = min; k < max; k++) { ------------------ | Branch (82:28): [True: 150, False: 150] ------------------ 83| 150| answer->array[answer->cardinality++] = k; 84| 150| } 85| 150| return answer; 86| 150|} array_container_clone: 90| 22.1k|array_container_t *array_container_clone(const array_container_t *src) { 91| 22.1k| array_container_t *newcontainer = 92| 22.1k| array_container_create_given_capacity(src->capacity); 93| 22.1k| if (newcontainer == NULL) return NULL; ------------------ | Branch (93:9): [True: 0, False: 22.1k] ------------------ 94| | 95| 22.1k| newcontainer->cardinality = src->cardinality; 96| | 97| 22.1k| memcpy(newcontainer->array, src->array, 98| 22.1k| src->cardinality * sizeof(uint16_t)); 99| | 100| 22.1k| return newcontainer; 101| 22.1k|} array_container_shrink_to_fit: 131| 3.63k|int array_container_shrink_to_fit(array_container_t *src) { 132| 3.63k| if (src->cardinality == src->capacity) return 0; // nothing to do ------------------ | Branch (132:9): [True: 475, False: 3.16k] ------------------ 133| 3.16k| int savings = src->capacity - src->cardinality; 134| 3.16k| src->capacity = src->cardinality; 135| 3.16k| if (src->capacity == ------------------ | Branch (135:9): [True: 0, False: 3.16k] ------------------ 136| 3.16k| 0) { // we do not want to rely on realloc for zero allocs 137| 0| roaring_free(src->array); 138| 0| src->array = NULL; 139| 3.16k| } else { 140| 3.16k| uint16_t *oldarray = src->array; 141| 3.16k| src->array = (uint16_t *)roaring_realloc( 142| 3.16k| oldarray, src->capacity * sizeof(uint16_t)); 143| 3.16k| if (src->array == NULL) roaring_free(oldarray); // should never happen? ------------------ | Branch (143:13): [True: 0, False: 3.16k] ------------------ 144| 3.16k| } 145| 3.16k| return savings; 146| 3.63k|} array_container_free: 149| 139k|void array_container_free(array_container_t *arr) { 150| 139k| if (arr == NULL) return; ------------------ | Branch (150:9): [True: 0, False: 139k] ------------------ 151| 139k| roaring_free(arr->array); 152| 139k| roaring_free(arr); 153| 139k|} array_container_grow: 167| 114k| bool preserve) { 168| 114k| int32_t max = (min <= DEFAULT_MAX_SIZE ? DEFAULT_MAX_SIZE : 65536); ------------------ | Branch (168:20): [True: 114k, False: 0] ------------------ 169| 114k| int32_t new_capacity = clamp(grow_capacity(container->capacity), min, max); 170| | 171| 114k| container->capacity = new_capacity; 172| 114k| uint16_t *array = container->array; 173| | 174| 114k| if (preserve) { ------------------ | Branch (174:9): [True: 105k, False: 9.59k] ------------------ 175| 105k| container->array = 176| 105k| (uint16_t *)roaring_realloc(array, new_capacity * sizeof(uint16_t)); 177| 105k| if (container->array == NULL) roaring_free(array); ------------------ | Branch (177:13): [True: 0, False: 105k] ------------------ 178| 105k| } else { 179| 9.59k| roaring_free(array); 180| 9.59k| container->array = 181| 9.59k| (uint16_t *)roaring_malloc(new_capacity * sizeof(uint16_t)); 182| 9.59k| } 183| | 184| | // if realloc fails, we have container->array == NULL. 185| 114k|} array_container_copy: 189| 34| array_container_t *dst) { 190| 34| const int32_t cardinality = src->cardinality; 191| 34| if (cardinality > dst->capacity) { ------------------ | Branch (191:9): [True: 34, False: 0] ------------------ 192| 34| array_container_grow(dst, cardinality, false); 193| 34| } 194| | 195| 34| dst->cardinality = cardinality; 196| 34| memcpy(dst->array, src->array, cardinality * sizeof(uint16_t)); 197| 34|} array_container_union: 211| 4.52k| array_container_t *out) { 212| 4.52k| const int32_t card_1 = array_1->cardinality, card_2 = array_2->cardinality; 213| 4.52k| const int32_t max_cardinality = card_1 + card_2; 214| | 215| 4.52k| if (out->capacity < max_cardinality) { ------------------ | Branch (215:9): [True: 0, False: 4.52k] ------------------ 216| | array_container_grow(out, max_cardinality, false); 217| 0| } 218| 4.52k| out->cardinality = (int32_t)fast_union_uint16( 219| 4.52k| array_1->array, card_1, array_2->array, card_2, out->array); 220| 4.52k|} array_container_andnot: 228| 5.70k| array_container_t *out) { 229| 5.70k| if (out->capacity < array_1->cardinality) ------------------ | Branch (229:9): [True: 2.27k, False: 3.43k] ------------------ 230| 2.27k| array_container_grow(out, array_1->cardinality, false); 231| 5.70k|#if CROARING_IS_X64 232| 5.70k| if ((croaring_hardware_support() & ROARING_SUPPORTS_AVX2) && ------------------ | Branch (232:9): [True: 5.70k, False: 0] ------------------ 233| 5.70k| (out != array_1) && (out != array_2)) { ------------------ | Branch (233:9): [True: 2.27k, False: 3.43k] | Branch (233:29): [True: 2.27k, False: 0] ------------------ 234| 2.27k| out->cardinality = difference_vector16( 235| 2.27k| array_1->array, array_1->cardinality, array_2->array, 236| 2.27k| array_2->cardinality, out->array); 237| 3.43k| } else { 238| 3.43k| out->cardinality = 239| 3.43k| difference_uint16(array_1->array, array_1->cardinality, 240| 3.43k| array_2->array, array_2->cardinality, out->array); 241| 3.43k| } 242| |#else 243| | out->cardinality = 244| | difference_uint16(array_1->array, array_1->cardinality, array_2->array, 245| | array_2->cardinality, out->array); 246| |#endif 247| 5.70k|} array_container_xor: 256| 7.70k| array_container_t *out) { 257| 7.70k| const int32_t card_1 = array_1->cardinality, card_2 = array_2->cardinality; 258| 7.70k| const int32_t max_cardinality = card_1 + card_2; 259| 7.70k| if (out->capacity < max_cardinality) { ------------------ | Branch (259:9): [True: 0, False: 7.70k] ------------------ 260| 0| array_container_grow(out, max_cardinality, false); 261| 0| } 262| | 263| 7.70k|#if CROARING_IS_X64 264| 7.70k| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (264:9): [True: 7.70k, False: 0] ------------------ 265| 7.70k| out->cardinality = 266| 7.70k| xor_vector16(array_1->array, array_1->cardinality, array_2->array, 267| 7.70k| array_2->cardinality, out->array); 268| 7.70k| } else { 269| 0| out->cardinality = 270| 0| xor_uint16(array_1->array, array_1->cardinality, array_2->array, 271| 0| array_2->cardinality, out->array); 272| 0| } 273| |#else 274| | out->cardinality = 275| | xor_uint16(array_1->array, array_1->cardinality, array_2->array, 276| | array_2->cardinality, out->array); 277| |#endif 278| 7.70k|} array_container_intersection: 290| 2.27k| array_container_t *out) { 291| 2.27k| int32_t card_1 = array1->cardinality, card_2 = array2->cardinality, 292| 2.27k| min_card = minimum_int32(card_1, card_2); 293| 2.27k| const int threshold = 64; // subject to tuning 294| 2.27k|#if CROARING_IS_X64 295| 2.27k| if (out->capacity < min_card) { ------------------ | Branch (295:9): [True: 2.27k, False: 0] ------------------ 296| 2.27k| array_container_grow(out, min_card + sizeof(__m128i) / sizeof(uint16_t), 297| 2.27k| false); 298| 2.27k| } 299| |#else 300| | if (out->capacity < min_card) { 301| | array_container_grow(out, min_card, false); 302| | } 303| |#endif 304| | 305| 2.27k| if (card_1 * threshold < card_2) { ------------------ | Branch (305:9): [True: 324, False: 1.94k] ------------------ 306| 324| out->cardinality = intersect_skewed_uint16( 307| 324| array1->array, card_1, array2->array, card_2, out->array); 308| 1.94k| } else if (card_2 * threshold < card_1) { ------------------ | Branch (308:16): [True: 202, False: 1.74k] ------------------ 309| 202| out->cardinality = intersect_skewed_uint16( 310| 202| array2->array, card_2, array1->array, card_1, out->array); 311| 1.74k| } else { 312| 1.74k|#if CROARING_IS_X64 313| 1.74k| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (313:13): [True: 1.74k, False: 0] ------------------ 314| 1.74k| out->cardinality = intersect_vector16( 315| 1.74k| array1->array, card_1, array2->array, card_2, out->array); 316| 1.74k| } else { 317| 0| out->cardinality = intersect_uint16( 318| 0| array1->array, card_1, array2->array, card_2, out->array); 319| 0| } 320| |#else 321| | out->cardinality = intersect_uint16(array1->array, card_1, 322| | array2->array, card_2, out->array); 323| |#endif 324| 1.74k| } 325| 2.27k|} array_container_intersection_cardinality: 330| 9.08k| const array_container_t *array2) { 331| 9.08k| int32_t card_1 = array1->cardinality, card_2 = array2->cardinality; 332| 9.08k| const int threshold = 64; // subject to tuning 333| 9.08k| if (card_1 * threshold < card_2) { ------------------ | Branch (333:9): [True: 1.29k, False: 7.78k] ------------------ 334| 1.29k| return intersect_skewed_uint16_cardinality(array1->array, card_1, 335| 1.29k| array2->array, card_2); 336| 7.78k| } else if (card_2 * threshold < card_1) { ------------------ | Branch (336:16): [True: 808, False: 6.98k] ------------------ 337| 808| return intersect_skewed_uint16_cardinality(array2->array, card_2, 338| 808| array1->array, card_1); 339| 6.98k| } else { 340| 6.98k|#if CROARING_IS_X64 341| 6.98k| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (341:13): [True: 6.98k, False: 0] ------------------ 342| 6.98k| return intersect_vector16_cardinality(array1->array, card_1, 343| 6.98k| array2->array, card_2); 344| 6.98k| } else { 345| 0| return intersect_uint16_cardinality(array1->array, card_1, 346| 0| array2->array, card_2); 347| 0| } 348| |#else 349| | return intersect_uint16_cardinality(array1->array, card_1, 350| | array2->array, card_2); 351| |#endif 352| 6.98k| } 353| 9.08k|} array_container_intersect: 356| 2.27k| const array_container_t *array2) { 357| 2.27k| int32_t card_1 = array1->cardinality, card_2 = array2->cardinality; 358| 2.27k| const int threshold = 64; // subject to tuning 359| 2.27k| if (card_1 * threshold < card_2) { ------------------ | Branch (359:9): [True: 324, False: 1.94k] ------------------ 360| 324| return intersect_skewed_uint16_nonempty(array1->array, card_1, 361| 324| array2->array, card_2); 362| 1.94k| } else if (card_2 * threshold < card_1) { ------------------ | Branch (362:16): [True: 202, False: 1.74k] ------------------ 363| 202| return intersect_skewed_uint16_nonempty(array2->array, card_2, 364| 202| array1->array, card_1); 365| 1.74k| } else { 366| | // we do not bother vectorizing 367| 1.74k| return intersect_uint16_nonempty(array1->array, card_1, array2->array, 368| 1.74k| card_2); 369| 1.74k| } 370| 2.27k|} array_container_intersection_inplace: 376| 4.64k| const array_container_t *src_2) { 377| 4.64k| int32_t card_1 = src_1->cardinality, card_2 = src_2->cardinality; 378| 4.64k| const int threshold = 64; // subject to tuning 379| 4.64k| if (card_1 * threshold < card_2) { ------------------ | Branch (379:9): [True: 0, False: 4.64k] ------------------ 380| 0| src_1->cardinality = intersect_skewed_uint16( 381| 0| src_1->array, card_1, src_2->array, card_2, src_1->array); 382| 4.64k| } else if (card_2 * threshold < card_1) { ------------------ | Branch (382:16): [True: 209, False: 4.43k] ------------------ 383| 209| src_1->cardinality = intersect_skewed_uint16( 384| 209| src_2->array, card_2, src_1->array, card_1, src_1->array); 385| 4.43k| } else { 386| 4.43k|#if CROARING_IS_X64 387| 4.43k| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (387:13): [True: 4.43k, False: 0] ------------------ 388| 4.43k| src_1->cardinality = intersect_vector16_inplace( 389| 4.43k| src_1->array, card_1, src_2->array, card_2); 390| 4.43k| } else { 391| 0| src_1->cardinality = intersect_uint16( 392| 0| src_1->array, card_1, src_2->array, card_2, src_1->array); 393| 0| } 394| |#else 395| | src_1->cardinality = intersect_uint16( 396| | src_1->array, card_1, src_2->array, card_2, src_1->array); 397| |#endif 398| 4.43k| } 399| 4.64k|} array_container_to_uint32_array: 403| 5.70k| uint32_t base) { 404| 5.70k|#if CROARING_IS_X64 405| 5.70k| int support = croaring_hardware_support(); 406| 5.70k|#if CROARING_COMPILER_SUPPORTS_AVX512 407| 5.70k| if (support & ROARING_SUPPORTS_AVX512) { ------------------ | Branch (407:9): [True: 0, False: 5.70k] ------------------ 408| 0| return avx512_array_container_to_uint32_array(vout, cont->array, 409| 0| cont->cardinality, base); 410| 0| } 411| 5.70k|#endif 412| 5.70k| if (support & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (412:9): [True: 5.70k, False: 0] ------------------ 413| 5.70k| return array_container_to_uint32_array_vector16( 414| 5.70k| vout, cont->array, cont->cardinality, base); 415| 5.70k| } 416| 0|#endif // CROARING_IS_X64 417| 0| int outpos = 0; 418| 0| uint32_t *out = (uint32_t *)vout; 419| 0| size_t i = 0; 420| 0| for (; i < (size_t)cont->cardinality; ++i) { ------------------ | Branch (420:12): [True: 0, False: 0] ------------------ 421| 0| const uint32_t val = base + cont->array[i]; 422| 0| memcpy(out + outpos, &val, 423| 0| sizeof(uint32_t)); // should be compiled as a MOV on x64 424| 0| outpos++; 425| 0| } 426| 0| return outpos; 427| 5.70k|} array_container_number_of_runs: 495| 12.3k|int32_t array_container_number_of_runs(const array_container_t *ac) { 496| | // Can SIMD work here? 497| 12.3k| int32_t nr_runs = 0; 498| 12.3k| int32_t prev = -2; 499| 1.54M| for (const uint16_t *p = ac->array; p != ac->array + ac->cardinality; ++p) { ------------------ | Branch (499:41): [True: 1.53M, False: 12.3k] ------------------ 500| 1.53M| if (*p != prev + 1) nr_runs++; ------------------ | Branch (500:13): [True: 835k, False: 699k] ------------------ 501| 1.53M| prev = *p; 502| 1.53M| } 503| 12.3k| return nr_runs; 504| 12.3k|} array_container_write: 512| 6.91k|int32_t array_container_write(const array_container_t *container, char *buf) { 513| |#if CROARING_IS_BIG_ENDIAN 514| | for (int32_t i = 0; i < container->cardinality; ++i) { 515| | uint16_t v_le = croaring_htole16(container->array[i]); 516| | memcpy(buf + i * sizeof(uint16_t), &v_le, sizeof(uint16_t)); 517| | } 518| |#else 519| 6.91k| memcpy(buf, container->array, container->cardinality * sizeof(uint16_t)); 520| 6.91k|#endif 521| 6.91k| return array_container_size_in_bytes(container); 522| 6.91k|} array_container_is_subset: 525| 5.54k| const array_container_t *container2) { 526| 5.54k| if (container1->cardinality > container2->cardinality) { ------------------ | Branch (526:9): [True: 620, False: 4.92k] ------------------ 527| 620| return false; 528| 620| } 529| 4.92k| int i1 = 0, i2 = 0; 530| 46.5k| while (i1 < container1->cardinality && i2 < container2->cardinality) { ------------------ | Branch (530:12): [True: 46.5k, False: 0] | Branch (530:44): [True: 46.5k, False: 25] ------------------ 531| 46.5k| if (container1->array[i1] == container2->array[i2]) { ------------------ | Branch (531:13): [True: 0, False: 46.5k] ------------------ 532| 0| i1++; 533| 0| i2++; 534| 46.5k| } else if (container1->array[i1] > container2->array[i2]) { ------------------ | Branch (534:20): [True: 41.6k, False: 4.90k] ------------------ 535| 41.6k| i2++; 536| 41.6k| } else { // container1->array[i1] < container2->array[i2] 537| 4.90k| return false; 538| 4.90k| } 539| 46.5k| } 540| 25| if (i1 == container1->cardinality) { ------------------ | Branch (540:9): [True: 0, False: 25] ------------------ 541| 0| return true; 542| 25| } else { 543| | return false; 544| 25| } 545| 25|} array_container_read: 548| 6.95k| const char *buf) { 549| 6.95k| if (container->capacity < cardinality) { ------------------ | Branch (549:9): [True: 0, False: 6.95k] ------------------ 550| 0| array_container_grow(container, cardinality, false); 551| 0| } 552| 6.95k| container->cardinality = cardinality; 553| |#if CROARING_IS_BIG_ENDIAN 554| | for (int32_t i = 0; i < cardinality; ++i) { 555| | uint16_t v_le; 556| | memcpy(&v_le, buf + i * sizeof(uint16_t), sizeof(uint16_t)); 557| | container->array[i] = croaring_letoh16(v_le); 558| | } 559| |#else 560| 6.95k| memcpy(container->array, buf, container->cardinality * sizeof(uint16_t)); 561| 6.95k|#endif 562| | 563| 6.95k| return array_container_size_in_bytes(container); 564| 6.95k|} array_container_iterate: 567| 6.25k| roaring_iterator iterator, void *ptr) { 568| 1.88M| for (int i = 0; i < cont->cardinality; i++) ------------------ | Branch (568:21): [True: 1.87M, False: 6.25k] ------------------ 569| 1.87M| if (!iterator(cont->array[i] + base, ptr)) return false; ------------------ | Branch (569:13): [True: 0, False: 1.87M] ------------------ 570| 6.25k| return true; 571| 6.25k|} array.c:clamp: 162| 114k|static inline int32_t clamp(int32_t val, int32_t min, int32_t max) { 163| 114k| return ((val < min) ? min : (val > max) ? max : val); ------------------ | Branch (163:13): [True: 25.6k, False: 89.1k] | Branch (163:33): [True: 223, False: 88.9k] ------------------ 164| 114k|} array.c:grow_capacity: 155| 114k|static inline int32_t grow_capacity(int32_t capacity) { 156| 114k| return (capacity <= 0) ? ARRAY_DEFAULT_INIT_SIZE ------------------ | Branch (156:12): [True: 25.0k, False: 89.6k] ------------------ 157| 114k| : capacity < 64 ? capacity * 2 ------------------ | Branch (157:14): [True: 65.4k, False: 24.2k] ------------------ 158| 89.6k| : capacity < 1024 ? capacity * 3 / 2 ------------------ | Branch (158:14): [True: 24.0k, False: 264] ------------------ 159| 24.2k| : capacity * 5 / 4; 160| 114k|} array.c:minimum_int32: 280| 2.27k|static inline int32_t minimum_int32(int32_t a, int32_t b) { 281| 2.27k| return (a < b) ? a : b; ------------------ | Branch (281:12): [True: 951, False: 1.32k] ------------------ 282| 2.27k|} bitset_container_clear: 51| 138k|void bitset_container_clear(bitset_container_t *bitset) { 52| 138k| memset(bitset->words, 0, sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); 53| 138k| bitset->cardinality = 0; 54| 138k|} bitset_container_set_all: 56| 151|void bitset_container_set_all(bitset_container_t *bitset) { 57| | memset(bitset->words, INT64_C(-1), 58| 151| sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); 59| 151| bitset->cardinality = (1 << 16); 60| 151|} bitset_container_create: 63| 138k|bitset_container_t *bitset_container_create(void) { 64| 138k| bitset_container_t *bitset = 65| 138k| (bitset_container_t *)roaring_malloc(sizeof(bitset_container_t)); 66| | 67| 138k| if (!bitset) { ------------------ | Branch (67:9): [True: 0, False: 138k] ------------------ 68| 0| return NULL; 69| 0| } 70| | 71| 138k| size_t align_size = 32; 72| 138k|#if CROARING_IS_X64 73| 138k| int support = croaring_hardware_support(); 74| 138k| if (support & ROARING_SUPPORTS_AVX512) { ------------------ | Branch (74:9): [True: 0, False: 138k] ------------------ 75| | // sizeof(__m512i) == 64 76| 0| align_size = 64; 77| 138k| } else { 78| | // sizeof(__m256i) == 32 79| 138k| align_size = 32; 80| 138k| } 81| 138k|#endif 82| 138k| bitset->words = (uint64_t *)roaring_aligned_malloc( 83| 138k| align_size, sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); 84| 138k| if (!bitset->words) { ------------------ | Branch (84:9): [True: 0, False: 138k] ------------------ 85| 0| roaring_free(bitset); 86| 0| return NULL; 87| 0| } 88| 138k| bitset_container_clear(bitset); 89| 138k| return bitset; 90| 138k|} bitset_container_copy: 94| 3.69k| bitset_container_t *dest) { 95| 3.69k| dest->cardinality = source->cardinality; 96| 3.69k| memcpy(dest->words, source->words, 97| 3.69k| sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); 98| 3.69k|} bitset_container_free: 129| 272k|void bitset_container_free(bitset_container_t *bitset) { 130| 272k| if (bitset == NULL) return; ------------------ | Branch (130:9): [True: 0, False: 272k] ------------------ 131| 272k| roaring_aligned_free(bitset->words); 132| 272k| roaring_free(bitset); 133| 272k|} bitset_container_clone: 137| 134k|bitset_container_t *bitset_container_clone(const bitset_container_t *src) { 138| 134k| bitset_container_t *bitset = 139| 134k| (bitset_container_t *)roaring_malloc(sizeof(bitset_container_t)); 140| | 141| 134k| if (!bitset) { ------------------ | Branch (141:9): [True: 0, False: 134k] ------------------ 142| 0| return NULL; 143| 0| } 144| | 145| 134k| size_t align_size = 32; 146| 134k|#if CROARING_IS_X64 147| 134k| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX512) { ------------------ | Branch (147:9): [True: 0, False: 134k] ------------------ 148| | // sizeof(__m512i) == 64 149| 0| align_size = 64; 150| 134k| } else { 151| | // sizeof(__m256i) == 32 152| 134k| align_size = 32; 153| 134k| } 154| 134k|#endif 155| 134k| bitset->words = (uint64_t *)roaring_aligned_malloc( 156| 134k| align_size, sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); 157| 134k| if (!bitset->words) { ------------------ | Branch (157:9): [True: 0, False: 134k] ------------------ 158| 0| roaring_free(bitset); 159| 0| return NULL; 160| 0| } 161| 134k| bitset->cardinality = src->cardinality; 162| 134k| memcpy(bitset->words, src->words, 163| 134k| sizeof(uint64_t) * BITSET_CONTAINER_SIZE_IN_WORDS); 164| 134k| return bitset; 165| 134k|} bitset_container_compute_cardinality: 269| 3.19k|int bitset_container_compute_cardinality(const bitset_container_t *bitset) { 270| 3.19k| int support = croaring_hardware_support(); 271| 3.19k|#if CROARING_COMPILER_SUPPORTS_AVX512 272| 3.19k| if (support & ROARING_SUPPORTS_AVX512) { ------------------ | Branch (272:9): [True: 0, False: 3.19k] ------------------ 273| 0| return (int)avx512_vpopcount( 274| 0| (const __m512i *)bitset->words, 275| 0| BITSET_CONTAINER_SIZE_IN_WORDS / (WORDS_IN_AVX512_REG)); ------------------ | | 253| 0|#define WORDS_IN_AVX512_REG sizeof(__m512i) / sizeof(uint64_t) ------------------ 276| 0| } else 277| 3.19k|#endif // CROARING_COMPILER_SUPPORTS_AVX512 278| 3.19k| if (support & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (278:13): [True: 3.19k, False: 0] ------------------ 279| 3.19k| return (int)avx2_harley_seal_popcount256( 280| 3.19k| (const __m256i *)bitset->words, 281| 3.19k| BITSET_CONTAINER_SIZE_IN_WORDS / (CROARING_WORDS_IN_AVX2_REG)); ------------------ | | 250| 3.19k|#define CROARING_WORDS_IN_AVX2_REG sizeof(__m256i) / sizeof(uint64_t) ------------------ 282| 3.19k| } else { 283| 0| return _scalar_bitset_container_compute_cardinality(bitset); 284| 0| } 285| 3.19k|} bitset_container_and_nocard: 724| 1.39k| bitset_container_t *dst) { \ 725| 1.39k| int support = croaring_hardware_support(); \ 726| 1.39k| if ( support & ROARING_SUPPORTS_AVX512 ) { \ ------------------ | Branch (726:10): [True: 0, False: 1.39k] ------------------ 727| 0| return _avx512_bitset_container_##opname##_nocard(src_1, src_2, dst); \ 728| 0| } \ 729| 1.39k| else if ( support & ROARING_SUPPORTS_AVX2 ) { \ ------------------ | Branch (729:15): [True: 1.39k, False: 0] ------------------ 730| 1.39k| return _avx2_bitset_container_##opname##_nocard(src_1, src_2, dst); \ 731| 1.39k| } else { \ 732| 0| return _scalar_bitset_container_##opname##_nocard(src_1, src_2, dst); \ 733| 0| } \ 734| 1.39k| } \ bitset_container_and_justcard: 736| 1.39k| const bitset_container_t *src_2) { \ 737| 1.39k| int support = croaring_hardware_support(); \ 738| 1.39k| if ( support & ROARING_SUPPORTS_AVX512 ) { \ ------------------ | Branch (738:10): [True: 0, False: 1.39k] ------------------ 739| 0| return _avx512_bitset_container_##opname##_justcard(src_1, src_2); \ 740| 0| } \ 741| 1.39k| else if ( support & ROARING_SUPPORTS_AVX2 ) { \ ------------------ | Branch (741:15): [True: 1.39k, False: 0] ------------------ 742| 1.39k| return _avx2_bitset_container_##opname##_justcard(src_1, src_2); \ 743| 1.39k| } else { \ 744| 0| return _scalar_bitset_container_##opname##_justcard(src_1, src_2); \ 745| 0| } \ 746| 1.39k| } bitset_container_xor: 711| 1.43k| bitset_container_t *dst) { \ 712| 1.43k| int support = croaring_hardware_support(); \ 713| 1.43k| if ( support & ROARING_SUPPORTS_AVX512 ) { \ ------------------ | Branch (713:10): [True: 0, False: 1.43k] ------------------ 714| 0| return _avx512_bitset_container_##opname(src_1, src_2, dst); \ 715| 0| } \ 716| 1.43k| else if ( support & ROARING_SUPPORTS_AVX2 ) { \ ------------------ | Branch (716:15): [True: 1.43k, False: 0] ------------------ 717| 1.43k| return _avx2_bitset_container_##opname(src_1, src_2, dst); \ 718| 1.43k| } else { \ 719| 0| return _scalar_bitset_container_##opname(src_1, src_2, dst); \ 720| 0| } \ 721| 1.43k| } \ bitset_container_to_uint32_array: 936| 1.39k|){ 937| 1.39k|#if CROARING_IS_X64 938| 1.39k| int support = croaring_hardware_support(); 939| 1.39k|#if CROARING_COMPILER_SUPPORTS_AVX512 940| 1.39k| if(( support & ROARING_SUPPORTS_AVX512 ) && (bc->cardinality >= 8192)) // heuristic ------------------ | Branch (940:7): [True: 0, False: 1.39k] | Branch (940:49): [True: 0, False: 0] ------------------ 941| 0| return (int) bitset_extract_setbits_avx512(bc->words, 942| 0| BITSET_CONTAINER_SIZE_IN_WORDS, out, bc->cardinality, base); 943| 1.39k| else 944| 1.39k|#endif 945| 1.39k| if(( support & ROARING_SUPPORTS_AVX2 ) && (bc->cardinality >= 8192)) // heuristic ------------------ | Branch (945:7): [True: 1.39k, False: 0] | Branch (945:47): [True: 932, False: 465] ------------------ 946| 932| return (int) bitset_extract_setbits_avx2(bc->words, 947| 932| BITSET_CONTAINER_SIZE_IN_WORDS, out, bc->cardinality, base); 948| 465| else 949| 465| return (int) bitset_extract_setbits(bc->words, 950| 465| BITSET_CONTAINER_SIZE_IN_WORDS, out, base); 951| |#else 952| | return (int) bitset_extract_setbits(bc->words, 953| | BITSET_CONTAINER_SIZE_IN_WORDS, out, base); 954| |#endif 955| 1.39k|} bitset_container_write: 1050| 66.7k| char *buf) { 1051| |#if CROARING_IS_BIG_ENDIAN 1052| | for (int32_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i) { 1053| | uint64_t w_le = croaring_htole64(container->words[i]); 1054| | memcpy(buf + i * sizeof(uint64_t), &w_le, sizeof(uint64_t)); 1055| | } 1056| |#else 1057| 66.7k| memcpy(buf, container->words, BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t)); 1058| 66.7k|#endif 1059| 66.7k| return bitset_container_size_in_bytes(container); 1060| 66.7k|} bitset_container_read: 1064| 66.7k| const char *buf) { 1065| 66.7k| container->cardinality = cardinality; 1066| |#if CROARING_IS_BIG_ENDIAN 1067| | for (int32_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i) { 1068| | uint64_t w_le; 1069| | memcpy(&w_le, buf + i * sizeof(uint64_t), sizeof(uint64_t)); 1070| | container->words[i] = croaring_letoh64(w_le); 1071| | } 1072| |#else 1073| 66.7k| memcpy(container->words, buf, BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t)); 1074| 66.7k|#endif 1075| 66.7k| return bitset_container_size_in_bytes(container); 1076| 66.7k|} bitset_container_iterate: 1078| 1.53k|bool bitset_container_iterate(const bitset_container_t *cont, uint32_t base, roaring_iterator iterator, void *ptr) { 1079| 1.57M| for (int32_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i ) { ------------------ | Branch (1079:23): [True: 1.57M, False: 1.53k] ------------------ 1080| 1.57M| uint64_t w = cont->words[i]; 1081| 56.7M| while (w != 0) { ------------------ | Branch (1081:12): [True: 55.1M, False: 1.57M] ------------------ 1082| 55.1M| uint64_t t = w & (~w + 1); 1083| 55.1M| int r = roaring_trailing_zeroes(w); 1084| 55.1M| if(!iterator(r + base, ptr)) return false; ------------------ | Branch (1084:10): [True: 0, False: 55.1M] ------------------ 1085| 55.1M| w ^= t; 1086| 55.1M| } 1087| 1.57M| base += 64; 1088| 1.57M| } 1089| 1.53k| return true; 1090| 1.53k|} bitset_container_equals: 1144| 66.7k|bool bitset_container_equals(const bitset_container_t *container1, const bitset_container_t *container2) { 1145| 66.7k| if((container1->cardinality != BITSET_UNKNOWN_CARDINALITY) && (container2->cardinality != BITSET_UNKNOWN_CARDINALITY)) { ------------------ | Branch (1145:6): [True: 66.7k, False: 0] | Branch (1145:65): [True: 66.7k, False: 0] ------------------ 1146| 66.7k| if(container1->cardinality != container2->cardinality) { ------------------ | Branch (1146:8): [True: 0, False: 66.7k] ------------------ 1147| 0| return false; 1148| 0| } 1149| 66.7k| if (container1->cardinality == INT32_C(0x10000)) { ------------------ | Branch (1149:9): [True: 62.9k, False: 3.77k] ------------------ 1150| 62.9k| return true; 1151| 62.9k| } 1152| 66.7k| } 1153| 3.77k|#if CROARING_IS_X64 1154| 3.77k| int support = croaring_hardware_support(); 1155| 3.77k|#if CROARING_COMPILER_SUPPORTS_AVX512 1156| 3.77k| if( support & ROARING_SUPPORTS_AVX512 ) { ------------------ | Branch (1156:7): [True: 0, False: 3.77k] ------------------ 1157| 0| return _avx512_bitset_container_equals(container1, container2); 1158| 0| } 1159| 3.77k| else 1160| 3.77k|#endif 1161| 3.77k| if( support & ROARING_SUPPORTS_AVX2 ) { ------------------ | Branch (1161:7): [True: 3.77k, False: 0] ------------------ 1162| 3.77k| return _avx2_bitset_container_equals(container1, container2); 1163| 3.77k| } 1164| 0|#endif 1165| 0| return memcmp(container1->words, 1166| 0| container2->words, 1167| 0| BITSET_CONTAINER_SIZE_IN_WORDS*sizeof(uint64_t)) == 0; 1168| 3.77k|} bitset_container_minimum: 1218| 1.16k|uint16_t bitset_container_minimum(const bitset_container_t *container) { 1219| 19.0k| for (int32_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i ) { ------------------ | Branch (1219:23): [True: 19.0k, False: 0] ------------------ 1220| 19.0k| uint64_t w = container->words[i]; 1221| 19.0k| if (w != 0) { ------------------ | Branch (1221:9): [True: 1.16k, False: 17.8k] ------------------ 1222| 1.16k| int r = roaring_trailing_zeroes(w); 1223| 1.16k| return r + i * 64; 1224| 1.16k| } 1225| 19.0k| } 1226| 0| return UINT16_MAX; 1227| 1.16k|} bitset_container_maximum: 1230| 4.03k|uint16_t bitset_container_maximum(const bitset_container_t *container) { 1231| 1.12M| for (int32_t i = BITSET_CONTAINER_SIZE_IN_WORDS - 1; i > 0; --i ) { ------------------ | Branch (1231:56): [True: 1.12M, False: 0] ------------------ 1232| 1.12M| uint64_t w = container->words[i]; 1233| 1.12M| if (w != 0) { ------------------ | Branch (1233:9): [True: 4.03k, False: 1.12M] ------------------ 1234| 4.03k| int r = roaring_leading_zeroes(w); 1235| 4.03k| return i * 64 + 63 - r; 1236| 4.03k| } 1237| 1.12M| } 1238| 0| return 0; 1239| 4.03k|} bitset_container_index_equalorlarger: 1299| 1.73k|int bitset_container_index_equalorlarger(const bitset_container_t *container, uint16_t x) { 1300| 1.73k| uint32_t x32 = x; 1301| 1.73k| uint32_t k = x32 / 64; 1302| 1.73k| uint64_t word = container->words[k]; 1303| 1.73k| const int diff = x32 - k * 64; // in [0,64) 1304| 1.73k| word = (word >> diff) << diff; // a mask is faster, but we don't care 1305| 7.83k| while(word == 0) { ------------------ | Branch (1305:9): [True: 6.10k, False: 1.73k] ------------------ 1306| 6.10k| k++; 1307| 6.10k| if(k == BITSET_CONTAINER_SIZE_IN_WORDS) return -1; ------------------ | Branch (1307:8): [True: 0, False: 6.10k] ------------------ 1308| 6.10k| word = container->words[k]; 1309| 6.10k| } 1310| 1.73k| return k * 64 + roaring_trailing_zeroes(word); 1311| 1.73k|} bitset.c:_avx2_bitset_container_and_nocard: 503| 1.39k| bitset_container_t *dst) { \ 504| 1.39k| const uint8_t *__restrict__ words_1 = (const uint8_t *)src_1->words; \ 505| 1.39k| const uint8_t *__restrict__ words_2 = (const uint8_t *)src_2->words; \ 506| 1.39k| /* not using the blocking optimization for some reason*/ \ 507| 1.39k| uint8_t *out = (uint8_t *)dst->words; \ 508| 1.39k| const int innerloop = 8; \ 509| 1.39k| for (size_t i = 0; \ 510| 46.1k| i < BITSET_CONTAINER_SIZE_IN_WORDS / (CROARING_WORDS_IN_AVX2_REG); \ ------------------ | | 250| 46.1k|#define CROARING_WORDS_IN_AVX2_REG sizeof(__m256i) / sizeof(uint64_t) ------------------ | Branch (510:10): [True: 44.7k, False: 1.39k] ------------------ 511| 44.7k| i += innerloop) { \ 512| 44.7k| __m256i A1, A2, AO; \ 513| 44.7k| A1 = _mm256_lddqu_si256((const __m256i *)(words_1)); \ 514| 44.7k| A2 = _mm256_lddqu_si256((const __m256i *)(words_2)); \ 515| 44.7k| AO = avx_intrinsic(A2, A1); \ 516| 44.7k| _mm256_storeu_si256((__m256i *)out, AO); \ 517| 44.7k| A1 = _mm256_lddqu_si256((const __m256i *)(words_1 + 32)); \ 518| 44.7k| A2 = _mm256_lddqu_si256((const __m256i *)(words_2 + 32)); \ 519| 44.7k| AO = avx_intrinsic(A2, A1); \ 520| 44.7k| _mm256_storeu_si256((__m256i *)(out + 32), AO); \ 521| 44.7k| A1 = _mm256_lddqu_si256((const __m256i *)(words_1 + 64)); \ 522| 44.7k| A2 = _mm256_lddqu_si256((const __m256i *)(words_2 + 64)); \ 523| 44.7k| AO = avx_intrinsic(A2, A1); \ 524| 44.7k| _mm256_storeu_si256((__m256i *)(out + 64), AO); \ 525| 44.7k| A1 = _mm256_lddqu_si256((const __m256i *)(words_1 + 96)); \ 526| 44.7k| A2 = _mm256_lddqu_si256((const __m256i *)(words_2 + 96)); \ 527| 44.7k| AO = avx_intrinsic(A2, A1); \ 528| 44.7k| _mm256_storeu_si256((__m256i *)(out + 96), AO); \ 529| 44.7k| A1 = _mm256_lddqu_si256((const __m256i *)(words_1 + 128)); \ 530| 44.7k| A2 = _mm256_lddqu_si256((const __m256i *)(words_2 + 128)); \ 531| 44.7k| AO = avx_intrinsic(A2, A1); \ 532| 44.7k| _mm256_storeu_si256((__m256i *)(out + 128), AO); \ 533| 44.7k| A1 = _mm256_lddqu_si256((const __m256i *)(words_1 + 160)); \ 534| 44.7k| A2 = _mm256_lddqu_si256((const __m256i *)(words_2 + 160)); \ 535| 44.7k| AO = avx_intrinsic(A2, A1); \ 536| 44.7k| _mm256_storeu_si256((__m256i *)(out + 160), AO); \ 537| 44.7k| A1 = _mm256_lddqu_si256((const __m256i *)(words_1 + 192)); \ 538| 44.7k| A2 = _mm256_lddqu_si256((const __m256i *)(words_2 + 192)); \ 539| 44.7k| AO = avx_intrinsic(A2, A1); \ 540| 44.7k| _mm256_storeu_si256((__m256i *)(out + 192), AO); \ 541| 44.7k| A1 = _mm256_lddqu_si256((const __m256i *)(words_1 + 224)); \ 542| 44.7k| A2 = _mm256_lddqu_si256((const __m256i *)(words_2 + 224)); \ 543| 44.7k| AO = avx_intrinsic(A2, A1); \ 544| 44.7k| _mm256_storeu_si256((__m256i *)(out + 224), AO); \ 545| 44.7k| out += 256; \ 546| 44.7k| words_1 += 256; \ 547| 44.7k| words_2 += 256; \ 548| 44.7k| } \ 549| 1.39k| dst->cardinality = BITSET_UNKNOWN_CARDINALITY; \ 550| 1.39k| return dst->cardinality; \ 551| 1.39k| } bitset.c:_avx2_bitset_container_and_justcard: 572| 1.39k| const bitset_container_t *src_1, const bitset_container_t *src_2) { \ 573| 1.39k| const __m256i *__restrict__ data1 = (const __m256i *)src_1->words; \ 574| 1.39k| const __m256i *__restrict__ data2 = (const __m256i *)src_2->words; \ 575| 1.39k| return (int)avx2_harley_seal_popcount256_##opname( \ 576| 1.39k| data2, data1, BITSET_CONTAINER_SIZE_IN_WORDS / (CROARING_WORDS_IN_AVX2_REG)); \ ------------------ | | 250| 1.39k|#define CROARING_WORDS_IN_AVX2_REG sizeof(__m256i) / sizeof(uint64_t) ------------------ 577| 1.39k| } bitset.c:_avx2_bitset_container_xor: 558| 1.43k| bitset_container_t *dst) { \ 559| 1.43k| const __m256i *__restrict__ words_1 = (const __m256i *)src_1->words; \ 560| 1.43k| const __m256i *__restrict__ words_2 = (const __m256i *)src_2->words; \ 561| 1.43k| __m256i *out = (__m256i *)dst->words; \ 562| 1.43k| dst->cardinality = (int32_t)avx2_harley_seal_popcount256andstore_##opname( \ 563| 1.43k| words_2, words_1, out, \ 564| 1.43k| BITSET_CONTAINER_SIZE_IN_WORDS / (CROARING_WORDS_IN_AVX2_REG)); \ ------------------ | | 250| 1.43k|#define CROARING_WORDS_IN_AVX2_REG sizeof(__m256i) / sizeof(uint64_t) ------------------ 565| 1.43k| return dst->cardinality; \ 566| 1.43k| } \ bitset.c:_avx2_bitset_container_equals: 1127| 3.77k|static inline bool _avx2_bitset_container_equals(const bitset_container_t *container1, const bitset_container_t *container2) { 1128| 3.77k| const __m256i *ptr1 = (const __m256i*)container1->words; 1129| 3.77k| const __m256i *ptr2 = (const __m256i*)container2->words; 1130| 970k| for (size_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS*sizeof(uint64_t)/32; i++) { ------------------ | Branch (1130:24): [True: 967k, False: 3.77k] ------------------ 1131| 967k| __m256i r1 = _mm256_loadu_si256(ptr1+i); 1132| 967k| __m256i r2 = _mm256_loadu_si256(ptr2+i); 1133| 967k| int mask = _mm256_movemask_epi8(_mm256_cmpeq_epi8(r1, r2)); 1134| 967k| if ((uint32_t)mask != UINT32_MAX) { ------------------ | Branch (1134:11): [True: 0, False: 967k] ------------------ 1135| 0| return false; 1136| 0| } 1137| 967k| } 1138| 3.77k| return true; 1139| 3.77k|} container_free: 58| 581k|void container_free(container_t *c, uint8_t type) { 59| 581k| switch (type) { 60| 270k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 270k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (60:9): [True: 270k, False: 311k] ------------------ 61| 270k| bitset_container_free(CAST_bitset(c)); ------------------ | | 52| 270k|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 270k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 62| 270k| break; 63| 126k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 126k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (63:9): [True: 126k, False: 455k] ------------------ 64| 126k| array_container_free(CAST_array(c)); ------------------ | | 54| 126k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 126k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 65| 126k| break; 66| 185k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 185k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (66:9): [True: 185k, False: 396k] ------------------ 67| 185k| run_container_free(CAST_run(c)); ------------------ | | 77| 185k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 185k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 68| 185k| break; 69| 0| case SHARED_CONTAINER_TYPE: ------------------ | | 51| 0|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (69:9): [True: 0, False: 581k] ------------------ 70| 0| shared_container_free(CAST_shared(c)); ------------------ | | 79| 0|#define CAST_shared(c) CAST(shared_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 71| 0| break; 72| 0| default: ------------------ | Branch (72:9): [True: 0, False: 581k] ------------------ 73| 0| assert(false); 74| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 75| 581k| } 76| 581k|} get_copy_of_container: 181| 253| bool copy_on_write) { 182| 253| if (copy_on_write) { ------------------ | Branch (182:9): [True: 0, False: 253] ------------------ 183| 0| shared_container_t *shared_container; 184| 0| if (*typecode == SHARED_CONTAINER_TYPE) { ------------------ | | 51| 0|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (184:13): [True: 0, False: 0] ------------------ 185| 0| shared_container = CAST_shared(c); ------------------ | | 79| 0|#define CAST_shared(c) CAST(shared_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 186| 0| croaring_refcount_inc(&shared_container->counter); 187| 0| return shared_container; 188| 0| } 189| 0| assert(*typecode != SHARED_CONTAINER_TYPE); 190| | 191| 0| if ((shared_container = (shared_container_t *)roaring_malloc( ------------------ | Branch (191:13): [True: 0, False: 0] ------------------ 192| 0| sizeof(shared_container_t))) == NULL) { 193| 0| return NULL; 194| 0| } 195| | 196| 0| shared_container->container = c; 197| 0| shared_container->typecode = *typecode; 198| | // At this point, we are creating new shared container 199| | // so there should be no other references, and setting 200| | // the counter to 2 - even non-atomically - is safe as 201| | // long as the value is set before the return statement. 