1 | // © 2016 and later: Unicode, Inc. and others. |
2 | // License & terms of use: http://www.unicode.org/copyright.html |
3 | /* |
4 | ******************************************************************************* |
5 | * |
6 | * Copyright (C) 2005-2016, International Business Machines |
7 | * Corporation and others. All Rights Reserved. |
8 | * |
9 | ******************************************************************************* |
10 | * file name: ucasemap.cpp |
11 | * encoding: UTF-8 |
12 | * tab size: 8 (not used) |
13 | * indentation:4 |
14 | * |
15 | * created on: 2005may06 |
16 | * created by: Markus W. Scherer |
17 | * |
18 | * Case mapping service object and functions using it. |
19 | */ |
20 | |
21 | #include "unicode/utypes.h" |
22 | #include "unicode/brkiter.h" |
23 | #include "unicode/bytestream.h" |
24 | #include "unicode/casemap.h" |
25 | #include "unicode/edits.h" |
26 | #include "unicode/stringoptions.h" |
27 | #include "unicode/stringpiece.h" |
28 | #include "unicode/ubrk.h" |
29 | #include "unicode/uloc.h" |
30 | #include "unicode/ustring.h" |
31 | #include "unicode/ucasemap.h" |
32 | #if !UCONFIG_NO_BREAK_ITERATION |
33 | #include "unicode/utext.h" |
34 | #endif |
35 | #include "unicode/utf.h" |
36 | #include "unicode/utf8.h" |
37 | #include "unicode/utf16.h" |
38 | #include "bytesinkutil.h" |
39 | #include "cmemory.h" |
40 | #include "cstring.h" |
41 | #include "uassert.h" |
42 | #include "ucase.h" |
43 | #include "ucasemap_imp.h" |
44 | #include "ustr_imp.h" |
45 | |
46 | U_NAMESPACE_USE |
47 | |
48 | /* UCaseMap service object -------------------------------------------------- */ |
49 | |
50 | UCaseMap::UCaseMap(const char *localeID, uint32_t opts, UErrorCode *pErrorCode) : |
51 | #if !UCONFIG_NO_BREAK_ITERATION |
52 | iter(nullptr), |
53 | #endif |
54 | caseLocale(UCASE_LOC_UNKNOWN), options(opts) { |
55 | ucasemap_setLocale(this, localeID, pErrorCode); |
56 | } |
57 | |
58 | UCaseMap::~UCaseMap() { |
59 | #if !UCONFIG_NO_BREAK_ITERATION |
60 | delete iter; |
61 | #endif |
62 | } |
63 | |
64 | U_CAPI UCaseMap * U_EXPORT2 |
65 | ucasemap_open(const char *locale, uint32_t options, UErrorCode *pErrorCode) { |
66 | if(U_FAILURE(*pErrorCode)) { |
67 | return nullptr; |
68 | } |
69 | UCaseMap *csm = new UCaseMap(locale, options, pErrorCode); |
70 | if(csm==nullptr) { |
71 | *pErrorCode = U_MEMORY_ALLOCATION_ERROR; |
72 | return nullptr; |
73 | } else if (U_FAILURE(*pErrorCode)) { |
74 | delete csm; |
75 | return nullptr; |
76 | } |
77 | return csm; |
78 | } |
79 | |
80 | U_CAPI void U_EXPORT2 |
81 | ucasemap_close(UCaseMap *csm) { |
82 | delete csm; |
83 | } |
84 | |
85 | U_CAPI const char * U_EXPORT2 |
86 | ucasemap_getLocale(const UCaseMap *csm) { |
87 | return csm->locale; |
88 | } |
89 | |
90 | U_CAPI uint32_t U_EXPORT2 |
91 | ucasemap_getOptions(const UCaseMap *csm) { |
92 | return csm->options; |
93 | } |
94 | |
95 | U_CAPI void U_EXPORT2 |
96 | ucasemap_setLocale(UCaseMap *csm, const char *locale, UErrorCode *pErrorCode) { |
97 | if(U_FAILURE(*pErrorCode)) { |
98 | return; |
99 | } |
100 | if (locale != nullptr && *locale == 0) { |
101 | csm->locale[0] = 0; |
102 | csm->caseLocale = UCASE_LOC_ROOT; |
103 | return; |
104 | } |
105 | |
106 | int32_t length=uloc_getName(locale, csm->locale, (int32_t)sizeof(csm->locale), pErrorCode); |
107 | if(*pErrorCode==U_BUFFER_OVERFLOW_ERROR || length==sizeof(csm->locale)) { |
108 | *pErrorCode=U_ZERO_ERROR; |
109 | /* we only really need the language code for case mappings */ |
110 | length=uloc_getLanguage(locale, csm->locale, (int32_t)sizeof(csm->locale), pErrorCode); |
111 | } |
112 | if(length==sizeof(csm->locale)) { |
113 | *pErrorCode=U_BUFFER_OVERFLOW_ERROR; |
114 | } |
115 | if(U_SUCCESS(*pErrorCode)) { |
116 | csm->caseLocale = ucase_getCaseLocale(csm->locale); |
117 | } else { |
118 | csm->locale[0]=0; |
119 | csm->caseLocale = UCASE_LOC_ROOT; |
120 | } |
121 | } |
122 | |
123 | U_CAPI void U_EXPORT2 |
124 | ucasemap_setOptions(UCaseMap *csm, uint32_t options, UErrorCode *pErrorCode) { |
125 | if(U_FAILURE(*pErrorCode)) { |
126 | return; |
127 | } |
128 | csm->options=options; |
129 | } |
130 | |
131 | /* UTF-8 string case mappings ----------------------------------------------- */ |
132 | |
133 | /* TODO(markus): Move to a new, separate utf8case.cpp file. */ |
134 | |
135 | namespace { |
136 | |
137 | /* append a full case mapping result, see UCASE_MAX_STRING_LENGTH */ |
138 | inline UBool |
