/src/abseil-cpp/absl/strings/internal/str_split_internal.h
Line | Count | Source (jump to first uncovered line) |
1 | | // Copyright 2017 The Abseil Authors. |
2 | | // |
3 | | // Licensed under the Apache License, Version 2.0 (the "License"); |
4 | | // you may not use this file except in compliance with the License. |
5 | | // You may obtain a copy of the License at |
6 | | // |
7 | | // https://www.apache.org/licenses/LICENSE-2.0 |
8 | | // |
9 | | // Unless required by applicable law or agreed to in writing, software |
10 | | // distributed under the License is distributed on an "AS IS" BASIS, |
11 | | // WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
12 | | // See the License for the specific language governing permissions and |
13 | | // limitations under the License. |
14 | | // |
15 | | |
16 | | // This file declares INTERNAL parts of the Split API that are inline/templated |
17 | | // or otherwise need to be available at compile time. The main abstractions |
18 | | // defined in here are |
19 | | // |
20 | | // - ConvertibleToStringView |
21 | | // - SplitIterator<> |
22 | | // - Splitter<> |
23 | | // |
24 | | // DO NOT INCLUDE THIS FILE DIRECTLY. Use this file by including |
25 | | // absl/strings/str_split.h. |
26 | | // |
27 | | // IWYU pragma: private, include "absl/strings/str_split.h" |
28 | | |
29 | | #ifndef ABSL_STRINGS_INTERNAL_STR_SPLIT_INTERNAL_H_ |
30 | | #define ABSL_STRINGS_INTERNAL_STR_SPLIT_INTERNAL_H_ |
31 | | |
32 | | #include <array> |
33 | | #include <initializer_list> |
34 | | #include <iterator> |
35 | | #include <tuple> |
36 | | #include <type_traits> |
37 | | #include <utility> |
38 | | #include <vector> |
39 | | |
40 | | #include "absl/base/macros.h" |
41 | | #include "absl/base/port.h" |
42 | | #include "absl/meta/type_traits.h" |
43 | | #include "absl/strings/string_view.h" |
44 | | |
45 | | #ifdef _GLIBCXX_DEBUG |
46 | | #include "absl/strings/internal/stl_type_traits.h" |
47 | | #endif // _GLIBCXX_DEBUG |
48 | | |
49 | | namespace absl { |
50 | | ABSL_NAMESPACE_BEGIN |
51 | | namespace strings_internal { |
52 | | |
53 | | // This class is implicitly constructible from everything that absl::string_view |
54 | | // is implicitly constructible from, except for rvalue strings. This means it |
55 | | // can be used as a function parameter in places where passing a temporary |
56 | | // string might cause memory lifetime issues. |
57 | | class ConvertibleToStringView { |
58 | | public: |
59 | | ConvertibleToStringView(const char* s) // NOLINT(runtime/explicit) |
60 | 0 | : value_(s) {} |
61 | 0 | ConvertibleToStringView(char* s) : value_(s) {} // NOLINT(runtime/explicit) |
62 | | ConvertibleToStringView(absl::string_view s) // NOLINT(runtime/explicit) |
63 | 13.6M | : value_(s) {} |
64 | | ConvertibleToStringView(const std::string& s) // NOLINT(runtime/explicit) |
65 | 0 | : value_(s) {} |
66 | | |
67 | | // Disable conversion from rvalue strings. |
68 | | ConvertibleToStringView(std::string&& s) = delete; |
69 | | ConvertibleToStringView(const std::string&& s) = delete; |
70 | | |
71 | 13.6M | absl::string_view value() const { return value_; } |
72 | | |
73 | | private: |
74 | | absl::string_view value_; |
75 | | }; |
76 | | |
77 | | // An iterator that enumerates the parts of a string from a Splitter. The text |
78 | | // to be split, the Delimiter, and the Predicate are all taken from the given |
79 | | // Splitter object. Iterators may only be compared if they refer to the same |
80 | | // Splitter instance. |
81 | | // |
82 | | // This class is NOT part of the public splitting API. |
83 | | template <typename Splitter> |
84 | | class SplitIterator { |
85 | | public: |
86 | | using iterator_category = std::input_iterator_tag; |
87 | | using value_type = absl::string_view; |
