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) 2004-2014, International Business Machines |
7 | * Corporation and others. All Rights Reserved. |
8 | * |
9 | ******************************************************************************* |
10 | * file name: ucase.cpp |
11 | * encoding: UTF-8 |
12 | * tab size: 8 (not used) |
13 | * indentation:4 |
14 | * |
15 | * created on: 2004aug30 |
16 | * created by: Markus W. Scherer |
17 | * |
18 | * Low-level Unicode character/string case mapping code. |
19 | * Much code moved here (and modified) from uchar.c. |
20 | */ |
21 | |
22 | #include "unicode/utypes.h" |
23 | #include "unicode/unistr.h" |
24 | #include "unicode/uset.h" |
25 | #include "unicode/utf16.h" |
26 | #include "cmemory.h" |
27 | #include "uassert.h" |
28 | #include "ucase.h" |
29 | #include "umutex.h" |
30 | #include "utrie2.h" |
31 | |
32 | /* ucase_props_data.h is machine-generated by genprops/casepropsbuilder.cpp */ |
33 | #define INCLUDED_FROM_UCASE_CPP |
34 | #include "ucase_props_data.h" |
35 | |
36 | /* set of property starts for UnicodeSet ------------------------------------ */ |
37 | |
38 | static UBool U_CALLCONV |
39 | _enumPropertyStartsRange(const void *context, UChar32 start, UChar32 /*end*/, uint32_t /*value*/) { |
40 | /* add the start code point to the USet */ |
41 | const USetAdder *sa=(const USetAdder *)context; |
42 | sa->add(sa->set, start); |
43 | return true; |
44 | } |
45 | |
46 | U_CFUNC void U_EXPORT2 |
47 | ucase_addPropertyStarts(const USetAdder *sa, UErrorCode *pErrorCode) { |
48 | if(U_FAILURE(*pErrorCode)) { |
49 | return; |
50 | } |
51 | |
52 | /* add the start code point of each same-value range of the trie */ |
53 | utrie2_enum(&ucase_props_singleton.trie, nullptr, _enumPropertyStartsRange, sa); |
54 | |
55 | /* add code points with hardcoded properties, plus the ones following them */ |
56 | |
57 | /* (none right now, see comment below) */ |
58 | |
59 | /* |
60 | * Omit code points with hardcoded specialcasing properties |
61 | * because we do not build property UnicodeSets for them right now. |
62 | */ |
63 | } |
64 | |
65 | /* data access primitives --------------------------------------------------- */ |
66 | |
67 | U_CAPI const struct UCaseProps * U_EXPORT2 |
68 | ucase_getSingleton(int32_t *pExceptionsLength, int32_t *pUnfoldLength) { |
69 | *pExceptionsLength = UPRV_LENGTHOF(ucase_props_exceptions); |
70 | *pUnfoldLength = UPRV_LENGTHOF(ucase_props_unfold); |
71 | return &ucase_props_singleton; |
72 | } |
73 | |
74 | U_CFUNC const UTrie2 * U_EXPORT2 |
75 | ucase_getTrie() { |
76 | return &ucase_props_singleton.trie; |
77 | } |
78 | |
79 | #define GET_EXCEPTIONS(csp, props) ((csp)->exceptions+((props)>>UCASE_EXC_SHIFT)) |
80 | |
81 | /* number of bits in an 8-bit integer value */ |
82 | static const uint8_t flagsOffset[256]={ |
83 | 0, 1, 1, 2, 1, 2, 2, 3, 1, 2, 2, 3, 2, 3, 3, 4, |
84 | 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, |
85 | 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, |
86 | 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, |
87 | 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, |
88 | 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, |
89 | 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, |
90 | 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, |
91 | 1, 2, 2, 3, 2, 3, 3, 4, 2, 3, 3, 4, 3, 4, 4, 5, |
92 | 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, |
93 | 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, |
94 | 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, |
95 | 2, 3, 3, 4, 3, 4, 4, 5, 3, 4, 4, 5, 4, 5, 5, 6, |
96 | 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, |
97 | 3, 4, 4, 5, 4, 5, 5, 6, 4, 5, 5, 6, 5, 6, 6, 7, |
98 | 4, 5, 5, 6, 5, 6, 6, 7, 5, 6, 6, 7, 6, 7, 7, 8 |
99 | }; |
100 | |
101 | #define HAS_SLOT(flags, idx) ((flags)&(1<<(idx))) |
102 | #define SLOT_OFFSET(flags, idx) flagsOffset[(flags)&((1<<(idx))-1)] |
