/src/skia/third_party/externals/icu/source/common/characterproperties.cpp
Line | Count | Source (jump to first uncovered line) |
1 | | // © 2018 and later: Unicode, Inc. and others. |
2 | | // License & terms of use: http://www.unicode.org/copyright.html |
3 | | |
4 | | // characterproperties.cpp |
5 | | // created: 2018sep03 Markus W. Scherer |
6 | | |
7 | | #include "unicode/utypes.h" |
8 | | #include "unicode/localpointer.h" |
9 | | #include "unicode/uchar.h" |
10 | | #include "unicode/ucpmap.h" |
11 | | #include "unicode/ucptrie.h" |
12 | | #include "unicode/umutablecptrie.h" |
13 | | #include "unicode/uniset.h" |
14 | | #include "unicode/uscript.h" |
15 | | #include "unicode/uset.h" |
16 | | #include "cmemory.h" |
17 | | #include "mutex.h" |
18 | | #include "normalizer2impl.h" |
19 | | #include "uassert.h" |
20 | | #include "ubidi_props.h" |
21 | | #include "ucase.h" |
22 | | #include "ucln_cmn.h" |
23 | | #include "umutex.h" |
24 | | #include "uprops.h" |
25 | | |
26 | | using icu::LocalPointer; |
27 | | #if !UCONFIG_NO_NORMALIZATION |
28 | | using icu::Normalizer2Factory; |
29 | | using icu::Normalizer2Impl; |
30 | | #endif |
31 | | using icu::UInitOnce; |
32 | | using icu::UnicodeSet; |
33 | | |
34 | | namespace { |
35 | | |
36 | | UBool U_CALLCONV characterproperties_cleanup(); |
37 | | |
38 | | constexpr int32_t NUM_INCLUSIONS = UPROPS_SRC_COUNT + UCHAR_INT_LIMIT - UCHAR_INT_START; |
39 | | |
40 | | struct Inclusion { |
41 | | UnicodeSet *fSet = nullptr; |
42 | | UInitOnce fInitOnce = U_INITONCE_INITIALIZER; |
43 | | }; |
44 | | Inclusion gInclusions[NUM_INCLUSIONS]; // cached getInclusions() |
45 | | |
46 | | UnicodeSet *sets[UCHAR_BINARY_LIMIT] = {}; |
47 | | |
48 | | UCPMap *maps[UCHAR_INT_LIMIT - UCHAR_INT_START] = {}; |
49 | | |
50 | | icu::UMutex cpMutex; |
51 | | |
52 | | //---------------------------------------------------------------- |
53 | | // Inclusions list |
54 | | //---------------------------------------------------------------- |
55 | | |
56 | | // USetAdder implementation |
57 | | // Does not use uset.h to reduce code dependencies |
58 | | void U_CALLCONV |
59 | 24.9k | _set_add(USet *set, UChar32 c) { |
60 | 24.9k | ((UnicodeSet *)set)->add(c); |
61 | 24.9k | } |
62 | | |
63 | | void U_CALLCONV |
64 | 0 | _set_addRange(USet *set, UChar32 start, UChar32 end) { |
65 | 0 | ((UnicodeSet *)set)->add(start, end); |
66 | 0 | } |
67 | | |
68 | | void U_CALLCONV |
69 | 0 | _set_addString(USet *set, const UChar *str, int32_t length) { |
70 | 0 | ((UnicodeSet *)set)->add(icu::UnicodeString((UBool)(length<0), str, length)); |
71 | 0 | } |
72 | | |
73 | 0 | UBool U_CALLCONV characterproperties_cleanup() { |
74 | 0 | for (Inclusion &in: gInclusions) { |
75 | 0 | delete in.fSet; |
76 | 0 | in.fSet = nullptr; |
77 | 0 | in.fInitOnce.reset(); |
78 | 0 | } |
79 | 0 | for (int32_t i = 0; i < UPRV_LENGTHOF(sets); ++i) { |
80 | 0 | delete sets[i]; |
81 | 0 | sets[i] = nullptr; |
82 | 0 | } |
83 | 0 | for (int32_t i = 0; i < UPRV_LENGTHOF(maps); ++i) { |