88 | | using difference_type = ptrdiff_t; |
89 | | using pointer = const value_type*; |
90 | | using reference = const value_type&; |
91 | | |
92 | | enum State { kInitState, kLastState, kEndState }; |
93 | | SplitIterator(State state, const Splitter* splitter) |
94 | | : pos_(0), |
95 | | state_(state), |
96 | | splitter_(splitter), |
97 | | delimiter_(splitter->delimiter()), |
98 | 5.98M | predicate_(splitter->predicate()) { |
99 | | // Hack to maintain backward compatibility. This one block makes it so an |
100 | | // empty absl::string_view whose .data() happens to be nullptr behaves |
101 | | // *differently* from an otherwise empty absl::string_view whose .data() is |
102 | | // not nullptr. This is an undesirable difference in general, but this |
103 | | // behavior is maintained to avoid breaking existing code that happens to |
104 | | // depend on this old behavior/bug. Perhaps it will be fixed one day. The |
105 | | // difference in behavior is as follows: |
106 | | // Split(absl::string_view(""), '-'); // {""} |
107 | | // Split(absl::string_view(), '-'); // {} |
108 | 5.98M | if (splitter_->text().data() == nullptr) { |
109 | 0 | state_ = kEndState; |
110 | 0 | pos_ = splitter_->text().size(); |
111 | 0 | return; |
112 | 0 | } |
113 | | |
114 | 5.98M | if (state_ == kEndState) { |
115 | 0 | pos_ = splitter_->text().size(); |
116 | 5.98M | } else { |
117 | 5.98M | ++(*this); |
118 | 5.98M | } |
119 | 5.98M | } |
120 | | |
121 | 26.9M | bool at_end() const { return state_ == kEndState; } |
122 | | |
123 | | reference operator*() const { return curr_; } |
124 | 30.0M | pointer operator->() const { return &curr_; } |
125 | | |
126 | 21.0M | SplitIterator& operator++() { |
127 | 21.0M | do { |
128 | 21.0M | if (state_ == kLastState) { |
129 | 5.98M | state_ = kEndState; |
130 | 5.98M | return *this; |
131 | 5.98M | } |
132 | 15.0M | const absl::string_view text = splitter_->text(); |
133 | 15.0M | const absl::string_view d = delimiter_.Find(text, pos_); |
134 | 15.0M | if (d.data() == text.data() + text.size()) state_ = kLastState; |
135 | 15.0M | curr_ = text.substr(pos_, |
136 | 15.0M | static_cast<size_t>(d.data() - (text.data() + pos_))); |
137 | 15.0M | pos_ += curr_.size() + d.size(); |
138 | 15.0M | } while (!predicate_(curr_)); |
139 | 15.0M | return *this; |
140 | 21.0M | } |
141 | | |
142 | | SplitIterator operator++(int) { |
143 | | SplitIterator old(*this); |
144 | | ++(*this); |
145 | | return old; |
146 | | } |
147 | | |
148 | | friend bool operator==(const SplitIterator& a, const SplitIterator& b) { |
149 | | return a.state_ == b.state_ && a.pos_ == b.pos_; |
150 | | } |
151 | | |
152 | | friend bool operator!=(const SplitIterator& a, const SplitIterator& b) { |
153 | | return !(a == b); |
154 | | } |
155 | | |
156 | | private: |
157 | | size_t pos_; |
158 | | State state_; |
159 | | absl::string_view curr_; |
160 | | const Splitter* splitter_; |
161 | | typename Splitter::DelimiterType delimiter_; |
162 | | typename Splitter::PredicateType predicate_; |
163 | | }; |
164 | | |
165 | | // HasMappedType<T>::value is true iff there exists a type T::mapped_type. |
166 | | template <typename T, typename = void> |
167 | | struct HasMappedType : std::false_type {}; |
168 | | template <typename T> |
169 | | struct HasMappedType<T, absl::void_t<typename T::mapped_type>> |
170 | | : std::true_type {}; |
171 | | |
172 | | // HasValueType<T>::value is true iff there exists a type T::value_type. |
173 | | template <typename T, typename = void> |
174 | | struct HasValueType : std::false_type {}; |
175 | | template <typename T> |
176 | | struct HasValueType<T, absl::void_t<typename T::value_type>> : std::true_type { |
177 | | }; |
178 | | |
179 | | // HasConstIterator<T>::value is true iff there exists a type T::const_iterator. |