202| 0| shared_container->counter = 2; 203| 0| *typecode = SHARED_CONTAINER_TYPE; ------------------ | | 51| 0|#define SHARED_CONTAINER_TYPE 4 ------------------ 204| | 205| 0| return shared_container; 206| 0| } // copy_on_write 207| | // otherwise, no copy on write... 208| 253| const container_t *actual_container = container_unwrap_shared(c, typecode); 209| | assert(*typecode != SHARED_CONTAINER_TYPE); 210| 253| return container_clone(actual_container, *typecode); 211| 253|} container_clone: 217| 319k|container_t *container_clone(const container_t *c, uint8_t typecode) { 218| | // We do not want to allow cloning of shared containers. 219| | // c = container_unwrap_shared(c, &typecode); 220| 319k| switch (typecode) { 221| 134k| case BITSET_CONTAINER_TYPE: ------------------ | | 48| 134k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (221:9): [True: 134k, False: 185k] ------------------ 222| 134k| return bitset_container_clone(const_CAST_bitset(c)); ------------------ | | 53| 134k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 134k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 223| 22.1k| case ARRAY_CONTAINER_TYPE: ------------------ | | 49| 22.1k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (223:9): [True: 22.1k, False: 297k] ------------------ 224| 22.1k| return array_container_clone(const_CAST_array(c)); ------------------ | | 55| 22.1k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 22.1k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 225| 163k| case RUN_CONTAINER_TYPE: ------------------ | | 50| 163k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (225:9): [True: 163k, False: 156k] ------------------ 226| 163k| return run_container_clone(const_CAST_run(c)); ------------------ | | 78| 163k|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 163k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 227| 0| case SHARED_CONTAINER_TYPE: ------------------ | | 51| 0|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (227:9): [True: 0, False: 319k] ------------------ 228| | // Shared containers are not cloneable. Are you mixing COW and 229| | // non-COW bitmaps? 230| 0| return NULL; 231| 0| default: ------------------ | Branch (231:9): [True: 0, False: 319k] ------------------ 232| 0| assert(false); 233| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 234| 0| return NULL; 235| 319k| } 236| 319k|} container_init_iterator: 304| 18.2k| uint16_t *value) { 305| 18.2k| switch (typecode) { 306| 3.53k| case BITSET_CONTAINER_TYPE: { ------------------ | | 48| 3.53k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (306:9): [True: 3.53k, False: 14.6k] ------------------ 307| 3.53k| const bitset_container_t *bc = const_CAST_bitset(c); ------------------ | | 53| 3.53k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 3.53k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 308| 3.53k| uint32_t wordindex = 0; 309| 3.53k| uint64_t word; 310| 33.9k| while ((word = bc->words[wordindex]) == 0) { ------------------ | Branch (310:20): [True: 30.4k, False: 3.53k] ------------------ 311| 30.4k| wordindex++; 312| 30.4k| } 313| | // word is non-zero 314| 3.53k| int32_t index = wordindex * 64 + roaring_trailing_zeroes(word); 315| 3.53k| *value = index; 316| 3.53k| return ROARING_INIT_ROARING_CONTAINER_ITERATOR_T{ ------------------ | | 13| 3.53k|#define ROARING_INIT_ROARING_CONTAINER_ITERATOR_T (roaring_container_iterator_t) ------------------ 317| 3.53k| .index = index, 318| 3.53k| }; 319| 0| } 320| 14.1k| case ARRAY_CONTAINER_TYPE: { ------------------ | | 49| 14.1k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (320:9): [True: 14.1k, False: 4.02k] ------------------ 321| 14.1k| const array_container_t *ac = const_CAST_array(c); ------------------ | | 55| 14.1k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 14.1k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 322| 14.1k| *value = ac->array[0]; 323| 14.1k| return ROARING_INIT_ROARING_CONTAINER_ITERATOR_T{ ------------------ | | 13| 14.1k|#define ROARING_INIT_ROARING_CONTAINER_ITERATOR_T (roaring_container_iterator_t) ------------------ 324| 14.1k| .index = 0, 325| 14.1k| }; 326| 0| } 327| 483| case RUN_CONTAINER_TYPE: { ------------------ | | 50| 483|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (327:9): [True: 483, False: 17.7k] ------------------ 328| 483| const run_container_t *rc = const_CAST_run(c); ------------------ | | 78| 483|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 483|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 329| 483| *value = rc->runs[0].value; 330| 483| return ROARING_INIT_ROARING_CONTAINER_ITERATOR_T{ ------------------ | | 13| 483|#define ROARING_INIT_ROARING_CONTAINER_ITERATOR_T (roaring_container_iterator_t) ------------------ 331| 483| .index = 0, 332| 483| }; 333| 0| } 334| 0| default: ------------------ | Branch (334:9): [True: 0, False: 18.2k] ------------------ 335| 0| assert(false); 336| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 337| 0| return ROARING_INIT_ROARING_CONTAINER_ITERATOR_T{0}; ------------------ | | 13| 0|#define ROARING_INIT_ROARING_CONTAINER_ITERATOR_T (roaring_container_iterator_t) ------------------ 338| 18.2k| } 339| 18.2k|} container_iterator_lower_bound: 386| 6.05k| uint16_t *value_out, uint16_t val) { 387| 6.05k| if (val > container_maximum(c, typecode)) { ------------------ | Branch (387:9): [True: 0, False: 6.05k] ------------------ 388| 0| return false; 389| 0| } 390| 6.05k| switch (typecode) { 391| 1.73k| case BITSET_CONTAINER_TYPE: { ------------------ | | 48| 1.73k|#define BITSET_CONTAINER_TYPE 1 ------------------ | Branch (391:9): [True: 1.73k, False: 4.32k] ------------------ 392| 1.73k| const bitset_container_t *bc = const_CAST_bitset(c); ------------------ | | 53| 1.73k|#define const_CAST_bitset(c) CAST(const bitset_container_t *, c) | | ------------------ | | | | 86| 1.73k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 393| 1.73k| it->index = bitset_container_index_equalorlarger(bc, val); 394| 1.73k| *value_out = it->index; 395| 1.73k| return true; 396| 0| } 397| 4.32k| case ARRAY_CONTAINER_TYPE: { ------------------ | | 49| 4.32k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (397:9): [True: 4.32k, False: 1.73k] ------------------ 398| 4.32k| const array_container_t *ac = const_CAST_array(c); ------------------ | | 55| 4.32k|#define const_CAST_array(c) CAST(const array_container_t *, c) | | ------------------ | | | | 86| 4.32k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 399| 4.32k| it->index = array_container_index_equalorlarger(ac, val); 400| 4.32k| *value_out = ac->array[it->index]; 401| 4.32k| return true; 402| 0| } 403| 0| case RUN_CONTAINER_TYPE: { ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (403:9): [True: 0, False: 6.05k] ------------------ 404| 0| const run_container_t *rc = const_CAST_run(c); ------------------ | | 78| 0|#define const_CAST_run(c) CAST(const run_container_t *, c) | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 405| 0| it->index = run_container_index_equalorlarger(rc, val); 406| 0| if (rc->runs[it->index].value <= val) { ------------------ | Branch (406:17): [True: 0, False: 0] ------------------ 407| 0| *value_out = val; 408| 0| } else { 409| 0| *value_out = rc->runs[it->index].value; 410| 0| } 411| 0| return true; 412| 0| } 413| 0| default: ------------------ | Branch (413:9): [True: 0, False: 6.05k] ------------------ 414| 0| assert(false); 415| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 416| 0| return false; 417| 6.05k| } 418| 6.05k|} bitset_container_from_array: 22| 2.16k|bitset_container_t *bitset_container_from_array(const array_container_t *ac) { 23| 2.16k| bitset_container_t *ans = bitset_container_create(); 24| 2.16k| int limit = array_container_cardinality(ac); 25| 2.07M| for (int i = 0; i < limit; ++i) bitset_container_set(ans, ac->array[i]); ------------------ | Branch (25:21): [True: 2.07M, False: 2.16k] ------------------ 26| 2.16k| return ans; 27| 2.16k|} bitset_container_from_run: 29| 495|bitset_container_t *bitset_container_from_run(const run_container_t *arr) { 30| 495| int card = run_container_cardinality(arr); 31| 495| bitset_container_t *answer = bitset_container_create(); 32| 16.3k| for (int rlepos = 0; rlepos < arr->n_runs; ++rlepos) { ------------------ | Branch (32:26): [True: 15.9k, False: 495] ------------------ 33| 15.9k| rle16_t vl = arr->runs[rlepos]; 34| 15.9k| bitset_set_lenrange(answer->words, vl.value, vl.length); 35| 15.9k| } 36| 495| answer->cardinality = card; 37| 495| return answer; 38| 495|} array_container_from_run: 40| 1.09k|array_container_t *array_container_from_run(const run_container_t *arr) { 41| 1.09k| array_container_t *answer = 42| 1.09k| array_container_create_given_capacity(run_container_cardinality(arr)); 43| 1.09k| answer->cardinality = 0; 44| 79.4k| for (int rlepos = 0; rlepos < arr->n_runs; ++rlepos) { ------------------ | Branch (44:26): [True: 78.3k, False: 1.09k] ------------------ 45| 78.3k| int run_start = arr->runs[rlepos].value; 46| 78.3k| int run_end = run_start + arr->runs[rlepos].length; 47| | 48| 393k| for (int run_value = run_start; run_value <= run_end; ++run_value) { ------------------ | Branch (48:41): [True: 314k, False: 78.3k] ------------------ 49| 314k| answer->array[answer->cardinality++] = (uint16_t)run_value; 50| 314k| } 51| 78.3k| } 52| 1.09k| return answer; 53| 1.09k|} array_container_from_bitset: 55| 2.57k|array_container_t *array_container_from_bitset(const bitset_container_t *bits) { 56| 2.57k| array_container_t *result = 57| 2.57k| array_container_create_given_capacity(bits->cardinality); 58| 2.57k| result->cardinality = bits->cardinality; 59| 2.57k|#if CROARING_IS_X64 60| 2.57k|#if CROARING_COMPILER_SUPPORTS_AVX512 61| 2.57k| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX512) { ------------------ | Branch (61:9): [True: 0, False: 2.57k] ------------------ 62| 0| bitset_extract_setbits_avx512_uint16( 63| 0| bits->words, BITSET_CONTAINER_SIZE_IN_WORDS, result->array, 64| 0| bits->cardinality, 0); 65| 0| } else 66| 2.57k|#endif 67| 2.57k| { 68| | // sse version ends up being slower here 69| | // (bitset_extract_setbits_sse_uint16) 70| | // because of the sparsity of the data 71| 2.57k| bitset_extract_setbits_uint16( 72| 2.57k| bits->words, BITSET_CONTAINER_SIZE_IN_WORDS, result->array, 0); 73| 2.57k| } 74| |#else 75| | // If the system is not x64, then we have no accelerated function. 76| | bitset_extract_setbits_uint16(bits->words, BITSET_CONTAINER_SIZE_IN_WORDS, 77| | result->array, 0); 78| |#endif 79| | 80| 2.57k| return result; 81| 2.57k|} convert_to_bitset_or_array_container: 120| 66.2k| uint8_t *resulttype) { 121| 66.2k| if (card <= DEFAULT_MAX_SIZE) { ------------------ | Branch (121:9): [True: 1.18k, False: 65.1k] ------------------ 122| 1.18k| array_container_t *answer = array_container_create_given_capacity(card); 123| 1.18k| answer->cardinality = 0; 124| 62.4k| for (int rlepos = 0; rlepos < rc->n_runs; ++rlepos) { ------------------ | Branch (124:30): [True: 61.2k, False: 1.18k] ------------------ 125| 61.2k| uint16_t run_start = rc->runs[rlepos].value; 126| 61.2k| uint16_t run_end = run_start + rc->runs[rlepos].length; 127| 658k| for (uint16_t run_value = run_start; run_value < run_end; ------------------ | Branch (127:50): [True: 597k, False: 61.2k] ------------------ 128| 597k| ++run_value) { 129| 597k| answer->array[answer->cardinality++] = run_value; 130| 597k| } 131| 61.2k| answer->array[answer->cardinality++] = run_end; 132| 61.2k| } 133| 1.18k| assert(card == answer->cardinality); 134| 1.18k| *resulttype = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 1.18k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 135| | // run_container_free(r); 136| 1.18k| return answer; 137| 1.18k| } 138| 65.1k| bitset_container_t *answer = bitset_container_create(); 139| 146k| for (int rlepos = 0; rlepos < rc->n_runs; ++rlepos) { ------------------ | Branch (139:26): [True: 81.5k, False: 65.1k] ------------------ 140| 81.5k| uint16_t run_start = rc->runs[rlepos].value; 141| 81.5k| bitset_set_lenrange(answer->words, run_start, rc->runs[rlepos].length); 142| 81.5k| } 143| 65.1k| answer->cardinality = card; 144| 65.1k| *resulttype = BITSET_CONTAINER_TYPE; ------------------ | | 48| 65.1k|#define BITSET_CONTAINER_TYPE 1 ------------------ 145| | // run_container_free(r); 146| 65.1k| return answer; 147| 66.2k|} convert_run_to_efficient_container: 155| 128k| uint8_t *typecode_after) { 156| 128k| int32_t size_as_run_container = 157| 128k| run_container_serialized_size_in_bytes(c->n_runs); 158| | 159| 128k| int32_t size_as_bitset_container = 160| 128k| bitset_container_serialized_size_in_bytes(); 161| 128k| int32_t card = run_container_cardinality(c); 162| 128k| int32_t size_as_array_container = 163| 128k| array_container_serialized_size_in_bytes(card); 164| | 165| 128k| int32_t min_size_non_run = 166| 128k| size_as_bitset_container < size_as_array_container ------------------ | Branch (166:9): [True: 57.3k, False: 70.8k] ------------------ 167| 128k| ? size_as_bitset_container 168| 128k| : size_as_array_container; 169| 128k| if (size_as_run_container <= min_size_non_run) { // no conversion ------------------ | Branch (169:9): [True: 65.5k, False: 62.6k] ------------------ 170| 65.5k| *typecode_after = RUN_CONTAINER_TYPE; ------------------ | | 50| 65.5k|#define RUN_CONTAINER_TYPE 3 ------------------ 171| 65.5k| return c; 172| 65.5k| } 173| 62.6k| if (card <= DEFAULT_MAX_SIZE) { ------------------ | Branch (173:9): [True: 62.6k, False: 0] ------------------ 174| | // to array 175| 62.6k| array_container_t *answer = array_container_create_given_capacity(card); 176| 62.6k| answer->cardinality = 0; 177| 203k| for (int rlepos = 0; rlepos < c->n_runs; ++rlepos) { ------------------ | Branch (177:30): [True: 140k, False: 62.6k] ------------------ 178| 140k| int run_start = c->runs[rlepos].value; 179| 140k| int run_end = run_start + c->runs[rlepos].length; 180| | 181| 375k| for (int run_value = run_start; run_value <= run_end; ++run_value) { ------------------ | Branch (181:45): [True: 235k, False: 140k] ------------------ 182| 235k| answer->array[answer->cardinality++] = (uint16_t)run_value; 183| 235k| } 184| 140k| } 185| 62.6k| *typecode_after = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 62.6k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 186| 62.6k| return answer; 187| 62.6k| } 188| | 189| | // else to bitset 190| 0| bitset_container_t *answer = bitset_container_create(); 191| | 192| 0| for (int rlepos = 0; rlepos < c->n_runs; ++rlepos) { ------------------ | Branch (192:26): [True: 0, False: 0] ------------------ 193| 0| int start = c->runs[rlepos].value; 194| 0| int end = start + c->runs[rlepos].length; 195| 0| bitset_set_range(answer->words, start, end + 1); 196| 0| } 197| 0| answer->cardinality = card; 198| 0| *typecode_after = BITSET_CONTAINER_TYPE; ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 199| 0| return answer; 200| 62.6k|} convert_run_to_efficient_container_and_free: 204| 116k| run_container_t *c, uint8_t *typecode_after) { 205| 116k| container_t *answer = convert_run_to_efficient_container(c, typecode_after); 206| 116k| if (answer != c) run_container_free(c); ------------------ | Branch (206:9): [True: 56.5k, False: 60.4k] ------------------ 207| 116k| return answer; 208| 116k|} convert_run_optimize: 218| 12.3k| uint8_t *typecode_after) { 219| 12.3k| if (typecode_original == RUN_CONTAINER_TYPE) { ------------------ | | 50| 12.3k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (219:9): [True: 0, False: 12.3k] ------------------ 220| 0| container_t *newc = 221| 0| convert_run_to_efficient_container(CAST_run(c), typecode_after); ------------------ | | 77| 0|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 222| 0| if (newc != c) { ------------------ | Branch (222:13): [True: 0, False: 0] ------------------ 223| 0| container_free(c, typecode_original); 224| 0| } 225| 0| return newc; 226| 12.3k| } else if (typecode_original == ARRAY_CONTAINER_TYPE) { ------------------ | | 49| 12.3k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (226:16): [True: 12.3k, False: 0] ------------------ 227| | // it might need to be converted to a run container. 228| 12.3k| array_container_t *c_qua_array = CAST_array(c); ------------------ | | 54| 12.3k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 12.3k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 229| 12.3k| int32_t n_runs = array_container_number_of_runs(c_qua_array); 230| 12.3k| int32_t size_as_run_container = 231| 12.3k| run_container_serialized_size_in_bytes(n_runs); 232| 12.3k| int32_t card = array_container_cardinality(c_qua_array); 233| 12.3k| int32_t size_as_array_container = 234| 12.3k| array_container_serialized_size_in_bytes(card); 235| | 236| 12.3k| if (size_as_run_container >= size_as_array_container) { ------------------ | Branch (236:13): [True: 8.44k, False: 3.86k] ------------------ 237| 8.44k| *typecode_after = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 8.44k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 238| 8.44k| return c; 239| 8.44k| } 240| | // else convert array to run container 241| 3.86k| run_container_t *answer = run_container_create_given_capacity(n_runs); 242| 3.86k| int prev = -2; 243| 3.86k| int run_start = -1; 244| | 245| 3.86k| assert(card > 0); 246| 624k| for (int i = 0; i < card; ++i) { ------------------ | Branch (246:25): [True: 620k, False: 3.86k] ------------------ 247| 620k| uint16_t cur_val = c_qua_array->array[i]; 248| 620k| if (cur_val != prev + 1) { ------------------ | Branch (248:17): [True: 252k, False: 368k] ------------------ 249| | // new run starts; flush old one, if any 250| 252k| if (run_start != -1) add_run(answer, run_start, prev); ------------------ | Branch (250:21): [True: 248k, False: 3.86k] ------------------ 251| 252k| run_start = cur_val; 252| 252k| } 253| 620k| prev = c_qua_array->array[i]; 254| 620k| } 255| 3.86k| assert(run_start >= 0); 256| | // now prev is the last seen value 257| 3.86k| add_run(answer, run_start, prev); 258| 3.86k| *typecode_after = RUN_CONTAINER_TYPE; ------------------ | | 50| 3.86k|#define RUN_CONTAINER_TYPE 3 ------------------ 259| 3.86k| array_container_free(c_qua_array); 260| 3.86k| return answer; 261| 12.3k| } else if (typecode_original == ------------------ | Branch (261:16): [True: 0, False: 0] ------------------ 262| 0| BITSET_CONTAINER_TYPE) { // run conversions on bitset ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 263| | // does bitset need conversion to run? 264| 0| bitset_container_t *c_qua_bitset = CAST_bitset(c); ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 265| 0| int32_t n_runs = bitset_container_number_of_runs(c_qua_bitset); 266| 0| int32_t size_as_run_container = 267| 0| run_container_serialized_size_in_bytes(n_runs); 268| 0| int32_t size_as_bitset_container = 269| 0| bitset_container_serialized_size_in_bytes(); 270| | 271| 0| if (size_as_bitset_container <= size_as_run_container) { ------------------ | Branch (271:13): [True: 0, False: 0] ------------------ 272| | // no conversion needed. 273| 0| *typecode_after = BITSET_CONTAINER_TYPE; ------------------ | | 48| 0|#define BITSET_CONTAINER_TYPE 1 ------------------ 274| 0| return c; 275| 0| } 276| | // bitset to runcontainer (ported from Java RunContainer( 277| | // BitmapContainer bc, int nbrRuns)) 278| 0| assert(n_runs > 0); // no empty bitmaps 279| 0| run_container_t *answer = run_container_create_given_capacity(n_runs); 280| | 281| 0| int long_ctr = 0; 282| 0| uint64_t cur_word = c_qua_bitset->words[0]; 283| 0| while (true) { ------------------ | Branch (283:16): [True: 0, Folded] ------------------ 284| 0| while (cur_word == UINT64_C(0) && ------------------ | Branch (284:20): [True: 0, False: 0] ------------------ 285| 0| long_ctr < BITSET_CONTAINER_SIZE_IN_WORDS - 1) ------------------ | Branch (285:20): [True: 0, False: 0] ------------------ 286| 0| cur_word = c_qua_bitset->words[++long_ctr]; 287| | 288| 0| if (cur_word == UINT64_C(0)) { ------------------ | Branch (288:17): [True: 0, False: 0] ------------------ 289| 0| bitset_container_free(c_qua_bitset); 290| 0| *typecode_after = RUN_CONTAINER_TYPE; ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 291| 0| return answer; 292| 0| } 293| | 294| 0| int local_run_start = roaring_trailing_zeroes(cur_word); 295| 0| int run_start = local_run_start + 64 * long_ctr; 296| 0| uint64_t cur_word_with_1s = cur_word | (cur_word - 1); 297| | 298| 0| int run_end = 0; 299| 0| while (cur_word_with_1s == UINT64_C(0xFFFFFFFFFFFFFFFF) && ------------------ | Branch (299:20): [True: 0, False: 0] ------------------ 300| 0| long_ctr < BITSET_CONTAINER_SIZE_IN_WORDS - 1) ------------------ | Branch (300:20): [True: 0, False: 0] ------------------ 301| 0| cur_word_with_1s = c_qua_bitset->words[++long_ctr]; 302| | 303| 0| if (cur_word_with_1s == UINT64_C(0xFFFFFFFFFFFFFFFF)) { ------------------ | Branch (303:17): [True: 0, False: 0] ------------------ 304| 0| run_end = 64 + long_ctr * 64; // exclusive, I guess 305| 0| add_run(answer, run_start, run_end - 1); 306| 0| bitset_container_free(c_qua_bitset); 307| 0| *typecode_after = RUN_CONTAINER_TYPE; ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 308| 0| return answer; 309| 0| } 310| 0| int local_run_end = roaring_trailing_zeroes(~cur_word_with_1s); 311| 0| run_end = local_run_end + long_ctr * 64; 312| 0| add_run(answer, run_start, run_end - 1); 313| 0| cur_word = cur_word_with_1s & (cur_word_with_1s + 1); 314| 0| } 315| 0| return answer; 316| 0| } else { 317| 0| assert(false); 318| 0| roaring_unreachable; ------------------ | | 233| 0|#define roaring_unreachable __builtin_unreachable() ------------------ 319| 0| return NULL; 320| 0| } 321| 12.3k|} convert.c:add_run: 84| 252k|static void add_run(run_container_t *rc, int s, int e) { 85| 252k| rc->runs[rc->n_runs].value = s; 86| 252k| rc->runs[rc->n_runs].length = e - s; 87| 252k| rc->n_runs++; 88| 252k|} array_bitset_container_andnot: 26| 553| array_container_t *dst) { 27| | // follows Java implementation as of June 2016 28| 553| if (dst->capacity < src_1->cardinality) { ------------------ | Branch (28:9): [True: 553, False: 0] ------------------ 29| 553| array_container_grow(dst, src_1->cardinality, false); 30| 553| } 31| 553| int32_t newcard = 0; 32| 553| const int32_t origcard = src_1->cardinality; 33| 93.2k| for (int i = 0; i < origcard; ++i) { ------------------ | Branch (33:21): [True: 92.7k, False: 553] ------------------ 34| 92.7k| uint16_t key = src_1->array[i]; 35| 92.7k| dst->array[newcard] = key; 36| 92.7k| newcard += 1 - bitset_container_contains(src_2, key); 37| 92.7k| } 38| 553| dst->cardinality = newcard; 39| 553|} bitset_array_container_iandnot: 83| 1.39k| container_t **dst) { 84| 1.39k| *dst = src_1; 85| 1.39k| src_1->cardinality = 86| 1.39k| (int32_t)bitset_clear_list(src_1->words, (uint64_t)src_1->cardinality, 87| 1.39k| src_2->array, (uint64_t)src_2->cardinality); 88| | 89| 1.39k| if (src_1->cardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (89:9): [True: 166, False: 1.22k] ------------------ 90| 166| *dst = array_container_from_bitset(src_1); 91| 166| bitset_container_free(src_1); 92| 166| return false; // not bitset 93| 166| } else 94| 1.22k| return true; 95| 1.39k|} run_bitset_container_andnot: 106| 606| container_t **dst) { 107| | // follows the Java implementation as of June 2016 108| 606| int card = run_container_cardinality(src_1); 109| 606| if (card <= DEFAULT_MAX_SIZE) { ------------------ | Branch (109:9): [True: 606, False: 0] ------------------ 110| | // must be an array 111| 606| array_container_t *answer = array_container_create_given_capacity(card); 112| 606| answer->cardinality = 0; 113| 51.5k| for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { ------------------ | Branch (113:34): [True: 50.9k, False: 606] ------------------ 114| 50.9k| rle16_t rle = src_1->runs[rlepos]; 115| 183k| for (int run_value = rle.value; run_value <= rle.value + rle.length; ------------------ | Branch (115:45): [True: 132k, False: 50.9k] ------------------ 116| 132k| ++run_value) { 117| 132k| if (!bitset_container_get(src_2, (uint16_t)run_value)) { ------------------ | Branch (117:21): [True: 37.2k, False: 95.1k] ------------------ 118| 37.2k| answer->array[answer->cardinality++] = (uint16_t)run_value; 119| 37.2k| } 120| 132k| } 121| 50.9k| } 122| 606| *dst = answer; 123| 606| return false; 124| 606| } else { // we guess it will be a bitset, though have to check guess when 125| | // done 126| 0| bitset_container_t *answer = bitset_container_clone(src_2); 127| | 128| 0| uint32_t last_pos = 0; 129| 0| for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { ------------------ | Branch (129:34): [True: 0, False: 0] ------------------ 130| 0| rle16_t rle = src_1->runs[rlepos]; 131| | 132| 0| uint32_t start = rle.value; 133| 0| uint32_t end = start + rle.length + 1; 134| 0| bitset_reset_range(answer->words, last_pos, start); 135| 0| bitset_flip_range(answer->words, start, end); 136| 0| last_pos = end; 137| 0| } 138| 0| bitset_reset_range(answer->words, last_pos, (uint32_t)(1 << 16)); 139| | 140| 0| answer->cardinality = bitset_container_compute_cardinality(answer); 141| | 142| 0| if (answer->cardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (142:13): [True: 0, False: 0] ------------------ 143| 0| *dst = array_container_from_bitset(answer); 144| 0| bitset_container_free(answer); 145| 0| return false; // not bitset 146| 0| } 147| 0| *dst = answer; 148| | return true; // bitset 149| 0| } 150| 606|} run_array_container_andnot: 279| 1.89k| container_t **dst) { 280| | // follows the Java impl as of June 2016 281| | 282| 1.89k| int card = run_container_cardinality(src_1); 283| 1.89k| const int arbitrary_threshold = 32; 284| | 285| 1.89k| if (card <= arbitrary_threshold) { ------------------ | Branch (285:9): [True: 350, False: 1.54k] ------------------ 286| 350| if (src_2->cardinality == 0) { ------------------ | Branch (286:13): [True: 0, False: 350] ------------------ 287| 0| *dst = run_container_clone(src_1); 288| 0| return RUN_CONTAINER_TYPE; ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 289| 0| } 290| | // Java's "lazyandNot.toEfficientContainer" thing 291| 350| run_container_t *answer = run_container_create_given_capacity( 292| 350| card + array_container_cardinality(src_2)); 293| | 294| 350| int rlepos = 0; 295| 350| int xrlepos = 0; // "x" is src_2 296| 350| rle16_t rle = src_1->runs[rlepos]; 297| 350| int32_t start = rle.value; 298| 350| int32_t end = start + rle.length + 1; 299| 350| int32_t xstart = src_2->array[xrlepos]; 300| | 301| 11.2k| while ((rlepos < src_1->n_runs) && (xrlepos < src_2->cardinality)) { ------------------ | Branch (301:16): [True: 10.9k, False: 281] | Branch (301:44): [True: 10.8k, False: 69] ------------------ 302| 10.8k| if (end <= xstart) { ------------------ | Branch (302:17): [True: 605, False: 10.2k] ------------------ 303| | // output the first run 304| 605| answer->runs[answer->n_runs++] = 305| 605| CROARING_MAKE_RLE16(start, end - start - 1); ------------------ | | 60| 605| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 306| 605| rlepos++; 307| 605| if (rlepos < src_1->n_runs) { ------------------ | Branch (307:21): [True: 568, False: 37] ------------------ 308| 568| start = src_1->runs[rlepos].value; 309| 568| end = start + src_1->runs[rlepos].length + 1; 310| 568| } 311| 10.2k| } else if (xstart + 1 <= start) { ------------------ | Branch (311:24): [True: 6.31k, False: 3.94k] ------------------ 312| | // exit the second run 313| 6.31k| xrlepos++; 314| 6.31k| if (xrlepos < src_2->cardinality) { ------------------ | Branch (314:21): [True: 6.24k, False: 69] ------------------ 315| 6.24k| xstart = src_2->array[xrlepos]; 316| 6.24k| } 317| 6.31k| } else { 318| 3.94k| if (start < xstart) { ------------------ | Branch (318:21): [True: 347, False: 3.59k] ------------------ 319| 347| answer->runs[answer->n_runs++] = 320| 347| CROARING_MAKE_RLE16(start, xstart - start - 1); ------------------ | | 60| 347| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 321| 347| } 322| 3.94k| if (xstart + 1 < end) { ------------------ | Branch (322:21): [True: 2.42k, False: 1.51k] ------------------ 323| 2.42k| start = xstart + 1; 324| 2.42k| } else { 325| 1.51k| rlepos++; 326| 1.51k| if (rlepos < src_1->n_runs) { ------------------ | Branch (326:25): [True: 1.27k, False: 244] ------------------ 327| 1.27k| start = src_1->runs[rlepos].value; 328| 1.27k| end = start + src_1->runs[rlepos].length + 1; 329| 1.27k| } 330| 1.51k| } 331| 3.94k| } 332| 10.8k| } 333| 350| if (rlepos < src_1->n_runs) { ------------------ | Branch (333:13): [True: 69, False: 281] ------------------ 334| 69| answer->runs[answer->n_runs++] = 335| 69| CROARING_MAKE_RLE16(start, end - start - 1); ------------------ | | 60| 69| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 336| 69| rlepos++; 337| 69| if (rlepos < src_1->n_runs) { ------------------ | Branch (337:17): [True: 25, False: 44] ------------------ 338| 25| memcpy(answer->runs + answer->n_runs, src_1->runs + rlepos, 339| 25| (src_1->n_runs - rlepos) * sizeof(rle16_t)); 340| 25| answer->n_runs += (src_1->n_runs - rlepos); 341| 25| } 342| 69| } 343| 350| uint8_t return_type; 344| 350| *dst = convert_run_to_efficient_container(answer, &return_type); 345| 350| if (answer != *dst) run_container_free(answer); ------------------ | Branch (345:13): [True: 229, False: 121] ------------------ 346| 350| return return_type; 347| 350| } 348| | // else it's a bitmap or array 349| | 350| 1.54k| if (card <= DEFAULT_MAX_SIZE) { ------------------ | Branch (350:9): [True: 1.26k, False: 276] ------------------ 351| 1.26k| array_container_t *ac = array_container_create_given_capacity(card); 352| | // nb Java code used a generic iterator-based merge to compute 353| | // difference 354| 1.26k| ac->cardinality = run_array_array_subtract(src_1, src_2, ac); 355| 1.26k| *dst = ac; 356| 1.26k| return ARRAY_CONTAINER_TYPE; ------------------ | | 49| 1.26k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 357| 1.26k| } 358| 276| bitset_container_t *ans = bitset_container_from_run(src_1); 359| 276| bool result_is_bitset = bitset_array_container_iandnot(ans, src_2, dst); 360| 276| return (result_is_bitset ? BITSET_CONTAINER_TYPE : ARRAY_CONTAINER_TYPE); ------------------ | | 48| 263|#define BITSET_CONTAINER_TYPE 1 ------------------ return (result_is_bitset ? BITSET_CONTAINER_TYPE : ARRAY_CONTAINER_TYPE); ------------------ | | 49| 13|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (360:13): [True: 263, False: 13] ------------------ 361| 1.54k|} run_array_container_iandnot: 372| 692| container_t **dst) { 373| | // dummy implementation same as June 2016 Java 374| 692| int ans = run_array_container_andnot(src_1, src_2, dst); 375| 692| run_container_free(src_1); 376| 692| return ans; 377| 692|} array_run_container_andnot: 383| 735| array_container_t *dst) { 384| | // basically following Java impl as of June 2016 385| 735| if (src_1->cardinality > dst->capacity) { ------------------ | Branch (385:9): [True: 689, False: 46] ------------------ 386| 689| array_container_grow(dst, src_1->cardinality, false); 387| 689| } 388| | 389| 735| if (src_2->n_runs == 0) { ------------------ | Branch (389:9): [True: 0, False: 735] ------------------ 390| 0| memmove(dst->array, src_1->array, 391| 0| sizeof(uint16_t) * src_1->cardinality); 392| 0| dst->cardinality = src_1->cardinality; 393| 0| return; 394| 0| } 395| 735| int32_t run_start = src_2->runs[0].value; 396| 735| int32_t run_end = run_start + src_2->runs[0].length; 397| 735| int which_run = 0; 398| | 399| 735| uint16_t val = 0; 400| 735| int dest_card = 0; 401| 157k| for (int i = 0; i < src_1->cardinality; ++i) { ------------------ | Branch (401:21): [True: 156k, False: 735] ------------------ 402| 156k| val = src_1->array[i]; 403| 156k| if (val < run_start) ------------------ | Branch (403:13): [True: 40.7k, False: 116k] ------------------ 404| 40.7k| dst->array[dest_card++] = val; 405| 116k| else if (val <= run_end) { ------------------ | Branch (405:18): [True: 83.8k, False: 32.2k] ------------------ 406| 83.8k| ; // omitted item 407| 83.8k| } else { 408| 44.4k| do { 409| 44.4k| if (which_run + 1 < src_2->n_runs) { ------------------ | Branch (409:21): [True: 44.3k, False: 163] ------------------ 410| 44.3k| ++which_run; 411| 44.3k| run_start = src_2->runs[which_run].value; 412| 44.3k| run_end = run_start + src_2->runs[which_run].length; 413| | 414| 44.3k| } else 415| 163| run_start = run_end = (1 << 16) + 1; 416| 44.4k| } while (val > run_end); ------------------ | Branch (416:22): [True: 12.2k, False: 32.2k] ------------------ 417| 32.2k| --i; 418| 32.2k| } 419| 156k| } 420| 735| dst->cardinality = dest_card; 421| 735|} array_run_container_iandnot: 428| 46| const run_container_t *src_2) { 429| 46| array_run_container_andnot(src_1, src_2, src_1); 430| 46|} run_run_container_andnot: 437| 1.10k| const run_container_t *src_2, container_t **dst) { 438| 1.10k| run_container_t *ans = run_container_create(); 439| 1.10k| run_container_andnot(src_1, src_2, ans); 440| 1.10k| uint8_t typecode_after; 441| 1.10k| *dst = convert_run_to_efficient_container_and_free(ans, &typecode_after); 442| 1.10k| return typecode_after; 443| 1.10k|} run_run_container_iandnot: 453| 439| const run_container_t *src_2, container_t **dst) { 454| | // following Java impl as of June 2016 (dummy) 455| 439| int ans = run_run_container_andnot(src_1, src_2, dst); 456| 439| run_container_free(src_1); 457| 439| return ans; 458| 439|} array_array_container_andnot: 466| 2.27k| array_container_t *dst) { 467| 2.27k| array_container_andnot(src_1, src_2, dst); 468| 2.27k|} array_array_container_iandnot: 473| 3.43k| const array_container_t *src_2) { 474| 3.43k| array_container_andnot(src_1, src_2, src_1); 475| 3.43k|} mixed_andnot.c:run_array_array_subtract: 235| 1.26k| array_container_t *a_out) { 236| 1.26k| int out_card = 0; 237| 1.26k| int32_t in_array_pos = 238| 1.26k| -1; // since advanceUntil always assumes we start the search AFTER this 239| | 240| 91.1k| for (int rlepos = 0; rlepos < rc->n_runs; rlepos++) { ------------------ | Branch (240:26): [True: 89.9k, False: 1.26k] ------------------ 241| 89.9k| int32_t start = rc->runs[rlepos].value; 242| 89.9k| int32_t end = start + rc->runs[rlepos].length + 1; 243| | 244| 89.9k| in_array_pos = advanceUntil(a_in->array, in_array_pos, 245| 89.9k| a_in->cardinality, (uint16_t)start); 246| | 247| 89.9k| if (in_array_pos >= a_in->cardinality) { // run has no items subtracted ------------------ | Branch (247:13): [True: 5.21k, False: 84.7k] ------------------ 248| 66.1k| for (int32_t i = start; i < end; ++i) ------------------ | Branch (248:37): [True: 60.8k, False: 5.21k] ------------------ 249| 60.8k| a_out->array[out_card++] = (uint16_t)i; 250| 84.7k| } else { 251| 84.7k| uint16_t next_nonincluded = a_in->array[in_array_pos]; 252| 84.7k| if (next_nonincluded >= end) { ------------------ | Branch (252:17): [True: 28.6k, False: 56.0k] ------------------ 253| | // another case when run goes unaltered 254| 77.6k| for (int32_t i = start; i < end; ++i) ------------------ | Branch (254:41): [True: 49.0k, False: 28.6k] ------------------ 255| 49.0k| a_out->array[out_card++] = (uint16_t)i; 256| 28.6k| in_array_pos--; // ensure we see this item again if necessary 257| 56.0k| } else { 258| 281k| for (int32_t i = start; i < end; ++i) ------------------ | Branch (258:41): [True: 225k, False: 56.0k] ------------------ 259| 225k| if (i != next_nonincluded) ------------------ | Branch (259:25): [True: 98.3k, False: 126k] ------------------ 260| 98.3k| a_out->array[out_card++] = (uint16_t)i; 261| 126k| else // 0 should ensure we don't match 262| 126k| next_nonincluded = 263| 126k| (in_array_pos + 1 >= a_in->cardinality) ------------------ | Branch (263:29): [True: 1.00k, False: 125k] ------------------ 264| 126k| ? 