1 | /* |
2 | * Copyright (C) 2006 Lars Knoll <[email protected]> |
3 | * Copyright (C) 2007-2016 Apple Inc. All rights reserved. |
4 | * |
5 | * This library is free software; you can redistribute it and/or |
6 | * modify it under the terms of the GNU Library General Public |
7 | * License as published by the Free Software Foundation; either |
8 | * version 2 of the License, or (at your option) any later version. |
9 | * |
10 | * This library is distributed in the hope that it will be useful, |
11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of |
12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU |
13 | * Library General Public License for more details. |
14 | * |
15 | * You should have received a copy of the GNU Library General Public License |
16 | * along with this library; see the file COPYING.LIB. If not, write to |
17 | * the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, |
18 | * Boston, MA 02110-1301, USA. |
19 | * |
20 | */ |
21 | |
22 | #pragma once |
23 | |
24 | #include <wtf/NeverDestroyed.h> |
25 | #include <wtf/Variant.h> |
26 | #include <wtf/text/StringView.h> |
27 | #include <wtf/text/icu/TextBreakIteratorICU.h> |
28 | |
29 | #if PLATFORM(MAC) || PLATFORM(IOS_FAMILY) |
30 | #include <wtf/text/cf/TextBreakIteratorCF.h> |
31 | #else |
32 | #include <wtf/text/NullTextBreakIterator.h> |
33 | #endif |
34 | |
35 | namespace WTF { |
36 | |
37 | #if PLATFORM(MAC) || PLATFORM(IOS_FAMILY) |
38 | typedef TextBreakIteratorCF TextBreakIteratorPlatform; |
39 | #else |
40 | typedef NullTextBreakIterator TextBreakIteratorPlatform; |
41 | #endif |
42 | |
43 | class TextBreakIteratorCache; |
44 | |
45 | class TextBreakIterator { |
46 | public: |
47 | enum class Mode { |
48 | Line, |
49 | Caret, |
50 | Delete |
51 | }; |
52 | |
53 | TextBreakIterator() = delete; |
54 | TextBreakIterator(const TextBreakIterator&) = delete; |
55 | TextBreakIterator(TextBreakIterator&&) = default; |
56 | TextBreakIterator& operator=(const TextBreakIterator&) = delete; |
57 | TextBreakIterator& operator=(TextBreakIterator&&) = default; |
58 | |
59 | Optional<unsigned> preceding(unsigned location) const |
60 | { |
61 | return switchOn(m_backing, [&](const auto& iterator) { |
62 | return iterator.preceding(location); |
63 | }); |
64 | } |
65 | |
66 | Optional<unsigned> following(unsigned location) const |
67 | { |
68 | return switchOn(m_backing, [&](const auto& iterator) { |
69 | return iterator.following(location); |
70 | }); |
71 | } |
72 | |
73 | bool isBoundary(unsigned location) const |
74 | { |
75 | return switchOn(m_backing, [&](const auto& iterator) { |
76 | return iterator.isBoundary(location); |
77 | }); |
78 | } |
79 | |
80 | private: |
81 | friend class TextBreakIteratorCache; |
82 | |
83 | // Use CachedTextBreakIterator instead of constructing one of these directly. |
84 | WTF_EXPORT TextBreakIterator(StringView, Mode, const AtomString& locale); |
85 | |
86 | void setText(StringView string) |
87 | { |
88 | return switchOn(m_backing, [&](auto& iterator) { |
89 | return iterator.setText(string); |
90 | }); |
91 | } |
92 | |
93 | Mode mode() const |
94 | { |
95 | return m_mode; |
96 | } |
97 | |
98 | const AtomString& locale() const |
99 | { |
100 | return m_locale; |
101 | } |
102 | |
103 | Variant<TextBreakIteratorICU, TextBreakIteratorPlatform> m_backing; |