180 | | template <typename T, typename = void> |
181 | | struct HasConstIterator : std::false_type {}; |
182 | | template <typename T> |
183 | | struct HasConstIterator<T, absl::void_t<typename T::const_iterator>> |
184 | | : std::true_type {}; |
185 | | |
186 | | // HasEmplace<T>::value is true iff there exists a method T::emplace(). |
187 | | template <typename T, typename = void> |
188 | | struct HasEmplace : std::false_type {}; |
189 | | template <typename T> |
190 | | struct HasEmplace<T, absl::void_t<decltype(std::declval<T>().emplace())>> |
191 | | : std::true_type {}; |
192 | | |
193 | | // IsInitializerList<T>::value is true iff T is an std::initializer_list. More |
194 | | // details below in Splitter<> where this is used. |
195 | | std::false_type IsInitializerListDispatch(...); // default: No |
196 | | template <typename T> |
197 | | std::true_type IsInitializerListDispatch(std::initializer_list<T>*); |
198 | | template <typename T> |
199 | | struct IsInitializerList |
200 | | : decltype(IsInitializerListDispatch(static_cast<T*>(nullptr))) {}; |
201 | | |
202 | | // A SplitterIsConvertibleTo<C>::type alias exists iff the specified condition |
203 | | // is true for type 'C'. |
204 | | // |
205 | | // Restricts conversion to container-like types (by testing for the presence of |
206 | | // a const_iterator member type) and also to disable conversion to an |
207 | | // std::initializer_list (which also has a const_iterator). Otherwise, code |
208 | | // compiled in C++11 will get an error due to ambiguous conversion paths (in |
209 | | // C++11 std::vector<T>::operator= is overloaded to take either a std::vector<T> |
210 | | // or an std::initializer_list<T>). |
211 | | |
212 | | template <typename C, bool has_value_type, bool has_mapped_type> |
213 | | struct SplitterIsConvertibleToImpl : std::false_type {}; |
214 | | |
215 | | template <typename C> |
216 | | struct SplitterIsConvertibleToImpl<C, true, false> |
217 | | : std::is_constructible<typename C::value_type, absl::string_view> {}; |
218 | | |
219 | | template <typename C> |
220 | | struct SplitterIsConvertibleToImpl<C, true, true> |
221 | | : absl::conjunction< |
222 | | std::is_constructible<typename C::key_type, absl::string_view>, |
223 | | std::is_constructible<typename C::mapped_type, absl::string_view>> {}; |
224 | | |
225 | | template <typename C> |
226 | | struct SplitterIsConvertibleTo |
227 | | : SplitterIsConvertibleToImpl< |
228 | | C, |
229 | | #ifdef _GLIBCXX_DEBUG |
230 | | !IsStrictlyBaseOfAndConvertibleToSTLContainer<C>::value && |
231 | | #endif // _GLIBCXX_DEBUG |
232 | | !IsInitializerList< |
233 | | typename std::remove_reference<C>::type>::value && |
234 | | HasValueType<C>::value && HasConstIterator<C>::value, |
235 | | HasMappedType<C>::value> { |
236 | | }; |
237 | | |
238 | | template <typename StringType, typename Container, typename = void> |
239 | | struct ShouldUseLifetimeBound : std::false_type {}; |
240 | | |
241 | | template <typename StringType, typename Container> |
242 | | struct ShouldUseLifetimeBound< |
243 | | StringType, Container, |
244 | | std::enable_if_t< |
245 | | std::is_same<StringType, std::string>::value && |
246 | | std::is_same<typename Container::value_type, absl::string_view>::value>> |
247 | | : std::true_type {}; |
248 | | |
249 | | template <typename StringType, typename First, typename Second> |
250 | | using ShouldUseLifetimeBoundForPair = std::integral_constant< |
251 | | bool, std::is_same<StringType, std::string>::value && |
252 | | (std::is_same<First, absl::string_view>::value || |
253 | | std::is_same<Second, absl::string_view>::value)>; |
254 | | |
255 | | |
256 | | // This class implements the range that is returned by absl::StrSplit(). This |