139 | appendResult(int32_t cpLength, int32_t result, const char16_t *s, |
140 | ByteSink &sink, uint32_t options, icu::Edits *edits, UErrorCode &errorCode) { |
141 | U_ASSERT(U_SUCCESS(errorCode)); |
142 | |
143 | /* decode the result */ |
144 | if(result<0) { |
145 | /* (not) original code point */ |
146 | if(edits!=nullptr) { |
147 | edits->addUnchanged(cpLength); |
148 | } |
149 | if((options & U_OMIT_UNCHANGED_TEXT) == 0) { |
150 | ByteSinkUtil::appendCodePoint(cpLength, ~result, sink); |
151 | } |
152 | } else { |
153 | if(result<=UCASE_MAX_STRING_LENGTH) { |
154 | // string: "result" is the UTF-16 length |
155 | return ByteSinkUtil::appendChange(cpLength, s, result, sink, edits, errorCode); |
156 | } else { |
157 | ByteSinkUtil::appendCodePoint(cpLength, result, sink, edits); |
158 | } |
159 | } |
160 | return true; |
161 | } |
162 | |
163 | // See unicode/utf8.h U8_APPEND_UNSAFE(). |
164 | inline uint8_t getTwoByteLead(UChar32 c) { return (uint8_t)((c >> 6) | 0xc0); } |
165 | inline uint8_t getTwoByteTrail(UChar32 c) { return (uint8_t)((c & 0x3f) | 0x80); } |
166 | |
167 | UChar32 U_CALLCONV |
168 | utf8_caseContextIterator(void *context, int8_t dir) { |
169 | UCaseContext *csc=(UCaseContext *)context; |
170 | UChar32 c; |
171 | |
172 | if(dir<0) { |
173 | /* reset for backward iteration */ |
174 | csc->index=csc->cpStart; |
175 | csc->dir=dir; |
176 | } else if(dir>0) { |
177 | /* reset for forward iteration */ |
178 | csc->index=csc->cpLimit; |
179 | csc->dir=dir; |
180 | } else { |
181 | /* continue current iteration direction */ |
182 | dir=csc->dir; |
183 | } |
184 | |
185 | if(dir<0) { |
186 | if(csc->start<csc->index) { |
187 | U8_PREV((const uint8_t *)csc->p, csc->start, csc->index, c); |
188 | return c; |
189 | } |
190 | } else { |
191 | if(csc->index<csc->limit) { |
192 | U8_NEXT((const uint8_t *)csc->p, csc->index, csc->limit, c); |
193 | return c; |
194 | } |
195 | } |
196 | return U_SENTINEL; |
197 | } |
198 | |
199 | /** |
200 | * caseLocale >= 0: Lowercases [srcStart..srcLimit[ but takes context [0..srcLength[ into account. |
201 | * caseLocale < 0: Case-folds [srcStart..srcLimit[. |
202 | */ |
203 | void toLower(int32_t caseLocale, uint32_t options, |
204 | const uint8_t *src, UCaseContext *csc, int32_t srcStart, int32_t srcLimit, |
205 | icu::ByteSink &sink, icu::Edits *edits, UErrorCode &errorCode) { |
206 | const int8_t *latinToLower; |
207 | if (caseLocale == UCASE_LOC_ROOT || |
208 | (caseLocale >= 0 ? |
209 | !(caseLocale == UCASE_LOC_TURKISH || caseLocale == UCASE_LOC_LITHUANIAN) : |
210 | (options & _FOLD_CASE_OPTIONS_MASK) == U_FOLD_CASE_DEFAULT)) { |
211 | latinToLower = LatinCase::TO_LOWER_NORMAL; |
212 | } else { |
213 | latinToLower = LatinCase::TO_LOWER_TR_LT; |
214 | } |
215 | const UTrie2 *trie = ucase_getTrie(); |
216 | int32_t prev = srcStart; |
217 | int32_t srcIndex = srcStart; |
218 | for (;;) { |
219 | // fast path for simple cases |
220 | int32_t cpStart; |
221 | UChar32 c; |
222 | for (;;) { |
223 | if (U_FAILURE(errorCode) || srcIndex >= srcLimit) { |
224 | c = U_SENTINEL; |
225 | break; |
226 | } |
227 | uint8_t lead = src[srcIndex++]; |
228 | if (lead <= 0x7f) { |
229 | int8_t d = latinToLower[lead]; |
230 | if (d == LatinCase::EXC) { |
231 | cpStart = srcIndex - 1; |
232 | c = lead; |
233 | break; |
234 | } |
235 | if (d == 0) { continue; } |
236 | ByteSinkUtil::appendUnchanged(src + prev, srcIndex - 1 - prev, |
237 | sink, options, edits, errorCode); |
238 | char ascii = (char)(lead + d); |
239 | sink.Append(&ascii, 1); |
240 | if (edits != nullptr) { |
241 | edits->addReplace(1, 1); |
242 | } |
243 | prev = srcIndex; |
244 | continue; |
245 | } else if (lead < 0xe3) { |
246 | uint8_t t; |
247 | if (0xc2 <= lead && lead <= 0xc5 && srcIndex < srcLimit && |
248 | (t = src[srcIndex] - 0x80) <= 0x3f) { |
249 | // U+0080..U+017F |
250 | ++srcIndex; |
251 | c = ((lead - 0xc0) << 6) | t; |
252 | int8_t d = latinToLower[c]; |
253 | if (d == LatinCase::EXC) { |
254 | cpStart = srcIndex - 2; |
255 | break; |
256 | } |
257 | if (d == 0) { continue; } |
258 | ByteSinkUtil::appendUnchanged(src + prev, srcIndex - 2 - prev, |
259 | sink, options, edits, errorCode); |
260 | ByteSinkUtil::appendTwoBytes(c + d, sink); |
261 | if (edits != nullptr) { |
262 | edits->addReplace(2, 2); |
263 | } |
264 | prev = srcIndex; |
265 | continue; |
266 | } |
267 | } else if ((lead <= 0xe9 || lead == 0xeb || lead == 0xec) && |