103 | |
104 | /* |
105 | * Get the value of an optional-value slot where HAS_SLOT(excWord, idx). |
106 | * |
107 | * @param excWord (in) initial exceptions word |
108 | * @param idx (in) desired slot index |
109 | * @param pExc16 (in/out) const uint16_t * after excWord=*pExc16++; |
110 | * moved to the last uint16_t of the value, use +1 for beginning of next slot |
111 | * @param value (out) int32_t or uint32_t output if hasSlot, otherwise not modified |
112 | */ |
113 | #define GET_SLOT_VALUE(excWord, idx, pExc16, value) UPRV_BLOCK_MACRO_BEGIN { \ |
114 | if(((excWord)&UCASE_EXC_DOUBLE_SLOTS)==0) { \ |
115 | (pExc16)+=SLOT_OFFSET(excWord, idx); \ |
116 | (value)=*pExc16; \ |
117 | } else { \ |
118 | (pExc16)+=2*SLOT_OFFSET(excWord, idx); \ |
119 | (value)=*pExc16++; \ |
120 | (value)=((value)<<16)|*pExc16; \ |
121 | } \ |
122 | } UPRV_BLOCK_MACRO_END |
123 | |
124 | /* simple case mappings ----------------------------------------------------- */ |
125 | |
126 | U_CAPI UChar32 U_EXPORT2 |
127 | ucase_tolower(UChar32 c) { |
128 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
129 | if(!UCASE_HAS_EXCEPTION(props)) { |
130 | if(UCASE_IS_UPPER_OR_TITLE(props)) { |
131 | c+=UCASE_GET_DELTA(props); |
132 | } |
133 | } else { |
134 | const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); |
135 | uint16_t excWord=*pe++; |
136 | if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_IS_UPPER_OR_TITLE(props)) { |
137 | int32_t delta; |
138 | GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); |
139 | return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; |
140 | } |
141 | if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { |
142 | GET_SLOT_VALUE(excWord, UCASE_EXC_LOWER, pe, c); |
143 | } |
144 | } |
145 | return c; |
146 | } |
147 | |
148 | U_CAPI UChar32 U_EXPORT2 |
149 | ucase_toupper(UChar32 c) { |
150 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
151 | if(!UCASE_HAS_EXCEPTION(props)) { |
152 | if(UCASE_GET_TYPE(props)==UCASE_LOWER) { |
153 | c+=UCASE_GET_DELTA(props); |
154 | } |
155 | } else { |
156 | const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); |
157 | uint16_t excWord=*pe++; |
158 | if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_GET_TYPE(props)==UCASE_LOWER) { |
159 | int32_t delta; |
160 | GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); |
161 | return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; |
162 | } |
163 | if(HAS_SLOT(excWord, UCASE_EXC_UPPER)) { |
164 | GET_SLOT_VALUE(excWord, UCASE_EXC_UPPER, pe, c); |
165 | } |
166 | } |
167 | return c; |
168 | } |
169 | |
170 | U_CAPI UChar32 U_EXPORT2 |
171 | ucase_totitle(UChar32 c) { |
172 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
173 | if(!UCASE_HAS_EXCEPTION(props)) { |
174 | if(UCASE_GET_TYPE(props)==UCASE_LOWER) { |
175 | c+=UCASE_GET_DELTA(props); |
176 | } |
177 | } else { |
178 | const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); |
179 | uint16_t excWord=*pe++; |
180 | if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_GET_TYPE(props)==UCASE_LOWER) { |
181 | int32_t delta; |
182 | GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); |
183 | return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; |
184 | } |
185 | int32_t idx; |
186 | if(HAS_SLOT(excWord, UCASE_EXC_TITLE)) { |
187 | idx=UCASE_EXC_TITLE; |
188 | } else if(HAS_SLOT(excWord, UCASE_EXC_UPPER)) { |
189 | idx=UCASE_EXC_UPPER; |
190 | } else { |
191 | return c; |
192 | } |
193 | GET_SLOT_VALUE(excWord, idx, pe, c); |
194 | } |
195 | return c; |
196 | } |
197 | |
198 | static const char16_t iDot[2] = { 0x69, 0x307 }; |
199 | static const char16_t jDot[2] = { 0x6a, 0x307 }; |
200 | static const char16_t iOgonekDot[3] = { 0x12f, 0x307 }; |
201 | static const char16_t iDotGrave[3] = { 0x69, 0x307, 0x300 }; |
202 | static const char16_t iDotAcute[3] = { 0x69, 0x307, 0x301 }; |