0 265| 126k| : a_in->array[++in_array_pos]; 266| 56.0k| in_array_pos--; // see again 267| 56.0k| } 268| 84.7k| } 269| 89.9k| } 270| 1.26k| return out_card; 271| 1.26k|} array_container_equal_bitset: 10| 403| const bitset_container_t* container2) { 11| 403| if (container2->cardinality != BITSET_UNKNOWN_CARDINALITY) { ------------------ | Branch (11:9): [True: 403, False: 0] ------------------ 12| 403| if (container2->cardinality != container1->cardinality) { ------------------ | Branch (12:13): [True: 403, False: 0] ------------------ 13| 403| return false; 14| 403| } 15| 403| } 16| 0| int32_t pos = 0; 17| 0| for (int32_t i = 0; i < BITSET_CONTAINER_SIZE_IN_WORDS; ++i) { ------------------ | Branch (17:25): [True: 0, False: 0] ------------------ 18| 0| uint64_t w = container2->words[i]; 19| 0| while (w != 0) { ------------------ | Branch (19:16): [True: 0, False: 0] ------------------ 20| 0| uint64_t t = w & (~w + 1); 21| 0| uint16_t r = i * 64 + roaring_trailing_zeroes(w); 22| 0| if (pos >= container1->cardinality) { ------------------ | Branch (22:17): [True: 0, False: 0] ------------------ 23| 0| return false; 24| 0| } 25| 0| if (container1->array[pos] != r) { ------------------ | Branch (25:17): [True: 0, False: 0] ------------------ 26| 0| return false; 27| 0| } 28| 0| ++pos; 29| 0| w ^= t; 30| 0| } 31| 0| } 32| 0| return (pos == container1->cardinality); 33| 0|} run_container_equals_array: 36| 839| const array_container_t* container2) { 37| 839| if (run_container_cardinality(container1) != container2->cardinality) ------------------ | Branch (37:9): [True: 794, False: 45] ------------------ 38| 794| return false; 39| 45| int32_t pos = 0; 40| 45| for (int i = 0; i < container1->n_runs; ++i) { ------------------ | Branch (40:21): [True: 45, False: 0] ------------------ 41| 45| const uint32_t run_start = container1->runs[i].value; 42| 45| const uint32_t le = container1->runs[i].length; 43| | 44| 45| if (container2->array[pos] != run_start) { ------------------ | Branch (44:13): [True: 45, False: 0] ------------------ 45| 45| return false; 46| 45| } 47| | 48| 0| if (container2->array[pos + le] != run_start + le) { ------------------ | Branch (48:13): [True: 0, False: 0] ------------------ 49| 0| return false; 50| 0| } 51| | 52| 0| pos += le + 1; 53| 0| } 54| 0| return true; 55| 45|} array_bitset_container_intersection: 21| 602| array_container_t *dst) { 22| 602| if (dst->capacity < src_1->cardinality) { ------------------ | Branch (22:9): [True: 602, False: 0] ------------------ 23| 602| array_container_grow(dst, src_1->cardinality, false); 24| 602| } 25| 602| int32_t newcard = 0; // dst could be src_1 26| 602| const int32_t origcard = src_1->cardinality; 27| 293k| for (int i = 0; i < origcard; ++i) { ------------------ | Branch (27:21): [True: 292k, False: 602] ------------------ 28| 292k| uint16_t key = src_1->array[i]; 29| | // this branchless approach is much faster... 30| 292k| dst->array[newcard] = key; 31| 292k| newcard += bitset_container_contains(src_2, key); 32| | /** 33| | * we could do it this way instead... 34| | * if (bitset_container_contains(src_2, key)) { 35| | * dst->array[newcard++] = key; 36| | * } 37| | * but if the result is unpredictible, the processor generates 38| | * many mispredicted branches. 39| | * Difference can be huge (from 3 cycles when predictible all the way 40| | * to 16 cycles when unpredictible. 41| | * See 42| | * https://github.com/lemire/Code-used-on-Daniel-Lemire-s-blog/blob/master/extra/bitset/c/arraybitsetintersection.c 43| | */ 44| 292k| } 45| 602| dst->cardinality = newcard; 46| 602|} array_bitset_container_intersection_cardinality: 50| 2.21k| const array_container_t *src_1, const bitset_container_t *src_2) { 51| 2.21k| int32_t newcard = 0; 52| 2.21k| const int32_t origcard = src_1->cardinality; 53| 373k| for (int i = 0; i < origcard; ++i) { ------------------ | Branch (53:21): [True: 370k, False: 2.21k] ------------------ 54| 370k| uint16_t key = src_1->array[i]; 55| 370k| newcard += bitset_container_contains(src_2, key); 56| 370k| } 57| 2.21k| return newcard; 58| 2.21k|} array_bitset_container_intersect: 61| 553| const bitset_container_t *src_2) { 62| 553| const int32_t origcard = src_1->cardinality; 63| 5.85k| for (int i = 0; i < origcard; ++i) { ------------------ | Branch (63:21): [True: 5.82k, False: 28] ------------------ 64| 5.82k| uint16_t key = src_1->array[i]; 65| 5.82k| if (bitset_container_contains(src_2, key)) return true; ------------------ | Branch (65:13): [True: 525, False: 5.29k] ------------------ 66| 5.82k| } 67| 28| return false; 68| 553|} array_run_container_intersection: 75| 3.13k| array_container_t *dst) { 76| 3.13k| if (run_container_is_full(src_2)) { ------------------ | Branch (76:9): [True: 34, False: 3.09k] ------------------ 77| 34| if (dst != src_1) array_container_copy(src_1, dst); ------------------ | Branch (77:13): [True: 34, False: 0] ------------------ 78| 34| return; 79| 34| } 80| 3.09k| if (dst->capacity < src_1->cardinality) { ------------------ | Branch (80:9): [True: 3.09k, False: 0] ------------------ 81| 3.09k| array_container_grow(dst, src_1->cardinality, false); 82| 3.09k| } 83| 3.09k| if (src_2->n_runs == 0) { ------------------ | Branch (83:9): [True: 0, False: 3.09k] ------------------ 84| 0| return; 85| 0| } 86| 3.09k| int32_t rlepos = 0; 87| 3.09k| int32_t arraypos = 0; 88| 3.09k| rle16_t rle = src_2->runs[rlepos]; 89| 3.09k| int32_t newcard = 0; 90| 410k| while (arraypos < src_1->cardinality) { ------------------ | Branch (90:12): [True: 407k, False: 2.58k] ------------------ 91| 407k| const uint16_t arrayval = src_1->array[arraypos]; 92| 550k| while (rle.value + rle.length < ------------------ | Branch (92:16): [True: 143k, False: 407k] ------------------ 93| 550k| arrayval) { // this will frequently be false 94| 143k| ++rlepos; 95| 143k| if (rlepos == src_2->n_runs) { ------------------ | Branch (95:17): [True: 514, False: 143k] ------------------ 96| 514| dst->cardinality = newcard; 97| 514| return; // we are done 98| 514| } 99| 143k| rle = src_2->runs[rlepos]; 100| 143k| } 101| 407k| if (rle.value > arrayval) { ------------------ | Branch (101:13): [True: 28.2k, False: 379k] ------------------ 102| 28.2k| arraypos = advanceUntil(src_1->array, arraypos, src_1->cardinality, 103| 28.2k| rle.value); 104| 379k| } else { 105| 379k| dst->array[newcard] = arrayval; 106| 379k| newcard++; 107| 379k| arraypos++; 108| 379k| } 109| 407k| } 110| 2.58k| dst->cardinality = newcard; 111| 2.58k|} run_bitset_container_intersection: 119| 606| container_t **dst) { 120| 606| if (run_container_is_full(src_1)) { ------------------ | Branch (120:9): [True: 0, False: 606] ------------------ 121| 0| if (*dst != src_2) *dst = bitset_container_clone(src_2); ------------------ | Branch (121:13): [True: 0, False: 0] ------------------ 122| 0| return true; 123| 0| } 124| 606| int32_t card = run_container_cardinality(src_1); 125| 606| if (card <= DEFAULT_MAX_SIZE) { ------------------ | Branch (125:9): [True: 606, False: 0] ------------------ 126| | // result can only be an array (assuming that we never make a 127| | // RunContainer) 128| 606| if (card > src_2->cardinality) { ------------------ | Branch (128:13): [True: 0, False: 606] ------------------ 129| 0| card = src_2->cardinality; 130| 0| } 131| 606| array_container_t *answer = array_container_create_given_capacity(card); 132| 606| *dst = answer; 133| 606| if (*dst == NULL) { ------------------ | Branch (133:13): [True: 0, False: 606] ------------------ 134| 0| return false; 135| 0| } 136| 51.5k| for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { ------------------ | Branch (136:34): [True: 50.9k, False: 606] ------------------ 137| 50.9k| rle16_t rle = src_1->runs[rlepos]; 138| 50.9k| uint32_t endofrun = (uint32_t)rle.value + rle.length; 139| 183k| for (uint32_t runValue = rle.value; runValue <= endofrun; ------------------ | Branch (139:49): [True: 132k, False: 50.9k] ------------------ 140| 132k| ++runValue) { 141| 132k| answer->array[answer->cardinality] = (uint16_t)runValue; 142| 132k| answer->cardinality += 143| 132k| bitset_container_contains(src_2, runValue); 144| 132k| } 145| 50.9k| } 146| 606| return false; 147| 606| } 148| 0| if (*dst == src_2) { // we attempt in-place ------------------ | Branch (148:9): [True: 0, False: 0] ------------------ 149| 0| bitset_container_t *answer = CAST_bitset(*dst); ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 150| 0| uint32_t start = 0; 151| 0| for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { ------------------ | Branch (151:34): [True: 0, False: 0] ------------------ 152| 0| const rle16_t rle = src_1->runs[rlepos]; 153| 0| uint32_t end = rle.value; 154| 0| bitset_reset_range(src_2->words, start, end); 155| | 156| 0| start = end + rle.length + 1; 157| 0| } 158| 0| bitset_reset_range(src_2->words, start, UINT32_C(1) << 16); 159| 0| answer->cardinality = bitset_container_compute_cardinality(answer); 160| 0| if (src_2->cardinality > DEFAULT_MAX_SIZE) { ------------------ | Branch (160:13): [True: 0, False: 0] ------------------ 161| 0| return true; 162| 0| } else { 163| 0| array_container_t *newanswer = array_container_from_bitset(src_2); 164| 0| if (newanswer == NULL) { ------------------ | Branch (164:17): [True: 0, False: 0] ------------------ 165| 0| *dst = NULL; 166| 0| return false; 167| 0| } 168| 0| *dst = newanswer; 169| 0| return false; 170| 0| } 171| 0| } else { // no inplace 172| | // we expect the answer to be a bitmap (if we are lucky) 173| 0| bitset_container_t *answer = bitset_container_clone(src_2); 174| | 175| 0| *dst = answer; 176| 0| if (answer == NULL) { ------------------ | Branch (176:13): [True: 0, False: 0] ------------------ 177| 0| return true; 178| 0| } 179| 0| uint32_t start = 0; 180| 0| for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { ------------------ | Branch (180:34): [True: 0, False: 0] ------------------ 181| 0| const rle16_t rle = src_1->runs[rlepos]; 182| 0| uint32_t end = rle.value; 183| 0| bitset_reset_range(answer->words, start, end); 184| 0| start = end + rle.length + 1; 185| 0| } 186| 0| bitset_reset_range(answer->words, start, UINT32_C(1) << 16); 187| 0| answer->cardinality = bitset_container_compute_cardinality(answer); 188| | 189| 0| if (answer->cardinality > DEFAULT_MAX_SIZE) { ------------------ | Branch (189:13): [True: 0, False: 0] ------------------ 190| 0| return true; 191| 0| } else { 192| 0| array_container_t *newanswer = array_container_from_bitset(answer); 193| 0| bitset_container_free(CAST_bitset(*dst)); ------------------ | | 52| 0|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 194| 0| if (newanswer == NULL) { ------------------ | Branch (194:17): [True: 0, False: 0] ------------------ 195| 0| *dst = NULL; 196| 0| return false; 197| 0| } 198| 0| *dst = newanswer; 199| | return false; 200| 0| } 201| 0| } 202| 0|} array_run_container_intersection_cardinality: 206| 7.70k| const run_container_t *src_2) { 207| 7.70k| if (run_container_is_full(src_2)) { ------------------ | Branch (207:9): [True: 136, False: 7.56k] ------------------ 208| 136| return src_1->cardinality; 209| 136| } 210| 7.56k| if (src_2->n_runs == 0) { ------------------ | Branch (210:9): [True: 0, False: 7.56k] ------------------ 211| 0| return 0; 212| 0| } 213| 7.56k| int32_t rlepos = 0; 214| 7.56k| int32_t arraypos = 0; 215| 7.56k| rle16_t rle = src_2->runs[rlepos]; 216| 7.56k| int32_t newcard = 0; 217| 734k| while (arraypos < src_1->cardinality) { ------------------ | Branch (217:12): [True: 728k, False: 5.88k] ------------------ 218| 728k| const uint16_t arrayval = src_1->array[arraypos]; 219| 1.08M| while (rle.value + rle.length < ------------------ | Branch (219:16): [True: 360k, False: 726k] ------------------ 220| 1.08M| arrayval) { // this will frequently be false 221| 360k| ++rlepos; 222| 360k| if (rlepos == src_2->n_runs) { ------------------ | Branch (222:17): [True: 1.68k, False: 358k] ------------------ 223| 1.68k| return newcard; // we are done 224| 1.68k| } 225| 358k| rle = src_2->runs[rlepos]; 226| 358k| } 227| 726k| if (rle.value > arrayval) { ------------------ | Branch (227:13): [True: 98.6k, False: 627k] ------------------ 228| 98.6k| arraypos = advanceUntil(src_1->array, arraypos, src_1->cardinality, 229| 98.6k| rle.value); 230| 627k| } else { 231| 627k| newcard++; 232| 627k| arraypos++; 233| 627k| } 234| 726k| } 235| 5.88k| return newcard; 236| 7.56k|} run_bitset_container_intersection_cardinality: 241| 2.42k| const run_container_t *src_1, const bitset_container_t *src_2) { 242| 2.42k| if (run_container_is_full(src_1)) { ------------------ | Branch (242:9): [True: 0, False: 2.42k] ------------------ 243| 0| return bitset_container_cardinality(src_2); 244| 0| } 245| 2.42k| int answer = 0; 246| 206k| for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { ------------------ | Branch (246:30): [True: 203k, False: 2.42k] ------------------ 247| 203k| rle16_t rle = src_1->runs[rlepos]; 248| 203k| answer += 249| 203k| bitset_lenrange_cardinality(src_2->words, rle.value, rle.length); 250| 203k| } 251| 2.42k| return answer; 252| 2.42k|} array_run_container_intersect: 255| 1.92k| const run_container_t *src_2) { 256| 1.92k| if (run_container_is_full(src_2)) { ------------------ | Branch (256:9): [True: 34, False: 1.89k] ------------------ 257| 34| return !array_container_empty(src_1); 258| 34| } 259| 1.89k| if (src_2->n_runs == 0) { ------------------ | Branch (259:9): [True: 0, False: 1.89k] ------------------ 260| 0| return false; 261| 0| } 262| 1.89k| int32_t rlepos = 0; 263| 1.89k| int32_t arraypos = 0; 264| 1.89k| rle16_t rle = src_2->runs[rlepos]; 265| 3.27k| while (arraypos < src_1->cardinality) { ------------------ | Branch (265:12): [True: 3.22k, False: 47] ------------------ 266| 3.22k| const uint16_t arrayval = src_1->array[arraypos]; 267| 8.00k| while (rle.value + rle.length < ------------------ | Branch (267:16): [True: 4.81k, False: 3.18k] ------------------ 268| 8.00k| arrayval) { // this will frequently be false 269| 4.81k| ++rlepos; 270| 4.81k| if (rlepos == src_2->n_runs) { ------------------ | Branch (270:17): [True: 40, False: 4.77k] ------------------ 271| 40| return false; // we are done 272| 40| } 273| 4.77k| rle = src_2->runs[rlepos]; 274| 4.77k| } 275| 3.18k| if (rle.value > arrayval) { ------------------ | Branch (275:13): [True: 1.37k, False: 1.80k] ------------------ 276| 1.37k| arraypos = advanceUntil(src_1->array, arraypos, src_1->cardinality, 277| 1.37k| rle.value); 278| 1.80k| } else { 279| 1.80k| return true; 280| 1.80k| } 281| 3.18k| } 282| 47| return false; 283| 1.89k|} run_bitset_container_intersect: 288| 606| const bitset_container_t *src_2) { 289| 606| if (run_container_is_full(src_1)) { ------------------ | Branch (289:9): [True: 0, False: 606] ------------------ 290| 0| return !bitset_container_empty(src_2); 291| 0| } 292| 3.59k| for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { ------------------ | Branch (292:30): [True: 3.58k, False: 12] ------------------ 293| 3.58k| rle16_t rle = src_1->runs[rlepos]; 294| 3.58k| if (!bitset_lenrange_empty(src_2->words, rle.value, rle.length)) ------------------ | Branch (294:13): [True: 594, False: 2.99k] ------------------ 295| 594| return true; 296| 3.58k| } 297| 12| return false; 298| 606|} bitset_bitset_container_intersection_inplace: 329| 1.39k| container_t **dst) { 330| 1.39k| const int newCardinality = bitset_container_and_justcard(src_1, src_2); 331| 1.39k| if (newCardinality > DEFAULT_MAX_SIZE) { ------------------ | Branch (331:9): [True: 1.39k, False: 0] ------------------ 332| 1.39k| *dst = src_1; 333| 1.39k| bitset_container_and_nocard(src_1, src_2, src_1); 334| 1.39k| CAST_bitset(*dst)->cardinality = newCardinality; ------------------ | | 52| 1.39k|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 1.39k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 335| 1.39k| return true; // it is a bitset 336| 1.39k| } 337| 0| *dst = array_container_create_given_capacity(newCardinality); 338| 0| if (*dst != NULL) { ------------------ | Branch (338:9): [True: 0, False: 0] ------------------ 339| 0| CAST_array(*dst)->cardinality = newCardinality; ------------------ | | 54| 0|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 340| 0| bitset_extract_intersection_setbits_uint16( 341| 0| src_1->words, src_2->words, BITSET_CONTAINER_SIZE_IN_WORDS, 342| 0| CAST_array(*dst)->array, 0); ------------------ | | 54| 0|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 343| 0| } 344| | return false; // not a bitset 345| 1.39k|} array_container_negation: 33| 151| bitset_container_t *dst) { 34| 151| uint64_t card = UINT64_C(1 << 16); 35| 151| bitset_container_set_all(dst); 36| | 37| 151| if (src->cardinality == 0) { ------------------ | Branch (37:9): [True: 0, False: 151] ------------------ 38| 0| return; 39| 0| } 40| | 41| 151| dst->cardinality = (int32_t)bitset_clear_list(dst->words, card, src->array, 42| 151| (uint64_t)src->cardinality); 43| 151|} bitset_container_negation_inplace: 67| 57| container_t **dst) { 68| 57| return bitset_container_negation_range_inplace(src, 0, (1 << 16), dst); 69| 57|} run_container_negation_inplace: 88| 5.59k|int run_container_negation_inplace(run_container_t *src, container_t **dst) { 89| 5.59k| return run_container_negation_range_inplace(src, 0, (1 << 16), dst); 90| 5.59k|} array_container_negation_range: 99| 3.41k| container_t **dst) { 100| | /* close port of the Java implementation */ 101| 3.41k| if (range_start >= range_end) { ------------------ | Branch (101:9): [True: 0, False: 3.41k] ------------------ 102| 0| *dst = array_container_clone(src); 103| 0| return false; 104| 0| } 105| | 106| 3.41k| int32_t start_index = 107| 3.41k| binarySearch(src->array, src->cardinality, (uint16_t)range_start); 108| 3.41k| if (start_index < 0) start_index = -start_index - 1; ------------------ | Branch (108:9): [True: 3.11k, False: 296] ------------------ 109| | 110| 3.41k| int32_t last_index = 111| 3.41k| binarySearch(src->array, src->cardinality, (uint16_t)(range_end - 1)); 112| 3.41k| if (last_index < 0) last_index = -last_index - 2; ------------------ | Branch (112:9): [True: 3.17k, False: 244] ------------------ 113| | 114| 3.41k| const int32_t current_values_in_range = last_index - start_index + 1; 115| 3.41k| const int32_t span_to_be_flipped = range_end - range_start; 116| 3.41k| const int32_t new_values_in_range = 117| 3.41k| span_to_be_flipped - current_values_in_range; 118| 3.41k| const int32_t cardinality_change = 119| 3.41k| new_values_in_range - current_values_in_range; 120| 3.41k| const int32_t new_cardinality = src->cardinality + cardinality_change; 121| | 122| 3.41k| if (new_cardinality > DEFAULT_MAX_SIZE) { ------------------ | Branch (122:9): [True: 121, False: 3.29k] ------------------ 123| 121| bitset_container_t *temp = bitset_container_from_array(src); 124| 121| bitset_flip_range(temp->words, (uint32_t)range_start, 125| 121| (uint32_t)range_end); 126| 121| temp->cardinality = new_cardinality; 127| 121| *dst = temp; 128| 121| return true; 129| 121| } 130| | 131| 3.29k| array_container_t *arr = 132| 3.29k| array_container_create_given_capacity(new_cardinality); 133| 3.29k| *dst = (container_t *)arr; 134| 3.29k| if (new_cardinality == 0) { ------------------ | Branch (134:9): [True: 18, False: 3.27k] ------------------ 135| 18| arr->cardinality = new_cardinality; 136| 18| return false; // we are done. 137| 18| } 138| | // copy stuff before the active area 139| 3.27k| memcpy(arr->array, src->array, start_index * sizeof(uint16_t)); 140| | 141| | // work on the range 142| 3.27k| int32_t out_pos = start_index, in_pos = start_index; 143| 3.27k| int32_t val_in_range = range_start; 144| 44.6k| for (; val_in_range < range_end && in_pos <= last_index; ++val_in_range) { ------------------ | Branch (144:12): [True: 44.4k, False: 225] | Branch (144:40): [True: 41.4k, False: 3.05k] ------------------ 145| 41.4k| if ((uint16_t)val_in_range != src->array[in_pos]) { ------------------ | Branch (145:13): [True: 26.1k, False: 15.2k] ------------------ 146| 26.1k| arr->array[out_pos++] = (uint16_t)val_in_range; 147| 26.1k| } else { 148| 15.2k| ++in_pos; 149| 15.2k| } 150| 41.4k| } 151| 47.3k| for (; val_in_range < range_end; ++val_in_range) ------------------ | Branch (151:12): [True: 44.1k, False: 3.27k] ------------------ 152| 44.1k| arr->array[out_pos++] = (uint16_t)val_in_range; 153| | 154| | // content after the active range 155| 3.27k| memcpy(arr->array + out_pos, src->array + (last_index + 1), 156| 3.27k| (src->cardinality - (last_index + 1)) * sizeof(uint16_t)); 157| 3.27k| arr->cardinality = new_cardinality; 158| | return false; 159| 3.29k|} array_container_negation_range_inplace: 168| 3.41k| container_t **dst) { 169| 3.41k| bool ans = array_container_negation_range(src, range_start, range_end, dst); 170| | // TODO : try a real inplace version 171| 3.41k| array_container_free(src); 172| 3.41k| return ans; 173| 3.41k|} bitset_container_negation_range_inplace: 217| 1.30k| container_t **dst) { 218| 1.30k| bitset_flip_range(src->words, (uint32_t)range_start, (uint32_t)range_end); 219| 1.30k| src->cardinality = bitset_container_compute_cardinality(src); 220| 1.30k| if (src->cardinality > DEFAULT_MAX_SIZE) { ------------------ | Branch (220:9): [True: 1.24k, False: 56] ------------------ 221| 1.24k| *dst = src; 222| 1.24k| return true; 223| 1.24k| } 224| 56| *dst = array_container_from_bitset(src); 225| 56| bitset_container_free(src); 226| | return false; 227| 1.30k|} run_container_negation_range: 237| 167| container_t **dst) { 238| 167| uint8_t return_typecode; 239| | 240| | // follows the Java implementation 241| 167| if (range_end <= range_start) { ------------------ | Branch (241:9): [True: 0, False: 167] ------------------ 242| 0| *dst = run_container_clone(src); 243| 0| return RUN_CONTAINER_TYPE; ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 244| 0| } 245| | 246| 167| run_container_t *ans = run_container_create_given_capacity( 247| 167| src->n_runs + 1); // src->n_runs + 1); 248| 167| int k = 0; 249| 3.09k| for (; k < src->n_runs && src->runs[k].value < range_start; ++k) { ------------------ | Branch (249:12): [True: 3.04k, False: 49] | Branch (249:31): [True: 2.92k, False: 118] ------------------ 250| 2.92k| ans->runs[k] = src->runs[k]; 251| 2.92k| ans->n_runs++; 252| 2.92k| } 253| | 254| 167| run_container_smart_append_exclusive( 255| 167| ans, (uint16_t)range_start, (uint16_t)(range_end - range_start - 1)); 256| | 257| 4.09k| for (; k < src->n_runs; ++k) { ------------------ | Branch (257:12): [True: 3.92k, False: 167] ------------------ 258| 3.92k| run_container_smart_append_exclusive(ans, src->runs[k].value, 259| 3.92k| src->runs[k].length); 260| 3.92k| } 261| | 262| 167| *dst = convert_run_to_efficient_container(ans, &return_typecode); 263| 167| if (return_typecode != RUN_CONTAINER_TYPE) run_container_free(ans); ------------------ | | 50| 167|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (263:9): [True: 8, False: 159] ------------------ 264| | 265| 167| return return_typecode; 266| 167|} run_container_negation_range_inplace: 278| 7.22k| container_t **dst) { 279| 7.22k| uint8_t return_typecode; 280| | 281| 7.22k| if (range_end <= range_start) { ------------------ | Branch (281:9): [True: 0, False: 7.22k] ------------------ 282| 0| *dst = src; 283| 0| return RUN_CONTAINER_TYPE; ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 284| 0| } 285| | 286| | // TODO: efficient special case when range is 0 to 65535 inclusive 287| | 288| 7.22k| if (src->capacity == src->n_runs) { ------------------ | Branch (288:9): [True: 6.14k, False: 1.08k] ------------------ 289| | // no excess room. More checking to see if result can fit 290| 6.14k| bool last_val_before_range = false; 291| 6.14k| bool first_val_in_range = false; 292| 6.14k| bool last_val_in_range = false; 293| 6.14k| bool first_val_past_range = false; 294| | 295| 6.14k| if (range_start > 0) ------------------ | Branch (295:13): [True: 174, False: 5.96k] ------------------ 296| 174| last_val_before_range = 297| 174| run_container_contains(src, (uint16_t)(range_start - 1)); 298| 6.14k| first_val_in_range = run_container_contains(src, (uint16_t)range_start); 299| | 300| 6.14k| if (last_val_before_range == first_val_in_range) { ------------------ | Branch (300:13): [True: 478, False: 5.66k] ------------------ 301| 478| last_val_in_range = 302| 478| run_container_contains(src, (uint16_t)(range_end - 1)); 303| 478| if (range_end != 0x10000) ------------------ | Branch (303:17): [True: 303, False: 175] ------------------ 304| 303| first_val_past_range = 305| 303| run_container_contains(src, (uint16_t)range_end); 306| | 307| 478| if (last_val_in_range == ------------------ | Branch (307:17): [True: 167, False: 311] ------------------ 308| 478| first_val_past_range) { // no space for inplace 309| 167| int ans = run_container_negation_range(src, range_start, 310| 167| range_end, dst); 311| 167| run_container_free(src); 312| 167| return ans; 313| 167| } 314| 478| } 315| 6.14k| } 316| | // all other cases: result will fit 317| | 318| 7.05k| run_container_t *ans = src; 319| 7.05k| int my_nbr_runs = src->n_runs; 320| | 321| 7.05k| ans->n_runs = 0; 322| 7.05k| int k = 0; 323| 9.14k| for (; (k < my_nbr_runs) && (src->runs[k].value < range_start); ++k) { ------------------ | Branch (323:12): [True: 9.06k, False: 83] | Branch (323:33): [True: 2.09k, False: 6.97k] ------------------ 324| | // ans->runs[k] = src->runs[k]; (would be self-copy) 325| 2.09k| ans->n_runs++; 326| 2.09k| } 327| | 328| | // as with Java implementation, use locals to give self a buffer of depth 1 329| 7.05k| rle16_t buffered = CROARING_MAKE_RLE16(0, 0); ------------------ | | 60| 7.05k| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 330| 7.05k| rle16_t next = buffered; 331| 7.05k| if (k < my_nbr_runs) buffered = src->runs[k]; ------------------ | Branch (331:9): [True: 6.97k, False: 83] ------------------ 332| | 333| 7.05k| run_container_smart_append_exclusive( 334| 7.05k| ans, (uint16_t)range_start, (uint16_t)(range_end - range_start - 1)); 335| | 336| 88.5k| for (; k < my_nbr_runs; ++k) { ------------------ | Branch (336:12): [True: 81.5k, False: 7.05k] ------------------ 337| 81.5k| if (k + 1 < my_nbr_runs) next = src->runs[k + 1]; ------------------ | Branch (337:13): [True: 74.5k, False: 6.97k] ------------------ 338| | 339| 81.5k| run_container_smart_append_exclusive(ans, buffered.value, 340| 81.5k| buffered.length); 341| 81.5k| buffered = next; 342| 81.5k| } 343| | 344| 7.05k| *dst = convert_run_to_efficient_container(ans, &return_typecode); 345| 7.05k| if (return_typecode != RUN_CONTAINER_TYPE) run_container_free(ans); ------------------ | | 50| 7.05k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (345:9): [True: 5.47k, False: 1.58k] ------------------ 346| | 347| 7.05k| return return_typecode; 348| 7.22k|} array_container_is_subset_bitset: 11| 2.24k| const bitset_container_t* container2) { 12| 2.24k| if (container2->cardinality != BITSET_UNKNOWN_CARDINALITY) { ------------------ | Branch (12:9): [True: 2.24k, False: 0] ------------------ 13| 2.24k| if (container2->cardinality < container1->cardinality) { ------------------ | Branch (13:13): [True: 0, False: 2.24k] ------------------ 14| 0| return false; 15| 0| } 16| 2.24k| } 17| 2.24k| for (int i = 0; i < container1->cardinality; ++i) { ------------------ | Branch (17:21): [True: 2.24k, False: 0] ------------------ 18| 2.24k| if (!bitset_container_contains(container2, container1->array[i])) { ------------------ | Branch (18:13): [True: 2.24k, False: 0] ------------------ 19| 2.24k| return false; 20| 2.24k| } 21| 2.24k| } 22| 0| return true; 23| 2.24k|} run_container_is_subset_array: 26| 374| const array_container_t* container2) { 27| 374| if (run_container_cardinality(container1) > container2->cardinality) ------------------ | Branch (27:9): [True: 184, False: 190] ------------------ 28| 184| return false; 29| 190| int32_t start_pos = -1, stop_pos = -1; 30| 190| for (int i = 0; i < container1->n_runs; ++i) { ------------------ | Branch (30:21): [True: 190, False: 0] ------------------ 31| 190| int32_t start = container1->runs[i].value; 32| 190| int32_t stop = start + container1->runs[i].length; 33| 190| start_pos = advanceUntil(container2->array, stop_pos, 34| 190| container2->cardinality, start); 35| 190| stop_pos = advanceUntil(container2->array, stop_pos, 36| 190| container2->cardinality, stop); 37| 190| if (stop_pos == container2->cardinality) { ------------------ | Branch (37:13): [True: 5, False: 185] ------------------ 38| 5| return false; 39| 185| } else if (stop_pos - start_pos != stop - start || ------------------ | Branch (39:20): [True: 74, False: 111] ------------------ 40| 111| container2->array[start_pos] != start || ------------------ | Branch (40:20): [True: 111, False: 0] ------------------ 41| 185| container2->array[stop_pos] != stop) { ------------------ | Branch (41:20): [True: 0, False: 0] ------------------ 42| 185| return false; 43| 185| } 44| 190| } 45| 0| return true; 46| 190|} array_container_is_subset_run: 49| 2.12k| const run_container_t* container2) { 50| 2.12k| if (container1->cardinality > run_container_cardinality(container2)) ------------------ | Branch (50:9): [True: 154, False: 1.96k] ------------------ 51| 154| return false; 52| 1.96k| int i_array = 0, i_run = 0; 53| 5.59k| while (i_array < container1->cardinality && i_run < container2->n_runs) { ------------------ | Branch (53:12): [True: 5.59k, False: 0] | Branch (53:49): [True: 5.57k, False: 19] ------------------ 54| 5.57k| uint32_t start = container2->runs[i_run].value; 55| 5.57k| uint32_t stop = start + container2->runs[i_run].length; 56| 5.57k| if (container1->array[i_array] < start) { ------------------ | Branch (56:13): [True: 1.95k, False: 3.62k] ------------------ 57| 1.95k| return false; 58| 3.62k| } else if (container1->array[i_array] > stop) { ------------------ | Branch (58:20): [True: 3.62k, False: 0] ------------------ 59| 3.62k| i_run++; 60| 3.62k| } else { // the value of the array is in the run 61| 0| i_array++; 62| 0| } 63| 5.57k| } 64| 19| if (i_array == container1->cardinality) { ------------------ | Branch (64:9): [True: 0, False: 19] ------------------ 65| 0| return true; 66| 19| } else { 67| | return false; 68| 19| } 69| 19|} array_bitset_container_union: 24| 1.15k| bitset_container_t *dst) { 25| 1.15k| if (src_2 != dst) bitset_container_copy(src_2, dst); ------------------ | Branch (25:9): [True: 1.15k, False: 0] ------------------ 26| 1.15k| dst->cardinality = (int32_t)bitset_set_list_withcard( 27| 1.15k| dst->words, dst->cardinality, src_1->array, src_1->cardinality); 28| 1.15k|} run_bitset_container_union: 43| 1.21k| bitset_container_t *dst) { 44| 1.21k| assert(!run_container_is_full(src_1)); // catch this case upstream 45| 1.21k| if (src_2 != dst) bitset_container_copy(src_2, dst); ------------------ | Branch (45:9): [True: 1.21k, False: 0] ------------------ 46| 103k| for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { ------------------ | Branch (46:30): [True: 101k, False: 1.21k] ------------------ 47| 101k| rle16_t rle = src_1->runs[rlepos]; 48| 101k| bitset_set_lenrange(dst->words, rle.value, rle.length); 49| 101k| } 50| 1.21k| dst->cardinality = bitset_container_compute_cardinality(dst); 51| 1.21k|} array_run_container_union: 68| 2.50k| run_container_t *dst) { 69| 2.50k| if (run_container_is_full(src_2)) { ------------------ | Branch (69:9): [True: 106, False: 2.40k] ------------------ 70| 106| run_container_copy(src_2, dst); 71| 106| return; 72| 106| } 73| | // TODO: see whether the "2*" is spurious 74| 2.40k| run_container_grow(dst, 2 * (src_1->cardinality + src_2->n_runs), false); 75| 2.40k| int32_t rlepos = 0; 76| 2.40k| int32_t arraypos = 0; 77| 2.40k| rle16_t previousrle; 78| 2.40k| if (src_2->runs[rlepos].value <= src_1->array[arraypos]) { ------------------ | Branch (78:9): [True: 1.83k, False: 565] ------------------ 79| 1.83k| previousrle = run_container_append_first(dst, src_2->runs[rlepos]); 80| 1.83k| rlepos++; 81| 1.83k| } else { 82| 565| previousrle = 83| 565| run_container_append_value_first(dst, src_1->array[arraypos]); 84| 565| arraypos++; 85| 565| } 86| 422k| while ((rlepos < src_2->n_runs) && (arraypos < src_1->cardinality)) { ------------------ | Branch (86:12): [True: 420k, False: 1.54k] | Branch (86:40): [True: 419k, False: 858] ------------------ 87| 419k| if (src_2->runs[rlepos].value <= src_1->array[arraypos]) { ------------------ | Branch (87:13): [True: 136k, False: 283k] ------------------ 88| 136k| run_container_append(dst, src_2->runs[rlepos], &previousrle); 89| 136k| rlepos++; 90| 283k| } else { 91| 283k| run_container_append_value(dst, src_1->array[arraypos], 92| 283k| &previousrle); 93| 283k| arraypos++; 94| 283k| } 95| 419k| } 96| 2.40k| if (arraypos < src_1->cardinality) { ------------------ | Branch (96:9): [True: 1.54k, False: 858] ------------------ 97| 147k| while (arraypos < src_1->cardinality) { ------------------ | Branch (97:16): [True: 145k, False: 1.54k] ------------------ 98| 145k| run_container_append_value(dst, src_1->array[arraypos], 99| 145k| &previousrle); 100| 145k| arraypos++; 101| 145k| } 102| 1.54k| } else { 103| 6.15k| while (rlepos < src_2->n_runs) { ------------------ | Branch (103:16): [True: 5.29k, False: 858] ------------------ 104| 5.29k| run_container_append(dst, src_2->runs[rlepos], &previousrle); 105| 5.29k| rlepos++; 106| 5.29k| } 107| 858| } 108| 2.40k|} array_run_container_inplace_union: 111| 233| run_container_t *src_2) { 112| 233| if (run_container_is_full(src_2)) { ------------------ | Branch (112:9): [True: 0, False: 233] ------------------ 113| 0| return; 114| 0| } 115| 233| const int32_t maxoutput = src_1->cardinality + src_2->n_runs; 116| 233| const int32_t neededcapacity = maxoutput + src_2->n_runs; 117| 233| if (src_2->capacity < neededcapacity) ------------------ | Branch (117:9): [True: 233, False: 0] ------------------ 118| 233| run_container_grow(src_2, neededcapacity, true); 119| 233| memmove(src_2->runs + maxoutput, src_2->runs, 120| 233| src_2->n_runs * sizeof(rle16_t)); 121| 233| rle16_t *inputsrc2 = src_2->runs + maxoutput; 122| 233| int32_t rlepos = 0; 123| 233| int32_t arraypos = 0; 124| 233| int src2nruns = src_2->n_runs; 125| 233| src_2->n_runs = 0; 126| | 127| 233| rle16_t previousrle; 128| | 129| 233| if (inputsrc2[rlepos].value <= src_1->array[arraypos]) { ------------------ | Branch (129:9): [True: 199, False: 34] ------------------ 130| 199| previousrle = run_container_append_first(src_2, inputsrc2[rlepos]); 131| 199| rlepos++; 132| 199| } else { 133| 34| previousrle = 134| 34| run_container_append_value_first(src_2, src_1->array[arraypos]); 135| 34| arraypos++; 136| 34| } 137| | 138| 49.4k| while ((rlepos < src2nruns) && (arraypos < src_1->cardinality)) { ------------------ | Branch (138:12): [True: 49.2k, False: 233] | Branch (138:36): [True: 49.2k, False: 0] ------------------ 139| 49.2k| if (inputsrc2[rlepos].value <= src_1->array[arraypos]) { ------------------ | Branch (139:13): [True: 12.8k, False: 36.4k] ------------------ 140| 12.8k| run_container_append(src_2, inputsrc2[rlepos], &previousrle); 141| 12.8k| rlepos++; 142| 36.4k| } else { 143| 36.4k| run_container_append_value(src_2, src_1->array[arraypos], 144| 36.4k| &previousrle); 145| 36.4k| arraypos++; 146| 36.4k| } 147| 49.2k| } 148| 233| if (arraypos < src_1->cardinality) { ------------------ | Branch (148:9): [True: 233, False: 0] ------------------ 149| 2.10k| while (arraypos < src_1->cardinality) { ------------------ | Branch (149:16): [True: 1.87k, False: 233] ------------------ 150| 1.87k| run_container_append_value(src_2, src_1->array[arraypos], 151| 1.87k| &previousrle); 152| 1.87k| arraypos++; 153| 1.87k| } 154| 233| } else { 155| 0| while (rlepos < src2nruns) { ------------------ | Branch (155:16): [True: 0, False: 0] ------------------ 156| 0| run_container_append(src_2, inputsrc2[rlepos], &previousrle); 157| 0| rlepos++; 158| 0| } 159| 0| } 160| 233|} array_array_container_union: 164| 2.27k| container_t **dst) { 165| 2.27k| int totalCardinality = src_1->cardinality + src_2->cardinality; 166| 2.27k| if (totalCardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (166:9): [True: 2.15k, False: 114] ------------------ 167| 2.15k| *dst = array_container_create_given_capacity(totalCardinality); 168| 2.15k| if (*dst != NULL) { ------------------ | Branch (168:13): [True: 2.15k, False: 0] ------------------ 169| 2.15k| array_container_union(src_1, src_2, CAST_array(*dst)); ------------------ | | 54| 2.15k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 2.15k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 170| 2.15k| } else { 171| 0| return true; // otherwise failure won't be caught 172| 0| } 173| 2.15k| return false; // not a bitset 174| 2.15k| } 175| 114| *dst = bitset_container_create(); 176| 114| bool returnval = true; // expect a bitset 177| 114| if (*dst != NULL) { ------------------ | Branch (177:9): [True: 114, False: 0] ------------------ 178| 114| bitset_container_t *ourbitset = CAST_bitset(*dst); ------------------ | | 52| 114|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 114|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 179| 114| bitset_set_list(ourbitset->words, src_1->array, src_1->cardinality); 180| 114| ourbitset->cardinality = (int32_t)bitset_set_list_withcard( 181| 114| ourbitset->words, src_1->cardinality, src_2->array, 182| 114| src_2->cardinality); 183| 114| if (ourbitset->cardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (183:13): [True: 65, False: 49] ------------------ 184| | // need to convert! 