84 | 0 | ucptrie_close(reinterpret_cast<UCPTrie *>(maps[i])); |
85 | 0 | maps[i] = nullptr; |
86 | 0 | } |
87 | 0 | return TRUE; |
88 | 0 | } |
89 | | |
90 | 4 | void U_CALLCONV initInclusion(UPropertySource src, UErrorCode &errorCode) { |
91 | | // This function is invoked only via umtx_initOnce(). |
92 | 4 | U_ASSERT(0 <= src && src < UPROPS_SRC_COUNT); |
93 | 4 | if (src == UPROPS_SRC_NONE) { |
94 | 0 | errorCode = U_INTERNAL_PROGRAM_ERROR; |
95 | 0 | return; |
96 | 0 | } |
97 | 4 | U_ASSERT(gInclusions[src].fSet == nullptr); |
98 | | |
99 | 4 | LocalPointer<UnicodeSet> incl(new UnicodeSet()); |
100 | 4 | if (incl.isNull()) { |
101 | 0 | errorCode = U_MEMORY_ALLOCATION_ERROR; |
102 | 0 | return; |
103 | 0 | } |
104 | 4 | USetAdder sa = { |
105 | 4 | (USet *)incl.getAlias(), |
106 | 4 | _set_add, |
107 | 4 | _set_addRange, |
108 | 4 | _set_addString, |
109 | 4 | nullptr, // don't need remove() |
110 | 4 | nullptr // don't need removeRange() |
111 | 4 | }; |
112 | | |
113 | 4 | switch(src) { |
114 | 2 | case UPROPS_SRC_CHAR: |
115 | 2 | uchar_addPropertyStarts(&sa, &errorCode); |
116 | 2 | break; |
117 | 2 | case UPROPS_SRC_PROPSVEC: |
118 | 2 | upropsvec_addPropertyStarts(&sa, &errorCode); |
119 | 2 | break; |
120 | 0 | case UPROPS_SRC_CHAR_AND_PROPSVEC: |
121 | 0 | uchar_addPropertyStarts(&sa, &errorCode); |
122 | 0 | upropsvec_addPropertyStarts(&sa, &errorCode); |
123 | 0 | break; |
124 | 0 | #if !UCONFIG_NO_NORMALIZATION |
125 | 0 | case UPROPS_SRC_CASE_AND_NORM: { |
126 | 0 | const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode); |
127 | 0 | if(U_SUCCESS(errorCode)) { |
128 | 0 | impl->addPropertyStarts(&sa, errorCode); |
129 | 0 | } |
130 | 0 | ucase_addPropertyStarts(&sa, &errorCode); |
131 | 0 | break; |
132 | 0 | } |
133 | 0 | case UPROPS_SRC_NFC: { |
134 | 0 | const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode); |
135 | 0 | if(U_SUCCESS(errorCode)) { |
136 | 0 | impl->addPropertyStarts(&sa, errorCode); |
137 | 0 | } |
138 | 0 | break; |
139 | 0 | } |
140 | 0 | case UPROPS_SRC_NFKC: { |
141 | 0 | const Normalizer2Impl *impl=Normalizer2Factory::getNFKCImpl(errorCode); |
142 | 0 | if(U_SUCCESS(errorCode)) { |
143 | 0 | impl->addPropertyStarts(&sa, errorCode); |
144 | 0 | } |
145 | 0 | break; |
146 | 0 | } |
147 | 0 | case UPROPS_SRC_NFKC_CF: { |
148 | 0 | const Normalizer2Impl *impl=Normalizer2Factory::getNFKC_CFImpl(errorCode); |
149 | 0 | if(U_SUCCESS(errorCode)) { |
150 | 0 | impl->addPropertyStarts(&sa, errorCode); |
151 | 0 | } |
152 | 0 | break; |
153 | 0 | } |
154 | 0 | case UPROPS_SRC_NFC_CANON_ITER: { |
155 | 0 | const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode); |
156 | 0 | if(U_SUCCESS(errorCode)) { |
157 | 0 | impl->addCanonIterPropertyStarts(&sa, errorCode); |
158 | 0 | } |
159 | 0 | break; |
160 | 0 | } |
161 | 0 | #endif |
162 | 0 | case UPROPS_SRC_CASE: |
163 | 0 | ucase_addPropertyStarts(&sa, &errorCode); |
164 | 0 | break; |
165 | 0 | case UPROPS_SRC_BIDI: |