268 | (srcIndex + 2) <= srcLimit && |
269 | U8_IS_TRAIL(src[srcIndex]) && U8_IS_TRAIL(src[srcIndex + 1])) { |
270 | // most of CJK: no case mappings |
271 | srcIndex += 2; |
272 | continue; |
273 | } |
274 | cpStart = --srcIndex; |
275 | U8_NEXT(src, srcIndex, srcLimit, c); |
276 | if (c < 0) { |
277 | // ill-formed UTF-8 |
278 | continue; |
279 | } |
280 | uint16_t props = UTRIE2_GET16(trie, c); |
281 | if (UCASE_HAS_EXCEPTION(props)) { break; } |
282 | int32_t delta; |
283 | if (!UCASE_IS_UPPER_OR_TITLE(props) || (delta = UCASE_GET_DELTA(props)) == 0) { |
284 | continue; |
285 | } |
286 | ByteSinkUtil::appendUnchanged(src + prev, cpStart - prev, |
287 | sink, options, edits, errorCode); |
288 | ByteSinkUtil::appendCodePoint(srcIndex - cpStart, c + delta, sink, edits); |
289 | prev = srcIndex; |
290 | } |
291 | if (c < 0) { |
292 | break; |
293 | } |
294 | // slow path |
295 | const char16_t *s; |
296 | if (caseLocale >= 0) { |
297 | csc->cpStart = cpStart; |
298 | csc->cpLimit = srcIndex; |
299 | c = ucase_toFullLower(c, utf8_caseContextIterator, csc, &s, caseLocale); |
300 | } else { |
301 | c = ucase_toFullFolding(c, &s, options); |
302 | } |
303 | if (c >= 0) { |
304 | ByteSinkUtil::appendUnchanged(src + prev, cpStart - prev, |
305 | sink, options, edits, errorCode); |
306 | appendResult(srcIndex - cpStart, c, s, sink, options, edits, errorCode); |
307 | prev = srcIndex; |
308 | } |
309 | } |
310 | ByteSinkUtil::appendUnchanged(src + prev, srcIndex - prev, |
311 | sink, options, edits, errorCode); |
312 | } |
313 | |
314 | void toUpper(int32_t caseLocale, uint32_t options, |
315 | const uint8_t *src, UCaseContext *csc, int32_t srcLength, |
316 | icu::ByteSink &sink, icu::Edits *edits, UErrorCode &errorCode) { |
317 | const int8_t *latinToUpper; |
318 | if (caseLocale == UCASE_LOC_TURKISH) { |
319 | latinToUpper = LatinCase::TO_UPPER_TR; |
320 | } else { |
321 | latinToUpper = LatinCase::TO_UPPER_NORMAL; |
322 | } |
323 | const UTrie2 *trie = ucase_getTrie(); |
324 | int32_t prev = 0; |
325 | int32_t srcIndex = 0; |
326 | for (;;) { |
327 | // fast path for simple cases |
328 | int32_t cpStart; |
329 | UChar32 c; |
330 | for (;;) { |
331 | if (U_FAILURE(errorCode) || srcIndex >= srcLength) { |
332 | c = U_SENTINEL; |
333 | break; |
334 | } |
335 | uint8_t lead = src[srcIndex++]; |
336 | if (lead <= 0x7f) { |
337 | int8_t d = latinToUpper[lead]; |
338 | if (d == LatinCase::EXC) { |
339 | cpStart = srcIndex - 1; |
340 | c = lead; |
341 | break; |
342 | } |
343 | if (d == 0) { continue; } |
344 | ByteSinkUtil::appendUnchanged(src + prev, srcIndex - 1 - prev, |
345 | sink, options, edits, errorCode); |
346 | char ascii = (char)(lead + d); |
347 | sink.Append(&ascii, 1); |
348 | if (edits != nullptr) { |
349 | edits->addReplace(1, 1); |
350 | } |
351 | prev = srcIndex; |
352 | continue; |
353 | } else if (lead < 0xe3) { |
354 | uint8_t t; |
355 | if (0xc2 <= lead && lead <= 0xc5 && srcIndex < srcLength && |
356 | (t = src[srcIndex] - 0x80) <= 0x3f) { |
357 | // U+0080..U+017F |
358 | ++srcIndex; |
359 | c = ((lead - 0xc0) << 6) | t; |
360 | int8_t d = latinToUpper[c]; |
361 | if (d == LatinCase::EXC) { |
362 | cpStart = srcIndex - 2; |
363 | break; |
364 | } |
365 | if (d == 0) { continue; } |
366 | ByteSinkUtil::appendUnchanged(src + prev, srcIndex - 2 - prev, |
367 | sink, options, edits, errorCode); |
368 | ByteSinkUtil::appendTwoBytes(c + d, sink); |
369 | if (edits != nullptr) { |
370 | edits->addReplace(2, 2); |
371 | } |
372 | prev = srcIndex; |
373 | continue; |
374 | } |
375 | } else if ((lead <= 0xe9 || lead == 0xeb || lead == 0xec) && |
376 | (srcIndex + 2) <= srcLength && |
377 | U8_IS_TRAIL(src[srcIndex]) && U8_IS_TRAIL(src[srcIndex + 1])) { |
378 | // most of CJK: no case mappings |
379 | srcIndex += 2; |
380 | continue; |
381 | } |
382 | cpStart = --srcIndex; |
383 | U8_NEXT(src, srcIndex, srcLength, c); |
384 | if (c < 0) { |
385 | // ill-formed UTF-8 |
386 | continue; |
387 | } |
388 | uint16_t props = UTRIE2_GET16(trie, c); |
389 | if (UCASE_HAS_EXCEPTION(props)) { break; } |
390 | int32_t delta; |
391 | if (UCASE_GET_TYPE(props) != UCASE_LOWER || (delta = UCASE_GET_DELTA(props)) == 0) { |
392 | continue; |
393 | } |
394 | ByteSinkUtil::appendUnchanged(src + prev, cpStart - prev, |
395 | sink, options, edits, errorCode); |
396 | ByteSinkUtil::appendCodePoint(srcIndex - cpStart, c + delta, sink, edits); |
397 | prev = srcIndex; |
398 | } |