185| 65| *dst = array_container_from_bitset(ourbitset); 186| 65| bitset_container_free(ourbitset); 187| | returnval = false; // not going to be a bitset 188| 65| } 189| 114| } 190| 114| return returnval; 191| 2.27k|} array_array_container_inplace_union: 195| 2.45k| container_t **dst) { 196| 2.45k| int totalCardinality = src_1->cardinality + src_2->cardinality; 197| 2.45k| *dst = NULL; 198| 2.45k| if (totalCardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (198:9): [True: 2.37k, False: 81] ------------------ 199| 2.37k| if (src_1->capacity < totalCardinality) { ------------------ | Branch (199:13): [True: 2.37k, False: 0] ------------------ 200| 2.37k| *dst = array_container_create_given_capacity( 201| 2.37k| 2 * totalCardinality); // be purposefully generous 202| 2.37k| if (*dst != NULL) { ------------------ | Branch (202:17): [True: 2.37k, False: 0] ------------------ 203| 2.37k| array_container_union(src_1, src_2, CAST_array(*dst)); ------------------ | | 54| 2.37k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 2.37k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 204| 2.37k| } else { 205| 0| return true; // otherwise failure won't be caught 206| 0| } 207| 2.37k| return false; // not a bitset 208| 2.37k| } else { 209| 0| memmove(src_1->array + src_2->cardinality, src_1->array, 210| 0| src_1->cardinality * sizeof(uint16_t)); 211| | // In theory, we could use fast_union_uint16, but it is unsafe. It 212| | // fails with Intel compilers in particular. 213| | // https://github.com/RoaringBitmap/CRoaring/pull/452 214| | // See report https://github.com/RoaringBitmap/CRoaring/issues/476 215| 0| src_1->cardinality = (int32_t)union_uint16( 216| 0| src_1->array + src_2->cardinality, src_1->cardinality, 217| 0| src_2->array, src_2->cardinality, src_1->array); 218| 0| return false; // not a bitset 219| 0| } 220| 2.37k| } 221| 81| *dst = bitset_container_create(); 222| 81| bool returnval = true; // expect a bitset 223| 81| if (*dst != NULL) { ------------------ | Branch (223:9): [True: 81, False: 0] ------------------ 224| 81| bitset_container_t *ourbitset = CAST_bitset(*dst); ------------------ | | 52| 81|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 81|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 225| 81| bitset_set_list(ourbitset->words, src_1->array, src_1->cardinality); 226| 81| ourbitset->cardinality = (int32_t)bitset_set_list_withcard( 227| 81| ourbitset->words, src_1->cardinality, src_2->array, 228| 81| src_2->cardinality); 229| 81| if (ourbitset->cardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (229:13): [True: 81, False: 0] ------------------ 230| | // need to convert! 231| 81| if (src_1->capacity < ourbitset->cardinality) { ------------------ | Branch (231:17): [True: 81, False: 0] ------------------ 232| 81| array_container_grow(src_1, ourbitset->cardinality, false); 233| 81| } 234| | 235| 81| bitset_extract_setbits_uint16(ourbitset->words, 236| 81| BITSET_CONTAINER_SIZE_IN_WORDS, 237| 81| src_1->array, 0); 238| 81| src_1->cardinality = ourbitset->cardinality; 239| 81| *dst = src_1; 240| 81| bitset_container_free(ourbitset); 241| | returnval = false; // not going to be a bitset 242| 81| } 243| 81| } 244| 81| return returnval; 245| 2.45k|} array_bitset_container_xor: 25| 642| container_t **dst) { 26| 642| bitset_container_t *result = bitset_container_create(); 27| 642| bitset_container_copy(src_2, result); 28| 642| result->cardinality = (int32_t)bitset_flip_list_withcard( 29| 642| result->words, result->cardinality, src_1->array, src_1->cardinality); 30| | 31| | // do required type conversions. 32| 642| if (result->cardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (32:9): [True: 124, False: 518] ------------------ 33| 124| *dst = array_container_from_bitset(result); 34| 124| bitset_container_free(result); 35| 124| return false; // not bitset 36| 124| } 37| 518| *dst = result; 38| | return true; // bitset 39| 642|} run_bitset_container_xor: 63| 684| container_t **dst) { 64| 684| bitset_container_t *result = bitset_container_create(); 65| | 66| 684| bitset_container_copy(src_2, result); 67| 55.3k| for (int32_t rlepos = 0; rlepos < src_1->n_runs; ++rlepos) { ------------------ | Branch (67:30): [True: 54.6k, False: 684] ------------------ 68| 54.6k| rle16_t rle = src_1->runs[rlepos]; 69| 54.6k| bitset_flip_range(result->words, rle.value, 70| 54.6k| rle.value + rle.length + UINT32_C(1)); 71| 54.6k| } 72| 684| result->cardinality = bitset_container_compute_cardinality(result); 73| | 74| 684| if (result->cardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (74:9): [True: 125, False: 559] ------------------ 75| 125| *dst = array_container_from_bitset(result); 76| 125| bitset_container_free(result); 77| 125| return false; // not bitset 78| 125| } 79| 559| *dst = result; 80| | return true; // bitset 81| 684|} array_run_container_xor: 105| 2.23k| const run_container_t *src_2, container_t **dst) { 106| | // semi following Java XOR implementation as of May 2016 107| | // the C OR implementation works quite differently and can return a run 108| | // container 109| | // TODO could optimize for full run containers. 110| | 111| | // use of lazy following Java impl. 112| 2.23k| const int arbitrary_threshold = 32; 113| 2.23k| if (src_1->cardinality < arbitrary_threshold) { ------------------ | Branch (113:9): [True: 915, False: 1.31k] ------------------ 114| 915| run_container_t *ans = run_container_create(); 115| 915| array_run_container_lazy_xor(src_1, src_2, ans); // keeps runs. 116| 915| uint8_t typecode_after; 117| 915| *dst = 118| 915| convert_run_to_efficient_container_and_free(ans, &typecode_after); 119| 915| return typecode_after; 120| 915| } 121| | 122| 1.31k| int card = run_container_cardinality(src_2); 123| 1.31k| if (card <= DEFAULT_MAX_SIZE) { ------------------ | Branch (123:9): [True: 1.09k, False: 219] ------------------ 124| | // Java implementation works with the array, xoring the run elements via 125| | // iterator 126| 1.09k| array_container_t *temp = array_container_from_run(src_2); 127| 1.09k| bool ret_is_bitset = array_array_container_xor(temp, src_1, dst); 128| 1.09k| array_container_free(temp); 129| 1.09k| return ret_is_bitset ? BITSET_CONTAINER_TYPE : ARRAY_CONTAINER_TYPE; ------------------ | | 48| 8|#define BITSET_CONTAINER_TYPE 1 ------------------ return ret_is_bitset ? BITSET_CONTAINER_TYPE : ARRAY_CONTAINER_TYPE; ------------------ | | 49| 1.09k|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (129:16): [True: 8, False: 1.09k] ------------------ 130| | 131| 1.09k| } else { // guess that it will end up as a bitset 132| 219| bitset_container_t *result = bitset_container_from_run(src_2); 133| 219| bool is_bitset = bitset_array_container_ixor(result, src_1, dst); 134| | // any necessary type conversion has been done by the ixor 135| 219| int retval = (is_bitset ? BITSET_CONTAINER_TYPE : ARRAY_CONTAINER_TYPE); ------------------ | | 48| 206|#define BITSET_CONTAINER_TYPE 1 ------------------ int retval = (is_bitset ? BITSET_CONTAINER_TYPE : ARRAY_CONTAINER_TYPE); ------------------ | | 49| 13|#define ARRAY_CONTAINER_TYPE 2 ------------------ | Branch (135:23): [True: 206, False: 13] ------------------ 136| 219| return retval; 137| 219| } 138| 1.31k|} array_run_container_lazy_xor: 147| 915| run_container_t *dst) { 148| 915| run_container_grow(dst, src_1->cardinality + src_2->n_runs, false); 149| 915| int32_t rlepos = 0; 150| 915| int32_t arraypos = 0; 151| 915| dst->n_runs = 0; 152| | 153| 28.8k| while ((rlepos < src_2->n_runs) && (arraypos < src_1->cardinality)) { ------------------ | Branch (153:12): [True: 28.3k, False: 544] | Branch (153:40): [True: 27.9k, False: 371] ------------------ 154| 27.9k| if (src_2->runs[rlepos].value <= src_1->array[arraypos]) { ------------------ | Branch (154:13): [True: 17.0k, False: 10.8k] ------------------ 155| 17.0k| run_container_smart_append_exclusive(dst, src_2->runs[rlepos].value, 156| 17.0k| src_2->runs[rlepos].length); 157| 17.0k| rlepos++; 158| 17.0k| } else { 159| 10.8k| run_container_smart_append_exclusive(dst, src_1->array[arraypos], 160| 10.8k| 0); 161| 10.8k| arraypos++; 162| 10.8k| } 163| 27.9k| } 164| 3.14k| while (arraypos < src_1->cardinality) { ------------------ | Branch (164:12): [True: 2.22k, False: 915] ------------------ 165| 2.22k| run_container_smart_append_exclusive(dst, src_1->array[arraypos], 0); 166| 2.22k| arraypos++; 167| 2.22k| } 168| 7.99k| while (rlepos < src_2->n_runs) { ------------------ | Branch (168:12): [True: 7.07k, False: 915] ------------------ 169| 7.07k| run_container_smart_append_exclusive(dst, src_2->runs[rlepos].value, 170| 7.07k| src_2->runs[rlepos].length); 171| 7.07k| rlepos++; 172| 7.07k| } 173| 915|} run_run_container_xor: 180| 55.3k| const run_container_t *src_2, container_t **dst) { 181| 55.3k| run_container_t *ans = run_container_create(); 182| 55.3k| run_container_xor(src_1, src_2, ans); 183| 55.3k| uint8_t typecode_after; 184| 55.3k| *dst = convert_run_to_efficient_container_and_free(ans, &typecode_after); 185| 55.3k| return typecode_after; 186| 55.3k|} array_array_container_xor: 198| 8.19k| container_t **dst) { 199| 8.19k| int totalCardinality = 200| 8.19k| src_1->cardinality + src_2->cardinality; // upper bound 201| 8.19k| if (totalCardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (201:9): [True: 7.70k, False: 485] ------------------ 202| 7.70k| *dst = array_container_create_given_capacity(totalCardinality); 203| 7.70k| array_container_xor(src_1, src_2, CAST_array(*dst)); ------------------ | | 54| 7.70k|#define CAST_array(c) CAST(array_container_t *, c) // safer downcast | | ------------------ | | | | 86| 7.70k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 204| 7.70k| return false; // not a bitset 205| 7.70k| } 206| 485| *dst = bitset_container_from_array(src_1); 207| 485| bool returnval = true; // expect a bitset 208| 485| bitset_container_t *ourbitset = CAST_bitset(*dst); ------------------ | | 52| 485|#define CAST_bitset(c) CAST(bitset_container_t *, c) // safer downcast | | ------------------ | | | | 86| 485|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 209| 485| ourbitset->cardinality = (uint32_t)bitset_flip_list_withcard( 210| 485| ourbitset->words, src_1->cardinality, src_2->array, src_2->cardinality); 211| 485| if (ourbitset->cardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (211:9): [True: 468, False: 17] ------------------ 212| | // need to convert! 213| 468| *dst = array_container_from_bitset(ourbitset); 214| 468| bitset_container_free(ourbitset); 215| 468| returnval = false; // not going to be a bitset 216| 468| } 217| | 218| 485| return returnval; 219| 8.19k|} bitset_array_container_ixor: 284| 219| container_t **dst) { 285| 219| *dst = src_1; 286| 219| src_1->cardinality = (uint32_t)bitset_flip_list_withcard( 287| 219| src_1->words, src_1->cardinality, src_2->array, src_2->cardinality); 288| | 289| 219| if (src_1->cardinality <= DEFAULT_MAX_SIZE) { ------------------ | Branch (289:9): [True: 13, False: 206] ------------------ 290| 13| *dst = array_container_from_bitset(src_1); 291| 13| bitset_container_free(src_1); 292| 13| return false; // not bitset 293| 13| } else 294| 206| return true; 295| 219|} bitset_bitset_container_ixor: 304| 1.43k| container_t **dst) { 305| 1.43k| int card = bitset_container_xor(src_1, src_2, src_1); 306| 1.43k| if (card <= DEFAULT_MAX_SIZE) { ------------------ | Branch (306:9): [True: 1.43k, False: 0] ------------------ 307| 1.43k| *dst = array_container_from_bitset(src_1); 308| 1.43k| bitset_container_free(src_1); 309| 1.43k| return false; // not bitset 310| 1.43k| } else { 311| 0| *dst = src_1; 312| | return true; 313| 0| } 314| 1.43k|} array_bitset_container_ixor: 318| 89| container_t **dst) { 319| | bool ans = array_bitset_container_xor(src_1, src_2, dst); 320| 89| array_container_free(src_1); 321| 89| return ans; 322| 89|} run_bitset_container_ixor: 333| 10| container_t **dst) { 334| | bool ans = run_bitset_container_xor(src_1, src_2, dst); 335| 10| run_container_free(src_1); 336| 10| return ans; 337| 10|} bitset_run_container_ixor: 341| 68| container_t **dst) { 342| | bool ans = run_bitset_container_xor(src_2, src_1, dst); 343| 68| bitset_container_free(src_1); 344| 68| return ans; 345| 68|} array_run_container_ixor: 352| 253| const run_container_t *src_2, container_t **dst) { 353| 253| int ans = array_run_container_xor(src_1, src_2, dst); 354| 253| array_container_free(src_1); 355| 253| return ans; 356| 253|} run_array_container_ixor: 360| 53| container_t **dst) { 361| 53| int ans = array_run_container_xor(src_2, src_1, dst); 362| 53| run_container_free(src_1); 363| 53| return ans; 364| 53|} array_array_container_ixor: 368| 4.82k| container_t **dst) { 369| | bool ans = array_array_container_xor(src_1, src_2, dst); 370| 4.82k| array_container_free(src_1); 371| 4.82k| return ans; 372| 4.82k|} run_run_container_ixor: 375| 54.7k| container_t **dst) { 376| 54.7k| int ans = run_run_container_xor(src_1, src_2, dst); 377| 54.7k| run_container_free(src_1); 378| 54.7k| return ans; 379| 54.7k|} run_container_add: 39| 390k|bool run_container_add(run_container_t *run, uint16_t pos) { 40| 390k| int32_t index = interleavedBinarySearch(run->runs, run->n_runs, pos); 41| 390k| if (index >= 0) return false; // already there ------------------ | Branch (41:9): [True: 174k, False: 216k] ------------------ 42| 216k| index = -index - 2; // points to preceding value, possibly -1 43| 216k| if (index >= 0) { // possible match ------------------ | Branch (43:9): [True: 215k, False: 361] ------------------ 44| 215k| int32_t offset = pos - run->runs[index].value; 45| 215k| int32_t le = run->runs[index].length; 46| 215k| if (offset <= le) return false; // already there ------------------ | Branch (46:13): [True: 127k, False: 88.1k] ------------------ 47| 88.1k| if (offset == le + 1) { ------------------ | Branch (47:13): [True: 13.8k, False: 74.3k] ------------------ 48| | // we may need to fuse 49| 13.8k| if (index + 1 < run->n_runs) { ------------------ | Branch (49:17): [True: 13.6k, False: 274] ------------------ 50| 13.6k| if (run->runs[index + 1].value == pos + 1) { ------------------ | Branch (50:21): [True: 6.31k, False: 7.28k] ------------------ 51| | // indeed fusion is needed 52| 6.31k| run->runs[index].length = run->runs[index + 1].value + 53| 6.31k| run->runs[index + 1].length - 54| 6.31k| run->runs[index].value; 55| 6.31k| recoverRoomAtIndex(run, (uint16_t)(index + 1)); 56| 6.31k| return true; 57| 6.31k| } 58| 13.6k| } 59| 7.56k| run->runs[index].length++; 60| 7.56k| return true; 61| 13.8k| } 62| 74.3k| if (index + 1 < run->n_runs) { ------------------ | Branch (62:13): [True: 72.5k, False: 1.74k] ------------------ 63| | // we may need to fuse 64| 72.5k| if (run->runs[index + 1].value == pos + 1) { ------------------ | Branch (64:17): [True: 31.9k, False: 40.6k] ------------------ 65| | // indeed fusion is needed 66| 31.9k| run->runs[index + 1].value = pos; 67| 31.9k| run->runs[index + 1].length = run->runs[index + 1].length + 1; 68| 31.9k| return true; 69| 31.9k| } 70| 72.5k| } 71| 74.3k| } 72| 42.7k| if (index == -1) { ------------------ | Branch (72:9): [True: 361, False: 42.3k] ------------------ 73| | // we may need to extend the first run 74| 361| if (0 < run->n_runs) { ------------------ | Branch (74:13): [True: 361, False: 0] ------------------ 75| 361| if (run->runs[0].value == pos + 1) { ------------------ | Branch (75:17): [True: 234, False: 127] ------------------ 76| 234| run->runs[0].length++; 77| 234| run->runs[0].value--; 78| 234| return true; 79| 234| } 80| 361| } 81| 361| } 82| 42.5k| makeRoomAtIndex(run, (uint16_t)(index + 1)); 83| 42.5k| run->runs[index + 1].value = pos; 84| 42.5k| run->runs[index + 1].length = 0; 85| | return true; 86| 42.7k|} run_container_create_given_capacity: 89| 370k|run_container_t *run_container_create_given_capacity(int32_t size) { 90| 370k| run_container_t *run; 91| | /* Allocate the run container itself. */ 92| 370k| if ((run = (run_container_t *)roaring_malloc(sizeof(run_container_t))) == ------------------ | Branch (92:9): [True: 0, False: 370k] ------------------ 93| 370k| NULL) { 94| 0| return NULL; 95| 0| } 96| 370k| if (size <= 0) { // we don't want to rely on malloc(0) ------------------ | Branch (96:9): [True: 116k, False: 253k] ------------------ 97| 116k| run->runs = NULL; 98| 253k| } else if ((run->runs = (rle16_t *)roaring_malloc(sizeof(rle16_t) * ------------------ | Branch (98:16): [True: 0, False: 253k] ------------------ 99| 253k| size)) == NULL) { 100| 0| roaring_free(run); 101| 0| return NULL; 102| 0| } 103| 370k| run->capacity = size; 104| 370k| run->n_runs = 0; 105| 370k| return run; 106| 370k|} run_container_shrink_to_fit: 108| 2.51k|int run_container_shrink_to_fit(run_container_t *src) { 109| 2.51k| if (src->n_runs == src->capacity) return 0; // nothing to do ------------------ | Branch (109:9): [True: 2.51k, False: 0] ------------------ 110| 0| int savings = src->capacity - src->n_runs; 111| 0| src->capacity = src->n_runs; 112| 0| rle16_t *oldruns = src->runs; 113| 0| src->runs = 114| 0| (rle16_t *)roaring_realloc(oldruns, src->capacity * sizeof(rle16_t)); 115| 0| if (src->runs == NULL) roaring_free(oldruns); // should never happen? ------------------ | Branch (115:9): [True: 0, False: 0] ------------------ 116| 0| return savings; 117| 2.51k|} run_container_create: 119| 116k|run_container_t *run_container_create(void) { 120| 116k| return run_container_create_given_capacity(RUN_DEFAULT_INIT_SIZE); 121| 116k|} run_container_clone: 124| 163k|run_container_t *run_container_clone(const run_container_t *src) { 125| 163k| run_container_t *run = run_container_create_given_capacity(src->capacity); 126| 163k| if (run == NULL) return NULL; ------------------ | Branch (126:9): [True: 0, False: 163k] ------------------ 127| 163k| run->capacity = src->capacity; 128| 163k| run->n_runs = src->n_runs; 129| 163k| memcpy(run->runs, src->runs, src->n_runs * sizeof(rle16_t)); 130| 163k| return run; 131| 163k|} run_container_free: 193| 370k|void run_container_free(run_container_t *run) { 194| 370k| if (run == NULL) return; ------------------ | Branch (194:9): [True: 0, False: 370k] ------------------ 195| 370k| roaring_free(run->runs); 196| 370k| roaring_free(run); 197| 370k|} run_container_grow: 199| 121k|void run_container_grow(run_container_t *run, int32_t min, bool copy) { 200| 121k| int32_t newCapacity = (run->capacity == 0) ? RUN_DEFAULT_INIT_SIZE ------------------ | Branch (200:27): [True: 116k, False: 4.92k] ------------------ 201| 121k| : run->capacity < 64 ? run->capacity * 2 ------------------ | Branch (201:29): [True: 3.40k, False: 1.51k] ------------------ 202| 4.92k| : run->capacity < 1024 ? run->capacity * 3 / 2 ------------------ | Branch (202:29): [True: 1.51k, False: 0] ------------------ 203| 1.51k| : run->capacity * 5 / 4; 204| 121k| if (newCapacity < min) newCapacity = min; ------------------ | Branch (204:9): [True: 118k, False: 3.04k] ------------------ 205| 121k| run->capacity = newCapacity; 206| 121k| assert(run->capacity >= min); 207| 121k| if (copy) { ------------------ | Branch (207:9): [True: 4.92k, False: 116k] ------------------ 208| 4.92k| rle16_t *oldruns = run->runs; 209| 4.92k| run->runs = (rle16_t *)roaring_realloc(oldruns, 210| 4.92k| run->capacity * sizeof(rle16_t)); 211| 4.92k| if (run->runs == NULL) roaring_free(oldruns); ------------------ | Branch (211:13): [True: 0, False: 4.92k] ------------------ 212| 116k| } else { 213| 116k| roaring_free(run->runs); 214| 116k| run->runs = (rle16_t *)roaring_malloc(run->capacity * sizeof(rle16_t)); 215| 116k| } 216| | // We may have run->runs == NULL. 217| 121k|} run_container_copy: 220| 53.0k|void run_container_copy(const run_container_t *src, run_container_t *dst) { 221| 53.0k| const int32_t n_runs = src->n_runs; 222| 53.0k| if (src->n_runs > dst->capacity) { ------------------ | Branch (222:9): [True: 53.0k, False: 30] ------------------ 223| | run_container_grow(dst, n_runs, false); 224| 53.0k| } 225| 53.0k| dst->n_runs = n_runs; 226| 53.0k| memcpy(dst->runs, src->runs, sizeof(rle16_t) * n_runs); 227| 53.0k|} run_container_union: 232| 677| const run_container_t *src_2, run_container_t *dst) { 233| | // TODO: this could be a lot more efficient 234| | 235| | // we start out with inexpensive checks 236| 677| const bool if1 = run_container_is_full(src_1); 237| 677| const bool if2 = run_container_is_full(src_2); 238| 677| if (if1 || if2) { ------------------ | Branch (238:9): [True: 0, False: 677] | Branch (238:16): [True: 14, False: 663] ------------------ 239| 14| if (if1) { ------------------ | Branch (239:13): [True: 0, False: 14] ------------------ 240| 0| run_container_copy(src_1, dst); 241| 0| return; 242| 0| } 243| 14| if (if2) { ------------------ | Branch (243:13): [True: 14, False: 0] ------------------ 244| 14| run_container_copy(src_2, dst); 245| 14| return; 246| 14| } 247| 14| } 248| 663| const int32_t neededcapacity = src_1->n_runs + src_2->n_runs; 249| 663| if (dst->capacity < neededcapacity) ------------------ | Branch (249:9): [True: 663, False: 0] ------------------ 250| 663| run_container_grow(dst, neededcapacity, false); 251| 663| dst->n_runs = 0; 252| 663| int32_t rlepos = 0; 253| 663| int32_t xrlepos = 0; 254| | 255| 663| rle16_t previousrle; 256| 663| if (src_1->runs[rlepos].value <= src_2->runs[xrlepos].value) { ------------------ | Branch (256:9): [True: 610, False: 53] ------------------ 257| 610| previousrle = run_container_append_first(dst, src_1->runs[rlepos]); 258| 610| rlepos++; 259| 610| } else { 260| 53| previousrle = run_container_append_first(dst, src_2->runs[xrlepos]); 261| 53| xrlepos++; 262| 53| } 263| | 264| 81.5k| while ((xrlepos < src_2->n_runs) && (rlepos < src_1->n_runs)) { ------------------ | Branch (264:12): [True: 81.2k, False: 291] | Branch (264:41): [True: 80.8k, False: 372] ------------------ 265| 80.8k| rle16_t newrl; 266| 80.8k| if (src_1->runs[rlepos].value <= src_2->runs[xrlepos].value) { ------------------ | Branch (266:13): [True: 43.8k, False: 36.9k] ------------------ 267| 43.8k| newrl = src_1->runs[rlepos]; 268| 43.8k| rlepos++; 269| 43.8k| } else { 270| 36.9k| newrl = src_2->runs[xrlepos]; 271| 36.9k| xrlepos++; 272| 36.9k| } 273| 80.8k| run_container_append(dst, newrl, &previousrle); 274| 80.8k| } 275| 1.69k| while (xrlepos < src_2->n_runs) { ------------------ | Branch (275:12): [True: 1.03k, False: 663] ------------------ 276| 1.03k| run_container_append(dst, src_2->runs[xrlepos], &previousrle); 277| 1.03k| xrlepos++; 278| 1.03k| } 279| 13.7k| while (rlepos < src_1->n_runs) { ------------------ | Branch (279:12): [True: 13.0k, False: 663] ------------------ 280| 13.0k| run_container_append(dst, src_1->runs[rlepos], &previousrle); 281| 13.0k| rlepos++; 282| 13.0k| } 283| 663|} run_container_union_inplace: 288| 1.68k| const run_container_t *src_2) { 289| | // TODO: this could be a lot more efficient 290| | 291| | // we start out with inexpensive checks 292| 1.68k| const bool if1 = run_container_is_full(src_1); 293| 1.68k| const bool if2 = run_container_is_full(src_2); 294| 1.68k| if (if1 || if2) { ------------------ | Branch (294:9): [True: 0, False: 1.68k] | Branch (294:16): [True: 30, False: 1.65k] ------------------ 295| 30| if (if1) { ------------------ | Branch (295:13): [True: 0, False: 30] ------------------ 296| 0| return; 297| 0| } 298| 30| if (if2) { ------------------ | Branch (298:13): [True: 30, False: 0] ------------------ 299| 30| run_container_copy(src_2, src_1); 300| 30| return; 301| 30| } 302| 30| } 303| | // we move the data to the end of the current array 304| 1.65k| const int32_t maxoutput = src_1->n_runs + src_2->n_runs; 305| 1.65k| const int32_t neededcapacity = maxoutput + src_1->n_runs; 306| 1.65k| if (src_1->capacity < neededcapacity) ------------------ | Branch (306:9): [True: 1.65k, False: 0] ------------------ 307| 1.65k| run_container_grow(src_1, neededcapacity, true); 308| 1.65k| memmove(src_1->runs + maxoutput, src_1->runs, 309| 1.65k| src_1->n_runs * sizeof(rle16_t)); 310| 1.65k| rle16_t *inputsrc1 = src_1->runs + maxoutput; 311| 1.65k| const int32_t input1nruns = src_1->n_runs; 312| 1.65k| src_1->n_runs = 0; 313| 1.65k| int32_t rlepos = 0; 314| 1.65k| int32_t xrlepos = 0; 315| | 316| 1.65k| rle16_t previousrle; 317| 1.65k| if (inputsrc1[rlepos].value <= src_2->runs[xrlepos].value) { ------------------ | Branch (317:9): [True: 1.56k, False: 85] ------------------ 318| 1.56k| previousrle = run_container_append_first(src_1, inputsrc1[rlepos]); 319| 1.56k| rlepos++; 320| 1.56k| } else { 321| 85| previousrle = run_container_append_first(src_1, src_2->runs[xrlepos]); 322| 85| xrlepos++; 323| 85| } 324| 189k| while ((xrlepos < src_2->n_runs) && (rlepos < input1nruns)) { ------------------ | Branch (324:12): [True: 188k, False: 141] | Branch (324:41): [True: 187k, False: 1.50k] ------------------ 325| 187k| rle16_t newrl; 326| 187k| if (inputsrc1[rlepos].value <= src_2->runs[xrlepos].value) { ------------------ | Branch (326:13): [True: 89.2k, False: 98.2k] ------------------ 327| 89.2k| newrl = inputsrc1[rlepos]; 328| 89.2k| rlepos++; 329| 98.2k| } else { 330| 98.2k| newrl = src_2->runs[xrlepos]; 331| 98.2k| xrlepos++; 332| 98.2k| } 333| 187k| run_container_append(src_1, newrl, &previousrle); 334| 187k| } 335| 4.23k| while (xrlepos < src_2->n_runs) { ------------------ | Branch (335:12): [True: 2.58k, False: 1.65k] ------------------ 336| 2.58k| run_container_append(src_1, src_2->runs[xrlepos], &previousrle); 337| 2.58k| xrlepos++; 338| 2.58k| } 339| 9.32k| while (rlepos < input1nruns) { ------------------ | Branch (339:12): [True: 7.67k, False: 1.65k] ------------------ 340| 7.67k| run_container_append(src_1, inputsrc1[rlepos], &previousrle); 341| 7.67k| rlepos++; 342| 7.67k| } 343| 1.65k|} run_container_xor: 349| 55.3k| const run_container_t *src_2, run_container_t *dst) { 350| | // don't bother to convert xor with full range into negation 351| | // since negation is implemented similarly 352| | 353| 55.3k| const int32_t neededcapacity = src_1->n_runs + src_2->n_runs; 354| 55.3k| if (dst->capacity < neededcapacity) ------------------ | Branch (354:9): [True: 55.3k, False: 0] ------------------ 355| 55.3k| run_container_grow(dst, neededcapacity, false); 356| | 357| 55.3k| int32_t pos1 = 0; 358| 55.3k| int32_t pos2 = 0; 359| 55.3k| dst->n_runs = 0; 360| | 361| 228k| while ((pos1 < src_1->n_runs) && (pos2 < src_2->n_runs)) { ------------------ | Branch (361:12): [True: 173k, False: 55.0k] | Branch (361:38): [True: 173k, False: 313] ------------------ 362| 173k| if (src_1->runs[pos1].value <= src_2->runs[pos2].value) { ------------------ | Branch (362:13): [True: 115k, False: 57.3k] ------------------ 363| 115k| run_container_smart_append_exclusive(dst, src_1->runs[pos1].value, 364| 115k| src_1->runs[pos1].length); 365| 115k| pos1++; 366| 115k| } else { 367| 57.3k| run_container_smart_append_exclusive(dst, src_2->runs[pos2].value, 368| 57.3k| src_2->runs[pos2].length); 369| 57.3k| pos2++; 370| 57.3k| } 371| 173k| } 372| 69.6k| while (pos1 < src_1->n_runs) { ------------------ | Branch (372:12): [True: 14.2k, False: 55.3k] ------------------ 373| 14.2k| run_container_smart_append_exclusive(dst, src_1->runs[pos1].value, 374| 14.2k| src_1->runs[pos1].length); 375| 14.2k| pos1++; 376| 14.2k| } 377| | 378| 112k| while (pos2 < src_2->n_runs) { ------------------ | Branch (378:12): [True: 57.0k, False: 55.3k] ------------------ 379| 57.0k| run_container_smart_append_exclusive(dst, src_2->runs[pos2].value, 380| 57.0k| src_2->runs[pos2].length); 381| 57.0k| pos2++; 382| 57.0k| } 383| 55.3k|} run_container_intersection: 389| 56.3k| run_container_t *dst) { 390| 56.3k| const bool if1 = run_container_is_full(src_1); 391| 56.3k| const bool if2 = run_container_is_full(src_2); 392| 56.3k| if (if1 || if2) { ------------------ | Branch (392:9): [True: 52.8k, False: 3.44k] | Branch (392:16): [True: 14, False: 3.43k] ------------------ 393| 52.9k| if (if1) { ------------------ | Branch (393:13): [True: 52.8k, False: 14] ------------------ 394| 52.8k| run_container_copy(src_2, dst); 395| 52.8k| return; 396| 52.8k| } 397| 14| if (if2) { ------------------ | Branch (397:13): [True: 14, False: 0] ------------------ 398| 14| run_container_copy(src_1, dst); 399| 14| return; 400| 14| } 401| 14| } 402| | // TODO: this could be a lot more efficient, could use SIMD optimizations 403| 3.43k| const int32_t neededcapacity = src_1->n_runs + src_2->n_runs; 404| 3.43k| if (dst->capacity < neededcapacity) ------------------ | Branch (404:9): [True: 3.43k, False: 0] ------------------ 405| 3.43k| run_container_grow(dst, neededcapacity, false); 406| 3.43k| dst->n_runs = 0; 407| 3.43k| int32_t rlepos = 0; 408| 3.43k| int32_t xrlepos = 0; 409| 3.43k| int32_t start = src_1->runs[rlepos].value; 410| 3.43k| int32_t end = start + src_1->runs[rlepos].length + 1; 411| 3.43k| int32_t xstart = src_2->runs[xrlepos].value; 412| 3.43k| int32_t xend = xstart + src_2->runs[xrlepos].length + 1; 413| 174k| while ((rlepos < src_1->n_runs) && (xrlepos < src_2->n_runs)) { ------------------ | Branch (413:12): [True: 171k, False: 2.96k] | Branch (413:40): [True: 170k, False: 473] ------------------ 414| 170k| if (end <= xstart) { ------------------ | Branch (414:13): [True: 42.1k, False: 128k] ------------------ 415| 42.1k| ++rlepos; 416| 42.1k| if (rlepos < src_1->n_runs) { ------------------ | Branch (416:17): [True: 42.0k, False: 82] ------------------ 417| 42.0k| start = src_1->runs[rlepos].value; 418| 42.0k| end = start + src_1->runs[rlepos].length + 1; 419| 42.0k| } 420| 128k| } else if (xend <= start) { ------------------ | Branch (420:20): [True: 12.4k, False: 116k] ------------------ 421| 12.4k| ++xrlepos; 422| 12.4k| if (xrlepos < src_2->n_runs) { ------------------ | Branch (422:17): [True: 12.3k, False: 54] ------------------ 423| 12.3k| xstart = src_2->runs[xrlepos].value; 424| 12.3k| xend = xstart + src_2->runs[xrlepos].length + 1; 425| 12.3k| } 426| 116k| } else { // they overlap 427| 116k| const int32_t lateststart = start > xstart ? start : xstart; ------------------ | Branch (427:41): [True: 15.8k, False: 100k] ------------------ 428| 116k| int32_t earliestend; 429| 116k| if (end == xend) { // improbable ------------------ | Branch (429:17): [True: 86.4k, False: 29.8k] ------------------ 430| 86.4k| earliestend = end; 431| 86.4k| rlepos++; 432| 86.4k| xrlepos++; 433| 86.4k| if (rlepos < src_1->n_runs) { ------------------ | Branch (433:21): [True: 83.7k, False: 2.73k] ------------------ 434| 83.7k| start = src_1->runs[rlepos].value; 435| 83.7k| end = start + src_1->runs[rlepos].length + 1; 436| 83.7k| } 437| 86.4k| if (xrlepos < src_2->n_runs) { ------------------ | Branch (437:21): [True: 83.5k, False: 2.95k] ------------------ 438| 83.5k| xstart = src_2->runs[xrlepos].value; 439| 83.5k| xend = xstart + src_2->runs[xrlepos].length + 1; 440| 83.5k| } 441| 86.4k| } else if (end < xend) { ------------------ | Branch (441:24): [True: 16.2k, False: 13.6k] ------------------ 442| 16.2k| earliestend = end; 443| 16.2k| rlepos++; 444| 16.2k| if (rlepos < src_1->n_runs) { ------------------ | Branch (444:21): [True: 16.0k, False: 142] ------------------ 445| 16.0k| start = src_1->runs[rlepos].value; 446| 16.0k| end = start + src_1->runs[rlepos].length + 1; 447| 16.0k| } 448| | 449| 16.2k| } else { // end > xend 450| 13.6k| earliestend = xend; 451| 13.6k| xrlepos++; 452| 13.6k| if (xrlepos < src_2->n_runs) { ------------------ | Branch (452:21): [True: 13.4k, False: 136] ------------------ 453| 13.4k| xstart = src_2->runs[xrlepos].value; 454| 13.4k| xend = xstart + src_2->runs[xrlepos].length + 1; 455| 13.4k| } 456| 13.6k| } 457| 116k| dst->runs[dst->n_runs].value = (uint16_t)lateststart; 458| 116k| dst->runs[dst->n_runs].length = 459| 116k| (uint16_t)(earliestend - lateststart - 1); 460| 116k| dst->n_runs++; 461| 116k| } 462| 170k| } 463| 3.43k|} run_container_intersection_cardinality: 467| 2.70k| const run_container_t *src_2) { 468| 2.70k| const bool if1 = run_container_is_full(src_1); 469| 2.70k| const bool if2 = run_container_is_full(src_2); 470| 2.70k| if (if1 || if2) { ------------------ | Branch (470:9): [True: 0, False: 2.70k] | Branch (470:16): [True: 56, False: 2.65k] ------------------ 471| 56| if (if1) { ------------------ | Branch (471:13): [True: 0, False: 56] ------------------ 472| 0| return run_container_cardinality(src_2); 473| 0| } 474| 56| if (if2) { ------------------ | Branch (474:13): [True: 56, False: 0] ------------------ 475| 56| return