104 | Mode m_mode; |
105 | AtomString m_locale; |
106 | }; |
107 | |
108 | class CachedTextBreakIterator; |
109 | |
110 | class TextBreakIteratorCache { |
111 | // Use CachedTextBreakIterator instead of dealing with the cache directly. |
112 | private: |
113 | friend class NeverDestroyed<TextBreakIteratorCache>; |
114 | friend class CachedTextBreakIterator; |
115 | |
116 | static TextBreakIteratorCache& singleton() |
117 | { |
118 | static NeverDestroyed<TextBreakIteratorCache> cache; |
119 | return cache.get(); |
120 | } |
121 | |
122 | TextBreakIteratorCache(const TextBreakIteratorCache&) = delete; |
123 | TextBreakIteratorCache(TextBreakIteratorCache&&) = delete; |
124 | TextBreakIteratorCache& operator=(const TextBreakIteratorCache&) = delete; |
125 | TextBreakIteratorCache& operator=(TextBreakIteratorCache&&) = delete; |
126 | |
127 | TextBreakIterator take(StringView string, TextBreakIterator::Mode mode, const AtomString& locale) |
128 | { |
129 | auto iter = std::find_if(m_unused.begin(), m_unused.end(), [&](TextBreakIterator& candidate) { |
130 | return candidate.mode() == mode && candidate.locale() == locale; |
131 | }); |
132 | if (iter == m_unused.end()) |
133 | return TextBreakIterator(string, mode, locale); |
134 | auto result = WTFMove(*iter); |
135 | m_unused.remove(iter - m_unused.begin()); |
136 | result.setText(string); |
137 | return result; |
138 | |
139 | } |
140 | |
141 | void put(TextBreakIterator&& iterator) |
142 | { |
143 | m_unused.append(WTFMove(iterator)); |
144 | if (m_unused.size() > capacity) |
145 | m_unused.remove(0); |
146 | } |
147 | |
148 | TextBreakIteratorCache() |
149 | { |
150 | } |
151 | |
152 | static constexpr int capacity = 2; |
153 | // FIXME: Break this up into different Vectors per mode. |
154 | Vector<TextBreakIterator, capacity> m_unused; |
155 | }; |
156 | |
157 | // RAII for TextBreakIterator and TextBreakIteratorCache. |
158 | class CachedTextBreakIterator { |
159 | public: |
160 | CachedTextBreakIterator(StringView string, TextBreakIterator::Mode mode, const AtomString& locale) |
161 | : m_backing(TextBreakIteratorCache::singleton().take(string, mode, locale)) |
162 | { |
163 | } |
164 | |
165 | ~CachedTextBreakIterator() |
166 | { |
167 | TextBreakIteratorCache::singleton().put(WTFMove(m_backing)); |
168 | } |
169 | |
170 | CachedTextBreakIterator() = delete; |
171 | CachedTextBreakIterator(const CachedTextBreakIterator&) = delete; |
172 | CachedTextBreakIterator(CachedTextBreakIterator&&) = default; |
173 | CachedTextBreakIterator& operator=(const CachedTextBreakIterator&) = delete; |
174 | CachedTextBreakIterator& operator=(CachedTextBreakIterator&&) = default; |
175 | |
176 | Optional<unsigned> preceding(unsigned location) const |
177 | { |
178 | return m_backing.preceding(location); |
179 | } |
180 | |
181 | Optional<unsigned> following(unsigned location) const |
182 | { |
183 | return m_backing.following(location); |
184 | } |
185 | |
186 | bool isBoundary(unsigned location) const |
187 | { |
188 | return m_backing.isBoundary(location); |
189 | } |
190 | |
191 | private: |
192 | TextBreakIterator m_backing; |
193 | }; |
194 | |
195 | // Note: The returned iterator is good only until you get another iterator, with the exception of acquireLineBreakIterator. |
196 | |