203 | static const char16_t iDotTilde[3] = { 0x69, 0x307, 0x303 }; |
204 | |
205 | |
206 | U_CFUNC void U_EXPORT2 |
207 | ucase_addCaseClosure(UChar32 c, const USetAdder *sa) { |
208 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
209 | if(!UCASE_HAS_EXCEPTION(props)) { |
210 | if(UCASE_GET_TYPE(props)!=UCASE_NONE) { |
211 | /* add the one simple case mapping, no matter what type it is */ |
212 | int32_t delta=UCASE_GET_DELTA(props); |
213 | if(delta!=0) { |
214 | sa->add(sa->set, c+delta); |
215 | } |
216 | } |
217 | } else { |
218 | /* |
219 | * c has exceptions, so there may be multiple simple and/or |
220 | * full case mappings. Add them all. |
221 | */ |
222 | const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); |
223 | uint16_t excWord=*pe++; |
224 | const uint16_t *pe0=pe; |
225 | |
226 | // Hardcode the case closure of i and its relatives and ignore the |
227 | // data file data for these characters. |
228 | // The Turkic dotless i and dotted I with their case mapping conditions |
229 | // and case folding option make the related characters behave specially. |
230 | // This code matches their closure behavior to their case folding behavior. |
231 | if (excWord&UCASE_EXC_CONDITIONAL_FOLD) { |
232 | // These characters have Turkic case foldings. Hardcode their closure. |
233 | if (c == 0x49) { |
234 | // Regular i and I are in one equivalence class. |
235 | sa->add(sa->set, 0x69); |
236 | return; |
237 | } else if (c == 0x130) { |
238 | // Dotted I is in a class with <0069 0307> |
239 | // (for canonical equivalence with <0049 0307>). |
240 | sa->addString(sa->set, iDot, 2); |
241 | return; |
242 | } |
243 | } else if (c == 0x69) { |
244 | sa->add(sa->set, 0x49); |
245 | return; |
246 | } else if (c == 0x131) { |
247 | // Dotless i is in a class by itself. |
248 | return; |
249 | } |
250 | |
251 | /* add all simple case mappings */ |
252 | for(int32_t idx=UCASE_EXC_LOWER; idx<=UCASE_EXC_TITLE; ++idx) { |
253 | if(HAS_SLOT(excWord, idx)) { |
254 | pe=pe0; |
255 | UChar32 mapping; |
256 | GET_SLOT_VALUE(excWord, idx, pe, mapping); |
257 | sa->add(sa->set, mapping); |
258 | } |
259 | } |
260 | if(HAS_SLOT(excWord, UCASE_EXC_DELTA)) { |
261 | pe=pe0; |
262 | int32_t delta; |
263 | GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); |
264 | sa->add(sa->set, (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta); |
265 | } |
266 | |
267 | /* get the closure string pointer & length */ |
268 | const char16_t *closure; |
269 | int32_t closureLength; |
270 | if(HAS_SLOT(excWord, UCASE_EXC_CLOSURE)) { |
271 | pe=pe0; |
272 | GET_SLOT_VALUE(excWord, UCASE_EXC_CLOSURE, pe, closureLength); |
273 | closureLength&=UCASE_CLOSURE_MAX_LENGTH; /* higher bits are reserved */ |
274 | closure=(const char16_t *)pe+1; /* behind this slot, unless there are full case mappings */ |
275 | } else { |
276 | closureLength=0; |
277 | closure=nullptr; |
278 | } |
279 | |
280 | /* add the full case folding */ |
281 | if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { |
282 | pe=pe0; |
283 | int32_t fullLength; |
284 | GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, fullLength); |
285 | |
286 | /* start of full case mapping strings */ |
287 | ++pe; |
288 | |
289 | fullLength&=0xffff; /* bits 16 and higher are reserved */ |
290 | |
291 | /* skip the lowercase result string */ |
292 | pe+=fullLength&UCASE_FULL_LOWER; |
293 | fullLength>>=4; |
294 | |
295 | /* add the full case folding string */ |
296 | int32_t length=fullLength&0xf; |
297 | if(length!=0) { |
298 | sa->addString(sa->set, (const char16_t *)pe, length); |
299 | pe+=length; |
300 | } |
301 | |
302 | /* skip the uppercase and titlecase strings */ |
303 | fullLength>>=4; |
304 | pe+=fullLength&0xf; |
305 | fullLength>>=4; |
306 | pe+=fullLength; |
307 | |
308 | closure=(const char16_t *)pe; /* behind full case mappings */ |
309 | } |
310 | |
311 | /* add each code point in the closure string */ |