run_container_cardinality(src_1); 476| 56| } 477| 56| } 478| 2.65k| int answer = 0; 479| 2.65k| int32_t rlepos = 0; 480| 2.65k| int32_t xrlepos = 0; 481| 2.65k| int32_t start = src_1->runs[rlepos].value; 482| 2.65k| int32_t end = start + src_1->runs[rlepos].length + 1; 483| 2.65k| int32_t xstart = src_2->runs[xrlepos].value; 484| 2.65k| int32_t xend = xstart + src_2->runs[xrlepos].length + 1; 485| 282k| while ((rlepos < src_1->n_runs) && (xrlepos < src_2->n_runs)) { ------------------ | Branch (485:12): [True: 281k, False: 1.80k] | Branch (485:40): [True: 280k, False: 852] ------------------ 486| 280k| if (end <= xstart) { ------------------ | Branch (486:13): [True: 66.2k, False: 214k] ------------------ 487| 66.2k| ++rlepos; 488| 66.2k| if (rlepos < src_1->n_runs) { ------------------ | Branch (488:17): [True: 65.9k, False: 328] ------------------ 489| 65.9k| start = src_1->runs[rlepos].value; 490| 65.9k| end = start + src_1->runs[rlepos].length + 1; 491| 65.9k| } 492| 214k| } else if (xend <= start) { ------------------ | Branch (492:20): [True: 49.7k, False: 164k] ------------------ 493| 49.7k| ++xrlepos; 494| 49.7k| if (xrlepos < src_2->n_runs) { ------------------ | Branch (494:17): [True: 49.5k, False: 216] ------------------ 495| 49.5k| xstart = src_2->runs[xrlepos].value; 496| 49.5k| xend = xstart + src_2->runs[xrlepos].length + 1; 497| 49.5k| } 498| 164k| } else { // they overlap 499| 164k| const int32_t lateststart = start > xstart ? start : xstart; ------------------ | Branch (499:41): [True: 63.3k, False: 100k] ------------------ 500| 164k| int32_t earliestend; 501| 164k| if (end == xend) { // improbable ------------------ | Branch (501:17): [True: 81.5k, False: 82.6k] ------------------ 502| 81.5k| earliestend = end; 503| 81.5k| rlepos++; 504| 81.5k| xrlepos++; 505| 81.5k| if (rlepos < src_1->n_runs) { ------------------ | Branch (505:21): [True: 80.6k, False: 904] ------------------ 506| 80.6k| start = src_1->runs[rlepos].value; 507| 80.6k| end = start + src_1->runs[rlepos].length + 1; 508| 80.6k| } 509| 81.5k| if (xrlepos < src_2->n_runs) { ------------------ | Branch (509:21): [True: 80.4k, False: 1.06k] ------------------ 510| 80.4k| xstart = src_2->runs[xrlepos].value; 511| 80.4k| xend = xstart + src_2->runs[xrlepos].length + 1; 512| 80.4k| } 513| 82.6k| } else if (end < xend) { ------------------ | Branch (513:24): [True: 64.8k, False: 17.8k] ------------------ 514| 64.8k| earliestend = end; 515| 64.8k| rlepos++; 516| 64.8k| if (rlepos < src_1->n_runs) { ------------------ | Branch (516:21): [True: 64.2k, False: 568] ------------------ 517| 64.2k| start = src_1->runs[rlepos].value; 518| 64.2k| end = start + src_1->runs[rlepos].length + 1; 519| 64.2k| } 520| | 521| 64.8k| } else { // end > xend 522| 17.8k| earliestend = xend; 523| 17.8k| xrlepos++; 524| 17.8k| if (xrlepos < src_2->n_runs) { ------------------ | Branch (524:21): [True: 17.6k, False: 204] ------------------ 525| 17.6k| xstart = src_2->runs[xrlepos].value; 526| 17.6k| xend = xstart + src_2->runs[xrlepos].length + 1; 527| 17.6k| } 528| 17.8k| } 529| 164k| answer += earliestend - lateststart; 530| 164k| } 531| 280k| } 532| 2.65k| return answer; 533| 2.70k|} run_container_intersect: 536| 677| const run_container_t *src_2) { 537| 677| const bool if1 = run_container_is_full(src_1); 538| 677| const bool if2 = run_container_is_full(src_2); 539| 677| if (if1 || if2) { ------------------ | Branch (539:9): [True: 0, False: 677] | Branch (539:16): [True: 14, False: 663] ------------------ 540| 14| if (if1) { ------------------ | Branch (540:13): [True: 0, False: 14] ------------------ 541| 0| return !run_container_empty(src_2); 542| 0| } 543| 14| if (if2) { ------------------ | Branch (543:13): [True: 14, False: 0] ------------------ 544| 14| return !run_container_empty(src_1); 545| 14| } 546| 14| } 547| 663| int32_t rlepos = 0; 548| 663| int32_t xrlepos = 0; 549| 663| int32_t start = src_1->runs[rlepos].value; 550| 663| int32_t end = start + src_1->runs[rlepos].length + 1; 551| 663| int32_t xstart = src_2->runs[xrlepos].value; 552| 663| int32_t xend = xstart + src_2->runs[xrlepos].length + 1; 553| 3.82k| while ((rlepos < src_1->n_runs) && (xrlepos < src_2->n_runs)) { ------------------ | Branch (553:12): [True: 3.80k, False: 19] | Branch (553:40): [True: 3.79k, False: 7] ------------------ 554| 3.79k| if (end <= xstart) { ------------------ | Branch (554:13): [True: 2.70k, False: 1.09k] ------------------ 555| 2.70k| ++rlepos; 556| 2.70k| if (rlepos < src_1->n_runs) { ------------------ | Branch (556:17): [True: 2.68k, False: 19] ------------------ 557| 2.68k| start = src_1->runs[rlepos].value; 558| 2.68k| end = start + src_1->runs[rlepos].length + 1; 559| 2.68k| } 560| 2.70k| } else if (xend <= start) { ------------------ | Branch (560:20): [True: 455, False: 637] ------------------ 561| 455| ++xrlepos; 562| 455| if (xrlepos < src_2->n_runs) { ------------------ | Branch (562:17): [True: 448, False: 7] ------------------ 563| 448| xstart = src_2->runs[xrlepos].value; 564| 448| xend = xstart + src_2->runs[xrlepos].length + 1; 565| 448| } 566| 637| } else { // they overlap 567| 637| return true; 568| 637| } 569| 3.79k| } 570| 26| return false; 571| 663|} run_container_andnot: 576| 1.10k| const run_container_t *src_2, run_container_t *dst) { 577| | // following Java implementation as of June 2016 578| | 579| 1.10k| if (dst->capacity < src_1->n_runs + src_2->n_runs) ------------------ | Branch (579:9): [True: 1.10k, False: 0] ------------------ 580| 1.10k| run_container_grow(dst, src_1->n_runs + src_2->n_runs, false); 581| | 582| 1.10k| dst->n_runs = 0; 583| | 584| 1.10k| int rlepos1 = 0; 585| 1.10k| int rlepos2 = 0; 586| 1.10k| int32_t start = src_1->runs[rlepos1].value; 587| 1.10k| int32_t end = start + src_1->runs[rlepos1].length + 1; 588| 1.10k| int32_t start2 = src_2->runs[rlepos2].value; 589| 1.10k| int32_t end2 = start2 + src_2->runs[rlepos2].length + 1; 590| | 591| 155k| while ((rlepos1 < src_1->n_runs) && (rlepos2 < src_2->n_runs)) { ------------------ | Branch (591:12): [True: 155k, False: 663] | Branch (591:41): [True: 154k, False: 439] ------------------ 592| 154k| if (end <= start2) { ------------------ | Branch (592:13): [True: 23.1k, False: 131k] ------------------ 593| | // output the first run 594| 23.1k| dst->runs[dst->n_runs++] = 595| 23.1k| CROARING_MAKE_RLE16(start, end - start - 1); ------------------ | | 60| 23.1k| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 596| 23.1k| rlepos1++; 597| 23.1k| if (rlepos1 < src_1->n_runs) { ------------------ | Branch (597:17): [True: 23.0k, False: 82] ------------------ 598| 23.0k| start = src_1->runs[rlepos1].value; 599| 23.0k| end = start + src_1->runs[rlepos1].length + 1; 600| 23.0k| } 601| 131k| } else if (end2 <= start) { ------------------ | Branch (601:20): [True: 63.7k, False: 67.9k] ------------------ 602| | // exit the second run 603| 63.7k| rlepos2++; 604| 63.7k| if (rlepos2 < src_2->n_runs) { ------------------ | Branch (604:17): [True: 63.3k, False: 439] ------------------ 605| 63.3k| start2 = src_2->runs[rlepos2].value; 606| 63.3k| end2 = start2 + src_2->runs[rlepos2].length + 1; 607| 63.3k| } 608| 67.9k| } else { 609| 67.9k| if (start < start2) { ------------------ | Branch (609:17): [True: 19.1k, False: 48.8k] ------------------ 610| 19.1k| dst->runs[dst->n_runs++] = 611| 19.1k| CROARING_MAKE_RLE16(start, start2 - start - 1); ------------------ | | 60| 19.1k| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 612| 19.1k| } 613| 67.9k| if (end2 < end) { ------------------ | Branch (613:17): [True: 18.6k, False: 49.3k] ------------------ 614| 18.6k| start = end2; 615| 49.3k| } else { 616| 49.3k| rlepos1++; 617| 49.3k| if (rlepos1 < src_1->n_runs) { ------------------ | Branch (617:21): [True: 48.7k, False: 581] ------------------ 618| 48.7k| start = src_1->runs[rlepos1].value; 619| 48.7k| end = start + src_1->runs[rlepos1].length + 1; 620| 48.7k| } 621| 49.3k| } 622| 67.9k| } 623| 154k| } 624| 1.10k| if (rlepos1 < src_1->n_runs) { ------------------ | Branch (624:9): [True: 439, False: 663] ------------------ 625| 439| dst->runs[dst->n_runs++] = CROARING_MAKE_RLE16(start, end - start - 1); ------------------ | | 60| 439| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 626| 439| rlepos1++; 627| 439| if (rlepos1 < src_1->n_runs) { ------------------ | Branch (627:13): [True: 194, False: 245] ------------------ 628| 194| memcpy(dst->runs + dst->n_runs, src_1->runs + rlepos1, 629| 194| sizeof(rle16_t) * (src_1->n_runs - rlepos1)); 630| 194| dst->n_runs += src_1->n_runs - rlepos1; 631| 194| } 632| 439| } 633| 1.10k|} run_container_read: 739| 59| const char *buf) { 740| 59| (void)cardinality; 741| 59| uint16_t cast_16; 742| 59| memcpy(&cast_16, buf, sizeof(uint16_t)); 743| 59| container->n_runs = croaring_letoh16(cast_16); ------------------ | | 516| 59|#define croaring_letoh16(x) croaring_htole16(x) | | ------------------ | | | | 510| 59|#define croaring_htole16(x) (x) | | ------------------ ------------------ 744| 59| if (container->n_runs > container->capacity) ------------------ | Branch (744:9): [True: 8, False: 51] ------------------ 745| 8| run_container_grow(container, container->n_runs, false); 746| 59| if (container->n_runs > 0) { ------------------ | Branch (746:9): [True: 8, False: 51] ------------------ 747| |#if CROARING_IS_BIG_ENDIAN 748| | const char *in = buf + sizeof(uint16_t); 749| | for (int32_t i = 0; i < container->n_runs; ++i) { 750| | uint16_t v_le, l_le; 751| | memcpy(&v_le, in, sizeof(uint16_t)); 752| | memcpy(&l_le, in + sizeof(uint16_t), sizeof(uint16_t)); 753| | container->runs[i].value = croaring_letoh16(v_le); 754| | container->runs[i].length = croaring_letoh16(l_le); 755| | in += sizeof(rle16_t); 756| | } 757| |#else 758| 8| memcpy(container->runs, buf + sizeof(uint16_t), 759| 8| container->n_runs * sizeof(rle16_t)); 760| 8|#endif 761| 8| } 762| 59| return run_container_size_in_bytes(container); 763| 59|} run_container_iterate: 766| 55.2k| roaring_iterator iterator, void *ptr) { 767| 153k| for (int i = 0; i < cont->n_runs; ++i) { ------------------ | Branch (767:21): [True: 98.4k, False: 55.2k] ------------------ 768| 98.4k| uint32_t run_start = base + cont->runs[i].value; 769| 98.4k| uint16_t le = cont->runs[i].length; 770| | 771| 3.53G| for (int j = 0; j <= le; ++j) ------------------ | Branch (771:25): [True: 3.53G, False: 98.4k] ------------------ 772| 3.53G| if (!iterator(run_start + j, ptr)) return false; ------------------ | Branch (772:17): [True: 0, False: 3.53G] ------------------ 773| 98.4k| } 774| 55.2k| return true; 775| 55.2k|} run_container_is_subset: 792| 242| const run_container_t *container2) { 793| 242| int i1 = 0, i2 = 0; 794| 760| while (i1 < container1->n_runs && i2 < container2->n_runs) { ------------------ | Branch (794:12): [True: 760, False: 0] | Branch (794:39): [True: 741, False: 19] ------------------ 795| 741| int start1 = container1->runs[i1].value; 796| 741| int stop1 = start1 + container1->runs[i1].length; 797| 741| int start2 = container2->runs[i2].value; 798| 741| int stop2 = start2 + container2->runs[i2].length; 799| 741| if (start1 < start2) { ------------------ | Branch (799:13): [True: 223, False: 518] ------------------ 800| 223| return false; 801| 518| } else { // start1 >= start2 802| 518| if (stop1 < stop2) { ------------------ | Branch (802:17): [True: 0, False: 518] ------------------ 803| 0| i1++; 804| 518| } else if (stop1 == stop2) { ------------------ | Branch (804:24): [True: 0, False: 518] ------------------ 805| 0| i1++; 806| 0| i2++; 807| 518| } else { // stop1 > stop2 808| 518| i2++; 809| 518| } 810| 518| } 811| 741| } 812| 19| if (i1 == container1->n_runs) { ------------------ | Branch (812:9): [True: 0, False: 19] ------------------ 813| 0| return true; 814| 19| } else { 815| | return false; 816| 19| } 817| 19|} run_container_smart_append_exclusive: 826| 374k| const uint16_t length) { 827| 374k| int old_end; 828| 374k| rle16_t *last_run = src->n_runs ? src->runs + (src->n_runs - 1) : NULL; ------------------ | Branch (828:25): [True: 305k, False: 68.7k] ------------------ 829| 374k| rle16_t *appended_last_run = src->runs + src->n_runs; 830| | 831| 374k| if (!src->n_runs || ------------------ | Branch (831:9): [True: 68.7k, False: 305k] ------------------ 832| 305k| (start > (old_end = last_run->value + last_run->length + 1))) { ------------------ | Branch (832:9): [True: 166k, False: 139k] ------------------ 833| 235k| *appended_last_run = CROARING_MAKE_RLE16(start, length); ------------------ | | 60| 235k| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 834| 235k| src->n_runs++; 835| 235k| return; 836| 235k| } 837| 139k| if (old_end == start) { ------------------ | Branch (837:9): [True: 3.79k, False: 135k] ------------------ 838| | // we merge 839| 3.79k| last_run->length += (length + 1); 840| 3.79k| return; 841| 3.79k| } 842| 135k| int new_end = start + length + 1; 843| | 844| 135k| if (start == last_run->value) { ------------------ | Branch (844:9): [True: 106k, False: 28.6k] ------------------ 845| | // wipe out previous 846| 106k| if (new_end < old_end) { ------------------ | Branch (846:13): [True: 6.77k, False: 99.8k] ------------------ 847| 6.77k| *last_run = CROARING_MAKE_RLE16(new_end, old_end - new_end - 1); ------------------ | | 60| 6.77k| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 848| 6.77k| return; 849| 99.8k| } else if (new_end > old_end) { ------------------ | Branch (849:20): [True: 2.31k, False: 97.5k] ------------------ 850| 2.31k| *last_run = CROARING_MAKE_RLE16(old_end, new_end - old_end - 1); ------------------ | | 60| 2.31k| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 851| 2.31k| return; 852| 97.5k| } else { 853| 97.5k| src->n_runs--; 854| 97.5k| return; 855| 97.5k| } 856| 106k| } 857| 28.6k| last_run->length = start - last_run->value - 1; 858| 28.6k| if (new_end < old_end) { ------------------ | Branch (858:9): [True: 22.8k, False: 5.79k] ------------------ 859| 22.8k| *appended_last_run = 860| 22.8k| CROARING_MAKE_RLE16(new_end, old_end - new_end - 1); ------------------ | | 60| 22.8k| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 861| 22.8k| src->n_runs++; 862| 22.8k| } else if (new_end > old_end) { ------------------ | Branch (862:16): [True: 1.02k, False: 4.76k] ------------------ 863| 1.02k| *appended_last_run = 864| 1.02k| CROARING_MAKE_RLE16(old_end, new_end - old_end - 1); ------------------ | | 60| 1.02k| (rle16_t) { .value = (uint16_t)(val), .length = (uint16_t)(len) } ------------------ 865| 1.02k| src->n_runs++; 866| 1.02k| } 867| 28.6k|} run_container_select: 871| 2.51k| uint32_t *element) { 872| 14.1k| for (int i = 0; i < container->n_runs; i++) { ------------------ | Branch (872:21): [True: 14.0k, False: 104] ------------------ 873| 14.0k| uint16_t length = container->runs[i].length; 874| 14.0k| if (rank <= *start_rank + length) { ------------------ | Branch (874:13): [True: 2.41k, False: 11.5k] ------------------ 875| 2.41k| uint16_t value = container->runs[i].value; 876| 2.41k| *element = value + rank - (*start_rank); 877| 2.41k| return true; 878| 2.41k| } else 879| 11.5k| *start_rank += length + 1; 880| 14.0k| } 881| 104| return false; 882| 2.51k|} run_container_rank: 884| 2.51k|int run_container_rank(const run_container_t *container, uint16_t x) { 885| 2.51k| int sum = 0; 886| 2.51k| uint32_t x32 = x; 887| 11.2k| for (int i = 0; i < container->n_runs; i++) { ------------------ | Branch (887:21): [True: 11.1k, False: 59] ------------------ 888| 11.1k| uint32_t startpoint = container->runs[i].value; 889| 11.1k| uint32_t length = container->runs[i].length; 890| 11.1k| uint32_t endpoint = length + startpoint; 891| 11.1k| if (x <= endpoint) { ------------------ | Branch (891:13): [True: 2.46k, False: 8.72k] ------------------ 892| 2.46k| if (x < startpoint) break; ------------------ | Branch (892:17): [True: 270, False: 2.19k] ------------------ 893| 2.19k| return sum + (x32 - startpoint) + 1; 894| 8.72k| } else { 895| 8.72k| sum += length + 1; 896| 8.72k| } 897| 11.1k| } 898| 329| return sum; 899| 2.51k|} _avx2_run_container_to_uint32_array: 1021| 55.8k| uint32_t base) { 1022| 55.8k| int outpos = 0; 1023| 55.8k| uint32_t *out = (uint32_t *)vout; 1024| | 1025| 192k| for (int i = 0; i < cont->n_runs; ++i) { ------------------ | Branch (1025:21): [True: 136k, False: 55.8k] ------------------ 1026| 136k| uint32_t run_start = base + cont->runs[i].value; 1027| 136k| uint16_t le = cont->runs[i].length; 1028| 136k| if (le < 8) { ------------------ | Branch (1028:13): [True: 77.0k, False: 59.2k] ------------------ 1029| 199k| for (int j = 0; j <= le; ++j) { ------------------ | Branch (1029:29): [True: 122k, False: 77.0k] ------------------ 1030| 122k| uint32_t val = run_start + j; 1031| 122k| memcpy(out + outpos, &val, 1032| 122k| sizeof(uint32_t)); // should be compiled as a MOV on x64 1033| 122k| outpos++; 1034| 122k| } 1035| 77.0k| } else { 1036| 59.2k| int j = 0; 1037| 59.2k| __m256i run_start_v = _mm256_set1_epi32(run_start); 1038| | // [8,8,8,8....] 1039| 59.2k| __m256i inc = _mm256_set1_epi32(8); 1040| | // used for generate sequence: 1041| | // [0, 1, 2, 3...], [8, 9, 10,...] 1042| 59.2k| __m256i delta = _mm256_setr_epi32(0, 1, 2, 3, 4, 5, 6, 7); 1043| 443M| for (j = 0; j + 8 <= le; j += 8) { ------------------ | Branch (1043:25): [True: 443M, False: 59.2k] ------------------ 1044| 443M| __m256i val_v = _mm256_add_epi32(run_start_v, delta); 1045| 443M| _mm256_storeu_si256((__m256i *)(out + outpos), val_v); 1046| 443M| delta = _mm256_add_epi32(inc, delta); 1047| 443M| outpos += 8; 1048| 443M| } 1049| 509k| for (; j <= le; ++j) { ------------------ | Branch (1049:20): [True: 450k, False: 59.2k] ------------------ 1050| 450k| uint32_t val = run_start + j; 1051| 450k| memcpy(out + outpos, &val, 1052| 450k| sizeof(uint32_t)); // should be compiled as a MOV on x64 1053| 450k| outpos++; 1054| 450k| } 1055| 59.2k| } 1056| 136k| } 1057| 55.8k| return outpos; 1058| 55.8k|} run_container_cardinality: 1077| 498k|int run_container_cardinality(const run_container_t *run) { 1078| | // Empirically AVX-512 is not always faster than AVX2 1079| |#if CROARING_COMPILER_SUPPORTS_AVX512 && \ 1080| | CROARING_ENABLE_AVX512_RUN_CONTAINER_CARDINALITY 1081| | if (croaring_hardware_support() & ROARING_SUPPORTS_AVX512) { 1082| | return _avx512_run_container_cardinality(run); 1083| | } else 1084| |#endif 1085| 498k| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (1085:13): [True: 498k, False: 0] ------------------ 1086| 498k| return _avx2_run_container_cardinality(run); 1087| 498k| } else { 1088| 0| return _scalar_run_container_cardinality(run); 1089| 0| } 1090| 498k|} run_container_to_uint32_array: 1111| 55.8k| uint32_t base) { 1112| 55.8k| if (croaring_hardware_support() & ROARING_SUPPORTS_AVX2) { ------------------ | Branch (1112:9): [True: 55.8k, False: 0] ------------------ 1113| 55.8k| return _avx2_run_container_to_uint32_array(vout, cont, base); 1114| 55.8k| } else { 1115| 0| return _scalar_run_container_to_uint32_array(vout, cont, base); 1116| 0| } 1117| 55.8k|} run.c:_avx2_run_container_cardinality: 991| 498k|static inline int _avx2_run_container_cardinality(const run_container_t *run) { 992| 498k| const int32_t n_runs = run->n_runs; 993| 498k| const rle16_t *runs = run->runs; 994| | 995| | /* by initializing with n_runs, we omit counting the +1 for each pair. */ 996| 498k| int sum = n_runs; 997| 498k| int32_t k = 0; 998| 498k| const int32_t step = sizeof(__m256i) / sizeof(rle16_t); 999| 498k| if (n_runs > step) { ------------------ | Branch (999:9): [True: 31.8k, False: 466k] ------------------ 1000| 31.8k| __m256i total = _mm256_setzero_si256(); 1001| 356k| for (; k + step <= n_runs; k += step) { ------------------ | Branch (1001:16): [True: 324k, False: 31.8k] ------------------ 1002| 324k| __m256i ymm1 = _mm256_lddqu_si256((const __m256i *)(runs + k)); 1003| 324k| __m256i justlengths = _mm256_srli_epi32(ymm1, 16); 1004| 324k| total = _mm256_add_epi32(total, justlengths); 1005| 324k| } 1006| | // a store might be faster than extract? 1007| 31.8k| uint32_t buffer[sizeof(__m256i) / sizeof(rle16_t)]; 1008| 31.8k| _mm256_storeu_si256((__m256i *)buffer, total); 1009| 31.8k| sum += (buffer[0] + buffer[1]) + (buffer[2] + buffer[3]) + 1010| 31.8k| (buffer[4] + buffer[5]) + (buffer[6] + buffer[7]); 1011| 31.8k| } 1012| 1.04M| for (; k < n_runs; ++k) { ------------------ | Branch (1012:12): [True: 543k, False: 498k] ------------------ 1013| 543k| sum += runs[k].length; 1014| 543k| } 1015| | 1016| 498k| return sum; 1017| 498k|} croaring_hardware_support: 321| 889k|int croaring_hardware_support(void) { 322| 889k| static 323| 889k|#if CROARING_ATOMIC_IMPL == CROARING_ATOMIC_IMPL_C 324| 889k| _Atomic 325| 889k|#endif 326| 889k| int support = 0xFFFFFFF; 327| 889k| if (support == 0xFFFFFFF) { ------------------ | Branch (327:9): [True: 1, False: 889k] ------------------ 328| 1| bool has_avx2 = (croaring_detect_supported_architectures() & 329| 1| CROARING_AVX2) == CROARING_AVX2; 330| 1| bool has_avx512 = false; 331| 1|#if CROARING_COMPILER_SUPPORTS_AVX512 332| 1| has_avx512 = (croaring_detect_supported_architectures() & 333| 1| CROARING_AVX512_REQUIRED) == CROARING_AVX512_REQUIRED; 334| 1|#endif // CROARING_COMPILER_SUPPORTS_AVX512 335| 1| support = (has_avx2 ? ROARING_SUPPORTS_AVX2 : 0) | ------------------ | Branch (335:20): [True: 1, False: 0] ------------------ 336| 1| (has_avx512 ? ROARING_SUPPORTS_AVX512 : 0); ------------------ | Branch (336:20): [True: 0, False: 1] ------------------ 337| 1| } 338| 889k| return support; 339| 889k|} isadetection.c:croaring_detect_supported_architectures: 269| 2|static uint32_t croaring_detect_supported_architectures(void) { 270| | // we use an atomic for thread safety 271| 2| static _Atomic uint32_t buffer = CROARING_UNINITIALIZED; 272| 2| if (buffer == CROARING_UNINITIALIZED) { ------------------ | Branch (272:9): [True: 1, False: 1] ------------------ 273| | // atomicity is sufficient 274| 1| buffer = dynamic_croaring_detect_supported_architectures(); 275| 1| } 276| 2| return buffer; 277| 2|} isadetection.c:dynamic_croaring_detect_supported_architectures: 154| 1|static inline uint32_t dynamic_croaring_detect_supported_architectures(void) { 155| 1| uint32_t eax, ebx, ecx, edx; 156| 1| uint32_t host_isa = 0x0; 157| | // Can be found on Intel ISA Reference for CPUID 158| 1| static uint32_t cpuid_avx2_bit = 159| 1| 1 << 5; ///< @private Bit 5 of EBX for EAX=0x7 160| 1| static uint32_t cpuid_bmi1_bit = 161| 1| 1 << 3; ///< @private bit 3 of EBX for EAX=0x7 162| 1| static uint32_t cpuid_bmi2_bit = 163| 1| 1 << 8; ///< @private bit 8 of EBX for EAX=0x7 164| 1| static uint32_t cpuid_avx512f_bit = 165| 1| 1 << 16; ///< @private bit 16 of EBX for EAX=0x7 166| 1| static uint32_t cpuid_avx512dq_bit = 167| 1| 1 << 17; ///< @private bit 17 of EBX for EAX=0x7 168| 1| static uint32_t cpuid_avx512bw_bit = 169| 1| 1 << 30; ///< @private bit 30 of EBX for EAX=0x7 170| 1| static uint32_t cpuid_avx512vbmi2_bit = 171| 1| 1 << 6; ///< @private bit 6 of ECX for EAX=0x7 172| 1| static uint32_t cpuid_avx512bitalg_bit = 173| 1| 1 << 12; ///< @private bit 12 of ECX for EAX=0x7 174| 1| static uint32_t cpuid_avx512vpopcntdq_bit = 175| 1| 1 << 14; ///< @private bit 14 of ECX for EAX=0x7 176| 1| static uint64_t cpuid_avx256_saved = 1 << 2; ///< @private bit 2 = AVX 177| 1| static uint64_t cpuid_avx512_saved = 178| 1| 7 << 5; ///< @private bits 5,6,7 = opmask, ZMM_hi256, hi16_ZMM 179| 1| static uint32_t cpuid_sse42_bit = 180| 1| 1 << 20; ///< @private bit 20 of ECX for EAX=0x1 181| 1| static uint32_t cpuid_osxsave = 182| 1| (1 << 26) | (1 << 27); ///< @private bits 26+27 of ECX for EAX=0x1 183| 1| static uint32_t cpuid_pclmulqdq_bit = 184| 1| 1 << 1; ///< @private bit 1 of ECX for EAX=0x1 185| | 186| | // EBX for EAX=0x1 187| 1| eax = 0x1; 188| 1| ecx = 0x0; 189| 1| cpuid(&eax, &ebx, &ecx, &edx); 190| | 191| 1| if (ecx & cpuid_sse42_bit) { ------------------ | Branch (191:9): [True: 1, False: 0] ------------------ 192| 1| host_isa |= CROARING_SSE42; 193| 1| } else { 194| 0| return host_isa; // everything after is redundant 195| 0| } 196| | 197| 1| if (ecx & cpuid_pclmulqdq_bit) { ------------------ | Branch (197:9): [True: 1, False: 0] ------------------ 198| 1| host_isa |= CROARING_PCLMULQDQ; 199| 1| } 200| | 201| 1| if ((ecx & cpuid_osxsave) != cpuid_osxsave) { ------------------ | Branch (201:9): [True: 0, False: 1] ------------------ 202| 0| return host_isa; 203| 0| } 204| | 205| | // xgetbv for checking if the OS saves registers 206| 1| uint64_t xcr0 = xgetbv(); 207| | 208| 1| if ((xcr0 & cpuid_avx256_saved) == 0) { ------------------ | Branch (208:9): [True: 0, False: 1] ------------------ 209| 0| return host_isa; 210| 0| } 211| | 212| | // ECX for EAX=0x7 213| 1| eax = 0x7; 214| 1| ecx = 0x0; 215| 1| cpuid(&eax, &ebx, &ecx, &edx); 216| 1| if (ebx & cpuid_avx2_bit) { ------------------ | Branch (216:9): [True: 1, False: 0] ------------------ 217| 1| host_isa |= CROARING_AVX2; 218| 1| } 219| 1| if (ebx & cpuid_bmi1_bit) { ------------------ | Branch (219:9): [True: 1, False: 0] ------------------ 220| 1| host_isa |= CROARING_BMI1; 221| 1| } 222| | 223| 1| if (ebx & cpuid_bmi2_bit) { ------------------ | Branch (223:9): [True: 1, False: 0] ------------------ 224| 1| host_isa |= CROARING_BMI2; 225| 1| } 226| | 227| 1| if (!((xcr0 & cpuid_avx512_saved) == cpuid_avx512_saved)) { ------------------ | Branch (227:9): [True: 1, False: 0] ------------------ 228| 1| return host_isa; 229| 1| } 230| | 231| 0| if (ebx & cpuid_avx512f_bit) { ------------------ | Branch (231:9): [True: 0, False: 0] ------------------ 232| 0| host_isa |= CROARING_AVX512F; 233| 0| } 234| | 235| 0| if (ebx & cpuid_avx512bw_bit) { ------------------ | Branch (235:9): [True: 0, False: 0] ------------------ 236| 0| host_isa |= CROARING_AVX512BW; 237| 0| } 238| | 239| 0| if (ebx & cpuid_avx512dq_bit) { ------------------ | Branch (239:9): [True: 0, False: 0] ------------------ 240| 0| host_isa |= CROARING_AVX512DQ; 241| 0| } 242| | 243| 0| if (ecx & cpuid_avx512vbmi2_bit) { ------------------ | Branch (243:9): [True: 0, False: 0] ------------------ 244| 0| host_isa |= CROARING_AVX512VBMI2; 245| 0| } 246| | 247| 0| if (ecx & cpuid_avx512bitalg_bit) { ------------------ | Branch (247:9): [True: 0, False: 0] ------------------ 248| 0| host_isa |= CROARING_AVX512BITALG; 249| 0| } 250| | 251| 0| if (ecx & cpuid_avx512vpopcntdq_bit) { ------------------ | Branch (251:9): [True: 0, False: 0] ------------------ 252| 0| host_isa |= CROARING_AVX512VPOPCNTDQ; 253| 0| } 254| | 255| 0| return host_isa; 256| 1|} isadetection.c:cpuid: 115| 2| uint32_t *edx) { 116| |#if CROARING_REGULAR_VISUAL_STUDIO 117| | int cpu_info[4]; 118| | __cpuidex(cpu_info, *eax, *ecx); 119| | *eax = cpu_info[0]; 120| | *ebx = cpu_info[1]; 121| | *ecx = cpu_info[2]; 122| | *edx = cpu_info[3]; 123| |#elif (defined(HAVE_GCC_GET_CPUID) && defined(USE_GCC_GET_CPUID)) || \ 124| | defined(__FILC__) 125| | uint32_t level = *eax; 126| | __get_cpuid(level, eax, ebx, ecx, edx); 127| |#else 128| 2| uint32_t a = *eax, b, c = *ecx, d; 129| 2| __asm__("cpuid\n\t" : "+a"(a), "=b"(b), "+c"(c), "=d"(d)); 130| 2| *eax = a; 131| 2| *ebx = b; 132| 2| *ecx = c; 133| 2| *edx = d; 134| 2|#endif 135| 2|} isadetection.c:xgetbv: 137| 1|static inline uint64_t xgetbv(void) { 138| |#if defined(_MSC_VER) 139| | return _xgetbv(0); 140| |#elif defined(__FILC__) 141| | return zxgetbv(); 142| |#else 143| 1| uint32_t xcr0_lo, xcr0_hi; 144| 1| __asm__("xgetbv\n\t" : "=a"(xcr0_lo), "=d"(xcr0_hi) : "c"(0)); 145| 1| return xcr0_lo | ((uint64_t)xcr0_hi << 32); 146| 1|#endif 147| 1|} roaring_malloc: 48| 1.32M|void* roaring_malloc(size_t n) { return global_memory_hook.malloc(n); } roaring_realloc: 50| 113k|void* roaring_realloc(void* p, size_t new_sz) { 51| 113k| return global_memory_hook.realloc(p, new_sz); 52| 113k|} roaring_free: 58| 1.55M|void roaring_free(void* p) { global_memory_hook.free(p); } roaring_aligned_malloc: 60| 272k|void* roaring_aligned_malloc(size_t alignment, size_t size) { 61| 272k| return global_memory_hook.aligned_malloc(alignment, size); 62| 272k|} roaring_aligned_free: 64| 272k|void roaring_aligned_free(void* p) { global_memory_hook.aligned_free(p); } memory.c:roaring_bitmap_aligned_malloc: 11| 272k|static void* roaring_bitmap_aligned_malloc(size_t alignment, size_t size) { 12| 272k| void* p; 13| |#ifdef _MSC_VER 14| | p = _aligned_malloc(size, alignment); 15| |#elif defined(__MINGW32__) || defined(__MINGW64__) 16| | p = __mingw_aligned_malloc(size, alignment); 17| |#else 18| | // somehow, if this is used before including "x86intrin.h", it creates an 19| | // implicit defined warning. 20| 272k| if (posix_memalign(&p, alignment, size) != 0) return NULL; ------------------ | Branch (20:9): [True: 0, False: 272k] ------------------ 21| 272k|#endif 22| 272k| return p; 23| 272k|} memory.c:roaring_bitmap_aligned_free: 25| 272k|static void roaring_bitmap_aligned_free(void* memblock) { 26| |#ifdef _MSC_VER 27| | _aligned_free(memblock); 28| |#elif defined(__MINGW32__) || defined(__MINGW64__) 29| | __mingw_aligned_free(memblock); 30| |#else 31| 272k| free(memblock); 32| 272k|#endif 33| 272k|} roaring_bitmap_create_with_capacity: 86| 24.4k|roaring_bitmap_t *roaring_bitmap_create_with_capacity(uint32_t cap) { 87| 24.4k| roaring_bitmap_t *ans = 88| 24.4k| (roaring_bitmap_t *)roaring_malloc(sizeof(roaring_bitmap_t)); 89| 24.4k| if (!ans) { ------------------ | Branch (89:9): [True: 0, False: 24.4k] ------------------ 90| 0| return NULL; 91| 0| } 92| 24.4k| bool is_ok = ra_init_with_capacity(&ans->high_low_container, cap); 93| 24.4k| if (!is_ok) { ------------------ | Branch (93:9): [True: 0, False: 24.4k] ------------------ 94| 0| roaring_free(ans); 95| 0| return NULL; 96| 0| } 97| 24.4k| return ans; 98| 24.4k|} roaring_bitmap_init_with_capacity: 100| 30.7k|bool roaring_bitmap_init_with_capacity(roaring_bitmap_t *r, uint32_t cap) { 101| 30.7k| return ra_init_with_capacity(&r->high_low_container, cap); 102| 30.7k|} roaring_bitmap_add_many: 135| 18.4k| const uint32_t *vals) { 136| 18.4k| uint32_t val; 137| 18.4k| const uint32_t *start = vals; 138| 18.4k| const uint32_t *end = vals + n_args; 139| 18.4k| const uint32_t *current_val = start; 140| | 141| 18.4k| if (n_args == 0) { ------------------ | Branch (141:9): [True: 0, False: 18.4k] ------------------ 142| 0| return; 143| 0| } 144| | 145| 18.4k| uint8_t typecode; 146| 18.4k| int idx; 147| 18.4k| container_t *container; 148| 18.4k| val = *current_val; 149| 18.4k| container = containerptr_roaring_bitmap_add(r, val, &typecode, &idx); 150| 18.4k| roaring_bulk_context_t context = {container, idx, (uint16_t)(val >> 16), 151| 18.4k| typecode}; 152| | 153| 7.71M| for (; current_val != end; current_val++) { ------------------ | Branch (153:12): [True: 7.69M, False: 18.4k] ------------------ 154| 7.69M| memcpy(&val, current_val, sizeof(val)); 155| 7.69M| add_bulk_impl(r, &context, val); 156| 7.69M| } 157| 18.4k|} roaring_bitmap_add_range_closed: 252| 2.66k| uint32_t max) { 253| 2.66k| if (min > max) { ------------------ | Branch (253:9): [True: 0, False: 2.66k] ------------------ 254| 0| return; 255| 0| } 256| | 257| 2.66k| roaring_array_t *ra = &r->high_low_container; 258| | 259| 2.66k| uint32_t min_key = min >> 16; 260| 2.66k| uint32_t max_key = max >> 16; 261| | 262| 2.66k| int32_t num_required_containers = max_key - min_key + 1; 263| 2.66k| int32_t suffix_length = 264| 2.66k| count_greater(ra->keys, ra->size, (uint16_t)max_key); 265| 2.66k| int32_t prefix_length = 266| 2.66k| count_less(ra->keys, ra->size - suffix_length, (uint16_t)min_key); 267| 2.66k| int32_t common_length = ra->size - prefix_length - suffix_length; 268| | 269| 2.66k| if (num_required_containers > common_length) { ------------------ | Branch (269:9): [True: 1.24k, False: 1.41k] ------------------ 270| 1.24k| ra_shift_tail(ra, suffix_length, 271| 1.24k| num_required_containers - common_length); 272| 1.24k| } 273| | 274| 2.66k| int32_t src = prefix_length + common_length - 1; 275| 2.66k| int32_t dst = ra->size - suffix_length - 1; 276| 74.4k| for (uint32_t key = max_key; key != min_key - 1; ------------------ | Branch (276:34): [True: 71.7k, False: 2.66k] ------------------ 277| 71.7k| key--) { // beware of min_key==0 278| 71.7k| uint32_t container_min = (min_key == key) ? (min & 0xffff) : 0; ------------------ | Branch (278:34): [True: 2.66k, False: 69.1k] ------------------ 279| 71.7k| uint32_t container_max = (max_key == key) ? (max & 0xffff) : 0xffff; ------------------ | Branch (279:34): [True: 2.66k, False: 69.1k] ------------------ 280| 71.7k| container_t *new_container; 281| 71.7k| uint8_t new_type; 282| | 283| 71.7k| if (src >= 0 && ra->keys[src] == key) { ------------------ | Branch (283:13): [True: 71.7k, False: 0] | Branch (283:25): [True: 3.06k, False: 68.7k] ------------------ 284| 3.06k| ra_unshare_container_at_index(ra, (uint16_t)src); 285| 3.06k| new_container = 286| 3.06k| container_add_range(ra->containers[src], ra->typecodes[src], 287| 3.06k| container_min, container_max, &new_type); 288| 3.06k| if (new_container != ra->containers[src]) { ------------------ | Branch (288:17): [True: 2.11k, False: 950] ------------------ 289| 2.11k| container_free(ra->containers[src], ra->typecodes[src]); 290| 2.11k| } 291| 3.06k| src--; 292| 68.7k| } else { 293| 68.7k| new_container = container_from_range(&new_type, container_min, 294| 68.7k| container_max + 1, 1); 295| 68.7k| } 296| 71.7k| ra_replace_key_and_container_at_index(ra, dst, (uint16_t)key, 297| 71.7k| new_container, new_type); 298| 71.7k| dst--; 299| 71.7k| } 300| 2.66k|} roaring_bitmap_remove_range_closed: 303| 7.12k| uint32_t max) { 304| 7.12k| if (min > max) { ------------------ | Branch (304:9): [True: 731, False: 6.39k] ------------------ 305| 731| return; 306| 731| } 307| | 308| 6.39k| roaring_array_t *ra = &r->high_low_container; 309| | 310| 6.39k| uint32_t min_key = min >> 16; 311| 6.39k| uint32_t max_key = max >> 16; 312| | 313| 6.39k| int32_t src = count_less(ra->keys, ra->size, (uint16_t)min_key); 314| 6.39k| int32_t dst = src; 315| 27.6k| while (src < ra->size && ra->keys[src] <= max_key) { ------------------ | Branch (315:12): [True: 23.4k, False: 4.21k] | Branch (315:30): [True: 21.2k, False: 2.17k] ------------------ 316| 21.2k| uint32_t container_min = 317| 21.2k| (min_key == ra->keys[src]) ? (min & 0xffff) : 0; ------------------ | Branch (317:13): [True: 5.93k, False: 15.3k] ------------------ 318| 21.2k| uint32_t container_max = 319| 21.2k| (max_key == ra->keys[src]) ? (max & 0xffff) : 0xffff; ------------------ | Branch (319:13): [True: 5.63k, False: 15.6k] ------------------ 320| 21.2k| ra_unshare_container_at_index(ra, (uint16_t)src); 321| 21.2k| container_t *new_container; 322| 21.2k| uint8_t new_type; 323| 21.2k| new_container = 324| 21.2k| container_remove_range(ra->containers[src], ra->typecodes[src], 325| 21.2k| container_min, container_max, &new_type); 326| 21.2k| if (new_container != ra->containers[src]) { ------------------ | Branch (326:13): [True: 15.3k, False: 5.90k] ------------------ 327| 15.3k| container_free(ra->containers[src], ra->typecodes[src]); 328| 15.3k| } 329| 21.2k| if (new_container) { ------------------ | Branch (329:13): [True: 6.18k, False: 15.0k] ------------------ 330| 6.18k| ra_replace_key_and_container_at_index(ra, dst, ra->keys[src], 331| 6.18k| new_container, new_type); 332| 6.18k| dst++; 333| 6.18k| } 334| 21.2k| src++; 335| 21.2k| } 336| 6.39k| if (src > dst) { ------------------ | Branch (336:9): [True: 699, False: 5.69k] ------------------ 337| 699| ra_shift_tail(ra, ra->size - src, dst - src); 338| 699| } 339| 6.39k|} roaring_bitmap_copy: 525| 141|roaring_bitmap_t *roaring_bitmap_copy(const roaring_bitmap_t *r) { 526| 141| roaring_bitmap_t *ans = 527| 141| (roaring_bitmap_t *)roaring_malloc(sizeof(roaring_bitmap_t)); 528| 141| if (!ans) { ------------------ | Branch (528:9): [True: 0, False: 141] ------------------ 529| 0| return NULL; 530| 0| } 531| 141| if (!ra_init_with_capacity( // allocation of list of containers can fail ------------------ | Branch (531:9): [True: 0, False: 141] ------------------ 532| 141| &ans->high_low_container, r->high_low_container.size)) { 