399 | if (c < 0) { |
400 | break; |
401 | } |
402 | // slow path |
403 | csc->cpStart = cpStart; |
404 | csc->cpLimit = srcIndex; |
405 | const char16_t *s; |
406 | c = ucase_toFullUpper(c, utf8_caseContextIterator, csc, &s, caseLocale); |
407 | if (c >= 0) { |
408 | ByteSinkUtil::appendUnchanged(src + prev, cpStart - prev, |
409 | sink, options, edits, errorCode); |
410 | appendResult(srcIndex - cpStart, c, s, sink, options, edits, errorCode); |
411 | prev = srcIndex; |
412 | } |
413 | } |
414 | ByteSinkUtil::appendUnchanged(src + prev, srcIndex - prev, |
415 | sink, options, edits, errorCode); |
416 | } |
417 | |
418 | } // namespace |
419 | |
420 | #if !UCONFIG_NO_BREAK_ITERATION |
421 | |
422 | namespace { |
423 | |
424 | constexpr uint8_t ACUTE_BYTE0 = u8"\u0301" [0]; |
425 | |
426 | constexpr uint8_t ACUTE_BYTE1 = u8"\u0301" [1]; |
427 | |
428 | /** |
429 | * Input: c is a letter I with or without acute accent. |
430 | * start is the index in src after c, and is less than segmentLimit. |
431 | * If a plain i/I is followed by a plain j/J, |
432 | * or an i/I with acute (precomposed or decomposed) is followed by a j/J with acute, |
433 | * then we output accordingly. |
434 | * |
435 | * @return the src index after the titlecased sequence, or the start index if no Dutch IJ |
436 | */ |
437 | int32_t maybeTitleDutchIJ(const uint8_t *src, UChar32 c, int32_t start, int32_t segmentLimit, |
438 | ByteSink &sink, uint32_t options, icu::Edits *edits, UErrorCode &errorCode) { |
439 | U_ASSERT(start < segmentLimit); |
440 | |
441 | int32_t index = start; |
442 | bool withAcute = false; |
443 | |
444 | // If the conditions are met, then the following variables tell us what to output. |
445 | int32_t unchanged1 = 0; // code units before the j, or the whole sequence (0..3) |
446 | bool doTitleJ = false; // true if the j needs to be titlecased |
447 | int32_t unchanged2 = 0; // after the j (0 or 1) |
448 | |
449 | // next character after the first letter |
450 | UChar32 c2; |
451 | c2 = src[index++]; |
452 | |
453 | // Is the first letter an i/I with accent? |
454 | if (c == u'I') { |
455 | if (c2 == ACUTE_BYTE0 && index < segmentLimit && src[index++] == ACUTE_BYTE1) { |
456 | withAcute = true; |
457 | unchanged1 = 2; // ACUTE is 2 code units in UTF-8 |
458 | if (index == segmentLimit) { return start; } |
459 | c2 = src[index++]; |
460 | } |
461 | } else { // Í |
462 | withAcute = true; |
463 | } |
464 | |
465 | // Is the next character a j/J? |
466 | if (c2 == u'j') { |
467 | doTitleJ = true; |
468 | } else if (c2 == u'J') { |
469 | ++unchanged1; |
470 | } else { |
471 | return start; |
472 | } |
473 | |
474 | // A plain i/I must be followed by a plain j/J. |
475 | // An i/I with acute must be followed by a j/J with acute. |
476 | if (withAcute) { |
477 | if ((index + 1) >= segmentLimit || src[index++] != ACUTE_BYTE0 || src[index++] != ACUTE_BYTE1) { |
478 | return start; |
479 | } |
480 | if (doTitleJ) { |
481 | unchanged2 = 2; // ACUTE is 2 code units in UTF-8 |
482 | } else { |
483 | unchanged1 = unchanged1 + 2; // ACUTE is 2 code units in UTF-8 |
484 | } |
485 | } |
486 | |
487 | // There must not be another combining mark. |
488 | if (index < segmentLimit) { |
489 | int32_t cp; |
490 | int32_t i = index; |
491 | U8_NEXT(src, i, segmentLimit, cp); |
492 | uint32_t typeMask = U_GET_GC_MASK(cp); |
493 | if ((typeMask & U_GC_M_MASK) != 0) { |
494 | return start; |
495 | } |
496 | } |
497 | |
498 | // Output the rest of the Dutch IJ. |
499 | ByteSinkUtil::appendUnchanged(src + start, unchanged1, sink, options, edits, errorCode); |
500 | start += unchanged1; |
501 | if (doTitleJ) { |
502 | ByteSinkUtil::appendCodePoint(1, u'J', sink, edits); |
503 | ++start; |
504 | } |
505 | ByteSinkUtil::appendUnchanged(src + start, unchanged2, sink, options, edits, errorCode); |
506 | |
507 | U_ASSERT(start + unchanged2 == index); |
508 | return index; |
509 | } |
510 | |
511 | } // namespace |
512 | |
513 | U_CFUNC void U_CALLCONV |
514 | ucasemap_internalUTF8ToTitle( |
515 | int32_t caseLocale, uint32_t options, BreakIterator *iter, |
516 | const uint8_t *src, int32_t srcLength, |
517 | ByteSink &sink, icu::Edits *edits, |
518 | UErrorCode &errorCode) { |
519 | if (!ustrcase_checkTitleAdjustmentOptions(options, errorCode)) { |
520 | return; |
521 | } |
522 | |
523 | /* set up local variables */ |
524 | UCaseContext csc=UCASECONTEXT_INITIALIZER; |
525 | csc.p=(void *)src; |
526 | csc.limit=srcLength; |