312 | for(int32_t idx=0; idx<closureLength;) { |
313 | UChar32 mapping; |
314 | U16_NEXT_UNSAFE(closure, idx, mapping); |
315 | sa->add(sa->set, mapping); |
316 | } |
317 | } |
318 | } |
319 | |
320 | namespace { |
321 | |
322 | /** |
323 | * Add the simple case closure mapping, |
324 | * except if there is not actually an scf relationship between the two characters. |
325 | * TODO: Unicode should probably add the corresponding scf mappings. |
326 | * See https://crbug.com/v8/13377 and Unicode-internal PAG issue #23. |
327 | * If & when those scf mappings are added, we should be able to remove all of these exceptions. |
328 | */ |
329 | void addOneSimpleCaseClosure(UChar32 c, UChar32 t, const USetAdder *sa) { |
330 | switch (c) { |
331 | case 0x0390: |
332 | if (t == 0x1FD3) { return; } |
333 | break; |
334 | case 0x03B0: |
335 | if (t == 0x1FE3) { return; } |
336 | break; |
337 | case 0x1FD3: |
338 | if (t == 0x0390) { return; } |
339 | break; |
340 | case 0x1FE3: |
341 | if (t == 0x03B0) { return; } |
342 | break; |
343 | case 0xFB05: |
344 | if (t == 0xFB06) { return; } |
345 | break; |
346 | case 0xFB06: |
347 | if (t == 0xFB05) { return; } |
348 | break; |
349 | default: |
350 | break; |
351 | } |
352 | sa->add(sa->set, t); |
353 | } |
354 | |
355 | } // namespace |
356 | |
357 | U_CFUNC void U_EXPORT2 |
358 | ucase_addSimpleCaseClosure(UChar32 c, const USetAdder *sa) { |
359 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
360 | if(!UCASE_HAS_EXCEPTION(props)) { |
361 | if(UCASE_GET_TYPE(props)!=UCASE_NONE) { |
362 | /* add the one simple case mapping, no matter what type it is */ |
363 | int32_t delta=UCASE_GET_DELTA(props); |
364 | if(delta!=0) { |
365 | sa->add(sa->set, c+delta); |
366 | } |
367 | } |
368 | } else { |
369 | // c has exceptions. Add the mappings relevant for scf=Simple_Case_Folding. |
370 | const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); |
371 | uint16_t excWord=*pe++; |
372 | const uint16_t *pe0=pe; |
373 | |
374 | // Hardcode the case closure of i and its relatives and ignore the |
375 | // data file data for these characters, like in ucase_addCaseClosure(). |
376 | if (excWord&UCASE_EXC_CONDITIONAL_FOLD) { |
377 | // These characters have Turkic case foldings. Hardcode their closure. |
378 | if (c == 0x49) { |
379 | // Regular i and I are in one equivalence class. |
380 | sa->add(sa->set, 0x69); |
381 | return; |
382 | } else if (c == 0x130) { |
383 | // For scf=Simple_Case_Folding, dotted I is in a class by itself. |
384 | return; |
385 | } |
386 | } else if (c == 0x69) { |
387 | sa->add(sa->set, 0x49); |
388 | return; |
389 | } else if (c == 0x131) { |
390 | // Dotless i is in a class by itself. |
391 | return; |
392 | } |
393 | |
394 | // Add all simple case mappings. |
395 | for(int32_t idx=UCASE_EXC_LOWER; idx<=UCASE_EXC_TITLE; ++idx) { |
396 | if(HAS_SLOT(excWord, idx)) { |
397 | pe=pe0; |
398 | UChar32 mapping; |
399 | GET_SLOT_VALUE(excWord, idx, pe, mapping); |
400 | addOneSimpleCaseClosure(c, mapping, sa); |
401 | } |
402 | } |
403 | if(HAS_SLOT(excWord, UCASE_EXC_DELTA)) { |
404 | pe=pe0; |
405 | int32_t delta; |
406 | GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); |
407 | UChar32 mapping = (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; |
408 | addOneSimpleCaseClosure(c, mapping, sa); |
409 | } |
410 | |
411 | /* get the closure string pointer & length */ |
412 | const char16_t *closure; |
413 | int32_t closureLength; |
414 | if(HAS_SLOT(excWord, UCASE_EXC_CLOSURE)) { |
415 | pe=pe0; |
416 | GET_SLOT_VALUE(excWord, UCASE_EXC_CLOSURE, pe, closureLength); |
417 | closureLength&=UCASE_CLOSURE_MAX_LENGTH; /* higher bits are reserved */ |
418 | closure=(const char16_t *)pe+1; /* behind this slot, unless there are full case mappings */ |
419 | } else { |
420 | closureLength=0; |
421 | closure=nullptr; |
422 | } |
423 | |
424 | // Skip the full case mappings. |