533| 0| roaring_free(ans); 534| 0| return NULL; 535| 0| } 536| 141| if (!ra_overwrite( // memory allocation of individual containers may fail ------------------ | Branch (536:9): [True: 0, False: 141] ------------------ 537| 141| &r->high_low_container, &ans->high_low_container, is_cow(r))) { 538| 0| roaring_bitmap_free(ans); // overwrite should leave in freeable state 539| 0| return NULL; 540| 0| } 541| 141| roaring_bitmap_set_copy_on_write(ans, is_cow(r)); 542| 141| return ans; 543| 141|} roaring_bitmap_overwrite: 546| 13.8k| const roaring_bitmap_t *src) { 547| 13.8k| roaring_bitmap_set_copy_on_write(dest, is_cow(src)); 548| 13.8k| return ra_overwrite(&src->high_low_container, &dest->high_low_container, 549| 13.8k| is_cow(src)); 550| 13.8k|} roaring_bitmap_clear: 562| 61.5k|void roaring_bitmap_clear(roaring_bitmap_t *r) { 563| 61.5k| ra_reset(&r->high_low_container); 564| 61.5k|} roaring_bitmap_add: 566| 6.15k|void roaring_bitmap_add(roaring_bitmap_t *r, uint32_t val) { 567| 6.15k| roaring_array_t *ra = &r->high_low_container; 568| | 569| 6.15k| const uint16_t hb = val >> 16; 570| 6.15k| const int i = ra_get_index(ra, hb); 571| 6.15k| uint8_t typecode; 572| 6.15k| if (i >= 0) { ------------------ | Branch (572:9): [True: 4.42k, False: 1.72k] ------------------ 573| 4.42k| ra_unshare_container_at_index(ra, (uint16_t)i); 574| 4.42k| container_t *container = 575| 4.42k| ra_get_container_at_index(ra, (uint16_t)i, &typecode); 576| 4.42k| uint8_t newtypecode = typecode; 577| 4.42k| container_t *container2 = 578| 4.42k| container_add(container, val & 0xFFFF, typecode, &newtypecode); 579| 4.42k| if (container2 != container) { ------------------ | Branch (579:13): [True: 0, False: 4.42k] ------------------ 580| 0| container_free(container, typecode); 581| 0| ra_set_container_at_index(&r->high_low_container, i, container2, 582| 0| newtypecode); 583| 0| } 584| 4.42k| } else { 585| 1.72k| array_container_t *newac = array_container_create(); 586| 1.72k| container_t *container = 587| 1.72k| container_add(newac, val & 0xFFFF, ARRAY_CONTAINER_TYPE, &typecode); ------------------ | | 49| 1.72k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 588| | // we could just assume that it stays an array container 589| 1.72k| ra_insert_new_key_value_at(&r->high_low_container, -i - 1, hb, 590| 1.72k| container, typecode); 591| 1.72k| } 592| 6.15k|} roaring_bitmap_add_checked: 594| 6.15k|bool roaring_bitmap_add_checked(roaring_bitmap_t *r, uint32_t val) { 595| 6.15k| const uint16_t hb = val >> 16; 596| 6.15k| const int i = ra_get_index(&r->high_low_container, hb); 597| 6.15k| uint8_t typecode; 598| 6.15k| bool result = false; 599| 6.15k| if (i >= 0) { ------------------ | Branch (599:9): [True: 4.65k, False: 1.49k] ------------------ 600| 4.65k| ra_unshare_container_at_index(&r->high_low_container, (uint16_t)i); 601| 4.65k| container_t *container = ra_get_container_at_index( 602| 4.65k| &r->high_low_container, (uint16_t)i, &typecode); 603| | 604| 4.65k| const int oldCardinality = 605| 4.65k| container_get_cardinality(container, typecode); 606| | 607| 4.65k| uint8_t newtypecode = typecode; 608| 4.65k| container_t *container2 = 609| 4.65k| container_add(container, val & 0xFFFF, typecode, &newtypecode); 610| 4.65k| if (container2 != container) { ------------------ | Branch (610:13): [True: 0, False: 4.65k] ------------------ 611| 0| container_free(container, typecode); 612| 0| ra_set_container_at_index(&r->high_low_container, i, container2, 613| 0| newtypecode); 614| 0| result = true; 615| 4.65k| } else { 616| 4.65k| const int newCardinality = 617| 4.65k| container_get_cardinality(container, newtypecode); 618| | 619| 4.65k| result = oldCardinality != newCardinality; 620| 4.65k| } 621| 4.65k| } else { 622| 1.49k| array_container_t *newac = array_container_create(); 623| 1.49k| container_t *container = 624| 1.49k| container_add(newac, val & 0xFFFF, ARRAY_CONTAINER_TYPE, &typecode); ------------------ | | 49| 1.49k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 625| | // we could just assume that it stays an array container 626| 1.49k| ra_insert_new_key_value_at(&r->high_low_container, -i - 1, hb, 627| 1.49k| container, typecode); 628| 1.49k| result = true; 629| 1.49k| } 630| | 631| 6.15k| return result; 632| 6.15k|} roaring_bitmap_remove: 634| 6.15k|void roaring_bitmap_remove(roaring_bitmap_t *r, uint32_t val) { 635| 6.15k| const uint16_t hb = val >> 16; 636| 6.15k| const int i = ra_get_index(&r->high_low_container, hb); 637| 6.15k| uint8_t typecode; 638| 6.15k| if (i >= 0) { ------------------ | Branch (638:9): [True: 5.48k, False: 667] ------------------ 639| 5.48k| ra_unshare_container_at_index(&r->high_low_container, (uint16_t)i); 640| 5.48k| container_t *container = ra_get_container_at_index( 641| 5.48k| &r->high_low_container, (uint16_t)i, &typecode); 642| 5.48k| uint8_t newtypecode = typecode; 643| 5.48k| container_t *container2 = 644| 5.48k| container_remove(container, val & 0xFFFF, typecode, &newtypecode); 645| 5.48k| if (container2 != container) { ------------------ | Branch (645:13): [True: 14, False: 5.47k] ------------------ 646| 14| container_free(container, typecode); 647| 14| ra_set_container_at_index(&r->high_low_container, i, container2, 648| 14| newtypecode); 649| 14| } 650| 5.48k| if (container_nonzero_cardinality(container2, newtypecode)) { ------------------ | Branch (650:13): [True: 5.42k, False: 63] ------------------ 651| 5.42k| ra_set_container_at_index(&r->high_low_container, i, container2, 652| 5.42k| newtypecode); 653| 5.42k| } else { 654| 63| ra_remove_at_index_and_free(&r->high_low_container, i); 655| 63| } 656| 5.48k| } 657| 6.15k|} roaring_bitmap_remove_checked: 659| 6.15k|bool roaring_bitmap_remove_checked(roaring_bitmap_t *r, uint32_t val) { 660| 6.15k| const uint16_t hb = val >> 16; 661| 6.15k| const int i = ra_get_index(&r->high_low_container, hb); 662| 6.15k| uint8_t typecode; 663| 6.15k| bool result = false; 664| 6.15k| if (i >= 0) { ------------------ | Branch (664:9): [True: 5.52k, False: 631] ------------------ 665| 5.52k| ra_unshare_container_at_index(&r->high_low_container, (uint16_t)i); 666| 5.52k| container_t *container = ra_get_container_at_index( 667| 5.52k| &r->high_low_container, (uint16_t)i, &typecode); 668| | 669| 5.52k| const int oldCardinality = 670| 5.52k| container_get_cardinality(container, typecode); 671| | 672| 5.52k| uint8_t newtypecode = typecode; 673| 5.52k| container_t *container2 = 674| 5.52k| container_remove(container, val & 0xFFFF, typecode, &newtypecode); 675| 5.52k| if (container2 != container) { ------------------ | Branch (675:13): [True: 3, False: 5.52k] ------------------ 676| 3| container_free(container, typecode); 677| 3| ra_set_container_at_index(&r->high_low_container, i, container2, 678| 3| newtypecode); 679| 3| } 680| | 681| 5.52k| const int newCardinality = 682| 5.52k| container_get_cardinality(container2, newtypecode); 683| | 684| 5.52k| if (newCardinality != 0) { ------------------ | Branch (684:13): [True: 5.51k, False: 10] ------------------ 685| 5.51k| ra_set_container_at_index(&r->high_low_container, i, container2, 686| 5.51k| newtypecode); 687| 5.51k| } else { 688| 10| ra_remove_at_index_and_free(&r->high_low_container, i); 689| 10| } 690| | 691| 5.52k| result = oldCardinality != newCardinality; 692| 5.52k| } 693| 6.15k| return result; 694| 6.15k|} roaring_bitmap_and: 732| 6.15k| const roaring_bitmap_t *x2) { 733| 6.15k| uint8_t result_type = 0; 734| 6.15k| const int length1 = x1->high_low_container.size, 735| 6.15k| length2 = x2->high_low_container.size; 736| 6.15k| uint32_t neededcap = length1 > length2 ? length2 : length1; ------------------ | Branch (736:26): [True: 47, False: 6.10k] ------------------ 737| 6.15k| roaring_bitmap_t *answer = roaring_bitmap_create_with_capacity(neededcap); 738| 6.15k| roaring_bitmap_set_copy_on_write(answer, is_cow(x1) || is_cow(x2)); ------------------ | Branch (738:46): [True: 0, False: 6.15k] | Branch (738:60): [True: 0, False: 6.15k] ------------------ 739| | 740| 6.15k| int pos1 = 0, pos2 = 0; 741| | 742| 12.2k| while (pos1 < length1 && pos2 < length2) { ------------------ | Branch (742:12): [True: 6.15k, False: 6.10k] | Branch (742:30): [True: 6.10k, False: 47] ------------------ 743| 6.10k| const uint16_t s1 = 744| 6.10k| ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 745| 6.10k| const uint16_t s2 = 746| 6.10k| ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 747| | 748| 6.10k| if (s1 == s2) { ------------------ | Branch (748:13): [True: 6.03k, False: 74] ------------------ 749| 6.03k| uint8_t type1, type2; 750| 6.03k| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 751| 6.03k| (uint16_t)pos1, &type1); 752| 6.03k| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 753| 6.03k| (uint16_t)pos2, &type2); 754| 6.03k| container_t *c = container_and(c1, type1, c2, type2, &result_type); 755| | 756| 6.03k| if (container_nonzero_cardinality(c, result_type)) { ------------------ | Branch (756:17): [True: 5.73k, False: 301] ------------------ 757| 5.73k| ra_append(&answer->high_low_container, s1, c, result_type); 758| 5.73k| } else { 759| 301| container_free(c, result_type); // otherwise: memory leak! 760| 301| } 761| 6.03k| ++pos1; 762| 6.03k| ++pos2; 763| 6.03k| } else if (s1 < s2) { // s1 < s2 ------------------ | Branch (763:20): [True: 74, False: 0] ------------------ 764| 74| pos1 = ra_advance_until(&x1->high_low_container, s2, pos1); 765| 74| } else { // s1 > s2 766| 0| pos2 = ra_advance_until(&x2->high_low_container, s1, pos2); 767| 0| } 768| 6.10k| } 769| 6.15k| return answer; 770| 6.15k|} roaring_bitmap_and_inplace: 813| 6.15k| const roaring_bitmap_t *x2) { 814| 6.15k| if (x1 == x2) return; ------------------ | Branch (814:9): [True: 0, False: 6.15k] ------------------ 815| 6.15k| int pos1 = 0, pos2 = 0, intersection_size = 0; 816| 6.15k| const int length1 = ra_get_size(&x1->high_low_container); 817| 6.15k| const int length2 = ra_get_size(&x2->high_low_container); 818| | 819| | // any skipped-over or newly emptied containers in x1 820| | // have to be freed. 821| 69.1k| while (pos1 < length1 && pos2 < length2) { ------------------ | Branch (821:12): [True: 63.0k, False: 6.10k] | Branch (821:30): [True: 63.0k, False: 47] ------------------ 822| 63.0k| const uint16_t s1 = 823| 63.0k| ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 824| 63.0k| const uint16_t s2 = 825| 63.0k| ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 826| | 827| 63.0k| if (s1 == s2) { ------------------ | Branch (827:13): [True: 62.9k, False: 74] ------------------ 828| 62.9k| uint8_t type1, type2, result_type; 829| 62.9k| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 830| 62.9k| (uint16_t)pos1, &type1); 831| 62.9k| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 832| 62.9k| (uint16_t)pos2, &type2); 833| | 834| | // We do the computation "in place" only when c1 is not a shared 835| | // container. Rationale: using a shared container safely with in 836| | // place computation would require making a copy and then doing the 837| | // computation in place which is likely less efficient than avoiding 838| | // in place entirely and always generating a new container. 839| 62.9k| container_t *c = 840| 62.9k| (type1 == SHARED_CONTAINER_TYPE) ------------------ | | 51| 62.9k|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (840:17): [True: 0, False: 62.9k] ------------------ 841| 62.9k| ? container_and(c1, type1, c2, type2, &result_type) 842| 62.9k| : container_iand(c1, type1, c2, type2, &result_type); 843| | 844| 62.9k| if (c != c1) { // in this instance a new container was created, and ------------------ | Branch (844:17): [True: 56.9k, False: 6.04k] ------------------ 845| | // we need to free the old one 846| 56.9k| container_free(c1, type1); 847| 56.9k| } 848| 62.9k| if (container_nonzero_cardinality(c, result_type)) { ------------------ | Branch (848:17): [True: 62.9k, False: 0] ------------------ 849| 62.9k| ra_replace_key_and_container_at_index(&x1->high_low_container, 850| 62.9k| intersection_size, s1, c, 851| 62.9k| result_type); 852| 62.9k| intersection_size++; 853| 62.9k| } else { 854| 0| container_free(c, result_type); 855| 0| } 856| 62.9k| ++pos1; 857| 62.9k| ++pos2; 858| 62.9k| } else if (s1 < s2) { ------------------ | Branch (858:20): [True: 74, False: 0] ------------------ 859| 74| pos1 = ra_advance_until_freeing(&x1->high_low_container, s2, pos1); 860| 74| } else { // s1 > s2 861| 0| pos2 = ra_advance_until(&x2->high_low_container, s1, pos2); 862| 0| } 863| 63.0k| } 864| | 865| | // if we ended early because x2 ran out, then all remaining in x1 should be 866| | // freed 867| 6.20k| while (pos1 < length1) { ------------------ | Branch (867:12): [True: 47, False: 6.15k] ------------------ 868| 47| container_free(x1->high_low_container.containers[pos1], 869| 47| x1->high_low_container.typecodes[pos1]); 870| 47| ++pos1; 871| 47| } 872| | 873| | // all containers after this have either been copied or freed 874| 6.15k| ra_downsize(&x1->high_low_container, intersection_size); 875| 6.15k|} roaring_bitmap_or: 878| 6.15k| const roaring_bitmap_t *x2) { 879| 6.15k| uint8_t result_type = 0; 880| 6.15k| const int length1 = x1->high_low_container.size, 881| 6.15k| length2 = x2->high_low_container.size; 882| 6.15k| if (0 == length1) { ------------------ | Branch (882:9): [True: 0, False: 6.15k] ------------------ 883| 0| return roaring_bitmap_copy(x2); 884| 0| } 885| 6.15k| if (0 == length2) { ------------------ | Branch (885:9): [True: 47, False: 6.10k] ------------------ 886| 47| return roaring_bitmap_copy(x1); 887| 47| } 888| 6.10k| roaring_bitmap_t *answer = 889| 6.10k| roaring_bitmap_create_with_capacity(length1 + length2); 890| 6.10k| roaring_bitmap_set_copy_on_write(answer, is_cow(x1) || is_cow(x2)); ------------------ | Branch (890:46): [True: 0, False: 6.10k] | Branch (890:60): [True: 0, False: 6.10k] ------------------ 891| 6.10k| int pos1 = 0, pos2 = 0; 892| 6.10k| uint8_t type1, type2; 893| 6.10k| uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 894| 6.10k| uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 895| 6.10k| while (true) { ------------------ | Branch (895:12): [True: 6.10k, Folded] ------------------ 896| 6.10k| if (s1 == s2) { ------------------ | Branch (896:13): [True: 6.03k, False: 74] ------------------ 897| 6.03k| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 898| 6.03k| (uint16_t)pos1, &type1); 899| 6.03k| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 900| 6.03k| (uint16_t)pos2, &type2); 901| 6.03k| container_t *c = container_or(c1, type1, c2, type2, &result_type); 902| | 903| | // since we assume that the initial containers are non-empty, the 904| | // result here 905| | // can only be non-empty 906| 6.03k| ra_append(&answer->high_low_container, s1, c, result_type); 907| 6.03k| ++pos1; 908| 6.03k| ++pos2; 909| 6.03k| if (pos1 == length1) break; ------------------ | Branch (909:17): [True: 6.03k, False: 0] ------------------ 910| 0| if (pos2 == length2) break; ------------------ | Branch (910:17): [True: 0, False: 0] ------------------ 911| 0| s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 912| 0| s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 913| | 914| 74| } else if (s1 < s2) { // s1 < s2 ------------------ | Branch (914:20): [True: 74, False: 0] ------------------ 915| 74| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 916| 74| (uint16_t)pos1, &type1); 917| | // c1 = container_clone(c1, type1); 918| 74| c1 = get_copy_of_container(c1, &type1, is_cow(x1)); 919| 74| if (is_cow(x1)) { ------------------ | Branch (919:17): [True: 0, False: 74] ------------------ 920| 0| ra_set_container_at_index(&x1->high_low_container, pos1, c1, 921| 0| type1); 922| 0| } 923| 74| ra_append(&answer->high_low_container, s1, c1, type1); 924| 74| pos1++; 925| 74| if (pos1 == length1) break; ------------------ | Branch (925:17): [True: 74, False: 0] ------------------ 926| 0| s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 927| | 928| 0| } else { // s1 > s2 929| 0| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 930| 0| (uint16_t)pos2, &type2); 931| | // c2 = container_clone(c2, type2); 932| 0| c2 = get_copy_of_container(c2, &type2, is_cow(x2)); 933| 0| if (is_cow(x2)) { ------------------ | Branch (933:17): [True: 0, False: 0] ------------------ 934| 0| ra_set_container_at_index(&x2->high_low_container, pos2, c2, 935| 0| type2); 936| 0| } 937| 0| ra_append(&answer->high_low_container, s2, c2, type2); 938| 0| pos2++; 939| 0| if (pos2 == length2) break; ------------------ | Branch (939:17): [True: 0, False: 0] ------------------ 940| 0| s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 941| 0| } 942| 6.10k| } 943| 6.10k| if (pos1 == length1) { ------------------ | Branch (943:9): [True: 6.10k, False: 0] ------------------ 944| 6.10k| ra_append_copy_range(&answer->high_low_container, 945| 6.10k| &x2->high_low_container, pos2, length2, 946| 6.10k| is_cow(x2)); 947| 6.10k| } else if (pos2 == length2) { ------------------ | Branch (947:16): [True: 0, False: 0] ------------------ 948| 0| ra_append_copy_range(&answer->high_low_container, 949| 0| &x1->high_low_container, pos1, length1, 950| 0| is_cow(x1)); 951| 0| } 952| 6.10k| return answer; 953| 6.15k|} roaring_bitmap_or_inplace: 957| 6.15k| const roaring_bitmap_t *x2) { 958| 6.15k| uint8_t result_type = 0; 959| 6.15k| int length1 = x1->high_low_container.size; 960| 6.15k| const int length2 = x2->high_low_container.size; 961| | 962| 6.15k| if (0 == length2) return; ------------------ | Branch (962:9): [True: 0, False: 6.15k] ------------------ 963| | 964| 6.15k| if (0 == length1) { ------------------ | Branch (964:9): [True: 0, False: 6.15k] ------------------ 965| 0| roaring_bitmap_overwrite(x1, x2); 966| 0| return; 967| 0| } 968| 6.15k| int pos1 = 0, pos2 = 0; 969| 6.15k| uint8_t type1, type2; 970| 6.15k| uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 971| 6.15k| uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 972| 6.15k| while (true) { ------------------ | Branch (972:12): [True: 6.15k, Folded] ------------------ 973| 6.15k| if (s1 == s2) { ------------------ | Branch (973:13): [True: 6.15k, False: 0] ------------------ 974| 6.15k| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 975| 6.15k| (uint16_t)pos1, &type1); 976| 6.15k| if (!container_is_full(c1, type1)) { ------------------ | Branch (976:17): [True: 6.15k, False: 0] ------------------ 977| 6.15k| container_t *c2 = ra_get_container_at_index( 978| 6.15k| &x2->high_low_container, (uint16_t)pos2, &type2); 979| 6.15k| container_t *c = 980| 6.15k| (type1 == SHARED_CONTAINER_TYPE) ------------------ | | 51| 6.15k|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (980:21): [True: 0, False: 6.15k] ------------------ 981| 6.15k| ? container_or(c1, type1, c2, type2, &result_type) 982| 6.15k| : container_ior(c1, type1, c2, type2, &result_type); 983| | 984| 6.15k| if (c != c1) { // in this instance a new container was created, ------------------ | Branch (984:21): [True: 4.39k, False: 1.76k] ------------------ 985| | // and we need to free the old one 986| 4.39k| container_free(c1, type1); 987| 4.39k| } 988| 6.15k| ra_set_container_at_index(&x1->high_low_container, pos1, c, 989| 6.15k| result_type); 990| 6.15k| } 991| 6.15k| ++pos1; 992| 6.15k| ++pos2; 993| 6.15k| if (pos1 == length1) break; ------------------ | Branch (993:17): [True: 6.15k, False: 0] ------------------ 994| 0| if (pos2 == length2) break; ------------------ | Branch (994:17): [True: 0, False: 0] ------------------ 995| 0| s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 996| 0| s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 997| | 998| 0| } else if (s1 < s2) { // s1 < s2 ------------------ | Branch (998:20): [True: 0, False: 0] ------------------ 999| 0| pos1++; 1000| 0| if (pos1 == length1) break; ------------------ | Branch (1000:17): [True: 0, False: 0] ------------------ 1001| 0| s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 1002| | 1003| 0| } else { // s1 > s2 1004| 0| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 1005| 0| (uint16_t)pos2, &type2); 1006| 0| c2 = get_copy_of_container(c2, &type2, is_cow(x2)); 1007| 0| if (is_cow(x2)) { ------------------ | Branch (1007:17): [True: 0, False: 0] ------------------ 1008| 0| ra_set_container_at_index(&x2->high_low_container, pos2, c2, 1009| 0| type2); 1010| 0| } 1011| | 1012| | // container_t *c2_clone = container_clone(c2, type2); 1013| 0| ra_insert_new_key_value_at(&x1->high_low_container, pos1, s2, c2, 1014| 0| type2); 1015| 0| pos1++; 1016| 0| length1++; 1017| 0| pos2++; 1018| 0| if (pos2 == length2) break; ------------------ | Branch (1018:17): [True: 0, False: 0] ------------------ 1019| 0| s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 1020| 0| } 1021| 6.15k| } 1022| 6.15k| if (pos1 == length1) { ------------------ | Branch (1022:9): [True: 6.15k, False: 0] ------------------ 1023| 6.15k| ra_append_copy_range(&x1->high_low_container, &x2->high_low_container, 1024| 6.15k| pos2, length2, is_cow(x2)); 1025| 6.15k| } 1026| 6.15k|} roaring_bitmap_xor: 1029| 6.15k| const roaring_bitmap_t *x2) { 1030| 6.15k| uint8_t result_type = 0; 1031| 6.15k| const int length1 = x1->high_low_container.size, 1032| 6.15k| length2 = x2->high_low_container.size; 1033| 6.15k| if (0 == length1) { ------------------ | Branch (1033:9): [True: 0, False: 6.15k] ------------------ 1034| 0| return roaring_bitmap_copy(x2); 1035| 0| } 1036| 6.15k| if (0 == length2) { ------------------ | Branch (1036:9): [True: 47, False: 6.10k] ------------------ 1037| 47| return roaring_bitmap_copy(x1); 1038| 47| } 1039| 6.10k| roaring_bitmap_t *answer = 1040| 6.10k| roaring_bitmap_create_with_capacity(length1 + length2); 1041| 6.10k| roaring_bitmap_set_copy_on_write(answer, is_cow(x1) || is_cow(x2)); ------------------ | Branch (1041:46): [True: 0, False: 6.10k] | Branch (1041:60): [True: 0, False: 6.10k] ------------------ 1042| 6.10k| int pos1 = 0, pos2 = 0; 1043| 6.10k| uint8_t type1, type2; 1044| 6.10k| uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 1045| 6.10k| uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 1046| 6.10k| while (true) { ------------------ | Branch (1046:12): [True: 6.10k, Folded] ------------------ 1047| 6.10k| if (s1 == s2) { ------------------ | Branch (1047:13): [True: 6.03k, False: 74] ------------------ 1048| 6.03k| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 1049| 6.03k| (uint16_t)pos1, &type1); 1050| 6.03k| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 1051| 6.03k| (uint16_t)pos2, &type2); 1052| 6.03k| container_t *c = container_xor(c1, type1, c2, type2, &result_type); 1053| | 1054| 6.03k| if (container_nonzero_cardinality(c, result_type)) { ------------------ | Branch (1054:17): [True: 6.01k, False: 17] ------------------ 1055| 6.01k| ra_append(&answer->high_low_container, s1, c, result_type); 1056| 6.01k| } else { 1057| 17| container_free(c, result_type); 1058| 17| } 1059| 6.03k| ++pos1; 1060| 6.03k| ++pos2; 1061| 6.03k| if (pos1 == length1) break; ------------------ | Branch (1061:17): [True: 6.03k, False: 0] ------------------ 1062| 0| if (pos2 == length2) break; ------------------ | Branch (1062:17): [True: 0, False: 0] ------------------ 1063| 0| s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 1064| 0| s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 1065| | 1066| 74| } else if (s1 < s2) { // s1 < s2 ------------------ | Branch (1066:20): [True: 74, False: 0] ------------------ 1067| 74| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 1068| 74| (uint16_t)pos1, &type1); 1069| 74| c1 = get_copy_of_container(c1, &type1, is_cow(x1)); 1070| 74| if (is_cow(x1)) { ------------------ | Branch (1070:17): [True: 0, False: 74] ------------------ 1071| 0| ra_set_container_at_index(&x1->high_low_container, pos1, c1, 1072| 0| type1); 1073| 0| } 1074| 74| ra_append(&answer->high_low_container, s1, c1, type1); 1075| 74| pos1++; 1076| 74| if (pos1 == length1) break; ------------------ | Branch (1076:17): [True: 74, False: 0] ------------------ 1077| 0| s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 1078| | 1079| 0| } else { // s1 > s2 1080| 0| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 1081| 0| (uint16_t)pos2, &type2); 1082| 0| c2 = get_copy_of_container(c2, &type2, is_cow(x2)); 1083| 0| if (is_cow(x2)) { ------------------ | Branch (1083:17): [True: 0, False: 0] ------------------ 1084| 0| ra_set_container_at_index(&x2->high_low_container, pos2, c2, 1085| 0| type2); 1086| 0| } 1087| 0| ra_append(&answer->high_low_container, s2, c2, type2); 1088| 0| pos2++; 1089| 0| if (pos2 == length2) break; ------------------ | Branch (1089:17): [True: 0, False: 0] ------------------ 1090| 0| s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 1091| 0| } 1092| 6.10k| } 1093| 6.10k| if (pos1 == length1) { ------------------ | Branch (1093:9): [True: 6.10k, False: 0] ------------------ 1094| 6.10k| ra_append_copy_range(&answer->high_low_container, 1095| 6.10k| &x2->high_low_container, pos2, length2, 1096| 6.10k| is_cow(x2)); 1097| 6.10k| } else if (pos2 == length2) { ------------------ | Branch (1097:16): [True: 0, False: 0] ------------------ 1098| 0| ra_append_copy_range(&answer->high_low_container, 1099| 0| &x1->high_low_container, pos1, length1, 1100| 0| is_cow(x1)); 1101| 0| } 1102| 6.10k| return answer; 1103| 6.15k|} roaring_bitmap_xor_inplace: 1108| 6.15k| const roaring_bitmap_t *x2) { 1109| 6.15k| assert(x1 != x2); 1110| 6.15k| uint8_t result_type = 0; 1111| 6.15k| int length1 = x1->high_low_container.size; 1112| 6.15k| const int length2 = x2->high_low_container.size; 1113| | 1114| 6.15k| if (0 == length2) return; ------------------ | Branch (1114:9): [True: 13, False: 6.14k] ------------------ 1115| | 1116| 6.14k| if (0 == length1) { ------------------ | Branch (1116:9): [True: 1.50k, False: 4.63k] ------------------ 1117| 1.50k| roaring_bitmap_overwrite(x1, x2); 1118| 1.50k| return; 1119| 1.50k| } 1120| | 1121| | // XOR can have new containers inserted from x2, but can also 1122| | // lose containers when x1 and x2 are nonempty and identical. 1123| | 1124| 4.63k| int pos1 = 0, pos2 = 0; 1125| 4.63k| uint8_t type1, type2; 1126| 4.63k| uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 1127| 4.63k| uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 1128| 61.5k| while (true) { ------------------ | Branch (1128:12): [True: 61.5k, Folded] ------------------ 1129| 61.5k| if (s1 == s2) { ------------------ | Branch (1129:13): [True: 61.4k, False: 105] ------------------ 1130| 61.4k| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 1131| 61.4k| (uint16_t)pos1, &type1); 1132| 61.4k| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 1133| 61.4k| (uint16_t)pos2, &type2); 1134| | 1135| | // We do the computation "in place" only when c1 is not a shared 1136| | // container. Rationale: using a shared container safely with in 1137| | // place computation would require making a copy and then doing the 1138| | // computation in place which is likely less efficient than avoiding 1139| | // in place entirely and always generating a new container. 1140| | 1141| 61.4k| container_t *c; 1142| 61.4k| if (type1 == SHARED_CONTAINER_TYPE) { ------------------ | | 51| 61.4k|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (1142:17): [True: 0, False: 61.4k] ------------------ 1143| 0| c = container_xor(c1, type1, c2, type2, &result_type); 1144| 0| shared_container_free(CAST_shared(c1)); // so release ------------------ | | 79| 0|#define CAST_shared(c) CAST(shared_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1145| 61.4k| } else { 1146| 61.4k| c = container_ixor(c1, type1, c2, type2, &result_type); 1147| 61.4k| } 1148| | 1149| 61.4k| if (container_nonzero_cardinality(c, result_type)) { ------------------ | Branch (1149:17): [True: 4.33k, False: 57.1k] ------------------ 1150| 4.33k| ra_set_container_at_index(&x1->high_low_container, pos1, c, 1151| 4.33k| result_type); 1152| 4.33k| ++pos1; 1153| 57.1k| } else { 1154| 57.1k| container_free(c, result_type); 1155| 57.1k| ra_remove_at_index(&x1->high_low_container, pos1); 1156| 57.1k| --length1; 1157| 57.1k| } 1158| | 1159| 61.4k| ++pos2; 1160| 61.4k| if (pos1 == length1) break; ------------------ | Branch (1160:17): [True: 4.63k, False: 56.8k] ------------------ 1161| 56.8k| if (pos2 == length2) break; ------------------ | Branch (1161:17): [True: 0, False: 56.8k] ------------------ 1162| 56.8k| s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 1163| 56.8k| s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 1164| | 1165| 56.8k| } else if (s1 < s2) { // s1 < s2 ------------------ | Branch (1165:20): [True: 0, False: 105] ------------------ 1166| 0| pos1++; 1167| 0| if (pos1 == length1) break; ------------------ | Branch (1167:17): [True: 0, False: 0] ------------------ 1168| 0| s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 1169| | 1170| 105| } else { // s1 > s2 1171| 105| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 1172| 105| (uint16_t)pos2, &type2); 1173| 105| c2 = get_copy_of_container(c2, &type2, is_cow(x2)); 1174| 105| if (is_cow(x2)) { ------------------ | Branch (1174:17): [True: 0, False: 105] ------------------ 1175| 0| ra_set_container_at_index(&x2->high_low_container, pos2, c2, 1176| 0| type2); 1177| 0| } 1178| | 1179| 105| ra_insert_new_key_value_at(&x1->high_low_container, pos1, s2, c2, 1180| 105| type2); 1181| 105| pos1++; 1182| 105| length1++; 1183| 105| pos2++; 1184| 105| if (pos2 == length2) break; ------------------ | Branch (1184:17): [True: 0, False: 105] ------------------ 1185| 105| s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 1186| 105| } 1187| 61.5k| } 1188| 4.63k| if (pos1 == length1) { ------------------ | Branch (1188:9): [True: 4.63k, False: 0] ------------------ 1189| 4.63k| ra_append_copy_range(&x1->high_low_container, &x2->high_low_container, 1190| 4.63k| pos2, length2, is_cow(x2)); 1191| 4.63k| } 1192| 4.63k|} roaring_bitmap_andnot: 1195| 6.15k| const roaring_bitmap_t *x2) { 1196| 6.15k| uint8_t result_type = 0; 1197| 6.15k| const int length1 = x1->high_low_container.size, 1198| 6.15k| length2 = x2->high_low_container.size; 1199| 6.15k| if (0 == length1) { ------------------ | Branch (1199:9): [True: 0, False: 6.15k] ------------------ 1200| 0| roaring_bitmap_t *empty_bitmap = roaring_bitmap_create(); 1201| 0| roaring_bitmap_set_copy_on_write(empty_bitmap, 1202| 0| is_cow(x1) || is_cow(x2)); ------------------ | Branch (1202:42): [True: 0, False: 0] | Branch (1202:56): [True: 0, False: 0] ------------------ 1203| 0| return empty_bitmap; 1204| 0| } 1205| 6.15k| if (0 == length2) { ------------------ | Branch (1205:9): [True: 47, False: 6.10k] ------------------ 1206| 47| return roaring_bitmap_copy(x1); 1207| 47| } 1208| 6.10k| roaring_bitmap_t *answer = roaring_bitmap_create_with_capacity(length1); 1209| 6.10k| roaring_bitmap_set_copy_on_write(answer, is_cow(x1) || is_cow(x2)); ------------------ | Branch (1209:46): [True: 0, False: 6.10k] | Branch (1209:60): [True: 0, False: 6.10k] ------------------ 1210| | 1211| 6.10k| int pos1 = 0, pos2 = 0; 1212| 6.10k| uint8_t type1, type2; 1213| 6.10k| uint16_t s1 = 0; 1214| 6.10k| uint16_t s2 = 0; 1215| 6.10k| while (true) { ------------------ | Branch (1215:12): [True: 6.10k, Folded] ------------------ 1216| 6.10k| s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 1217| 6.10k| s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 1218| | 1219| 6.10k| if (s1 == s2) { ------------------ | Branch (1219:13): [True: 6.03k, False: 74] ------------------ 1220| 6.03k| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 1221| 6.03k| (uint16_t)pos1, &type1); 1222| 6.03k| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 1223| 6.03k| (uint16_t)pos2, &type2); 1224| 6.03k| container_t *c = 1225| 6.03k| container_andnot(c1, type1, c2, type2, &result_type); 1226| | 1227| 6.03k| if (container_nonzero_cardinality(c, result_type)) { ------------------ | Branch (1227:17): [True: 5.82k, False: 212] ------------------ 1228| 5.82k| ra_append(&answer->high_low_container, s1, c, result_type); 1229| 5.82k| } else { 1230| 212| container_free(c, result_type); 1231| 212| } 1232| 6.03k| ++pos1; 1233| 6.03k| ++pos2; 1234| 6.03k| if (pos1 == length1) break; ------------------ | Branch (1234:17): [True: 6.03k, False: 0] ------------------ 1235| 0| if (pos2 == length2) break; ------------------ | Branch (1235:17): [True: 0, False: 0] ------------------ 1236| 74| } else if (s1 < s2) { // s1 < s2 ------------------ | Branch (1236:20): [True: 74, False: 0] ------------------ 1237| 74| const int next_pos1 = 1238| 74| ra_advance_until(&x1->high_low_container, s2, pos1); 1239| 74| ra_append_copy_range(&answer->high_low_container, 1240| 74| &x1->high_low_container, pos1, next_pos1, 1241| 74| is_cow(x1)); 1242| | // TODO : perhaps some of the copy_on_write should be based on 1243| | // answer rather than x1 (more stringent?). Many similar cases 1244| 74| pos1 = next_pos1; 1245| 74| if (pos1 == length1) break; ------------------ | Branch (1245:17): [True: 74, False: 0] ------------------ 1246| 74| } else { // s1 > s2 1247| 0| pos2 = ra_advance_until(&x2->high_low_container, s1, pos2); 1248| 0| if (pos2 == length2) break; ------------------ | Branch (1248:17): [True: 0, False: 0] ------------------ 1249| 0| } 1250| 6.10k| } 1251| 6.10k| if (pos2 == length2) { ------------------ | Branch (1251:9): [True: 3.81k, False: 2.29k] ------------------ 1252| 3.81k| ra_append_copy_range(&answer->high_low_container, 1253| 3.81k| &x1->high_low_container, pos1, length1, 1254| 3.81k| is_cow(x1)); 1255| 3.81k| } 1256| 6.10k| return answer; 1257| 6.15k|} roaring_bitmap_andnot_inplace: 1262| 6.15k| const roaring_bitmap_t *x2) { 1263| 6.15k| assert(x1 != x2); 1264| | 1265| 6.15k| uint8_t result_type = 0; 1266| 6.15k| int length1 = x1->high_low_container.size; 1267| 6.15k| const int length2 = x2->high_low_container.size; 1268| 6.15k| int intersection_size = 0; 1269| | 1270| 6.15k| if (0 == length2) return; ------------------ | Branch (1270:9): [True: 422, False: 5.73k] ------------------ 1271| | 1272| 5.73k| if (0 == length1) { ------------------ | Branch (1272:9): [True: 0, False: 5.73k] ------------------ 1273| 0| roaring_bitmap_clear(x1); 1274| 0| return; 1275| 0| } 1276| | 1277| 5.73k| int pos1 = 0, pos2 = 0; 1278| 5.73k| uint8_t type1, type2; 1279| 5.73k| uint16_t s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 1280| 5.73k| uint16_t s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 1281| 5.73k| while (true) { ------------------ | Branch (1281:12): [True: 5.73k, Folded] ------------------ 1282| 5.73k| if (s1 == s2) { ------------------ | Branch (1282:13): [True: 5.73k, False: 0] ------------------ 1283| 5.73k| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 1284| 5.73k| (uint16_t)pos1, &type1); 1285| 5.73k| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 1286| 5.73k| (uint16_t)pos2, &type2); 1287| | 1288| | // We do the computation "in place" only when c1 is not a shared 1289| | // container. Rationale: using a shared container safely with in 1290| | // place computation would require making a copy and then doing the 1291| | // computation in place which is likely less efficient than avoiding 1292| | // in place entirely and always generating a new container. 