166 | 0 | ubidi_addPropertyStarts(&sa, &errorCode); |
167 | 0 | break; |
168 | 0 | case UPROPS_SRC_INPC: |
169 | 0 | case UPROPS_SRC_INSC: |
170 | 0 | case UPROPS_SRC_VO: |
171 | 0 | uprops_addPropertyStarts((UPropertySource)src, &sa, &errorCode); |
172 | 0 | break; |
173 | 0 | default: |
174 | 0 | errorCode = U_INTERNAL_PROGRAM_ERROR; |
175 | 0 | break; |
176 | 4 | } |
177 | | |
178 | 4 | if (U_FAILURE(errorCode)) { |
179 | 0 | return; |
180 | 0 | } |
181 | 4 | if (incl->isBogus()) { |
182 | 0 | errorCode = U_MEMORY_ALLOCATION_ERROR; |
183 | 0 | return; |
184 | 0 | } |
185 | | // Compact for caching. |
186 | 4 | incl->compact(); |
187 | 4 | gInclusions[src].fSet = incl.orphan(); |
188 | 4 | ucln_common_registerCleanup(UCLN_COMMON_CHARACTERPROPERTIES, characterproperties_cleanup); |
189 | 4 | } |
190 | | |
191 | 12 | const UnicodeSet *getInclusionsForSource(UPropertySource src, UErrorCode &errorCode) { |
192 | 12 | if (U_FAILURE(errorCode)) { return nullptr; } |
193 | 12 | if (src < 0 || UPROPS_SRC_COUNT <= src) { |
194 | 0 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
195 | 0 | return nullptr; |
196 | 0 | } |
197 | 12 | Inclusion &i = gInclusions[src]; |
198 | 12 | umtx_initOnce(i.fInitOnce, &initInclusion, src, errorCode); |
199 | 12 | return i.fSet; |
200 | 12 | } |
201 | | |
202 | 4 | void U_CALLCONV initIntPropInclusion(UProperty prop, UErrorCode &errorCode) { |
203 | | // This function is invoked only via umtx_initOnce(). |
204 | 4 | U_ASSERT(UCHAR_INT_START <= prop && prop < UCHAR_INT_LIMIT); |
205 | 4 | int32_t inclIndex = UPROPS_SRC_COUNT + prop - UCHAR_INT_START; |
206 | 4 | U_ASSERT(gInclusions[inclIndex].fSet == nullptr); |
207 | 4 | UPropertySource src = uprops_getSource(prop); |
208 | 4 | const UnicodeSet *incl = getInclusionsForSource(src, errorCode); |
209 | 4 | if (U_FAILURE(errorCode)) { |
210 | 0 | return; |
211 | 0 | } |
212 | | |
213 | 4 | LocalPointer<UnicodeSet> intPropIncl(new UnicodeSet(0, 0)); |
214 | 4 | if (intPropIncl.isNull()) { |
215 | 0 | errorCode = U_MEMORY_ALLOCATION_ERROR; |
216 | 0 | return; |
217 | 0 | } |
218 | 4 | int32_t numRanges = incl->getRangeCount(); |
219 | 4 | int32_t prevValue = 0; |
220 | 11.7k | for (int32_t i = 0; i < numRanges; ++i) { |
221 | 11.7k | UChar32 rangeEnd = incl->getRangeEnd(i); |
222 | 39.0k | for (UChar32 c = incl->getRangeStart(i); c <= rangeEnd; ++c) { |
223 | | // TODO: Get a UCharacterProperty.IntProperty to avoid the property dispatch. |
224 | 27.2k | int32_t value = u_getIntPropertyValue(c, prop); |
225 | 27.2k | if (value != prevValue) { |
226 | 9.90k | intPropIncl->add(c); |
227 | 9.90k | prevValue = value; |
228 | 9.90k | } |
229 | 27.2k | } |
230 | 11.7k | } |
231 | | |
232 | 4 | if (intPropIncl->isBogus()) { |
233 | 0 | errorCode = U_MEMORY_ALLOCATION_ERROR; |
234 | 0 | return; |
235 | 0 | } |
236 | | // Compact for caching. |
237 | 4 | intPropIncl->compact(); |
238 | 4 | gInclusions[inclIndex].fSet = intPropIncl.orphan(); |
239 | 4 | ucln_common_registerCleanup(UCLN_COMMON_CHARACTERPROPERTIES, characterproperties_cleanup); |