257 | | // class has templated conversion operators that allow it to be implicitly |
258 | | // converted to a variety of types that the caller may have specified on the |
259 | | // left-hand side of an assignment. |
260 | | // |
261 | | // The main interface for interacting with this class is through its implicit |
262 | | // conversion operators. However, this class may also be used like a container |
263 | | // in that it has .begin() and .end() member functions. It may also be used |
264 | | // within a range-for loop. |
265 | | // |
266 | | // Output containers can be collections of any type that is constructible from |
267 | | // an absl::string_view. |
268 | | // |
269 | | // An Predicate functor may be supplied. This predicate will be used to filter |
270 | | // the split strings: only strings for which the predicate returns true will be |
271 | | // kept. A Predicate object is any unary functor that takes an absl::string_view |
272 | | // and returns bool. |
273 | | // |
274 | | // The StringType parameter can be either string_view or string, depending on |
275 | | // whether the Splitter refers to a string stored elsewhere, or if the string |
276 | | // resides inside the Splitter itself. |
277 | | template <typename Delimiter, typename Predicate, typename StringType> |
278 | | class Splitter { |
279 | | public: |
280 | | using DelimiterType = Delimiter; |
281 | | using PredicateType = Predicate; |
282 | | using const_iterator = strings_internal::SplitIterator<Splitter>; |
283 | | using value_type = typename std::iterator_traits<const_iterator>::value_type; |
284 | | |
285 | | Splitter(StringType input_text, Delimiter d, Predicate p) |
286 | | : text_(std::move(input_text)), |
287 | | delimiter_(std::move(d)), |
288 | 5.98M | predicate_(std::move(p)) {} |
289 | | |
290 | 21.0M | absl::string_view text() const { return text_; } |
291 | 5.98M | const Delimiter& delimiter() const { return delimiter_; } |
292 | 5.98M | const Predicate& predicate() const { return predicate_; } |
293 | | |
294 | | // Range functions that iterate the split substrings as absl::string_view |
295 | | // objects. These methods enable a Splitter to be used in a range-based for |
296 | | // loop. |
297 | 5.98M | const_iterator begin() const { return {const_iterator::kInitState, this}; } |
298 | | const_iterator end() const { return {const_iterator::kEndState, this}; } |
299 | | |
300 | | // An implicit conversion operator that is restricted to only those containers |
301 | | // that the splitter is convertible to. |
302 | | template < |
303 | | typename Container, |
304 | | std::enable_if_t<ShouldUseLifetimeBound<StringType, Container>::value && |
305 | | SplitterIsConvertibleTo<Container>::value, |
306 | | std::nullptr_t> = nullptr> |
307 | | // NOLINTNEXTLINE(google-explicit-constructor) |
308 | | operator Container() const ABSL_ATTRIBUTE_LIFETIME_BOUND { |
309 | | return ConvertToContainer<Container, typename Container::value_type, |
310 | | HasMappedType<Container>::value>()(*this); |
311 | | } |
312 | | |
313 | | template < |
314 | | typename Container, |
315 | | std::enable_if_t<!ShouldUseLifetimeBound<StringType, Container>::value && |
316 | | SplitterIsConvertibleTo<Container>::value, |
317 | | std::nullptr_t> = nullptr> |
318 | | // NOLINTNEXTLINE(google-explicit-constructor) |
319 | 5.98M | operator Container() const { |
320 | 5.98M | return ConvertToContainer<Container, typename Container::value_type, |
321 | 5.98M | HasMappedType<Container>::value>()(*this); |