527 | int32_t prev=0; |
528 | UBool isFirstIndex=true; |
529 | |
530 | /* titlecasing loop */ |
531 | while(prev<srcLength) { |
532 | /* find next index where to titlecase */ |
533 | int32_t index; |
534 | if(isFirstIndex) { |
535 | isFirstIndex=false; |
536 | index=iter->first(); |
537 | } else { |
538 | index=iter->next(); |
539 | } |
540 | if(index==UBRK_DONE || index>srcLength) { |
541 | index=srcLength; |
542 | } |
543 | |
544 | /* |
545 | * Segment [prev..index[ into 3 parts: |
546 | * a) skipped characters (copy as-is) [prev..titleStart[ |
547 | * b) first letter (titlecase) [titleStart..titleLimit[ |
548 | * c) subsequent characters (lowercase) [titleLimit..index[ |
549 | */ |
550 | if(prev<index) { |
551 | /* find and copy skipped characters [prev..titleStart[ */ |
552 | int32_t titleStart=prev; |
553 | int32_t titleLimit=prev; |
554 | UChar32 c; |
555 | U8_NEXT(src, titleLimit, index, c); |
556 | if ((options&U_TITLECASE_NO_BREAK_ADJUSTMENT)==0) { |
557 | // Adjust the titlecasing index to the next cased character, |
558 | // or to the next letter/number/symbol/private use. |
559 | // Stop with titleStart<titleLimit<=index |
560 | // if there is a character to be titlecased, |
561 | // or else stop with titleStart==titleLimit==index. |
562 | UBool toCased = (options&U_TITLECASE_ADJUST_TO_CASED) != 0; |
563 | while (toCased ? UCASE_NONE==ucase_getType(c) : !ustrcase_isLNS(c)) { |
564 | titleStart=titleLimit; |
565 | if(titleLimit==index) { |
566 | break; |
567 | } |
568 | U8_NEXT(src, titleLimit, index, c); |
569 | } |
570 | if (prev < titleStart) { |
571 | if (!ByteSinkUtil::appendUnchanged(src+prev, titleStart-prev, |
572 | sink, options, edits, errorCode)) { |
573 | return; |
574 | } |
575 | } |
576 | } |
577 | |
578 | if(titleStart<titleLimit) { |
579 | /* titlecase c which is from [titleStart..titleLimit[ */ |
580 | if(c>=0) { |
581 | csc.cpStart=titleStart; |
582 | csc.cpLimit=titleLimit; |
583 | const char16_t *s; |
584 | c=ucase_toFullTitle(c, utf8_caseContextIterator, &csc, &s, caseLocale); |
585 | if (!appendResult(titleLimit-titleStart, c, s, sink, options, edits, errorCode)) { |
586 | return; |
587 | } |
588 | } else { |
589 | // Malformed UTF-8. |
590 | if (!ByteSinkUtil::appendUnchanged(src+titleStart, titleLimit-titleStart, |
591 | sink, options, edits, errorCode)) { |
592 | return; |
593 | } |
594 | } |
595 | |
596 | /* Special case Dutch IJ titlecasing */ |
597 | if (titleLimit < index && |
598 | caseLocale == UCASE_LOC_DUTCH) { |
599 | if (c < 0) { |
600 | c = ~c; |
601 | } |
602 | |
603 | if (c == u'I' || c == u'Í') { |
604 | titleLimit = maybeTitleDutchIJ(src, c, titleLimit, index, sink, options, edits, errorCode); |
605 | } |
606 | } |
607 | |
608 | /* lowercase [titleLimit..index[ */ |
609 | if(titleLimit<index) { |
610 | if((options&U_TITLECASE_NO_LOWERCASE)==0) { |
611 | /* Normal operation: Lowercase the rest of the word. */ |
612 | toLower(caseLocale, options, |
613 | src, &csc, titleLimit, index, |
614 | sink, edits, errorCode); |
615 | if(U_FAILURE(errorCode)) { |
616 | return; |
617 | } |
618 | } else { |
619 | /* Optionally just copy the rest of the word unchanged. */ |
620 | if (!ByteSinkUtil::appendUnchanged(src+titleLimit, index-titleLimit, |
621 | sink, options, edits, errorCode)) { |
622 | return; |
623 | } |
624 | } |
625 | } |
626 | } |
627 | } |
628 | |
629 | prev=index; |
630 | } |
631 | } |
632 | |
633 | #endif |
634 | |
635 | U_NAMESPACE_BEGIN |
636 | namespace GreekUpper { |
637 | |
638 | UBool isFollowedByCasedLetter(const uint8_t *s, int32_t i, int32_t length) { |
639 | while (i < length) { |
640 | UChar32 c; |
641 | U8_NEXT(s, i, length, c); |
642 | int32_t type = ucase_getTypeOrIgnorable(c); |
643 | if ((type & UCASE_IGNORABLE) != 0) { |
644 | // Case-ignorable, continue with the loop. |
645 | } else if (type != UCASE_NONE) { |
646 | return true; // Followed by cased letter. |
647 | } else { |
648 | return false; // Uncased and not case-ignorable. |
649 | } |
650 | } |
651 | return false; // Not followed by cased letter. |
652 | } |
653 | |
654 | // Keep this consistent with the UTF-16 version in ustrcase.cpp and the Java version in CaseMap.java. |
655 | void toUpper(uint32_t options, |
656 | const uint8_t *src, int32_t srcLength, |
657 | ByteSink &sink, Edits *edits, |
658 | UErrorCode &errorCode) { |
659 | uint32_t state = 0; |
660 | for (int32_t i = 0; i < srcLength;) { |
661 | int32_t nextIndex = i; |