240 | 4 | } |
241 | | |
242 | | } // namespace |
243 | | |
244 | | U_NAMESPACE_BEGIN |
245 | | |
246 | | const UnicodeSet *CharacterProperties::getInclusionsForProperty( |
247 | 7.97k | UProperty prop, UErrorCode &errorCode) { |
248 | 7.97k | if (U_FAILURE(errorCode)) { return nullptr; } |
249 | 7.97k | if (UCHAR_INT_START <= prop && prop < UCHAR_INT_LIMIT) { |
250 | 7.96k | int32_t inclIndex = UPROPS_SRC_COUNT + prop - UCHAR_INT_START; |
251 | 7.96k | Inclusion &i = gInclusions[inclIndex]; |
252 | 7.96k | umtx_initOnce(i.fInitOnce, &initIntPropInclusion, prop, errorCode); |
253 | 7.96k | return i.fSet; |
254 | 8 | } else { |
255 | 8 | UPropertySource src = uprops_getSource(prop); |
256 | 8 | return getInclusionsForSource(src, errorCode); |
257 | 8 | } |
258 | 7.97k | } |
259 | | |
260 | | U_NAMESPACE_END |
261 | | |
262 | | namespace { |
263 | | |
264 | 0 | UnicodeSet *makeSet(UProperty property, UErrorCode &errorCode) { |
265 | 0 | if (U_FAILURE(errorCode)) { return nullptr; } |
266 | 0 | LocalPointer<UnicodeSet> set(new UnicodeSet()); |
267 | 0 | if (set.isNull()) { |
268 | 0 | errorCode = U_MEMORY_ALLOCATION_ERROR; |
269 | 0 | return nullptr; |
270 | 0 | } |
271 | 0 | const UnicodeSet *inclusions = |
272 | 0 | icu::CharacterProperties::getInclusionsForProperty(property, errorCode); |
273 | 0 | if (U_FAILURE(errorCode)) { return nullptr; } |
274 | 0 | int32_t numRanges = inclusions->getRangeCount(); |
275 | 0 | UChar32 startHasProperty = -1; |
276 | |
|
277 | 0 | for (int32_t i = 0; i < numRanges; ++i) { |
278 | 0 | UChar32 rangeEnd = inclusions->getRangeEnd(i); |
279 | 0 | for (UChar32 c = inclusions->getRangeStart(i); c <= rangeEnd; ++c) { |
280 | | // TODO: Get a UCharacterProperty.BinaryProperty to avoid the property dispatch. |
281 | 0 | if (u_hasBinaryProperty(c, property)) { |
282 | 0 | if (startHasProperty < 0) { |
283 | | // Transition from false to true. |
284 | 0 | startHasProperty = c; |
285 | 0 | } |
286 | 0 | } else if (startHasProperty >= 0) { |
287 | | // Transition from true to false. |
288 | 0 | set->add(startHasProperty, c - 1); |
289 | 0 | startHasProperty = -1; |
290 | 0 | } |
291 | 0 | } |
292 | 0 | } |
293 | 0 | if (startHasProperty >= 0) { |
294 | 0 | set->add(startHasProperty, 0x10FFFF); |
295 | 0 | } |
296 | 0 | set->freeze(); |
297 | 0 | return set.orphan(); |
298 | 0 | } |
299 | | |
300 | 0 | UCPMap *makeMap(UProperty property, UErrorCode &errorCode) { |
301 | 0 | if (U_FAILURE(errorCode)) { return nullptr; } |
302 | 0 | uint32_t nullValue = property == UCHAR_SCRIPT ? USCRIPT_UNKNOWN : 0; |
303 | 0 | icu::LocalUMutableCPTriePointer mutableTrie( |
304 | 0 | umutablecptrie_open(nullValue, nullValue, &errorCode)); |
305 | 0 | const UnicodeSet *inclusions = |
306 | 0 | icu::CharacterProperties::getInclusionsForProperty(property, errorCode); |
307 | 0 | if (U_FAILURE(errorCode)) { return nullptr; } |
308 | 0 | int32_t numRanges = inclusions->getRangeCount(); |
309 | 0 | UChar32 start = 0; |