425 | if(closureLength > 0 && HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { |
426 | pe=pe0; |
427 | int32_t fullLength; |
428 | GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, fullLength); |
429 | |
430 | /* start of full case mapping strings */ |
431 | ++pe; |
432 | |
433 | fullLength&=0xffff; /* bits 16 and higher are reserved */ |
434 | |
435 | // Skip all 4 full case mappings. |
436 | pe+=fullLength&UCASE_FULL_LOWER; |
437 | fullLength>>=4; |
438 | pe+=fullLength&0xf; |
439 | fullLength>>=4; |
440 | pe+=fullLength&0xf; |
441 | fullLength>>=4; |
442 | pe+=fullLength; |
443 | |
444 | closure=(const char16_t *)pe; /* behind full case mappings */ |
445 | } |
446 | |
447 | // Add each code point in the closure string whose scf maps back to c. |
448 | for(int32_t idx=0; idx<closureLength;) { |
449 | UChar32 mapping; |
450 | U16_NEXT_UNSAFE(closure, idx, mapping); |
451 | addOneSimpleCaseClosure(c, mapping, sa); |
452 | } |
453 | } |
454 | } |
455 | |
456 | /* |
457 | * compare s, which has a length, with t, which has a maximum length or is NUL-terminated |
458 | * must be length>0 and max>0 and length<=max |
459 | */ |
460 | static inline int32_t |
461 | strcmpMax(const char16_t *s, int32_t length, const char16_t *t, int32_t max) { |
462 | int32_t c1, c2; |
463 | |
464 | max-=length; /* we require length<=max, so no need to decrement max in the loop */ |
465 | do { |
466 | c1=*s++; |
467 | c2=*t++; |
468 | if(c2==0) { |
469 | return 1; /* reached the end of t but not of s */ |
470 | } |
471 | c1-=c2; |
472 | if(c1!=0) { |
473 | return c1; /* return difference result */ |
474 | } |
475 | } while(--length>0); |
476 | /* ends with length==0 */ |
477 | |
478 | if(max==0 || *t==0) { |
479 | return 0; /* equal to length of both strings */ |
480 | } else { |
481 | return -max; /* return length difference */ |
482 | } |
483 | } |
484 | |
485 | U_CFUNC UBool U_EXPORT2 |
486 | ucase_addStringCaseClosure(const char16_t *s, int32_t length, const USetAdder *sa) { |
487 | int32_t i, start, limit, result, unfoldRows, unfoldRowWidth, unfoldStringWidth; |
488 | |
489 | if(ucase_props_singleton.unfold==nullptr || s==nullptr) { |
490 | return false; /* no reverse case folding data, or no string */ |
491 | } |
492 | if(length<=1) { |
493 | /* the string is too short to find any match */ |
494 | /* |
495 | * more precise would be: |
496 | * if(!u_strHasMoreChar32Than(s, length, 1)) |
497 | * but this does not make much practical difference because |
498 | * a single supplementary code point would just not be found |
499 | */ |
500 | return false; |
501 | } |
502 | |
503 | const uint16_t *unfold=ucase_props_singleton.unfold; |
504 | unfoldRows=unfold[UCASE_UNFOLD_ROWS]; |
505 | unfoldRowWidth=unfold[UCASE_UNFOLD_ROW_WIDTH]; |
506 | unfoldStringWidth=unfold[UCASE_UNFOLD_STRING_WIDTH]; |
507 | unfold+=unfoldRowWidth; |
508 | |
509 | if(length>unfoldStringWidth) { |
510 | /* the string is too long to find any match */ |
511 | return false; |
512 | } |
513 | |
514 | /* do a binary search for the string */ |
515 | start=0; |
516 | limit=unfoldRows; |
517 | while(start<limit) { |
518 | i=(start+limit)/2; |
519 | const char16_t *p=reinterpret_cast<const char16_t *>(unfold+(i*unfoldRowWidth)); |
520 | result=strcmpMax(s, length, p, unfoldStringWidth); |
521 | |
522 | if(result==0) { |
523 | /* found the string: add each code point, and its case closure */ |
524 | UChar32 c; |
525 | |
526 | for(i=unfoldStringWidth; i<unfoldRowWidth && p[i]!=0;) { |
527 | U16_NEXT_UNSAFE(p, i, c); |
528 | sa->add(sa->set, c); |
529 | ucase_addCaseClosure(c, sa); |
530 | } |
531 | return true; |
532 | } else if(result<0) { |
533 | limit=i; |
534 | } else /* result>0 */ { |
535 | start=i+1; |
536 | } |
537 | } |
538 | |
539 | return false; /* string not found */ |
540 | } |
541 | |
542 | U_NAMESPACE_BEGIN |
543 | |
544 | FullCaseFoldingIterator::FullCaseFoldingIterator() |
545 | : unfold(reinterpret_cast<const char16_t *>(ucase_props_singleton.unfold)), |