1293| | 1294| 5.73k| container_t *c; 1295| 5.73k| if (type1 == SHARED_CONTAINER_TYPE) { ------------------ | | 51| 5.73k|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (1295:17): [True: 0, False: 5.73k] ------------------ 1296| 0| c = container_andnot(c1, type1, c2, type2, &result_type); 1297| 0| shared_container_free(CAST_shared(c1)); // release ------------------ | | 79| 0|#define CAST_shared(c) CAST(shared_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1298| 5.73k| } else { 1299| 5.73k| c = container_iandnot(c1, type1, c2, type2, &result_type); 1300| 5.73k| } 1301| | 1302| 5.73k| if (container_nonzero_cardinality(c, result_type)) { ------------------ | Branch (1302:17): [True: 4.22k, False: 1.50k] ------------------ 1303| 4.22k| ra_replace_key_and_container_at_index(&x1->high_low_container, 1304| 4.22k| intersection_size++, s1, 1305| 4.22k| c, result_type); 1306| 4.22k| } else { 1307| 1.50k| container_free(c, result_type); 1308| 1.50k| } 1309| | 1310| 5.73k| ++pos1; 1311| 5.73k| ++pos2; 1312| 5.73k| if (pos1 == length1) break; ------------------ | Branch (1312:17): [True: 3.61k, False: 2.11k] ------------------ 1313| 2.11k| if (pos2 == length2) break; ------------------ | Branch (1313:17): [True: 2.11k, False: 0] ------------------ 1314| 0| s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 1315| 0| s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 1316| | 1317| 0| } else if (s1 < s2) { // s1 < s2 ------------------ | Branch (1317:20): [True: 0, False: 0] ------------------ 1318| 0| if (pos1 != intersection_size) { ------------------ | Branch (1318:17): [True: 0, False: 0] ------------------ 1319| 0| container_t *c1 = ra_get_container_at_index( 1320| 0| &x1->high_low_container, (uint16_t)pos1, &type1); 1321| | 1322| 0| ra_replace_key_and_container_at_index( 1323| 0| &x1->high_low_container, intersection_size, s1, c1, type1); 1324| 0| } 1325| 0| intersection_size++; 1326| 0| pos1++; 1327| 0| if (pos1 == length1) break; ------------------ | Branch (1327:17): [True: 0, False: 0] ------------------ 1328| 0| s1 = ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 1329| | 1330| 0| } else { // s1 > s2 1331| 0| pos2 = ra_advance_until(&x2->high_low_container, s1, pos2); 1332| 0| if (pos2 == length2) break; ------------------ | Branch (1332:17): [True: 0, False: 0] ------------------ 1333| 0| s2 = ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 1334| 0| } 1335| 5.73k| } 1336| | 1337| 5.73k| if (pos1 < length1) { ------------------ | Branch (1337:9): [True: 2.11k, False: 3.61k] ------------------ 1338| | // all containers between intersection_size and 1339| | // pos1 are junk. However, they have either been moved 1340| | // (thus still referenced) or involved in an iandnot 1341| | // that will clean up all containers that could not be reused. 1342| | // Thus we should not free the junk containers between 1343| | // intersection_size and pos1. 1344| 2.11k| if (pos1 > intersection_size) { ------------------ | Branch (1344:13): [True: 35, False: 2.07k] ------------------ 1345| | // left slide of remaining items 1346| 35| ra_copy_range(&x1->high_low_container, pos1, length1, 1347| 35| intersection_size); 1348| 35| } 1349| | // else current placement is fine 1350| 2.11k| intersection_size += (length1 - pos1); 1351| 2.11k| } 1352| 5.73k| ra_downsize(&x1->high_low_container, intersection_size); 1353| 5.73k|} roaring_bitmap_get_cardinality: 1355| 61.5k|uint64_t roaring_bitmap_get_cardinality(const roaring_bitmap_t *r) { 1356| 61.5k| const roaring_array_t *ra = &r->high_low_container; 1357| | 1358| 61.5k| uint64_t card = 0; 1359| 406k| for (int i = 0; i < ra->size; ++i) ------------------ | Branch (1359:21): [True: 345k, False: 61.5k] ------------------ 1360| 345k| card += container_get_cardinality(ra->containers[i], ra->typecodes[i]); 1361| 61.5k| return card; 1362| 61.5k|} roaring_bitmap_is_empty: 1425| 12.3k|bool roaring_bitmap_is_empty(const roaring_bitmap_t *r) { 1426| 12.3k| return r->high_low_container.size == 0; 1427| 12.3k|} roaring_bitmap_to_uint32_array: 1429| 6.15k|void roaring_bitmap_to_uint32_array(const roaring_bitmap_t *r, uint32_t *ans) { 1430| 6.15k| ra_to_uint32_array(&r->high_low_container, ans); 1431| 6.15k|} roaring_bitmap_run_optimize: 1449| 12.3k|bool roaring_bitmap_run_optimize(roaring_bitmap_t *r) { 1450| 12.3k| bool answer = false; 1451| 24.6k| for (int i = 0; i < r->high_low_container.size; i++) { ------------------ | Branch (1451:21): [True: 12.3k, False: 12.3k] ------------------ 1452| 12.3k| uint8_t type_original, type_after; 1453| 12.3k| ra_unshare_container_at_index( 1454| 12.3k| &r->high_low_container, 1455| 12.3k| (uint16_t)i); // TODO: this introduces extra cloning! 1456| 12.3k| container_t *c = ra_get_container_at_index(&r->high_low_container, 1457| 12.3k| (uint16_t)i, &type_original); 1458| 12.3k| container_t *c1 = convert_run_optimize(c, type_original, &type_after); 1459| 12.3k| if (type_after == RUN_CONTAINER_TYPE) { ------------------ | | 50| 12.3k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (1459:13): [True: 3.86k, False: 8.44k] ------------------ 1460| | answer = true; 1461| 3.86k| } 1462| 12.3k| ra_set_container_at_index(&r->high_low_container, i, c1, type_after); 1463| 12.3k| } 1464| 12.3k| return answer; 1465| 12.3k|} roaring_bitmap_shrink_to_fit: 1467| 6.15k|size_t roaring_bitmap_shrink_to_fit(roaring_bitmap_t *r) { 1468| 6.15k| size_t answer = 0; 1469| 12.3k| for (int i = 0; i < r->high_low_container.size; i++) { ------------------ | Branch (1469:21): [True: 6.15k, False: 6.15k] ------------------ 1470| 6.15k| uint8_t type_original; 1471| 6.15k| container_t *c = ra_get_container_at_index(&r->high_low_container, 1472| 6.15k| (uint16_t)i, &type_original); 1473| 6.15k| answer += container_shrink_to_fit(c, type_original); 1474| 6.15k| } 1475| 6.15k| answer += ra_shrink_to_fit(&r->high_low_container); 1476| 6.15k| return answer; 1477| 6.15k|} roaring_bitmap_remove_run_compression: 1483| 6.15k|bool roaring_bitmap_remove_run_compression(roaring_bitmap_t *r) { 1484| 6.15k| bool answer = false; 1485| 79.7k| for (int i = 0; i < r->high_low_container.size; i++) { ------------------ | Branch (1485:21): [True: 73.6k, False: 6.15k] ------------------ 1486| 73.6k| uint8_t type_original, type_after; 1487| 73.6k| container_t *c = ra_get_container_at_index(&r->high_low_container, 1488| 73.6k| (uint16_t)i, &type_original); 1489| 73.6k| if (get_container_type(c, type_original) == RUN_CONTAINER_TYPE) { ------------------ | | 50| 73.6k|#define RUN_CONTAINER_TYPE 3 ------------------ | Branch (1489:13): [True: 66.2k, False: 7.34k] ------------------ 1490| 66.2k| answer = true; 1491| 66.2k| if (type_original == SHARED_CONTAINER_TYPE) { ------------------ | | 51| 66.2k|#define SHARED_CONTAINER_TYPE 4 ------------------ | Branch (1491:17): [True: 0, False: 66.2k] ------------------ 1492| 0| run_container_t *truec = CAST_run(CAST_shared(c)->container); ------------------ | | 77| 0|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1493| 0| int32_t card = run_container_cardinality(truec); 1494| 0| container_t *c1 = convert_to_bitset_or_array_container( 1495| 0| truec, card, &type_after); 1496| 0| shared_container_free(CAST_shared(c)); // frees run as needed ------------------ | | 79| 0|#define CAST_shared(c) CAST(shared_container_t *, c) // safer downcast | | ------------------ | | | | 86| 0|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1497| 0| ra_set_container_at_index(&r->high_low_container, i, c1, 1498| 0| type_after); 1499| | 1500| 66.2k| } else { 1501| 66.2k| int32_t card = run_container_cardinality(CAST_run(c)); ------------------ | | 77| 66.2k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 66.2k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1502| 66.2k| container_t *c1 = convert_to_bitset_or_array_container( 1503| 66.2k| CAST_run(c), card, &type_after); ------------------ | | 77| 66.2k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 66.2k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1504| 66.2k| run_container_free(CAST_run(c)); ------------------ | | 77| 66.2k|#define CAST_run(c) CAST(run_container_t *, c) // safer downcast | | ------------------ | | | | 86| 66.2k|#define CAST(type, value) ((type)value) | | ------------------ ------------------ 1505| 66.2k| ra_set_container_at_index(&r->high_low_container, i, c1, 1506| 66.2k| type_after); 1507| 66.2k| } 1508| 66.2k| } 1509| 73.6k| } 1510| 6.15k| return answer; 1511| 6.15k|} roaring_bitmap_portable_size_in_bytes: 1546| 12.3k|size_t roaring_bitmap_portable_size_in_bytes(const roaring_bitmap_t *r) { 1547| 12.3k| return ra_portable_size_in_bytes(&r->high_low_container); 1548| 12.3k|} roaring_bitmap_portable_deserialize_safe: 1551| 12.3k| size_t maxbytes) { 1552| 12.3k| roaring_bitmap_t *ans = 1553| 12.3k| (roaring_bitmap_t *)roaring_malloc(sizeof(roaring_bitmap_t)); 1554| 12.3k| if (ans == NULL) { ------------------ | Branch (1554:9): [True: 0, False: 12.3k] ------------------ 1555| 0| return NULL; 1556| 0| } 1557| 12.3k| size_t bytesread; 1558| 12.3k| bool is_ok = ra_portable_deserialize(&ans->high_low_container, buf, 1559| 12.3k| maxbytes, &bytesread); 1560| 12.3k| if (!is_ok) { ------------------ | Branch (1560:9): [True: 6.14k, False: 6.16k] ------------------ 1561| 6.14k| roaring_free(ans); 1562| 6.14k| return NULL; 1563| 6.14k| } 1564| 6.16k| roaring_bitmap_set_copy_on_write(ans, false); 1565| 6.16k| if (!is_ok) { ------------------ | Branch (1565:9): [True: 0, False: 6.16k] ------------------ 1566| 0| roaring_free(ans); 1567| 0| return NULL; 1568| 0| } 1569| 6.16k| return ans; 1570| 6.16k|} roaring_bitmap_portable_serialize: 1581| 6.15k|size_t roaring_bitmap_portable_serialize(const roaring_bitmap_t *r, char *buf) { 1582| 6.15k| return ra_portable_serialize(&r->high_low_container, buf); 1583| 6.15k|} roaring_iterate: 1664| 6.14k| void *ptr) { 1665| 6.14k| const roaring_array_t *ra = &r->high_low_container; 1666| | 1667| 69.2k| for (int i = 0; i < ra->size; ++i) ------------------ | Branch (1667:21): [True: 63.0k, False: 6.14k] ------------------ 1668| 63.0k| if (!container_iterate(ra->containers[i], ra->typecodes[i], ------------------ | Branch (1668:13): [True: 0, False: 63.0k] ------------------ 1669| 63.0k| ((uint32_t)ra->keys[i]) << 16, iterator, ptr)) { 1670| 0| return false; 1671| 0| } 1672| 6.14k| return true; 1673| 6.14k|} roaring_iterator_init: 1780| 12.3k| roaring_uint32_iterator_t *newit) { 1781| 12.3k| newit->parent = r; 1782| 12.3k| newit->container_index = 0; 1783| 12.3k| newit->has_value = loadfirstvalue(newit); 1784| 12.3k|} roaring_uint32_iterator_move_equalorlarger: 1812| 6.15k| uint32_t val) { 1813| 6.15k| uint16_t hb = val >> 16; 1814| 6.15k| const int i = ra_get_index(&it->parent->high_low_container, hb); 1815| 6.15k| if (i >= 0) { ------------------ | Branch (1815:9): [True: 6.06k, False: 89] ------------------ 1816| 6.06k| uint32_t lowvalue = 1817| 6.06k| container_maximum(it->parent->high_low_container.containers[i], 1818| 6.06k| it->parent->high_low_container.typecodes[i]); 1819| 6.06k| uint16_t lb = val & 0xFFFF; 1820| 6.06k| if (lowvalue < lb) { ------------------ | Branch (1820:13): [True: 9, False: 6.05k] ------------------ 1821| | // will have to load first value of next container 1822| 9| it->container_index = i + 1; 1823| 6.05k| } else { 1824| | // the value is necessarily within the range of the container 1825| 6.05k| it->container_index = i; 1826| 6.05k| it->has_value = loadfirstvalue_largeorequal(it, val); 1827| 6.05k| return it->has_value; 1828| 6.05k| } 1829| 6.06k| } else { 1830| | // there is no matching, so we are going for the next container 1831| 89| it->container_index = -i - 1; 1832| 89| } 1833| 98| it->has_value = loadfirstvalue(it); 1834| 98| return it->has_value; 1835| 6.15k|} roaring_uint32_iterator_advance: 1837| 320k|bool roaring_uint32_iterator_advance(roaring_uint32_iterator_t *it) { 1838| 320k| if (it->container_index >= it->parent->high_low_container.size) { ------------------ | Branch (1838:9): [True: 0, False: 320k] ------------------ 1839| 0| return (it->has_value = false); 1840| 0| } 1841| 320k| if (it->container_index < 0) { ------------------ | Branch (1841:9): [True: 0, False: 320k] ------------------ 1842| 0| it->container_index = 0; 1843| 0| return (it->has_value = loadfirstvalue(it)); 1844| 0| } 1845| 320k| uint16_t low16 = (uint16_t)it->current_value; 1846| 320k| if (container_iterator_next(it->container, it->typecode, &it->container_it, ------------------ | Branch (1846:9): [True: 314k, False: 5.94k] ------------------ 1847| 320k| &low16)) { 1848| 314k| it->current_value = it->highbits | low16; 1849| 314k| return (it->has_value = true); 1850| 314k| } 1851| 5.94k| it->container_index++; 1852| 5.94k| return (it->has_value = loadfirstvalue(it)); 1853| 320k|} roaring_bitmap_equals: 2048| 12.3k| const roaring_bitmap_t *r2) { 2049| 12.3k| const roaring_array_t *ra1 = &r1->high_low_container; 2050| 12.3k| const roaring_array_t *ra2 = &r2->high_low_container; 2051| | 2052| 12.3k| if (ra1->size != ra2->size) { ------------------ | Branch (2052:9): [True: 2.39k, False: 9.91k] ------------------ 2053| 2.39k| return false; 2054| 2.39k| } 2055| 87.2k| for (int i = 0; i < ra1->size; ++i) { ------------------ | Branch (2055:21): [True: 77.3k, False: 9.88k] ------------------ 2056| 77.3k| if (ra1->keys[i] != ra2->keys[i]) { ------------------ | Branch (2056:13): [True: 24, False: 77.3k] ------------------ 2057| 24| return false; 2058| 24| } 2059| 77.3k| } 2060| 83.5k| for (int i = 0; i < ra1->size; ++i) { ------------------ | Branch (2060:21): [True: 77.3k, False: 6.15k] ------------------ 2061| 77.3k| bool areequal = container_equals(ra1->containers[i], ra1->typecodes[i], 2062| 77.3k| ra2->containers[i], ra2->typecodes[i]); 2063| 77.3k| if (!areequal) { ------------------ | Branch (2063:13): [True: 3.73k, False: 73.6k] ------------------ 2064| 3.73k| return false; 2065| 3.73k| } 2066| 77.3k| } 2067| 6.15k| return true; 2068| 9.88k|} roaring_bitmap_is_subset: 2071| 11.1k| const roaring_bitmap_t *r2) { 2072| 11.1k| const roaring_array_t *ra1 = &r1->high_low_container; 2073| 11.1k| const roaring_array_t *ra2 = &r2->high_low_container; 2074| | 2075| 11.1k| const int length1 = ra1->size, length2 = ra2->size; 2076| | 2077| 11.1k| int pos1 = 0, pos2 = 0; 2078| | 2079| 11.1k| while (pos1 < length1 && pos2 < length2) { ------------------ | Branch (2079:12): [True: 10.6k, False: 424] | Branch (2079:30): [True: 10.6k, False: 47] ------------------ 2080| 10.6k| const uint16_t s1 = ra_get_key_at_index(ra1, (uint16_t)pos1); 2081| 10.6k| const uint16_t s2 = ra_get_key_at_index(ra2, (uint16_t)pos2); 2082| | 2083| 10.6k| if (s1 == s2) { ------------------ | Branch (2083:13): [True: 10.5k, False: 115] ------------------ 2084| 10.5k| uint8_t type1, type2; 2085| 10.5k| container_t *c1 = 2086| 10.5k| ra_get_container_at_index(ra1, (uint16_t)pos1, &type1); 2087| 10.5k| container_t *c2 = 2088| 10.5k| ra_get_container_at_index(ra2, (uint16_t)pos2, &type2); 2089| 10.5k| if (!container_is_subset(c1, type1, c2, type2)) return false; ------------------ | Branch (2089:17): [True: 10.5k, False: 0] ------------------ 2090| 0| ++pos1; 2091| 0| ++pos2; 2092| 115| } else if (s1 < s2) { // s1 < s2 ------------------ | Branch (2092:20): [True: 115, False: 0] ------------------ 2093| 115| return false; 2094| 115| } else { // s1 > s2 2095| 0| pos2 = ra_advance_until(ra2, s1, pos2); 2096| 0| } 2097| 10.6k| } 2098| 471| if (pos1 == length1) ------------------ | Branch (2098:9): [True: 424, False: 47] ------------------ 2099| 424| return true; 2100| 47| else 2101| 47| return false; 2102| 471|} roaring_bitmap_flip_inplace: 2271| 6.15k| uint64_t range_end) { 2272| 6.15k| if (range_start >= range_end || range_start > (uint64_t)UINT32_MAX + 1) { ------------------ | Branch (2272:9): [True: 5.68k, False: 465] | Branch (2272:37): [True: 0, False: 465] ------------------ 2273| 5.68k| return; 2274| 5.68k| } 2275| 465| roaring_bitmap_flip_inplace_closed(x1, (uint32_t)range_start, 2276| 465| (uint32_t)(range_end - 1)); 2277| 465|} roaring_bitmap_flip_inplace_closed: 2281| 6.61k| uint32_t range_end) { 2282| 6.61k| if (range_start > range_end) { ------------------ | Branch (2282:9): [True: 159, False: 6.46k] ------------------ 2283| 159| return; // empty range 2284| 159| } 2285| | 2286| 6.46k| uint16_t hb_start = (uint16_t)(range_start >> 16); 2287| 6.46k| const uint16_t lb_start = (uint16_t)range_start; 2288| 6.46k| uint16_t hb_end = (uint16_t)(range_end >> 16); 2289| 6.46k| const uint16_t lb_end = (uint16_t)range_end; 2290| | 2291| 6.46k| if (hb_start == hb_end) { ------------------ | Branch (2291:9): [True: 6.04k, False: 420] ------------------ 2292| 6.04k| inplace_flip_container(&x1->high_low_container, hb_start, lb_start, 2293| 6.04k| lb_end); 2294| 6.04k| } else { 2295| | // start and end containers are distinct 2296| 420| if (lb_start > 0) { ------------------ | Branch (2296:13): [True: 339, False: 81] ------------------ 2297| | // handle first (partial) container 2298| 339| inplace_flip_container(&x1->high_low_container, hb_start, lb_start, 2299| 339| 0xFFFF); 2300| 339| ++hb_start; // for the full containers. Can't wrap. 2301| 339| } 2302| | 2303| 420| if (lb_end != 0xFFFF) --hb_end; ------------------ | Branch (2303:13): [True: 398, False: 22] ------------------ 2304| | 2305| 21.7k| for (uint32_t hb = hb_start; hb <= hb_end; ++hb) { ------------------ | Branch (2305:38): [True: 21.3k, False: 420] ------------------ 2306| 21.3k| inplace_fully_flip_container(&x1->high_low_container, (uint16_t)hb); 2307| 21.3k| } 2308| | // handle a partial final container 2309| 420| if (lb_end != 0xFFFF) { ------------------ | Branch (2309:13): [True: 398, False: 22] ------------------ 2310| 398| inplace_flip_container(&x1->high_low_container, hb_end + 1, 0, 2311| 398| lb_end); 2312| 398| ++hb_end; 2313| 398| } 2314| 420| } 2315| 6.46k|} roaring_bitmap_rank: 2781| 6.15k|uint64_t roaring_bitmap_rank(const roaring_bitmap_t *bm, uint32_t x) { 2782| 6.15k| uint64_t size = 0; 2783| 6.15k| uint32_t xhigh = x >> 16; 2784| 6.15k| for (int i = 0; i < bm->high_low_container.size; i++) { ------------------ | Branch (2784:21): [True: 6.15k, False: 0] ------------------ 2785| 6.15k| uint32_t key = bm->high_low_container.keys[i]; 2786| 6.15k| if (xhigh > key) { ------------------ | Branch (2786:13): [True: 0, False: 6.15k] ------------------ 2787| 0| size += 2788| 0| container_get_cardinality(bm->high_low_container.containers[i], 2789| 0| bm->high_low_container.typecodes[i]); 2790| 6.15k| } else if (xhigh == key) { ------------------ | Branch (2790:20): [True: 6.15k, False: 0] ------------------ 2791| 6.15k| return size + container_rank(bm->high_low_container.containers[i], 2792| 6.15k| bm->high_low_container.typecodes[i], 2793| 6.15k| x & 0xFFFF); 2794| 6.15k| } else { 2795| 0| return size; 2796| 0| } 2797| 6.15k| } 2798| 0| return size; 2799| 6.15k|} roaring_bitmap_minimum: 2860| 6.15k|uint32_t roaring_bitmap_minimum(const roaring_bitmap_t *bm) { 2861| 6.15k| if (bm->high_low_container.size > 0) { ------------------ | Branch (2861:9): [True: 6.10k, False: 47] ------------------ 2862| 6.10k| container_t *c = bm->high_low_container.containers[0]; 2863| 6.10k| uint8_t type = bm->high_low_container.typecodes[0]; 2864| 6.10k| uint32_t key = bm->high_low_container.keys[0]; 2865| 6.10k| uint32_t lowvalue = container_minimum(c, type); 2866| 6.10k| return lowvalue | (key << 16); 2867| 6.10k| } 2868| 47| return UINT32_MAX; 2869| 6.15k|} roaring_bitmap_maximum: 2875| 6.15k|uint32_t roaring_bitmap_maximum(const roaring_bitmap_t *bm) { 2876| 6.15k| if (bm->high_low_container.size > 0) { ------------------ | Branch (2876:9): [True: 6.10k, False: 47] ------------------ 2877| 6.10k| container_t *container = 2878| 6.10k| bm->high_low_container.containers[bm->high_low_container.size - 1]; 2879| 6.10k| uint8_t typecode = 2880| 6.10k| bm->high_low_container.typecodes[bm->high_low_container.size - 1]; 2881| 6.10k| uint32_t key = 2882| 6.10k| bm->high_low_container.keys[bm->high_low_container.size - 1]; 2883| 6.10k| uint32_t lowvalue = container_maximum(container, typecode); 2884| 6.10k| return lowvalue | (key << 16); 2885| 6.10k| } 2886| 47| return 0; 2887| 6.15k|} roaring_bitmap_select: 2890| 6.15k| uint32_t *element) { 2891| 6.15k| container_t *container; 2892| 6.15k| uint8_t typecode; 2893| 6.15k| uint16_t key; 2894| 6.15k| uint32_t start_rank = 0; 2895| 6.15k| int i = 0; 2896| 6.15k| bool valid = false; 2897| 12.3k| while (!valid && i < bm->high_low_container.size) { ------------------ | Branch (2897:12): [True: 6.58k, False: 5.71k] | Branch (2897:22): [True: 6.15k, False: 435] ------------------ 2898| 6.15k| container = bm->high_low_container.containers[i]; 2899| 6.15k| typecode = bm->high_low_container.typecodes[i]; 2900| 6.15k| valid = 2901| 6.15k| container_select(container, typecode, &start_rank, rank, element); 2902| 6.15k| i++; 2903| 6.15k| } 2904| | 2905| 6.15k| if (valid) { ------------------ | Branch (2905:9): [True: 5.71k, False: 435] ------------------ 2906| 5.71k| key = bm->high_low_container.keys[i - 1]; 2907| 5.71k| *element |= (((uint32_t)key) << 16); // w/o cast, key promotes signed 2908| 5.71k| return true; 2909| 5.71k| } else 2910| 435| return false; 2911| 6.15k|} roaring_bitmap_intersect: 2914| 6.15k| const roaring_bitmap_t *x2) { 2915| 6.15k| const int length1 = x1->high_low_container.size, 2916| 6.15k| length2 = x2->high_low_container.size; 2917| 6.15k| uint64_t answer = 0; 2918| 6.15k| int pos1 = 0, pos2 = 0; 2919| | 2920| 6.52k| while (pos1 < length1 && pos2 < length2) { ------------------ | Branch (2920:12): [True: 6.15k, False: 375] | Branch (2920:30): [True: 6.10k, False: 47] ------------------ 2921| 6.10k| const uint16_t s1 = 2922| 6.10k| ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 2923| 6.10k| const uint16_t s2 = 2924| 6.10k| ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 2925| | 2926| 6.10k| if (s1 == s2) { ------------------ | Branch (2926:13): [True: 6.03k, False: 74] ------------------ 2927| 6.03k| uint8_t type1, type2; 2928| 6.03k| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 2929| 6.03k| (uint16_t)pos1, &type1); 2930| 6.03k| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 2931| 6.03k| (uint16_t)pos2, &type2); 2932| 6.03k| if (container_intersect(c1, type1, c2, type2)) return true; ------------------ | Branch (2932:17): [True: 5.73k, False: 301] ------------------ 2933| 301| ++pos1; 2934| 301| ++pos2; 2935| 301| } else if (s1 < s2) { // s1 < s2 ------------------ | Branch (2935:20): [True: 74, False: 0] ------------------ 2936| 74| pos1 = ra_advance_until(&x1->high_low_container, s2, pos1); 2937| 74| } else { // s1 > s2 2938| 0| pos2 = ra_advance_until(&x2->high_low_container, s1, pos2); 2939| 0| } 2940| 6.10k| } 2941| 422| return answer != 0; 2942| 6.15k|} roaring_bitmap_and_cardinality: 2964| 24.6k| const roaring_bitmap_t *x2) { 2965| 24.6k| const int length1 = x1->high_low_container.size, 2966| 24.6k| length2 = x2->high_low_container.size; 2967| 24.6k| uint64_t answer = 0; 2968| 24.6k| int pos1 = 0, pos2 = 0; 2969| 49.0k| while (pos1 < length1 && pos2 < length2) { ------------------ | Branch (2969:12): [True: 24.6k, False: 24.4k] | Branch (2969:30): [True: 24.4k, False: 188] ------------------ 2970| 24.4k| const uint16_t s1 = 2971| 24.4k| ra_get_key_at_index(&x1->high_low_container, (uint16_t)pos1); 2972| 24.4k| const uint16_t s2 = 2973| 24.4k| ra_get_key_at_index(&x2->high_low_container, (uint16_t)pos2); 2974| | 2975| 24.4k| if (s1 == s2) { ------------------ | Branch (2975:13): [True: 24.1k, False: 296] ------------------ 2976| 24.1k| uint8_t type1, type2; 2977| 24.1k| container_t *c1 = ra_get_container_at_index(&x1->high_low_container, 2978| 24.1k| (uint16_t)pos1, &type1); 2979| 24.1k| container_t *c2 = ra_get_container_at_index(&x2->high_low_container, 2980| 24.1k| (uint16_t)pos2, &type2); 2981| 24.1k| answer += container_and_cardinality(c1, type1, c2, type2); 2982| 24.1k| ++pos1; 2983| 24.1k| ++pos2; 2984| 24.1k| } else if (s1 < s2) { // s1 < s2 ------------------ | Branch (2984:20): [True: 296, False: 0] ------------------ 2985| 296| pos1 = ra_advance_until(&x1->high_low_container, s2, pos1); 2986| 296| } else { // s1 > s2 2987| 0| pos2 = ra_advance_until(&x2->high_low_container, s1, pos2); 2988| 0| } 2989| 24.4k| } 2990| 24.6k| return answer; 2991| 24.6k|} roaring_bitmap_jaccard_index: 2994| 6.15k| const roaring_bitmap_t *x2) { 2995| 6.15k| const uint64_t c1 = roaring_bitmap_get_cardinality(x1); 2996| 6.15k| const uint64_t c2 = roaring_bitmap_get_cardinality(x2); 2997| 6.15k| const uint64_t inter = roaring_bitmap_and_cardinality(x1, x2); 2998| 6.15k| return (double)inter / (double)(c1 + c2 - inter); 2999| 6.15k|} roaring_bitmap_or_cardinality: 3002| 6.15k| const roaring_bitmap_t *x2) { 3003| 6.15k| const uint64_t c1 = roaring_bitmap_get_cardinality(x1); 3004| 6.15k| const uint64_t c2 = roaring_bitmap_get_cardinality(x2); 3005| 6.15k| const uint64_t inter = roaring_bitmap_and_cardinality(x1, x2); 3006| 6.15k| return c1 + c2 - inter; 3007| 6.15k|} roaring_bitmap_andnot_cardinality: 3010| 6.15k| const roaring_bitmap_t *x2) { 3011| 6.15k| const uint64_t c1 = roaring_bitmap_get_cardinality(x1); 3012| 6.15k| const uint64_t inter = roaring_bitmap_and_cardinality(x1, x2); 3013| 6.15k| return c1 - inter; 3014| 6.15k|} roaring_bitmap_xor_cardinality: 3017| 6.15k| const roaring_bitmap_t *x2) { 3018| 6.15k| const uint64_t c1 = roaring_bitmap_get_cardinality(x1); 3019| 6.15k| const uint64_t c2 = roaring_bitmap_get_cardinality(x2); 3020| 6.15k| const uint64_t inter = roaring_bitmap_and_cardinality(x1, x2); 3021| 6.15k| return c1 + c2 - 2 * inter; 3022| 6.15k|} roaring_bitmap_contains_range: 3029| 6.15k| uint64_t range_start, uint64_t range_end) { 3030| 6.15k| if (range_start >= range_end || range_start > (uint64_t)UINT32_MAX + 1) { ------------------ | Branch (3030:9): [True: 5.37k, False: 782] | Branch (3030:37): [True: 0, False: 782] ------------------ 3031| 5.37k| return true; 3032| 5.37k| } 3033| 782| return roaring_bitmap_contains_range_closed(r, (uint32_t)range_start, 3034| 782| (uint32_t)(range_end - 1)); 3035| 6.15k|} roaring_bitmap_contains_range_closed: 3043| 782| uint32_t range_end) { 3044| 782| if (range_start > range_end) { ------------------ | Branch (3044:9): [True: 0, False: 782] ------------------ 3045| 0| return true; 3046| 0| } // empty range are always contained! 3047| 782| if (range_end == range_start) { ------------------ | Branch (3047:9): [True: 70, False: 712] ------------------ 3048| 70| return roaring_bitmap_contains(r, (uint32_t)range_start); 3049| 70| } 3050| 712| uint16_t hb_rs = (uint16_t)(range_start >> 16); 3051| 712| uint16_t hb_re = (uint16_t)(range_end >> 16); 3052| 712| const int32_t span = hb_re - hb_rs; 3053| 712| const int32_t hlc_sz = ra_get_size(&r->high_low_container); 3054| 712| if (hlc_sz < span + 1) { ------------------ | Branch (3054:9): [True: 193, False: 519] ------------------ 3055| 193| return false; 3056| 193| } 3057| 519| int32_t is = ra_get_index(&r->high_low_container, hb_rs); 3058| 519| int32_t ie = ra_get_index(&r->high_low_container, hb_re); 3059| 519| if ((ie < 0) || (is < 0) || ((ie - is) != span) || ie >= hlc_sz) { ------------------ | Branch (3059:9): [True: 43, False: 476] | Branch (3059:21): [True: 16, False: 460] | Branch (3059:33): [True: 7, False: 453] | Branch (3059:56): [True: 0, False: 453] ------------------ 3060| 66| return false; 3061| 66| } 3062| 453| const uint32_t lb_rs = range_start & 0xFFFF; 3063| 453| const uint32_t lb_re = (range_end & 0xFFFF) + 1; 3064| 453| uint8_t type; 3065| 453| container_t *c = 3066| 453| ra_get_container_at_index(&r->high_low_container, (uint16_t)is, &type); 3067| 453| if (hb_rs == hb_re) { ------------------ | Branch (3067:9): [True: 387, False: 66] ------------------ 3068| 387| return container_contains_range(c, lb_rs, lb_re, type); 3069| 387| } 3070| 66| if (!container_contains_range(c, lb_rs, 1 << 16, type)) { ------------------ | Branch (3070:9): [True: 12, False: 54] ------------------ 3071| 12| return false; 3072| 12| } 3073| 54| c = ra_get_container_at_index(&r->high_low_container, (uint16_t)ie, &type); 3074| 54| if (!container_contains_range(c, 0, lb_re, type)) { ------------------ | Branch (3074:9): [True: 6, False: 48] ------------------ 3075| 6| return false; 3076| 6| } 3077| 1.10k| for (int32_t i = is + 1; i < ie; ++i) { ------------------ | Branch (3077:30): [True: 1.07k, False: 29] ------------------ 3078| 1.07k| c = ra_get_container_at_index(&r->high_low_container, (uint16_t)i, 3079| 1.07k| &type); 3080| 1.07k| if (!container_is_full(c, type)) { ------------------ | Branch (3080:13): [True: 19, False: 1.06k] ------------------ 3081| 19| return false; 3082| 19| } 3083| 1.07k| } 3084| 29| return true; 3085| 48|} roaring_bitmap_is_strict_subset: 3088| 6.15k| const roaring_bitmap_t *r2) { 3089| 6.15k| return (roaring_bitmap_get_cardinality(r2) > ------------------ | Branch (3089:13): [True: 4.96k, False: 1.19k] ------------------ 3090| 6.15k| roaring_bitmap_get_cardinality(r1) && 3091| 4.96k| roaring_bitmap_is_subset(r1, r2)); ------------------ | Branch (3091:13): [True: 212, False: 4.74k] ------------------ 3092| 6.15k|} roaring.c:containerptr_roaring_bitmap_add: 56| 18.4k| int *index) { 57| 18.4k| roaring_array_t *ra = &r->high_low_container; 58| | 59| 18.4k| uint16_t hb = val >> 16; 60| 18.4k| const int i = ra_get_index(ra, hb); 61| 18.4k| if (i >= 0) { ------------------ | Branch (61:9): [True: 6.15k, False: 12.3k] ------------------ 62| 6.15k| ra_unshare_container_at_index(ra, (uint16_t)i); 63| 6.15k| container_t *c = ra_get_container_at_index(ra, (uint16_t)i, type); 64| 6.15k| uint8_t new_type = *type; 65| 6.15k| container_t *c2 = container_add(c, val & 0xFFFF, *type, &new_type); 66| 6.15k| *index = i; 67| 6.15k| if (c2 != c) { ------------------ | Branch (67:13): [True: 2, False: 6.15k] ------------------ 68| 2| container_free(c, *type); 69| 2| ra_set_container_at_index(ra, i, c2, new_type); 70| 2| *type = new_type; 71| 2| return c2; 72| 6.15k| } else { 73| 6.15k| return c; 74| 6.15k| } 75| 12.3k| } else { 76| 12.3k| array_container_t *new_ac = array_container_create(); 77| 12.3k| container_t *c = 78| 12.3k| container_add(new_ac, val & 0xFFFF, ARRAY_CONTAINER_TYPE, type); ------------------ | | 49| 12.3k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 79| | // we could just assume that it stays an array container 80| 12.3k| ra_insert_new_key_value_at(ra, -i - 1, hb, c, *type); 81| 12.3k| *index = -i - 1; 82| 12.3k| return c; 83| 12.3k| } 84| 18.4k|} roaring.c:add_bulk_impl: 106| 7.69M| uint32_t val) { 107| 7.69M| uint16_t key = val >> 16; 108| 7.69M| if (context->container == NULL || context->key != key) { ------------------ | Branch (108:9): [True: 0, False: 7.69M] | Branch (108:39): [True: 0, False: 7.69M] ------------------ 109| 0| uint8_t typecode; 110| 0| int idx; 111| 0| context->container = 112| 0| containerptr_roaring_bitmap_add(r, val, &typecode, &idx); 113| 0| context->typecode = typecode; 114| 0| context->idx = idx; 115| 0| context->key = key; 116| 7.69M| } else { 117| | // no need to seek the container, it is at hand 118| | // because we already have the container at hand, we can do the 119| | // insertion directly, bypassing the roaring_bitmap_add call 120| 7.69M| uint8_t new_typecode; 121| 7.69M| container_t *container2 = container_add( 122| 7.69M| context->container, val & 0xFFFF, context->typecode, &new_typecode); 123| 7.69M| if (container2 != context->container) { ------------------ | Branch (123:13): [True: 101, False: 7.69M] ------------------ 124| | // rare instance when we need to change the container type 125| 101| container_free(context->container, context->typecode); 126| 101| ra_set_container_at_index(&r->high_low_container, context->idx, 127| 101| container2, new_typecode); 128| 101| context->typecode = new_typecode; 129| 101| context->container = container2; 130| 101| } 131| 7.69M| } 132| 7.69M|} roaring.c:is_cow: 41| 104k|static inline bool is_cow(const roaring_bitmap_t *r) { 42| | return r->high_low_container.flags & ROARING_FLAG_COW; ------------------ | | 46| 104k|#define ROARING_FLAG_COW UINT8_C(0x1) ------------------ 43| 104k|} roaring.c:loadfirstvalue: 1728| 18.3k| roaring_uint32_iterator_t *newit) { 1729| 18.3k| if (iter_new_container_partial_init(newit)) { ------------------ | Branch (1729:9): [True: 12.1k, False: 6.19k] ------------------ 1730| 12.1k| uint16_t value = 0; 1731| 12.1k| newit->container_it = 1732| 12.1k| container_init_iterator(newit->container, newit->typecode, &value); 1733| 12.1k| newit->current_value = newit->highbits | value; 1734| 12.1k| } 1735| 18.3k| return newit->has_value; 1736| 18.3k|} roaring.c:iter_new_container_partial_init: 1698| 24.4k| roaring_uint32_iterator_t *newit) { 1699| 24.4k| newit->current_value = 0; 1700| 24.4k| if (newit->container_index >= newit->parent->high_low_container.size || ------------------ | Branch (1700:9): [True: 6.19k, False: 18.2k] ------------------ 1701| 18.2k| newit->container_index < 0) { ------------------ | Branch (1701:9): [True: 0, False: 18.2k] ------------------ 1702| 6.19k| newit->current_value = UINT32_MAX; 1703| 6.19k| return (newit->has_value = false); 1704| 6.19k| } 1705| 18.2k| newit->has_value = true; 1706| | // we precompute container, typecode and highbits so that successive 1707| | // iterators do not have to grab them from odd memory locations 1708| | // and have to worry about the (easily predicted) container_unwrap_shared 1709| | // call. 1710| 18.2k| newit->container = 1711| 18.2k| newit->parent->high_low_container.containers[newit->container_index]; 1712| 18.2k| newit->typecode = 1713| 18.2k| newit->parent->high_low_container.typecodes[newit->container_index]; 1714| 18.2k| newit->highbits = 1715| 18.2k| ((uint32_t) 1716| 18.2k| newit->parent->high_low_container.keys[newit->container_index]) 1717| 18.2k| << 16; 1718| 18.2k| newit->container = 1719| 18.2k| container_unwrap_shared(newit->container, &(newit->typecode)); 1720| | return true; 1721| 24.4k|} roaring.c:loadfirstvalue_largeorequal: 1759| 6.05k| roaring_uint32_iterator_t *newit, uint32_t val) { 1760| 6.05k| bool partial_init = iter_new_container_partial_init(newit); 1761| 6.05k| assert(partial_init); 1762| 6.05k| if (!partial_init) { ------------------ | Branch (1762:9): [True: 0, False: 6.05k] ------------------ 1763| 0| return false; 1764| 0| } 1765| 6.05k| uint16_t value = 0; 1766| 6.05k| newit->container_it = 1767| 6.05k| container_init_iterator(newit->container, newit->typecode, &value); 1768| 6.05k| bool found = container_iterator_lower_bound( 1769| 6.05k| newit->container, newit->typecode, &newit->container_it, &value, 1770| 6.05k| val & 0xFFFF); 1771| 6.05k| assert(found); 1772| 6.05k| if (!found) { ------------------ | Branch (1772:9): [True: 0, False: 6.05k] ------------------ 1773| 0| return false; 1774| 0| } 1775| 6.05k| newit->current_value = newit->highbits | value; 1776| | return true; 1777| 6.05k|} roaring.c:inplace_flip_container: 2133| 6.77k| uint16_t lb_start, uint16_t lb_end) { 2134| 6.77k| const int i = ra_get_index(x1_arr, hb); 2135| 6.77k| uint8_t ctype_in, ctype_out; 2136| 6.77k| container_t *flipped_container = NULL; 2137| 6.77k| if (i >= 0) { ------------------ | Branch (2137:9): [True: 6.28k, False: 497] ------------------ 2138| 6.28k| container_t *container_to_flip = 2139| 6.28k| ra_get_container_at_index(x1_arr, (uint16_t)i, &ctype_in); 2140| 6.28k| flipped_container = container_inot_range( 2141| 6.28k| container_to_flip, ctype_in, (uint32_t)lb_start, 2142| 6.28k| (uint32_t)(lb_end + 1), &ctype_out); 2143| | // if a new container was created, the old one was already freed 2144| 6.28k| if (container_nonzero_cardinality(flipped_container, ctype_out)) { ------------------ | Branch (2144:13): [True: 6.25k, False: 28] ------------------ 2145| 6.25k| ra_set_container_at_index(x1_arr, i, flipped_container, ctype_out); 2146| 6.25k| } else { 2147| 28| container_free(flipped_container, ctype_out); 2148| 28| ra_remove_at_index(x1_arr, i); 2149| 28| } 2150| | 2151| 6.28k| } else { 2152| 497| flipped_container = container_range_of_ones( 2153| 497| (uint32_t)lb_start, (uint32_t)(lb_end + 1), &ctype_out); 2154| 497| ra_insert_new_key_value_at(x1_arr, -i - 1, hb, flipped_container, 2155| 497| ctype_out); 2156| 497| } 2157| 6.77k|} roaring.c:inplace_fully_flip_container: 2184| 21.3k|static void inplace_fully_flip_container(roaring_array_t *x1_arr, uint16_t hb) { 2185| 21.3k| const int i = ra_get_index(x1_arr, hb); 2186| 21.3k| uint8_t ctype_in, ctype_out; 2187| 21.3k| container_t *flipped_container = NULL; 2188| 21.3k| if (i >= 0) { ------------------ | Branch (2188:9): [True: 5.80k, False: 15.5k] ------------------ 2189| 5.80k| container_t *container_to_flip = 2190| 5.80k| ra_get_container_at_index(x1_arr, (uint16_t)i, &ctype_in); 2191| 5.80k| flipped_container = 2192| 5.80k| container_inot(container_to_flip, ctype_in, &ctype_out); 2193| | 2194| 5.80k| if (container_nonzero_cardinality(flipped_container, ctype_out)) { ------------------ | Branch (2194:13): [True: 463, False: 5.34k] ------------------ 2195| 463| ra_set_container_at_index(x1_arr, i, flipped_container, ctype_out); 2196| 5.34k| } else { 2197| 5.34k| container_free(flipped_container, ctype_out); 2198| 5.34k| ra_remove_at_index(x1_arr, i); 2199| 5.34k| } 2200| | 2201| 15.5k| } else { 2202| 15.5k| flipped_container = container_range_of_ones(0U, 0x10000U, &ctype_out); 2203| 15.5k| ra_insert_new_key_value_at(x1_arr, -i - 1, hb, flipped_container, 2204| 15.5k| ctype_out); 2205| 15.5k| } 2206| 21.3k|} ra_init_with_capacity: 80| 61.5k|bool ra_init_with_capacity(roaring_array_t *new_ra, uint32_t cap) { 81| 61.5k| if (!new_ra) return false; ------------------ | Branch (81:9): [True: 0, False: 61.5k] ------------------ 82| 61.5k| ra_init(new_ra); 83| | 84| | // Containers hold 64Ki elements, so 64Ki containers is enough to hold 85| | // `0x10000 * 0x10000` (all 2^32) elements 86| 61.5k| if (cap > 0x10000) { ------------------ | Branch (86:9): [True: 0, False: 61.5k] ------------------ 87| 0| cap = 0x10000; 88| 0| } 89| | 90| 61.5k| if (cap > 0) { ------------------ | Branch (90:9): [True: 30.7k, False: 30.8k] ------------------ 91| 30.7k| void *bigalloc = roaring_malloc( 92| 30.7k| cap * (sizeof(uint16_t) + sizeof(container_t *) + sizeof(uint8_t))); 93| 30.7k| if (bigalloc == NULL) return false; ------------------ | Branch (93:13): [True: 0, False: 30.7k] ------------------ 94| 30.7k| new_ra->containers = (container_t **)bigalloc; 95| 30.7k| new_ra->keys = (uint16_t *)(new_ra->containers + cap); 96| 30.7k| new_ra->typecodes = (uint8_t *)(new_ra->keys + cap); 97| | // Narrowing is safe because of above check 98| 30.7k| new_ra->allocation_size = (int32_t)cap; 99| 30.7k| } 100| 61.5k| return true; 101| 61.5k|} ra_shrink_to_fit: 103| 67.7k|int ra_shrink_to_fit(roaring_array_t *ra) { 104| 67.7k| int savings = (ra->allocation_size - ra->size) * 105| 67.7k| (sizeof(uint16_t) + sizeof(container_t *) + sizeof(uint8_t)); 106| 67.7k| if (!realloc_array(ra, ra->size)) { ------------------ | Branch (106:9): [True: 0, False: 67.7k] ------------------ 107| 0| return 0; 108| 0| } 109| 67.7k| ra->allocation_size = ra->size; 110| 67.7k| return savings; 111| 67.7k|} ra_init: 113| 61.5k|void ra_init(roaring_array_t *new_ra) { 114| 61.5k| if (!new_ra) { ------------------ | Branch (114:9): [True: 0, False: 61.5k] ------------------ 115| 0| return; 116| 0| } 117| 61.5k| new_ra->keys = NULL; 118| 61.5k| new_ra->containers = NULL; 119| 61.5k| new_ra->typecodes = NULL; 120| | 121| 61.5k| new_ra->allocation_size = 0; 122| 61.5k| new_ra->size = 0; 123| 61.5k| new_ra->flags = 0; 124| 61.5k|} ra_overwrite: 127| 13.9k| bool copy_on_write) { 128| 13.9k| ra_clear_containers(dest); // we are going to overwrite them 129| 13.9k| if (source->size == 0) { // Note: can't call memcpy(NULL), even w/size ------------------ | Branch (129:9): [True: 22, False: 13.9k] ------------------ 130| 22| dest->size = 0; // <--- This is important. 