322 | 5.98M | } Unexecuted instantiation: absl::strings_internal::Splitter<absl::ByChar, absl::AllowEmpty, absl::string_view>::operator std::__1::vector<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> >, std::__1::allocator<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > > ><std::__1::vector<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> >, std::__1::allocator<std::__1::basic_string<char, std::__1::char_traits<char>, std::__1::allocator<char> > > >, (decltype(nullptr))0>() const absl::strings_internal::Splitter<absl::ByChar, absl::AllowEmpty, absl::string_view>::operator std::__1::vector<absl::string_view, std::__1::allocator<absl::string_view> ><std::__1::vector<absl::string_view, std::__1::allocator<absl::string_view> >, (decltype(nullptr))0>() const Line | Count | Source | 319 | 5.98M | operator Container() const { | 320 | 5.98M | return ConvertToContainer<Container, typename Container::value_type, | 321 | 5.98M | HasMappedType<Container>::value>()(*this); | 322 | 5.98M | } |
|
323 | | |
324 | | // Returns a pair with its .first and .second members set to the first two |
325 | | // strings returned by the begin() iterator. Either/both of .first and .second |
326 | | // will be constructed with empty strings if the iterator doesn't have a |
327 | | // corresponding value. |
328 | | template <typename First, typename Second, |
329 | | std::enable_if_t< |
330 | | ShouldUseLifetimeBoundForPair<StringType, First, Second>::value, |
331 | | std::nullptr_t> = nullptr> |
332 | | // NOLINTNEXTLINE(google-explicit-constructor) |
333 | | operator std::pair<First, Second>() const ABSL_ATTRIBUTE_LIFETIME_BOUND { |
334 | | return ConvertToPair<First, Second>(); |
335 | | } |
336 | | |
337 | | template <typename First, typename Second, |
338 | | std::enable_if_t<!ShouldUseLifetimeBoundForPair<StringType, First, |
339 | | Second>::value, |
340 | | std::nullptr_t> = nullptr> |
341 | | // NOLINTNEXTLINE(google-explicit-constructor) |
342 | | operator std::pair<First, Second>() const { |
343 | | return ConvertToPair<First, Second>(); |
344 | | } |
345 | | |
346 | | private: |
347 | | template <typename First, typename Second> |
348 | | std::pair<First, Second> ConvertToPair() const { |
349 | | absl::string_view first, second; |
350 | | auto it = begin(); |
351 | | if (it != end()) { |
352 | | first = *it; |
353 | | if (++it != end()) { |
354 | | second = *it; |
355 | | } |
356 | | } |
357 | | return {First(first), Second(second)}; |
358 | | } |
359 | | |
360 | | // ConvertToContainer is a functor converting a Splitter to the requested |
361 | | // Container of ValueType. It is specialized below to optimize splitting to |
362 | | // certain combinations of Container and ValueType. |
363 | | // |
364 | | // This base template handles the generic case of storing the split results in |
365 | | // the requested non-map-like container and converting the split substrings to |
366 | | // the requested type. |
367 | | template <typename Container, typename ValueType, bool is_map = false> |
368 | | struct ConvertToContainer { |
369 | | Container operator()(const Splitter& splitter) const { |
370 | | Container c; |
371 | | auto it = std::inserter(c, c.end()); |
372 | | for (const auto& sp : splitter) { |
373 | | *it++ = ValueType(sp); |
374 | | } |
375 | | return c; |
376 | | } |
377 | | }; |
378 | | |
379 | | // Partial specialization for a std::vector<absl::string_view>. |
380 | | // |
381 | | // Optimized for the common case of splitting to a |
382 | | // std::vector<absl::string_view>. In this case we first split the results to |
383 | | // a small array of absl::string_view on the stack, to reduce reallocations. |
384 | | template <typename A> |
385 | | struct ConvertToContainer<std::vector<absl::string_view, A>, |
386 | | absl::string_view, false> { |
387 | | std::vector<absl::string_view, A> operator()( |
388 | 5.98M | const Splitter& splitter) const { |
389 | 5.98M | struct raw_view { |
390 | 5.98M | const char* data; |
391 | 5.98M | size_t size; |