662 | UChar32 c; |
663 | U8_NEXT(src, nextIndex, srcLength, c); |
664 | uint32_t nextState = 0; |
665 | int32_t type = ucase_getTypeOrIgnorable(c); |
666 | if ((type & UCASE_IGNORABLE) != 0) { |
667 | // c is case-ignorable |
668 | nextState |= (state & AFTER_CASED); |
669 | } else if (type != UCASE_NONE) { |
670 | // c is cased |
671 | nextState |= AFTER_CASED; |
672 | } |
673 | uint32_t data = getLetterData(c); |
674 | if (data > 0) { |
675 | uint32_t upper = data & UPPER_MASK; |
676 | // Add a dialytika to this iota or ypsilon vowel |
677 | // if we removed a tonos from the previous vowel, |
678 | // and that previous vowel did not also have (or gain) a dialytika. |
679 | // Adding one only to the final vowel in a longer sequence |
680 | // (which does not occur in normal writing) would require lookahead. |
681 | // Set the same flag as for preserving an existing dialytika. |
682 | if ((data & HAS_VOWEL) != 0 && (state & AFTER_VOWEL_WITH_ACCENT) != 0 && |
683 | (upper == 0x399 || upper == 0x3A5)) { |
684 | data |= HAS_DIALYTIKA; |
685 | } |
686 | int32_t numYpogegrammeni = 0; // Map each one to a trailing, spacing, capital iota. |
687 | if ((data & HAS_YPOGEGRAMMENI) != 0) { |
688 | numYpogegrammeni = 1; |
689 | } |
690 | // Skip combining diacritics after this Greek letter. |
691 | int32_t nextNextIndex = nextIndex; |
692 | while (nextIndex < srcLength) { |
693 | UChar32 c2; |
694 | U8_NEXT(src, nextNextIndex, srcLength, c2); |
695 | uint32_t diacriticData = getDiacriticData(c2); |
696 | if (diacriticData != 0) { |
697 | data |= diacriticData; |
698 | if ((diacriticData & HAS_YPOGEGRAMMENI) != 0) { |
699 | ++numYpogegrammeni; |
700 | } |
701 | nextIndex = nextNextIndex; |
702 | } else { |
703 | break; // not a Greek diacritic |
704 | } |
705 | } |
706 | if ((data & HAS_VOWEL_AND_ACCENT_AND_DIALYTIKA) == HAS_VOWEL_AND_ACCENT) { |
707 | nextState |= AFTER_VOWEL_WITH_ACCENT; |
708 | } |
709 | // Map according to Greek rules. |
710 | UBool addTonos = false; |
711 | if (upper == 0x397 && |
712 | (data & HAS_ACCENT) != 0 && |
713 | numYpogegrammeni == 0 && |
714 | (state & AFTER_CASED) == 0 && |
715 | !isFollowedByCasedLetter(src, nextIndex, srcLength)) { |
716 | // Keep disjunctive "or" with (only) a tonos. |
717 | // We use the same "word boundary" conditions as for the Final_Sigma test. |
718 | if (i == nextIndex) { |
719 | upper = 0x389; // Preserve the precomposed form. |
720 | } else { |
721 | addTonos = true; |
722 | } |
723 | } else if ((data & HAS_DIALYTIKA) != 0) { |
724 | // Preserve a vowel with dialytika in precomposed form if it exists. |
725 | if (upper == 0x399) { |
726 | upper = 0x3AA; |
727 | data &= ~HAS_EITHER_DIALYTIKA; |
728 | } else if (upper == 0x3A5) { |
729 | upper = 0x3AB; |
730 | data &= ~HAS_EITHER_DIALYTIKA; |
731 | } |
732 | } |
733 | |
734 | UBool change; |
735 | if (edits == nullptr && (options & U_OMIT_UNCHANGED_TEXT) == 0) { |
736 | change = true; // common, simple usage |
737 | } else { |
738 | // Find out first whether we are changing the text. |
739 | U_ASSERT(0x370 <= upper && upper <= 0x3ff); // 2-byte UTF-8, main Greek block |
740 | change = (i + 2) > nextIndex || |
741 | src[i] != getTwoByteLead(upper) || src[i + 1] != getTwoByteTrail(upper) || |
742 | numYpogegrammeni > 0; |
743 | int32_t i2 = i + 2; |
744 | if ((data & HAS_EITHER_DIALYTIKA) != 0) { |
745 | change |= (i2 + 2) > nextIndex || |
746 | src[i2] != (uint8_t)u8"\u0308" [0] || |
747 | src[i2 + 1] != (uint8_t)u8"\u0308" [1]; |
748 | i2 += 2; |
749 | } |
750 | if (addTonos) { |
751 | change |= (i2 + 2) > nextIndex || |
752 | src[i2] != (uint8_t)u8"\u0301" [0] || |
753 | src[i2 + 1] != (uint8_t)u8"\u0301" [1]; |
754 | i2 += 2; |
755 | } |
756 | int32_t oldLength = nextIndex - i; |
757 | int32_t newLength = (i2 - i) + numYpogegrammeni * 2; // 2 bytes per U+0399 |
758 | change |= oldLength != newLength; |
759 | if (change) { |
760 | if (edits != nullptr) { |
761 | edits->addReplace(oldLength, newLength); |
762 | } |
763 | } else { |
764 | if (edits != nullptr) { |
765 | edits->addUnchanged(oldLength); |
766 | } |
767 | // Write unchanged text? |
768 | change = (options & U_OMIT_UNCHANGED_TEXT) == 0; |
769 | } |
770 | } |
771 | |
772 | if (change) { |
773 | ByteSinkUtil::appendTwoBytes(upper, sink); |
774 | if ((data & HAS_EITHER_DIALYTIKA) != 0) { |
775 | sink.AppendU8(u8"\u0308" , 2); // restore or add a dialytika |
776 | } |
777 | if (addTonos) { |