197 | enum class LineBreakIteratorMode { Default, Loose, Normal, Strict }; |
198 | |
199 | WTF_EXPORT_PRIVATE UBreakIterator* wordBreakIterator(StringView); |
200 | WTF_EXPORT_PRIVATE UBreakIterator* sentenceBreakIterator(StringView); |
201 | |
202 | WTF_EXPORT_PRIVATE UBreakIterator* acquireLineBreakIterator(StringView, const AtomString& locale, const UChar* priorContext, unsigned priorContextLength, LineBreakIteratorMode); |
203 | WTF_EXPORT_PRIVATE void releaseLineBreakIterator(UBreakIterator*); |
204 | UBreakIterator* openLineBreakIterator(const AtomString& locale); |
205 | void closeLineBreakIterator(UBreakIterator*&); |
206 | |
207 | WTF_EXPORT_PRIVATE bool isWordTextBreak(UBreakIterator*); |
208 | |
209 | class LazyLineBreakIterator { |
210 | public: |
211 | LazyLineBreakIterator() |
212 | { |
213 | resetPriorContext(); |
214 | } |
215 | |
216 | explicit LazyLineBreakIterator(StringView stringView, const AtomString& locale = AtomString(), LineBreakIteratorMode mode = LineBreakIteratorMode::Default) |
217 | : m_stringView(stringView) |
218 | , m_locale(locale) |
219 | , m_mode(mode) |
220 | { |
221 | resetPriorContext(); |
222 | } |
223 | |
224 | ~LazyLineBreakIterator() |
225 | { |
226 | if (m_iterator) |
227 | releaseLineBreakIterator(m_iterator); |
228 | } |
229 | |
230 | StringView stringView() const { return m_stringView; } |
231 | LineBreakIteratorMode mode() const { return m_mode; } |
232 | |
233 | UChar lastCharacter() const |
234 | { |
235 | static_assert(WTF_ARRAY_LENGTH(m_priorContext) == 2, "UBreakIterator unexpected prior context length" ); |
236 | return m_priorContext[1]; |
237 | } |
238 | |
239 | UChar secondToLastCharacter() const |
240 | { |
241 | static_assert(WTF_ARRAY_LENGTH(m_priorContext) == 2, "UBreakIterator unexpected prior context length" ); |
242 | return m_priorContext[0]; |
243 | } |
244 | |
245 | void setPriorContext(UChar last, UChar secondToLast) |
246 | { |
247 | static_assert(WTF_ARRAY_LENGTH(m_priorContext) == 2, "UBreakIterator unexpected prior context length" ); |
248 | m_priorContext[0] = secondToLast; |
249 | m_priorContext[1] = last; |
250 | } |
251 | |
252 | void updatePriorContext(UChar last) |
253 | { |
254 | static_assert(WTF_ARRAY_LENGTH(m_priorContext) == 2, "UBreakIterator unexpected prior context length" ); |
255 | m_priorContext[0] = m_priorContext[1]; |
256 | m_priorContext[1] = last; |
257 | } |
258 | |
259 | void resetPriorContext() |
260 | { |
261 | static_assert(WTF_ARRAY_LENGTH(m_priorContext) == 2, "UBreakIterator unexpected prior context length" ); |
262 | m_priorContext[0] = 0; |
263 | m_priorContext[1] = 0; |
264 | } |
265 | |
266 | unsigned priorContextLength() const |
267 | { |
268 | unsigned priorContextLength = 0; |
269 | static_assert(WTF_ARRAY_LENGTH(m_priorContext) == 2, "UBreakIterator unexpected prior context length" ); |
270 | if (m_priorContext[1]) { |
271 | ++priorContextLength; |
272 | if (m_priorContext[0]) |
273 | ++priorContextLength; |
274 | } |
275 | return priorContextLength; |
276 | } |
277 | |
278 | // Obtain text break iterator, possibly previously cached, where this iterator is (or has been) |
279 | // initialized to use the previously stored string as the primary breaking context and using |
280 | // previously stored prior context if non-empty. |
281 | UBreakIterator* get(unsigned priorContextLength) |