546 | unfoldRows(unfold[UCASE_UNFOLD_ROWS]), |
547 | unfoldRowWidth(unfold[UCASE_UNFOLD_ROW_WIDTH]), |
548 | unfoldStringWidth(unfold[UCASE_UNFOLD_STRING_WIDTH]), |
549 | currentRow(0), |
550 | rowCpIndex(unfoldStringWidth) { |
551 | unfold+=unfoldRowWidth; |
552 | } |
553 | |
554 | UChar32 |
555 | FullCaseFoldingIterator::next(UnicodeString &full) { |
556 | // Advance past the last-delivered code point. |
557 | const char16_t *p=unfold+(currentRow*unfoldRowWidth); |
558 | if(rowCpIndex>=unfoldRowWidth || p[rowCpIndex]==0) { |
559 | ++currentRow; |
560 | p+=unfoldRowWidth; |
561 | rowCpIndex=unfoldStringWidth; |
562 | } |
563 | if(currentRow>=unfoldRows) { return U_SENTINEL; } |
564 | // Set "full" to the NUL-terminated string in the first unfold column. |
565 | int32_t length=unfoldStringWidth; |
566 | while(length>0 && p[length-1]==0) { --length; } |
567 | full.setTo(false, p, length); |
568 | // Return the code point. |
569 | UChar32 c; |
570 | U16_NEXT_UNSAFE(p, rowCpIndex, c); |
571 | return c; |
572 | } |
573 | |
574 | namespace LatinCase { |
575 | |
576 | const int8_t TO_LOWER_NORMAL[LIMIT] = { |
577 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
578 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
579 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
580 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
581 | |
582 | 0, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, |
583 | 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 0, 0, 0, 0, 0, |
584 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
585 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
586 | |
587 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
588 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
589 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
590 | 0, 0, 0, 0, 0, EXC, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
591 | |
592 | 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, |
593 | 32, 32, 32, 32, 32, 32, 32, 0, 32, 32, 32, 32, 32, 32, 32, EXC, |
594 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
595 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
596 | |
597 | 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, |
598 | 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, |
599 | 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, |
600 | EXC, 0, 1, 0, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 1, |
601 | |
602 | 0, 1, 0, 1, 0, 1, 0, 1, 0, EXC, 1, 0, 1, 0, 1, 0, |
603 | 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, |
604 | 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, |
605 | 1, 0, 1, 0, 1, 0, 1, 0, -121, 1, 0, 1, 0, 1, 0, EXC |
606 | }; |
607 | |
608 | const int8_t TO_LOWER_TR_LT[LIMIT] = { |
609 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
610 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
611 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
612 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
613 | |
614 | 0, 32, 32, 32, 32, 32, 32, 32, 32, EXC, EXC, 32, 32, 32, 32, 32, |
615 | 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 0, 0, 0, 0, 0, |
616 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
617 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
618 | |
619 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
620 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
621 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
622 | 0, 0, 0, 0, 0, EXC, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
623 | |
624 | 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, 32, EXC, EXC, 32, 32, |
625 | 32, 32, 32, 32, 32, 32, 32, 0, 32, 32, 32, 32, 32, 32, 32, EXC, |
626 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
627 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
628 | |
629 | 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, |
630 | 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, |