778 | sink.AppendU8(u8"\u0301" , 2); |
779 | } |
780 | while (numYpogegrammeni > 0) { |
781 | sink.AppendU8(u8"\u0399" , 2); |
782 | --numYpogegrammeni; |
783 | } |
784 | } |
785 | } else if(c>=0) { |
786 | const char16_t *s; |
787 | c=ucase_toFullUpper(c, nullptr, nullptr, &s, UCASE_LOC_GREEK); |
788 | if (!appendResult(nextIndex - i, c, s, sink, options, edits, errorCode)) { |
789 | return; |
790 | } |
791 | } else { |
792 | // Malformed UTF-8. |
793 | if (!ByteSinkUtil::appendUnchanged(src+i, nextIndex-i, |
794 | sink, options, edits, errorCode)) { |
795 | return; |
796 | } |
797 | } |
798 | i = nextIndex; |
799 | state = nextState; |
800 | } |
801 | } |
802 | |
803 | } // namespace GreekUpper |
804 | U_NAMESPACE_END |
805 | |
806 | static void U_CALLCONV |
807 | ucasemap_internalUTF8ToLower(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED |
808 | const uint8_t *src, int32_t srcLength, |
809 | icu::ByteSink &sink, icu::Edits *edits, |
810 | UErrorCode &errorCode) { |
811 | UCaseContext csc=UCASECONTEXT_INITIALIZER; |
812 | csc.p=(void *)src; |
813 | csc.limit=srcLength; |
814 | toLower( |
815 | caseLocale, options, |
816 | src, &csc, 0, srcLength, |
817 | sink, edits, errorCode); |
818 | } |
819 | |
820 | static void U_CALLCONV |
821 | ucasemap_internalUTF8ToUpper(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED |
822 | const uint8_t *src, int32_t srcLength, |
823 | icu::ByteSink &sink, icu::Edits *edits, |
824 | UErrorCode &errorCode) { |
825 | if (caseLocale == UCASE_LOC_GREEK) { |
826 | GreekUpper::toUpper(options, src, srcLength, sink, edits, errorCode); |
827 | } else { |
828 | UCaseContext csc=UCASECONTEXT_INITIALIZER; |
829 | csc.p=(void *)src; |
830 | csc.limit=srcLength; |
831 | toUpper( |
832 | caseLocale, options, |
833 | src, &csc, srcLength, |
834 | sink, edits, errorCode); |
835 | } |
836 | } |
837 | |
838 | static void U_CALLCONV |
839 | ucasemap_internalUTF8Fold(int32_t /* caseLocale */, uint32_t options, UCASEMAP_BREAK_ITERATOR_UNUSED |
840 | const uint8_t *src, int32_t srcLength, |
841 | icu::ByteSink &sink, icu::Edits *edits, |
842 | UErrorCode &errorCode) { |
843 | toLower( |
844 | -1, options, |
845 | src, nullptr, 0, srcLength, |
846 | sink, edits, errorCode); |
847 | } |
848 | |
849 | void |
850 | ucasemap_mapUTF8(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM |
851 | const char *src, int32_t srcLength, |
852 | UTF8CaseMapper *stringCaseMapper, |
853 | icu::ByteSink &sink, icu::Edits *edits, |
854 | UErrorCode &errorCode) { |
855 | /* check argument values */ |
856 | if (U_FAILURE(errorCode)) { |
857 | return; |
858 | } |
859 | if ((src == nullptr && srcLength != 0) || srcLength < -1) { |
860 | errorCode = U_ILLEGAL_ARGUMENT_ERROR; |
861 | return; |
862 | } |
863 | |
864 | // Get the string length. |
865 | if (srcLength == -1) { |
866 | srcLength = (int32_t)uprv_strlen((const char *)src); |
867 | } |
868 | |
869 | if (edits != nullptr && (options & U_EDITS_NO_RESET) == 0) { |
870 | edits->reset(); |
871 | } |
872 | stringCaseMapper(caseLocale, options, UCASEMAP_BREAK_ITERATOR |
873 | (const uint8_t *)src, srcLength, sink, edits, errorCode); |
874 | sink.Flush(); |
875 | if (U_SUCCESS(errorCode)) { |
876 | if (edits != nullptr) { |
877 | edits->copyErrorTo(errorCode); |
878 | } |
879 | } |
880 | } |
881 | |
882 | int32_t |
883 | ucasemap_mapUTF8(int32_t caseLocale, uint32_t options, UCASEMAP_BREAK_ITERATOR_PARAM |
884 | char *dest, int32_t destCapacity, |
885 | const char *src, int32_t srcLength, |
886 | UTF8CaseMapper *stringCaseMapper, |
887 | icu::Edits *edits, |
888 | UErrorCode &errorCode) { |
889 | /* check argument values */ |
890 | if(U_FAILURE(errorCode)) { |
891 | return 0; |
892 | } |
893 | if( destCapacity<0 || |
894 | (dest==nullptr && destCapacity>0) || |
895 | (src==nullptr && srcLength!=0) || srcLength<-1 |
896 | ) { |
897 | errorCode=U_ILLEGAL_ARGUMENT_ERROR; |
898 | return 0; |
899 | } |
900 | |
901 | /* get the string length */ |
902 | if(srcLength==-1) { |
903 | srcLength=(int32_t)uprv_strlen((const char *)src); |
904 | } |
905 | |
906 | /* check for overlapping source and destination */ |
907 | if( dest!=nullptr && |
908 | ((src>=dest && src<(dest+destCapacity)) || |
909 | (dest>=src && dest<(src+srcLength))) |
910 | ) { |
911 | errorCode=U_ILLEGAL_ARGUMENT_ERROR; |
912 | return 0; |
913 | } |
914 | |
915 | CheckedArrayByteSink sink(dest, destCapacity); |
916 | if (edits != nullptr && (options & U_EDITS_NO_RESET) == 0) { |