392 | 15.0M | operator absl::string_view() const { // NOLINT(runtime/explicit) |
393 | 15.0M | return {data, size}; |
394 | 15.0M | } |
395 | 5.98M | }; |
396 | 5.98M | std::vector<absl::string_view, A> v; |
397 | 5.98M | std::array<raw_view, 16> ar; |
398 | 12.1M | for (auto it = splitter.begin(); !it.at_end();) { |
399 | 6.17M | size_t index = 0; |
400 | 15.0M | do { |
401 | 15.0M | ar[index].data = it->data(); |
402 | 15.0M | ar[index].size = it->size(); |
403 | 15.0M | ++it; |
404 | 15.0M | } while (++index != ar.size() && !it.at_end()); |
405 | 6.17M | v.insert(v.end(), ar.begin(), ar.begin() + index); |
406 | 6.17M | } |
407 | 5.98M | return v; |
408 | 5.98M | } |
409 | | }; |
410 | | |
411 | | // Partial specialization for a std::vector<std::string>. |
412 | | // |
413 | | // Optimized for the common case of splitting to a std::vector<std::string>. |
414 | | // In this case we first split the results to a std::vector<absl::string_view> |
415 | | // so the returned std::vector<std::string> can have space reserved to avoid |
416 | | // std::string moves. |
417 | | template <typename A> |
418 | | struct ConvertToContainer<std::vector<std::string, A>, std::string, false> { |
419 | 0 | std::vector<std::string, A> operator()(const Splitter& splitter) const { |
420 | 0 | const std::vector<absl::string_view> v = splitter; |
421 | 0 | return std::vector<std::string, A>(v.begin(), v.end()); |
422 | 0 | } |
423 | | }; |
424 | | |
425 | | // Partial specialization for containers of pairs (e.g., maps). |
426 | | // |
427 | | // The algorithm is to insert a new pair into the map for each even-numbered |
428 | | // item, with the even-numbered item as the key with a default-constructed |
429 | | // value. Each odd-numbered item will then be assigned to the last pair's |
430 | | // value. |
431 | | template <typename Container, typename First, typename Second> |
432 | | struct ConvertToContainer<Container, std::pair<const First, Second>, true> { |
433 | | using iterator = typename Container::iterator; |
434 | | |
435 | | Container operator()(const Splitter& splitter) const { |
436 | | Container m; |
437 | | iterator it; |
438 | | bool insert = true; |
439 | | for (const absl::string_view sv : splitter) { |
440 | | if (insert) { |
441 | | it = InsertOrEmplace(&m, sv); |
442 | | } else { |
443 | | it->second = Second(sv); |
444 | | } |
445 | | insert = !insert; |
446 | | } |
447 | | return m; |
448 | | } |
449 | | |
450 | | // Inserts the key and an empty value into the map, returning an iterator to |
451 | | // the inserted item. We use emplace() if available, otherwise insert(). |
452 | | template <typename M> |
453 | | static absl::enable_if_t<HasEmplace<M>::value, iterator> InsertOrEmplace( |
454 | | M* m, absl::string_view key) { |
455 | | // Use piecewise_construct to support old versions of gcc in which pair |
456 | | // constructor can't otherwise construct string from string_view. |
457 | | return ToIter(m->emplace(std::piecewise_construct, std::make_tuple(key), |
458 | | std::tuple<>())); |
459 | | } |
460 | | template <typename M> |
461 | | static absl::enable_if_t<!HasEmplace<M>::value, iterator> InsertOrEmplace( |
462 | | M* m, absl::string_view key) { |
463 | | return ToIter(m->insert(std::make_pair(First(key), Second("")))); |
464 | | } |
465 | | |
466 | | static iterator ToIter(std::pair<iterator, bool> pair) { |
467 | | return pair.first; |
468 | | } |
469 | | static iterator ToIter(iterator iter) { return iter; } |
470 | | }; |
471 | | |
472 | | StringType text_; |
473 | | Delimiter delimiter_; |
474 | | Predicate predicate_; |
475 | | }; |
476 | | |
477 | | } // namespace strings_internal |
478 | | ABSL_NAMESPACE_END |
479 | | } // namespace absl |
480 | | |
481 | | #endif // ABSL_STRINGS_INTERNAL_STR_SPLIT_INTERNAL_H_ |