631 | 1, 0, 1, 0, 1, 0, 1, 0, EXC, 0, 1, 0, 1, 0, EXC, 0, |
632 | EXC, 0, 1, 0, 1, 0, 1, 0, 0, 1, 0, 1, 0, 1, 0, 1, |
633 | |
634 | 0, 1, 0, 1, 0, 1, 0, 1, 0, EXC, 1, 0, 1, 0, 1, 0, |
635 | 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, |
636 | 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, |
637 | 1, 0, 1, 0, 1, 0, 1, 0, -121, 1, 0, 1, 0, 1, 0, EXC |
638 | }; |
639 | |
640 | const int8_t TO_UPPER_NORMAL[LIMIT] = { |
641 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
642 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
643 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
644 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
645 | |
646 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
647 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
648 | 0, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, |
649 | -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, 0, 0, 0, 0, 0, |
650 | |
651 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
652 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
653 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
654 | 0, 0, 0, 0, 0, EXC, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
655 | |
656 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
657 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, EXC, |
658 | -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, |
659 | -32, -32, -32, -32, -32, -32, -32, 0, -32, -32, -32, -32, -32, -32, -32, 121, |
660 | |
661 | 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, |
662 | 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, |
663 | 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, |
664 | 0, EXC, 0, -1, 0, -1, 0, -1, 0, 0, -1, 0, -1, 0, -1, 0, |
665 | |
666 | -1, 0, -1, 0, -1, 0, -1, 0, -1, EXC, 0, -1, 0, -1, 0, -1, |
667 | 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, |
668 | 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, |
669 | 0, -1, 0, -1, 0, -1, 0, -1, 0, 0, -1, 0, -1, 0, -1, EXC |
670 | }; |
671 | |
672 | const int8_t TO_UPPER_TR[LIMIT] = { |
673 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
674 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
675 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
676 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
677 | |
678 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
679 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
680 | 0, -32, -32, -32, -32, -32, -32, -32, -32, EXC, -32, -32, -32, -32, -32, -32, |
681 | -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, 0, 0, 0, 0, 0, |
682 | |
683 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
684 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
685 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
686 | 0, 0, 0, 0, 0, EXC, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
687 | |
688 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, |
689 | 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, EXC, |
690 | -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, -32, |
691 | -32, -32, -32, -32, -32, -32, -32, 0, -32, -32, -32, -32, -32, -32, -32, 121, |
692 | |
693 | 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, |
694 | 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, |
695 | 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, |
696 | 0, EXC, 0, -1, 0, -1, 0, -1, 0, 0, -1, 0, -1, 0, -1, 0, |
697 | |
698 | -1, 0, -1, 0, -1, 0, -1, 0, -1, EXC, 0, -1, 0, -1, 0, -1, |
699 | 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, |
700 | 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, 0, -1, |
701 | 0, -1, 0, -1, 0, -1, 0, -1, 0, 0, -1, 0, -1, 0, -1, EXC |
702 | }; |
703 | |
704 | } // namespace LatinCase |
705 | |
706 | U_NAMESPACE_END |
707 | |
708 | /** @return UCASE_NONE, UCASE_LOWER, UCASE_UPPER, UCASE_TITLE */ |
709 | U_CAPI int32_t U_EXPORT2 |
710 | ucase_getType(UChar32 c) { |