917 | edits->reset(); |
918 | } |
919 | stringCaseMapper(caseLocale, options, UCASEMAP_BREAK_ITERATOR |
920 | (const uint8_t *)src, srcLength, sink, edits, errorCode); |
921 | sink.Flush(); |
922 | if (U_SUCCESS(errorCode)) { |
923 | if (sink.Overflowed()) { |
924 | errorCode = U_BUFFER_OVERFLOW_ERROR; |
925 | } else if (edits != nullptr) { |
926 | edits->copyErrorTo(errorCode); |
927 | } |
928 | } |
929 | return u_terminateChars(dest, destCapacity, sink.NumberOfBytesAppended(), &errorCode); |
930 | } |
931 | |
932 | /* public API functions */ |
933 | |
934 | U_CAPI int32_t U_EXPORT2 |
935 | ucasemap_utf8ToLower(const UCaseMap *csm, |
936 | char *dest, int32_t destCapacity, |
937 | const char *src, int32_t srcLength, |
938 | UErrorCode *pErrorCode) { |
939 | return ucasemap_mapUTF8( |
940 | csm->caseLocale, csm->options, UCASEMAP_BREAK_ITERATOR_NULL |
941 | dest, destCapacity, |
942 | src, srcLength, |
943 | ucasemap_internalUTF8ToLower, nullptr, *pErrorCode); |
944 | } |
945 | |
946 | U_CAPI int32_t U_EXPORT2 |
947 | ucasemap_utf8ToUpper(const UCaseMap *csm, |
948 | char *dest, int32_t destCapacity, |
949 | const char *src, int32_t srcLength, |
950 | UErrorCode *pErrorCode) { |
951 | return ucasemap_mapUTF8( |
952 | csm->caseLocale, csm->options, UCASEMAP_BREAK_ITERATOR_NULL |
953 | dest, destCapacity, |
954 | src, srcLength, |
955 | ucasemap_internalUTF8ToUpper, nullptr, *pErrorCode); |
956 | } |
957 | |
958 | U_CAPI int32_t U_EXPORT2 |
959 | ucasemap_utf8FoldCase(const UCaseMap *csm, |
960 | char *dest, int32_t destCapacity, |
961 | const char *src, int32_t srcLength, |
962 | UErrorCode *pErrorCode) { |
963 | return ucasemap_mapUTF8( |
964 | UCASE_LOC_ROOT, csm->options, UCASEMAP_BREAK_ITERATOR_NULL |
965 | dest, destCapacity, |
966 | src, srcLength, |
967 | ucasemap_internalUTF8Fold, nullptr, *pErrorCode); |
968 | } |
969 | |
970 | U_NAMESPACE_BEGIN |
971 | |
972 | void CaseMap::utf8ToLower( |
973 | const char *locale, uint32_t options, |
974 | StringPiece src, ByteSink &sink, Edits *edits, |
975 | UErrorCode &errorCode) { |
976 | ucasemap_mapUTF8( |
977 | ustrcase_getCaseLocale(locale), options, UCASEMAP_BREAK_ITERATOR_NULL |
978 | src.data(), src.length(), |
979 | ucasemap_internalUTF8ToLower, sink, edits, errorCode); |
980 | } |
981 | |
982 | void CaseMap::utf8ToUpper( |
983 | const char *locale, uint32_t options, |
984 | StringPiece src, ByteSink &sink, Edits *edits, |
985 | UErrorCode &errorCode) { |
986 | ucasemap_mapUTF8( |
987 | ustrcase_getCaseLocale(locale), options, UCASEMAP_BREAK_ITERATOR_NULL |
988 | src.data(), src.length(), |
989 | ucasemap_internalUTF8ToUpper, sink, edits, errorCode); |
990 | } |
991 | |
992 | void CaseMap::utf8Fold( |
993 | uint32_t options, |
994 | StringPiece src, ByteSink &sink, Edits *edits, |
995 | UErrorCode &errorCode) { |
996 | ucasemap_mapUTF8( |
997 | UCASE_LOC_ROOT, options, UCASEMAP_BREAK_ITERATOR_NULL |
998 | src.data(), src.length(), |
999 | ucasemap_internalUTF8Fold, sink, edits, errorCode); |
1000 | } |
1001 | |
1002 | int32_t CaseMap::utf8ToLower( |
1003 | const char *locale, uint32_t options, |
1004 | const char *src, int32_t srcLength, |
1005 | char *dest, int32_t destCapacity, Edits *edits, |
1006 | UErrorCode &errorCode) { |
1007 | return ucasemap_mapUTF8( |
1008 | ustrcase_getCaseLocale(locale), options, UCASEMAP_BREAK_ITERATOR_NULL |
1009 | dest, destCapacity, |
1010 | src, srcLength, |
1011 | ucasemap_internalUTF8ToLower, edits, errorCode); |
1012 | } |
1013 | |
1014 | int32_t CaseMap::utf8ToUpper( |
1015 | const char *locale, uint32_t options, |
1016 | const char *src, int32_t srcLength, |
1017 | char *dest, int32_t destCapacity, Edits *edits, |
1018 | UErrorCode &errorCode) { |
1019 | return ucasemap_mapUTF8( |
1020 | ustrcase_getCaseLocale(locale), options, UCASEMAP_BREAK_ITERATOR_NULL |
1021 | dest, destCapacity, |
1022 | src, srcLength, |
1023 | ucasemap_internalUTF8ToUpper, edits, errorCode); |
1024 | } |
1025 | |
1026 | int32_t CaseMap::utf8Fold( |
1027 | uint32_t options, |
1028 | const char *src, int32_t srcLength, |
1029 | char *dest, int32_t destCapacity, Edits *edits, |
1030 | UErrorCode &errorCode) { |
1031 | return ucasemap_mapUTF8( |
1032 | UCASE_LOC_ROOT, options, UCASEMAP_BREAK_ITERATOR_NULL |
1033 | dest, destCapacity, |
1034 | src, srcLength, |
1035 | ucasemap_internalUTF8Fold, edits, errorCode); |
1036 | } |
1037 | |
1038 | U_NAMESPACE_END |
1039 | |