131| 22| return true; // output was just cleared, so they match 132| 22| } 133| 13.9k| if (dest->allocation_size < source->size) { ------------------ | Branch (133:9): [True: 12.2k, False: 1.64k] ------------------ 134| 12.2k| if (!realloc_array(dest, source->size)) { ------------------ | Branch (134:13): [True: 0, False: 12.2k] ------------------ 135| 0| return false; 136| 0| } 137| 12.2k| } 138| 13.9k| dest->size = source->size; 139| 13.9k| memcpy(dest->keys, source->keys, dest->size * sizeof(uint16_t)); 140| | // we go through the containers, turning them into shared containers... 141| 13.9k| if (copy_on_write) { ------------------ | Branch (141:9): [True: 0, False: 13.9k] ------------------ 142| 0| for (int32_t i = 0; i < dest->size; ++i) { ------------------ | Branch (142:29): [True: 0, False: 0] ------------------ 143| 0| source->containers[i] = get_copy_of_container( 144| 0| source->containers[i], &source->typecodes[i], copy_on_write); 145| 0| } 146| | // we do a shallow copy to the other bitmap 147| 0| memcpy(dest->containers, source->containers, 148| 0| dest->size * sizeof(container_t *)); 149| 0| memcpy(dest->typecodes, source->typecodes, 150| 0| dest->size * sizeof(uint8_t)); 151| 13.9k| } else { 152| 13.9k| memcpy(dest->typecodes, source->typecodes, 153| 13.9k| dest->size * sizeof(uint8_t)); 154| 162k| for (int32_t i = 0; i < dest->size; i++) { ------------------ | Branch (154:29): [True: 148k, False: 13.9k] ------------------ 155| 148k| dest->containers[i] = 156| 148k| container_clone(source->containers[i], source->typecodes[i]); 157| 148k| if (dest->containers[i] == NULL) { ------------------ | Branch (157:17): [True: 0, False: 148k] ------------------ 158| 0| for (int32_t j = 0; j < i; j++) { ------------------ | Branch (158:37): [True: 0, False: 0] ------------------ 159| 0| container_free(dest->containers[j], dest->typecodes[j]); 160| 0| } 161| 0| ra_clear_without_containers(dest); 162| 0| return false; 163| 0| } 164| 148k| } 165| 13.9k| } 166| 13.9k| return true; 167| 13.9k|} ra_clear_containers: 169| 75.5k|void ra_clear_containers(roaring_array_t *ra) { 170| 513k| for (int32_t i = 0; i < ra->size; ++i) { ------------------ | Branch (170:25): [True: 438k, False: 75.5k] ------------------ 171| 438k| container_free(ra->containers[i], ra->typecodes[i]); 172| 438k| } 173| 75.5k|} ra_reset: 175| 61.5k|void ra_reset(roaring_array_t *ra) { 176| 61.5k| ra_clear_containers(ra); 177| 61.5k| ra->size = 0; 178| 61.5k| ra_shrink_to_fit(ra); 179| 61.5k|} ra_clear_without_containers: 181| 45|void ra_clear_without_containers(roaring_array_t *ra) { 182| 45| roaring_free( 183| 45| ra->containers); // keys and typecodes are allocated with containers 184| 45| ra->size = 0; 185| 45| ra->allocation_size = 0; 186| 45| ra->containers = NULL; 187| 45| ra->keys = NULL; 188| | ra->typecodes = NULL; 189| 45|} ra_clear: 191| 45|void ra_clear(roaring_array_t *ra) { 192| 45| ra_clear_containers(ra); 193| 45| ra_clear_without_containers(ra); 194| 45|} extend_array: 196| 83.5k|bool extend_array(roaring_array_t *ra, int32_t k) { 197| 83.5k| int32_t desired_size = ra->size + k; 198| 83.5k| const int32_t max_containers = 65536; 199| 83.5k| assert(desired_size <= max_containers); 200| 83.5k| if (desired_size > ra->allocation_size) { ------------------ | Branch (200:9): [True: 17.5k, False: 66.0k] ------------------ 201| 17.5k| int32_t new_capacity = 202| 17.5k| (ra->size < 1024) ? 2 * desired_size : 5 * desired_size / 4; ------------------ | Branch (202:13): [True: 17.5k, False: 0] ------------------ 203| 17.5k| if (new_capacity > max_containers) { ------------------ | Branch (203:13): [True: 0, False: 17.5k] ------------------ 204| 0| new_capacity = max_containers; 205| 0| } 206| | 207| 17.5k| return realloc_array(ra, new_capacity); 208| 17.5k| } 209| 66.0k| return true; 210| 83.5k|} ra_append: 213| 23.7k| uint8_t typecode) { 214| 23.7k| extend_array(ra, 1); 215| 23.7k| const int32_t pos = ra->size; 216| | 217| 23.7k| ra->keys[pos] = key; 218| 23.7k| ra->containers[pos] = c; 219| 23.7k| ra->typecodes[pos] = typecode; 220| 23.7k| ra->size++; 221| 23.7k|} ra_append_copy_range: 254| 26.8k| bool copy_on_write) { 255| 26.8k| extend_array(ra, end_index - start_index); 256| 197k| for (int32_t i = start_index; i < end_index; ++i) { ------------------ | Branch (256:35): [True: 170k, False: 26.8k] ------------------ 257| 170k| const int32_t pos = ra->size; 258| 170k| ra->keys[pos] = sa->keys[i]; 259| 170k| if (copy_on_write) { ------------------ | Branch (259:13): [True: 0, False: 170k] ------------------ 260| 0| sa->containers[i] = get_copy_of_container( 261| 0| sa->containers[i], &sa->typecodes[i], copy_on_write); 262| 0| ra->containers[pos] = sa->containers[i]; 263| 0| ra->typecodes[pos] = sa->typecodes[i]; 264| 170k| } else { 265| 170k| ra->containers[pos] = 266| 170k| container_clone(sa->containers[i], sa->typecodes[i]); 267| 170k| ra->typecodes[pos] = sa->typecodes[i]; 268| 170k| } 269| 170k| ra->size++; 270| 170k| } 271| 26.8k|} ra_advance_until_freeing: 340| 74|int32_t ra_advance_until_freeing(roaring_array_t *ra, uint16_t x, int32_t pos) { 341| 148| while (pos < ra->size && ra->keys[pos] < x) { ------------------ | Branch (341:12): [True: 148, False: 0] | Branch (341:30): [True: 74, False: 74] ------------------ 342| 74| container_free(ra->containers[pos], ra->typecodes[pos]); 343| 74| ++pos; 344| 74| } 345| 74| return pos; 346| 74|} ra_insert_new_key_value_at: 349| 31.6k| container_t *c, uint8_t typecode) { 350| 31.6k| extend_array(ra, 1); 351| | // May be an optimization opportunity with DIY memmove 352| 31.6k| memmove(&(ra->keys[i + 1]), &(ra->keys[i]), 353| 31.6k| sizeof(uint16_t) * (ra->size - i)); 354| 31.6k| memmove(&(ra->containers[i + 1]), &(ra->containers[i]), 355| 31.6k| sizeof(container_t *) * (ra->size - i)); 356| 31.6k| memmove(&(ra->typecodes[i + 1]), &(ra->typecodes[i]), 357| 31.6k| sizeof(uint8_t) * (ra->size - i)); 358| 31.6k| ra->keys[i] = key; 359| 31.6k| ra->containers[i] = c; 360| 31.6k| ra->typecodes[i] = typecode; 361| 31.6k| ra->size++; 362| 31.6k|} ra_downsize: 369| 11.8k|void ra_downsize(roaring_array_t *ra, int32_t new_length) { 370| | assert(new_length <= ra->size); 371| 11.8k| ra->size = new_length; 372| 11.8k|} ra_remove_at_index: 374| 62.5k|void ra_remove_at_index(roaring_array_t *ra, int32_t i) { 375| 62.5k| memmove(&(ra->containers[i]), &(ra->containers[i + 1]), 376| 62.5k| sizeof(container_t *) * (ra->size - i - 1)); 377| 62.5k| memmove(&(ra->keys[i]), &(ra->keys[i + 1]), 378| 62.5k| sizeof(uint16_t) * (ra->size - i - 1)); 379| 62.5k| memmove(&(ra->typecodes[i]), &(ra->typecodes[i + 1]), 380| 62.5k| sizeof(uint8_t) * (ra->size - i - 1)); 381| 62.5k| ra->size--; 382| 62.5k|} ra_remove_at_index_and_free: 384| 73|void ra_remove_at_index_and_free(roaring_array_t *ra, int32_t i) { 385| 73| container_free(ra->containers[i], ra->typecodes[i]); 386| 73| ra_remove_at_index(ra, i); 387| 73|} ra_copy_range: 395| 35| uint32_t new_begin) { 396| 35| assert(begin <= end); 397| 35| assert(new_begin < begin); 398| | 399| 35| const int range = end - begin; 400| | 401| | // We ensure to previously have freed overwritten containers 402| | // that are not copied elsewhere 403| | 404| 35| memmove(&(ra->containers[new_begin]), &(ra->containers[begin]), 405| 35| sizeof(container_t *) * range); 406| 35| memmove(&(ra->keys[new_begin]), &(ra->keys[begin]), 407| 35| sizeof(uint16_t) * range); 408| 35| memmove(&(ra->typecodes[new_begin]), &(ra->typecodes[begin]), 409| 35| sizeof(uint8_t) * range); 410| 35|} ra_shift_tail: 412| 1.94k|void ra_shift_tail(roaring_array_t *ra, int32_t count, int32_t distance) { 413| 1.94k| if (distance > 0) { ------------------ | Branch (413:9): [True: 1.24k, False: 699] ------------------ 414| 1.24k| extend_array(ra, distance); 415| 1.24k| } 416| 1.94k| int32_t srcpos = ra->size - count; 417| 1.94k| int32_t dstpos = srcpos + distance; 418| 1.94k| memmove(&(ra->keys[dstpos]), &(ra->keys[srcpos]), sizeof(uint16_t) * count); 419| 1.94k| memmove(&(ra->containers[dstpos]), &(ra->containers[srcpos]), 420| 1.94k| sizeof(container_t *) * count); 421| 1.94k| memmove(&(ra->typecodes[dstpos]), &(ra->typecodes[srcpos]), 422| 1.94k| sizeof(uint8_t) * count); 423| 1.94k| ra->size += distance; 424| 1.94k|} ra_to_uint32_array: 426| 6.15k|void ra_to_uint32_array(const roaring_array_t *ra, uint32_t *ans) { 427| 6.15k| size_t ctr = 0; 428| 69.1k| for (int32_t i = 0; i < ra->size; ++i) { ------------------ | Branch (428:25): [True: 62.9k, False: 6.15k] ------------------ 429| 62.9k| int num_added = container_to_uint32_array( 430| 62.9k| ans + ctr, ra->containers[i], ra->typecodes[i], 431| 62.9k| ((uint32_t)ra->keys[i]) << 16); 432| 62.9k| ctr += num_added; 433| 62.9k| } 434| 6.15k|} ra_has_run_container: 436| 18.4k|bool ra_has_run_container(const roaring_array_t *ra) { 437| 169k| for (int32_t k = 0; k < ra->size; ++k) { ------------------ | Branch (437:25): [True: 153k, False: 15.9k] ------------------ 438| 153k| if (get_container_type(ra->containers[k], ra->typecodes[k]) == ------------------ | Branch (438:13): [True: 2.51k, False: 150k] ------------------ 439| 153k| RUN_CONTAINER_TYPE) ------------------ | | 50| 153k|#define RUN_CONTAINER_TYPE 3 ------------------ 440| 2.51k| return true; 441| 153k| } 442| 15.9k| return false; 443| 18.4k|} ra_portable_header_size: 445| 12.3k|uint32_t ra_portable_header_size(const roaring_array_t *ra) { 446| 12.3k| if (ra_has_run_container(ra)) { ------------------ | Branch (446:9): [True: 2.51k, False: 9.78k] ------------------ 447| 2.51k| if (ra->size < ------------------ | Branch (447:13): [True: 2.51k, False: 0] ------------------ 448| 2.51k| NO_OFFSET_THRESHOLD) { // for small bitmaps, we omit the offsets 449| 2.51k| return 4 + (ra->size + 7) / 8 + 4 * ra->size; 450| 2.51k| } 451| 0| return 4 + (ra->size + 7) / 8 + 452| 0| 8 * ra->size; // - 4 because we pack the size with the cookie 453| 9.78k| } else { 454| 9.78k| return 4 + 4 + 8 * ra->size; 455| 9.78k| } 456| 12.3k|} ra_portable_size_in_bytes: 458| 12.3k|size_t ra_portable_size_in_bytes(const roaring_array_t *ra) { 459| 12.3k| size_t count = ra_portable_header_size(ra); 460| | 461| 92.0k| for (int32_t k = 0; k < ra->size; ++k) { ------------------ | Branch (461:25): [True: 79.7k, False: 12.3k] ------------------ 462| 79.7k| count += container_size_in_bytes(ra->containers[k], ra->typecodes[k]); 463| 79.7k| } 464| 12.3k| return count; 465| 12.3k|} ra_portable_serialize: 469| 6.15k|size_t ra_portable_serialize(const roaring_array_t *ra, char *buf) { 470| 6.15k| char *initbuf = buf; 471| 6.15k| uint32_t startOffset = 0; 472| 6.15k| bool hasrun = ra_has_run_container(ra); 473| 6.15k| if (hasrun) { ------------------ | Branch (473:9): [True: 0, False: 6.15k] ------------------ 474| 0| uint32_t cookie = SERIAL_COOKIE | ((uint32_t)(ra->size - 1) << 16); 475| 0| uint32_t cookie_le = croaring_htole32(cookie); ------------------ | | 511| 0|#define croaring_htole32(x) (x) ------------------ 476| 0| memcpy(buf, &cookie_le, sizeof(cookie_le)); 477| 0| buf += sizeof(cookie_le); 478| 0| uint32_t s = (ra->size + 7) / 8; 479| 0| memset(buf, 0, s); 480| 0| for (int32_t i = 0; i < ra->size; ++i) { ------------------ | Branch (480:29): [True: 0, False: 0] ------------------ 481| 0| if (get_container_type(ra->containers[i], ra->typecodes[i]) == ------------------ | Branch (481:17): [True: 0, False: 0] ------------------ 482| 0| RUN_CONTAINER_TYPE) { ------------------ | | 50| 0|#define RUN_CONTAINER_TYPE 3 ------------------ 483| 0| buf[i / 8] |= 1 << (i % 8); 484| 0| } 485| 0| } 486| 0| buf += s; 487| 0| if (ra->size < NO_OFFSET_THRESHOLD) { ------------------ | Branch (487:13): [True: 0, False: 0] ------------------ 488| 0| startOffset = 4 + 4 * ra->size + s; 489| 0| } else { 490| 0| startOffset = 4 + 8 * ra->size + s; 491| 0| } 492| 6.15k| } else { // backwards compatibility 493| 6.15k| uint32_t cookie = SERIAL_COOKIE_NO_RUNCONTAINER; 494| 6.15k| uint32_t cookie_le = croaring_htole32(cookie); ------------------ | | 511| 6.15k|#define croaring_htole32(x) (x) ------------------ 495| 6.15k| memcpy(buf, &cookie_le, sizeof(cookie_le)); 496| 6.15k| buf += sizeof(cookie_le); 497| 6.15k| uint32_t size_le = croaring_htole32((uint32_t)ra->size); ------------------ | | 511| 6.15k|#define croaring_htole32(x) (x) ------------------ 498| 6.15k| memcpy(buf, &size_le, sizeof(size_le)); 499| 6.15k| buf += sizeof(size_le); 500| | 501| 6.15k| startOffset = 4 + 4 + 4 * ra->size + 4 * ra->size; 502| 6.15k| } 503| 79.7k| for (int32_t k = 0; k < ra->size; ++k) { ------------------ | Branch (503:25): [True: 73.6k, False: 6.15k] ------------------ 504| 73.6k| uint16_t key_le = croaring_htole16(ra->keys[k]); ------------------ | | 510| 73.6k|#define croaring_htole16(x) (x) ------------------ 505| 73.6k| memcpy(buf, &key_le, sizeof(key_le)); 506| 73.6k| buf += sizeof(key_le); 507| | // get_cardinality returns a value in [1,1<<16], subtracting one 508| | // we get [0,1<<16 - 1] which fits in 16 bits 509| 73.6k| uint16_t card = (uint16_t)(container_get_cardinality(ra->containers[k], 510| 73.6k| ra->typecodes[k]) - 511| 73.6k| 1); 512| 73.6k| uint16_t card_le = croaring_htole16(card); ------------------ | | 510| 73.6k|#define croaring_htole16(x) (x) ------------------ 513| 73.6k| memcpy(buf, &card_le, sizeof(card_le)); 514| 73.6k| buf += sizeof(card_le); 515| 73.6k| } 516| 6.15k| if ((!hasrun) || (ra->size >= NO_OFFSET_THRESHOLD)) { ------------------ | Branch (516:9): [True: 6.15k, False: 0] | Branch (516:22): [True: 0, False: 0] ------------------ 517| | // writing the containers offsets 518| 79.7k| for (int32_t k = 0; k < ra->size; k++) { ------------------ | Branch (518:29): [True: 73.6k, False: 6.15k] ------------------ 519| 73.6k| uint32_t off_le = croaring_htole32(startOffset); ------------------ | | 511| 73.6k|#define croaring_htole32(x) (x) ------------------ 520| 73.6k| memcpy(buf, &off_le, sizeof(off_le)); 521| 73.6k| buf += sizeof(off_le); 522| 73.6k| startOffset = 523| 73.6k| startOffset + 524| 73.6k| container_size_in_bytes(ra->containers[k], ra->typecodes[k]); 525| 73.6k| } 526| 6.15k| } 527| 79.7k| for (int32_t k = 0; k < ra->size; ++k) { ------------------ | Branch (527:25): [True: 73.6k, False: 6.15k] ------------------ 528| 73.6k| buf += container_write(ra->containers[k], ra->typecodes[k], buf); 529| 73.6k| } 530| 6.15k| return buf - initbuf; 531| 6.15k|} ra_portable_deserialize: 634| 12.3k| const size_t maxbytes, size_t *readbytes) { 635| 12.3k| *readbytes = sizeof(int32_t); // for cookie 636| 12.3k| if (*readbytes > maxbytes) { ------------------ | Branch (636:9): [True: 5.78k, False: 6.52k] ------------------ 637| | // Ran out of bytes while reading first 4 bytes. 638| 5.78k| return false; 639| 5.78k| } 640| 6.52k| uint32_t cookie; 641| 6.52k| memcpy(&cookie, buf, sizeof(int32_t)); 642| 6.52k| cookie = croaring_letoh32(cookie); ------------------ | | 517| 6.52k|#define croaring_letoh32(x) croaring_htole32(x) | | ------------------ | | | | 511| 6.52k|#define croaring_htole32(x) (x) | | ------------------ ------------------ 643| 6.52k| buf += sizeof(uint32_t); 644| 6.52k| if ((cookie & 0xFFFF) != SERIAL_COOKIE && ------------------ | Branch (644:9): [True: 6.46k, False: 53] ------------------ 645| 6.46k| cookie != SERIAL_COOKIE_NO_RUNCONTAINER) { ------------------ | Branch (645:9): [True: 238, False: 6.23k] ------------------ 646| | // "I failed to find one of the right cookies. 647| 238| return false; 648| 238| } 649| 6.28k| int32_t size; 650| | 651| 6.28k| if ((cookie & 0xFFFF) == SERIAL_COOKIE) ------------------ | Branch (651:9): [True: 53, False: 6.23k] ------------------ 652| 53| size = (cookie >> 16) + 1; 653| 6.23k| else { 654| 6.23k| *readbytes += sizeof(int32_t); 655| 6.23k| if (*readbytes > maxbytes) { ------------------ | Branch (655:13): [True: 5, False: 6.22k] ------------------ 656| | // Ran out of bytes while reading second part of the cookie. 657| 5| return false; 658| 5| } 659| 6.22k| uint32_t size_le; 660| 6.22k| memcpy(&size_le, buf, sizeof(int32_t)); 661| 6.22k| size = (int32_t)croaring_letoh32(size_le); ------------------ | | 517| 6.22k|#define croaring_letoh32(x) croaring_htole32(x) | | ------------------ | | | | 511| 6.22k|#define croaring_htole32(x) (x) | | ------------------ ------------------ 662| 6.22k| buf += sizeof(uint32_t); 663| 6.22k| } 664| 6.27k| if (size < 0) { ------------------ | Branch (664:9): [True: 27, False: 6.25k] ------------------ 665| | // You cannot have a negative number of containers, the data must be 666| | // corrupted. 667| 27| return false; 668| 27| } 669| 6.25k| if (size > (1 << 16)) { ------------------ | Branch (669:9): [True: 20, False: 6.23k] ------------------ 670| | // You cannot have so many containers, the data must be corrupted. 671| 20| return false; 672| 20| } 673| 6.23k| const char *bitmapOfRunContainers = NULL; 674| 6.23k| bool hasrun = (cookie & 0xFFFF) == SERIAL_COOKIE; 675| 6.23k| if (hasrun) { ------------------ | Branch (675:9): [True: 53, False: 6.17k] ------------------ 676| 53| int32_t s = (size + 7) / 8; 677| 53| *readbytes += s; 678| 53| if (*readbytes > maxbytes) { // data is corrupted? ------------------ | Branch (678:13): [True: 1, False: 52] ------------------ 679| | // Ran out of bytes while reading run bitmap. 680| 1| return false; 681| 1| } 682| 52| bitmapOfRunContainers = buf; 683| 52| buf += s; 684| 52| } 685| 6.23k| const char *keyscards = buf; 686| | 687| 6.23k| *readbytes += size * 2 * sizeof(uint16_t); 688| 6.23k| if (*readbytes > maxbytes) { ------------------ | Branch (688:9): [True: 22, False: 6.20k] ------------------ 689| | // Ran out of bytes while reading key-cardinality array. 690| 22| return false; 691| 22| } 692| 6.20k| buf += size * 2 * sizeof(uint16_t); 693| | 694| 6.20k| bool is_ok = ra_init_with_capacity(answer, size); 695| 6.20k| if (!is_ok) { ------------------ | Branch (695:9): [True: 0, False: 6.20k] ------------------ 696| | // Failed to allocate memory for roaring array. Bailing out. 697| 0| return false; 698| 0| } 699| | 700| 80.0k| for (int32_t k = 0; k < size; ++k) { ------------------ | Branch (700:25): [True: 73.8k, False: 6.20k] ------------------ 701| 73.8k| uint16_t tmp; 702| 73.8k| memcpy(&tmp, keyscards + 4 * k, sizeof(tmp)); 703| 73.8k| answer->keys[k] = croaring_letoh16(tmp); ------------------ | | 516| 73.8k|#define croaring_letoh16(x) croaring_htole16(x) | | ------------------ | | | | 510| 73.8k|#define croaring_htole16(x) (x) | | ------------------ ------------------ 704| 73.8k| } 705| 6.20k| if ((!hasrun) || (size >= NO_OFFSET_THRESHOLD)) { ------------------ | Branch (705:9): [True: 6.15k, False: 51] | Branch (705:22): [True: 22, False: 29] ------------------ 706| 6.18k| *readbytes += size * 4; 707| 6.18k| if (*readbytes > maxbytes) { // data is corrupted? ------------------ | Branch (707:13): [True: 6, False: 6.17k] ------------------ 708| | // Ran out of bytes while reading offsets. 709| 6| ra_clear(answer); // we need to clear the containers already 710| | // allocated, and the roaring array 711| 6| return false; 712| 6| } 713| | 714| | // skipping the offsets 715| 6.17k| buf += size * 4; 716| 6.17k| } 717| | // Reading the containers 718| 79.9k| for (int32_t k = 0; k < size; ++k) { ------------------ | Branch (718:25): [True: 73.7k, False: 6.16k] ------------------ 719| 73.7k| uint16_t tmp; 720| 73.7k| memcpy(&tmp, keyscards + 4 * k + 2, sizeof(tmp)); 721| 73.7k| tmp = croaring_letoh16(tmp); ------------------ | | 516| 73.7k|#define croaring_letoh16(x) croaring_htole16(x) | | ------------------ | | | | 510| 73.7k|#define croaring_htole16(x) (x) | | ------------------ ------------------ 722| 73.7k| uint32_t thiscard = tmp + 1; 723| 73.7k| bool isbitmap = (thiscard > DEFAULT_MAX_SIZE); 724| 73.7k| bool isrun = false; 725| 73.7k| if (hasrun) { ------------------ | Branch (725:13): [True: 137, False: 73.6k] ------------------ 726| 137| if ((bitmapOfRunContainers[k / 8] & (1 << (k % 8))) != 0) { ------------------ | Branch (726:17): [True: 75, False: 62] ------------------ 727| 75| isbitmap = false; 728| 75| isrun = true; 729| 75| } 730| 137| } 731| 73.7k| if (isbitmap) { ------------------ | Branch (731:13): [True: 66.7k, False: 7.03k] ------------------ 732| | // we check that the read is allowed 733| 66.7k| size_t containersize = 734| 66.7k| BITSET_CONTAINER_SIZE_IN_WORDS * sizeof(uint64_t); 735| 66.7k| *readbytes += containersize; 736| 66.7k| if (*readbytes > maxbytes) { ------------------ | Branch (736:17): [True: 10, False: 66.7k] ------------------ 737| | // Running out of bytes while reading a bitset container. 738| 10| ra_clear(answer); // we need to clear the containers already 739| | // allocated, and the roaring array 740| 10| return false; 741| 10| } 742| | // it is now safe to read 743| 66.7k| bitset_container_t *c = bitset_container_create(); 744| 66.7k| if (c == NULL) { // memory allocation failure ------------------ | Branch (744:17): [True: 0, False: 66.7k] ------------------ 745| | // Failed to allocate memory for a bitset container. 746| 0| ra_clear(answer); // we need to clear the containers already 747| | // allocated, and the roaring array 748| 0| return false; 749| 0| } 750| 66.7k| answer->size++; 751| 66.7k| buf += bitset_container_read(thiscard, c, buf); 752| 66.7k| answer->containers[k] = c; 753| 66.7k| answer->typecodes[k] = BITSET_CONTAINER_TYPE; ------------------ | | 48| 66.7k|#define BITSET_CONTAINER_TYPE 1 ------------------ 754| 66.7k| } else if (isrun) { ------------------ | Branch (754:20): [True: 75, False: 6.96k] ------------------ 755| | // we check that the read is allowed 756| 75| *readbytes += sizeof(uint16_t); 757| 75| if (*readbytes > maxbytes) { ------------------ | Branch (757:17): [True: 7, False: 68] ------------------ 758| | // Running out of bytes while reading a run container (header). 759| 7| ra_clear(answer); // we need to clear the containers already 760| | // allocated, and the roaring array 761| 7| return false; 762| 7| } 763| 68| uint16_t n_runs; 764| 68| memcpy(&n_runs, buf, sizeof(uint16_t)); 765| 68| n_runs = croaring_letoh16(n_runs); ------------------ | | 516| 68|#define croaring_letoh16(x) croaring_htole16(x) | | ------------------ | | | | 510| 68|#define croaring_htole16(x) (x) | | ------------------ ------------------ 766| 68| size_t containersize = n_runs * sizeof(rle16_t); 767| 68| *readbytes += containersize; 768| 68| if (*readbytes > maxbytes) { // data is corrupted? ------------------ | Branch (768:17): [True: 9, False: 59] ------------------ 769| | // Running out of bytes while reading a run container. 770| 9| ra_clear(answer); // we need to clear the containers already 771| | // allocated, and the roaring array 772| 9| return false; 773| 9| } 774| | // it is now safe to read 775| | 776| 59| run_container_t *c = run_container_create(); 777| 59| if (c == NULL) { // memory allocation failure ------------------ | Branch (777:17): [True: 0, False: 59] ------------------ 778| | // Failed to allocate memory for a run container. 779| 0| ra_clear(answer); // we need to clear the containers already 780| | // allocated, and the roaring array 781| 0| return false; 782| 0| } 783| 59| answer->size++; 784| 59| buf += run_container_read(thiscard, c, buf); 785| 59| answer->containers[k] = c; 786| 59| answer->typecodes[k] = RUN_CONTAINER_TYPE; ------------------ | | 50| 59|#define RUN_CONTAINER_TYPE 3 ------------------ 787| 6.96k| } else { 788| | // we check that the read is allowed 789| 6.96k| size_t containersize = thiscard * sizeof(uint16_t); 790| 6.96k| *readbytes += containersize; 791| 6.96k| if (*readbytes > maxbytes) { // data is corrupted? ------------------ | Branch (791:17): [True: 13, False: 6.95k] ------------------ 792| | // Running out of bytes while reading an array container. 793| 13| ra_clear(answer); // we need to clear the containers already 794| | // allocated, and the roaring array 795| 13| return false; 796| 13| } 797| | // it is now safe to read 798| 6.95k| array_container_t *c = 799| 6.95k| array_container_create_given_capacity(thiscard); 800| 6.95k| if (c == NULL) { // memory allocation failure ------------------ | Branch (800:17): [True: 0, False: 6.95k] ------------------ 801| | // Failed to allocate memory for an array container. 802| 0| ra_clear(answer); // we need to clear the containers already 803| | // allocated, and the roaring array 804| 0| return false; 805| 0| } 806| 6.95k| answer->size++; 807| 6.95k| buf += array_container_read(thiscard, c, buf); 808| 6.95k| answer->containers[k] = c; 809| 6.95k| answer->typecodes[k] = ARRAY_CONTAINER_TYPE; ------------------ | | 49| 6.95k|#define ARRAY_CONTAINER_TYPE 2 ------------------ 810| 6.95k| } 811| 73.7k| } 812| 6.16k| return true; 813| 6.20k|} roaring_array.c:realloc_array: 42| 97.4k|static bool realloc_array(roaring_array_t *ra, int32_t new_capacity) { 43| | // 44| | // Note: not implemented using C's realloc(), because the memory layout is 45| | // Struct-of-Arrays vs. Array-of-Structs: 46| | // https://github.com/RoaringBitmap/CRoaring/issues/256 47| | 48| 97.4k| if (new_capacity == 0) { ------------------ | Branch (48:9): [True: 61.5k, False: 35.9k] ------------------ 49| 61.5k| roaring_free(ra->containers); 50| 61.5k| ra->containers = NULL; 51| 61.5k| ra->keys = NULL; 52| 61.5k| ra->typecodes = NULL; 53| 61.5k| ra->allocation_size = 0; 54| 61.5k| return true; 55| 61.5k| } 56| 35.9k| const size_t memoryneeded = 57| 35.9k| new_capacity * 58| 35.9k| (sizeof(uint16_t) + sizeof(container_t *) + sizeof(uint8_t)); 59| 35.9k| void *bigalloc = roaring_malloc(memoryneeded); 60| 35.9k| if (!bigalloc) return false; ------------------ | Branch (60:9): [True: 0, False: 35.9k] ------------------ 61| 35.9k| void *oldbigalloc = ra->containers; 62| 35.9k| container_t **newcontainers = (container_t **)bigalloc; 63| 35.9k| uint16_t *newkeys = (uint16_t *)(newcontainers + new_capacity); 64| 35.9k| uint8_t *newtypecodes = (uint8_t *)(newkeys + new_capacity); 65| 35.9k| assert((char *)(newtypecodes + new_capacity) == 66| 35.9k| (char *)bigalloc + memoryneeded); 67| 35.9k| if (ra->size > 0) { ------------------ | Branch (67:9): [True: 11.3k, False: 24.5k] ------------------ 68| 11.3k| memcpy(newcontainers, ra->containers, sizeof(container_t *) * ra->size); 69| 11.3k| memcpy(newkeys, ra->keys, sizeof(uint16_t) * ra->size); 70| 11.3k| memcpy(newtypecodes, ra->typecodes, sizeof(uint8_t) * ra->size); 71| 11.3k| } 72| 35.9k| ra->containers = newcontainers; 73| 35.9k| ra->keys = newkeys; 74| 35.9k| ra->typecodes = newtypecodes; 75| 35.9k| ra->allocation_size = new_capacity; 76| 35.9k| roaring_free(oldbigalloc); 77| | return true; 78| 35.9k|} _Z17ConsumeVecInRangeR18FuzzedDataProvidermjj: 24| 12.3k| uint32_t max_value) { 25| 12.3k| std::vector result = {0}; 26| 12.3k| result.resize(length); 27| 12.3k| std::generate(result.begin(), result.end(), [&]() { 28| 12.3k| return fdp.ConsumeIntegralInRange(min_value, max_value); 29| 12.3k| }); 30| 12.3k| return result; 31| 12.3k|} LLVMFuzzerTestOneInput: 33| 6.15k|extern "C" int LLVMFuzzerTestOneInput(const uint8_t *data, size_t size) { 34| | /** 35| | * A bitmap may contain up to 2**32 elements. Later this function will 36| | * output the content to an array where each element uses 32 bits of 37| | * storage. That would use 16 GB. Thus this function is bound to run out of 38| | * memory. 39| | * 40| | * Even without the full serialization to a 32-bit array, a bitmap may still 41| | * use over 512 MB in the normal course of operation: that is to be expected 42| | * since it can represent all sets of integers in [0,2**32]. This function 43| | * may hold several bitmaps in memory at once, so it can require gigabytes 44| | * of memory (without bugs). Hence, unless it has a generous memory 45| | * capacity, this function will run out of memory almost certainly. 46| | * 47| | * For sanity, we may limit the range to, say, 10,000,000 which will use 38 48| | * MB or so. With such a limited range, if we run out of memory, then we can 49| | * almost certain that it has to do with a genuine bug. 50| | */ 51| | 52| 6.15k| uint32_t range_start = 0; 53| 6.15k| uint32_t range_end = 10'000'000; 54| | 55| | /** 56| | * We are not solely dependent on the range [range_start, range_end) because 57| | * ConsumeVecInRange below produce integers in a small range starting at 0. 58| | */ 59| | 60| 6.15k| FuzzedDataProvider fdp(data, size); 61| | /** 62| | * The next line was ConsumeVecInRange(fdp, 500, 0, 1000) but it would pick 63| | * 500 values at random from 0, 1000, making almost certain that all of the 64| | * values are picked. It seems more useful to pick 500 values in the range 65| | * 0,1000. 66| | */ 67| 6.15k| std::vector bitmap_data_a = ConsumeVecInRange(fdp, 500, 0, 1000); 68| 6.15k| roaring::Roaring a(bitmap_data_a.size(), bitmap_data_a.data()); 69| 6.15k| a.runOptimize(); 70| 6.15k| a.shrinkToFit(); 71| | 72| 6.15k| std::vector bitmap_data_b = ConsumeVecInRange(fdp, 500, 0, 1000); 73| 6.15k| roaring::Roaring b(bitmap_data_b.size(), bitmap_data_b.data()); 74| 6.15k| b.runOptimize(); 75| 6.15k| b.add(fdp.ConsumeIntegralInRange(range_start, range_end)); 76| 6.15k| b.addChecked(fdp.ConsumeIntegralInRange(range_start, range_end)); 77| 6.15k| b.addRange(fdp.ConsumeIntegralInRange(range_start, range_end), 78| 6.15k| fdp.ConsumeIntegralInRange(range_start, range_end)); 79| | // add half of a to b. 80| 6.15k| b.addMany(bitmap_data_a.size() / 2, bitmap_data_a.data()); 81| 6.15k| b.remove(fdp.ConsumeIntegralInRange(range_start, range_end)); 82| 6.15k| b.removeChecked( 83| 6.15k| fdp.ConsumeIntegralInRange(range_start, range_end)); 84| 6.15k| b.removeRange(fdp.ConsumeIntegralInRange(range_start, range_end), 85| 6.15k| fdp.ConsumeIntegralInRange(range_start, range_end)); 86| 6.15k| b.removeRangeClosed( 87| 6.15k| fdp.ConsumeIntegralInRange(range_start, range_end), 88| 6.15k| fdp.ConsumeIntegralInRange(range_start, range_end)); 89| 6.15k| b.maximum(); 90| 6.15k| b.minimum(); 91| 6.15k| b.contains(fdp.ConsumeIntegralInRange(range_start, range_end)); 92| 6.15k| b.containsRange( 93| 6.15k| fdp.ConsumeIntegralInRange(range_start, range_end), 94| 6.15k| fdp.ConsumeIntegralInRange(range_start, range_end)); 95| | 96| 6.15k| uint32_t element = 0; 97| 6.15k| a.select(fdp.ConsumeIntegralInRange(0, 1000), &element); 98| 6.15k| a.intersect(b); 99| 6.15k| a.jaccard_index(b); 100| 6.15k| a.or_cardinality(b); 101| 6.15k| a.andnot_cardinality(b); 102| 6.15k| a.xor_cardinality(b); 103| 6.15k| a.rank(fdp.ConsumeIntegralInRange(0, 5000)); 104| 6.15k| a.getSizeInBytes(); 105| | 106| 6.15k| roaring::Roaring c = a & b; 107| 6.15k| roaring::Roaring d = a - b; 108| 6.15k| roaring::Roaring e = a | b; 109| 6.15k| roaring::Roaring f = a ^ b; 110| 6.15k| a |= e; 111| 6.15k| a &= b; 112| 6.15k| a -= c; 113| 6.15k| a ^= f; 114| | 115| 6.15k| volatile bool is_equal = (a == b); 116| | 117| 6.15k| std::vector b_as_array = {0}; 118| 6.15k| b_as_array.resize(b.cardinality()); 119| 6.15k| b.isEmpty(); 120| 6.15k| b.toUint32Array(b_as_array.data()); 121| | 122| 6.15k| a.isSubset(b); 123| 6.15k| a.isStrictSubset(b); 124| 6.15k| b.flip(fdp.ConsumeIntegralInRange(range_start, range_end), 125| 6.15k| fdp.ConsumeIntegralInRange(range_start, range_end)); 126| 6.15k| b.flipClosed(fdp.ConsumeIntegralInRange(range_start, range_end), 127| 6.15k| fdp.ConsumeIntegralInRange(range_start, range_end)); 128| 6.15k| b.removeRunCompression(); 129| | 130| | // Move/copy constructors 131| 6.15k| roaring::Roaring copied = b; 132| 6.15k| roaring::Roaring moved = std::move(b); 133| | 134| | // Asignment operators 135| 6.15k| b = copied; 136| 6.15k| b = std::move(moved); 137| | 138| | // Safe read from serialized 139| 6.15k| std::vector read_buffer = fdp.ConsumeBytes(100); 140| 6.15k| try { 141| 6.15k| roaring::Roaring read_safely = 142| 6.15k| roaring::Roaring::readSafe(read_buffer.data(), read_buffer.size()); 143| | // The above is guaranteed to be safe. However, read_safely is maybe 144| | // in an improper state and it cannot be used safely (including for 145| | // reserialization). 146| 6.15k| } catch (...) { 147| 6.14k| } 148| | 149| | // The bitmap b can be serialized and re-read. 150| 6.15k| std::size_t expected_size_in_bytes = b.getSizeInBytes(); 151| 6.15k| std::vector buffer(expected_size_in_bytes); 152| 6.15k| std::size_t size_in_bytes = b.write(buffer.data()); 153| 6.15k| assert(expected_size_in_bytes == size_in_bytes); ------------------ | Branch (153:5): [True: 6.15k, False: 0] ------------------ 154| 6.15k| roaring::Roaring bread = 155| 6.15k| roaring::Roaring::readSafe(buffer.data(), size_in_bytes); 156| 6.15k| assert(bread == b); ------------------ | Branch (156:5): [True: 6.15k, False: 0] ------------------ 157| | 158| 6.15k| f.toString(); 159| | 160| 6.15k| volatile int unused = 0; 161| | 162| 326k| for (roaring::Roaring::const_iterator i = a.begin(); i != a.end(); i++) { ------------------ | Branch (162:58): [True: 320k, False: 6.15k] ------------------ 163| 320k| unused++; 164| 320k| } 165| | 166| 6.15k| roaring::Roaring::const_iterator b_iter = b.begin(); 167| 6.15k| b_iter.equalorlarger( 168| 6.15k| fdp.ConsumeIntegralInRange(range_start, range_end)); 169| | 170| 6.15k| return 0; 171| 6.15k|} croaring_fuzzer_cc.cc:_ZZ17ConsumeVecInRangeR18FuzzedDataProvidermjjENK3$_0clEv: 27| 6.15M| std::generate(result.begin(), result.end(), [&]() { 28| 6.15M| return fdp.ConsumeIntegralInRange(min_value, max_value); 29| 6.15M| });