310 | 0 | uint32_t value = nullValue; |
311 | |
|
312 | 0 | for (int32_t i = 0; i < numRanges; ++i) { |
313 | 0 | UChar32 rangeEnd = inclusions->getRangeEnd(i); |
314 | 0 | for (UChar32 c = inclusions->getRangeStart(i); c <= rangeEnd; ++c) { |
315 | | // TODO: Get a UCharacterProperty.IntProperty to avoid the property dispatch. |
316 | 0 | uint32_t nextValue = u_getIntPropertyValue(c, property); |
317 | 0 | if (value != nextValue) { |
318 | 0 | if (value != nullValue) { |
319 | 0 | umutablecptrie_setRange(mutableTrie.getAlias(), start, c - 1, value, &errorCode); |
320 | 0 | } |
321 | 0 | start = c; |
322 | 0 | value = nextValue; |
323 | 0 | } |
324 | 0 | } |
325 | 0 | } |
326 | 0 | if (value != 0) { |
327 | 0 | umutablecptrie_setRange(mutableTrie.getAlias(), start, 0x10FFFF, value, &errorCode); |
328 | 0 | } |
329 | |
|
330 | 0 | UCPTrieType type; |
331 | 0 | if (property == UCHAR_BIDI_CLASS || property == UCHAR_GENERAL_CATEGORY) { |
332 | 0 | type = UCPTRIE_TYPE_FAST; |
333 | 0 | } else { |
334 | 0 | type = UCPTRIE_TYPE_SMALL; |
335 | 0 | } |
336 | 0 | UCPTrieValueWidth valueWidth; |
337 | | // TODO: UCharacterProperty.IntProperty |
338 | 0 | int32_t max = u_getIntPropertyMaxValue(property); |
339 | 0 | if (max <= 0xff) { |
340 | 0 | valueWidth = UCPTRIE_VALUE_BITS_8; |
341 | 0 | } else if (max <= 0xffff) { |
342 | 0 | valueWidth = UCPTRIE_VALUE_BITS_16; |
343 | 0 | } else { |
344 | 0 | valueWidth = UCPTRIE_VALUE_BITS_32; |
345 | 0 | } |
346 | 0 | return reinterpret_cast<UCPMap *>( |
347 | 0 | umutablecptrie_buildImmutable(mutableTrie.getAlias(), type, valueWidth, &errorCode)); |
348 | 0 | } |
349 | | |
350 | | } // namespace |
351 | | |
352 | | U_NAMESPACE_USE |
353 | | |
354 | | U_CAPI const USet * U_EXPORT2 |
355 | 0 | u_getBinaryPropertySet(UProperty property, UErrorCode *pErrorCode) { |
356 | 0 | if (U_FAILURE(*pErrorCode)) { return nullptr; } |
357 | 0 | if (property < 0 || UCHAR_BINARY_LIMIT <= property) { |
358 | 0 | *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; |
359 | 0 | return nullptr; |
360 | 0 | } |
361 | 0 | Mutex m(&cpMutex); |
362 | 0 | UnicodeSet *set = sets[property]; |
363 | 0 | if (set == nullptr) { |
364 | 0 | sets[property] = set = makeSet(property, *pErrorCode); |
365 | 0 | } |
366 | 0 | if (U_FAILURE(*pErrorCode)) { return nullptr; } |
367 | 0 | return set->toUSet(); |
368 | 0 | } |
369 | | |
370 | | U_CAPI const UCPMap * U_EXPORT2 |
371 | 0 | u_getIntPropertyMap(UProperty property, UErrorCode *pErrorCode) { |
372 | 0 | if (U_FAILURE(*pErrorCode)) { return nullptr; } |
373 | 0 | if (property < UCHAR_INT_START || UCHAR_INT_LIMIT <= property) { |
374 | 0 | *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR; |
375 | 0 | return nullptr; |
376 | 0 | } |
377 | 0 | Mutex m(&cpMutex); |
378 | 0 | UCPMap *map = maps[property - UCHAR_INT_START]; |
379 | 0 | if (map == nullptr) { |
380 | 0 | maps[property - UCHAR_INT_START] = map = makeMap(property, *pErrorCode); |
381 | 0 | } |
382 | 0 | return map; |
383 | 0 | } |