282 | { |
283 | ASSERT(priorContextLength <= priorContextCapacity); |
284 | const UChar* priorContext = priorContextLength ? &m_priorContext[priorContextCapacity - priorContextLength] : 0; |
285 | if (!m_iterator) { |
286 | m_iterator = acquireLineBreakIterator(m_stringView, m_locale, priorContext, priorContextLength, m_mode); |
287 | m_cachedPriorContext = priorContext; |
288 | m_cachedPriorContextLength = priorContextLength; |
289 | } else if (priorContext != m_cachedPriorContext || priorContextLength != m_cachedPriorContextLength) { |
290 | resetStringAndReleaseIterator(m_stringView, m_locale, m_mode); |
291 | return this->get(priorContextLength); |
292 | } |
293 | return m_iterator; |
294 | } |
295 | |
296 | void resetStringAndReleaseIterator(StringView stringView, const AtomString& locale, LineBreakIteratorMode mode) |
297 | { |
298 | if (m_iterator) |
299 | releaseLineBreakIterator(m_iterator); |
300 | m_stringView = stringView; |
301 | m_locale = locale; |
302 | m_iterator = nullptr; |
303 | m_cachedPriorContext = nullptr; |
304 | m_mode = mode; |
305 | m_cachedPriorContextLength = 0; |
306 | } |
307 | |
308 | private: |
309 | static constexpr unsigned priorContextCapacity = 2; |
310 | StringView m_stringView; |
311 | AtomString m_locale; |
312 | UBreakIterator* m_iterator { nullptr }; |
313 | const UChar* m_cachedPriorContext { nullptr }; |
314 | LineBreakIteratorMode m_mode { LineBreakIteratorMode::Default }; |
315 | unsigned m_cachedPriorContextLength { 0 }; |
316 | UChar m_priorContext[priorContextCapacity]; |
317 | }; |
318 | |
319 | // Iterates over "extended grapheme clusters", as defined in UAX #29. |
320 | // Note that platform implementations may be less sophisticated - e.g. ICU prior to |
321 | // version 4.0 only supports "legacy grapheme clusters". |
322 | // Use this for general text processing, e.g. string truncation. |
323 | |
324 | class NonSharedCharacterBreakIterator { |
325 | WTF_MAKE_NONCOPYABLE(NonSharedCharacterBreakIterator); |
326 | public: |
327 | WTF_EXPORT_PRIVATE NonSharedCharacterBreakIterator(StringView); |
328 | WTF_EXPORT_PRIVATE ~NonSharedCharacterBreakIterator(); |
329 | |
330 | NonSharedCharacterBreakIterator(NonSharedCharacterBreakIterator&&); |
331 | |
332 | operator UBreakIterator*() const { return m_iterator; } |
333 | |
334 | private: |
335 | UBreakIterator* m_iterator; |
336 | }; |
337 | |
338 | // Counts the number of grapheme clusters. A surrogate pair or a sequence |
339 | // of a non-combining character and following combining characters is |
340 | // counted as 1 grapheme cluster. |
341 | WTF_EXPORT_PRIVATE unsigned numGraphemeClusters(StringView); |
342 | |
343 | // Returns the number of code units that create the specified number of |
344 | // grapheme clusters. If there are fewer clusters in the string than specified, |
345 | // the length of the string is returned. |
346 | WTF_EXPORT_PRIVATE unsigned numCodeUnitsInGraphemeClusters(StringView, unsigned); |
347 | |
348 | } |
349 | |
350 | using WTF::CachedTextBreakIterator; |
351 | using WTF::LazyLineBreakIterator; |
352 | using WTF::LineBreakIteratorMode; |
353 | using WTF::NonSharedCharacterBreakIterator; |
354 | using WTF::TextBreakIterator; |
355 | using WTF::TextBreakIteratorCache; |
356 | using WTF::isWordTextBreak; |
357 | |