711 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
712 | return UCASE_GET_TYPE(props); |
713 | } |
714 | |
715 | /** @return same as ucase_getType() and set bit 2 if c is case-ignorable */ |
716 | U_CAPI int32_t U_EXPORT2 |
717 | ucase_getTypeOrIgnorable(UChar32 c) { |
718 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
719 | return UCASE_GET_TYPE_AND_IGNORABLE(props); |
720 | } |
721 | |
722 | /** @return UCASE_NO_DOT, UCASE_SOFT_DOTTED, UCASE_ABOVE, UCASE_OTHER_ACCENT */ |
723 | static inline int32_t |
724 | getDotType(UChar32 c) { |
725 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
726 | if(!UCASE_HAS_EXCEPTION(props)) { |
727 | return props&UCASE_DOT_MASK; |
728 | } else { |
729 | const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); |
730 | return (*pe>>UCASE_EXC_DOT_SHIFT)&UCASE_DOT_MASK; |
731 | } |
732 | } |
733 | |
734 | U_CAPI UBool U_EXPORT2 |
735 | ucase_isSoftDotted(UChar32 c) { |
736 | return (UBool)(getDotType(c)==UCASE_SOFT_DOTTED); |
737 | } |
738 | |
739 | U_CAPI UBool U_EXPORT2 |
740 | ucase_isCaseSensitive(UChar32 c) { |
741 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
742 | if(!UCASE_HAS_EXCEPTION(props)) { |
743 | return (UBool)((props&UCASE_SENSITIVE)!=0); |
744 | } else { |
745 | const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); |
746 | return (UBool)((*pe&UCASE_EXC_SENSITIVE)!=0); |
747 | } |
748 | } |
749 | |
750 | /* string casing ------------------------------------------------------------ */ |
751 | |
752 | /* |
753 | * These internal functions form the core of string case mappings. |
754 | * They map single code points to result code points or strings and take |
755 | * all necessary conditions (context, locale ID, options) into account. |
756 | * |
757 | * They do not iterate over the source or write to the destination |
758 | * so that the same functions are useful for non-standard string storage, |
759 | * such as in a Replaceable (for Transliterator) or UTF-8/32 strings etc. |
760 | * For the same reason, the "surrounding text" context is passed in as a |
761 | * UCaseContextIterator which does not make any assumptions about |
762 | * the underlying storage. |
763 | * |
764 | * This section contains helper functions that check for conditions |
765 | * in the input text surrounding the current code point |
766 | * according to SpecialCasing.txt. |
767 | * |
768 | * Each helper function gets the index |
769 | * - after the current code point if it looks at following text |
770 | * - before the current code point if it looks at preceding text |
771 | * |
772 | * Unicode 3.2 UAX 21 "Case Mappings" defines the conditions as follows: |
773 | * |
774 | * Final_Sigma |
775 | * C is preceded by a sequence consisting of |
776 | * a cased letter and a case-ignorable sequence, |
777 | * and C is not followed by a sequence consisting of |
778 | * an ignorable sequence and then a cased letter. |
779 | * |
780 | * More_Above |
781 | * C is followed by one or more characters of combining class 230 (ABOVE) |
782 | * in the combining character sequence. |
783 | * |
784 | * After_Soft_Dotted |
785 | * The last preceding character with combining class of zero before C |
786 | * was Soft_Dotted, |
787 | * and there is no intervening combining character class 230 (ABOVE). |
788 | * |
789 | * Before_Dot |
790 | * C is followed by combining dot above (U+0307). |
791 | * Any sequence of characters with a combining class that is neither 0 nor 230 |
792 | * may intervene between the current character and the combining dot above. |
793 | * |
794 | * The erratum from 2002-10-31 adds the condition |
795 | * |
796 | * After_I |
797 | * The last preceding base character was an uppercase I, and there is no |
798 | * intervening combining character class 230 (ABOVE). |
799 | * |
800 | * (See Jitterbug 2344 and the comments on After_I below.) |
801 | * |
802 | * Helper definitions in Unicode 3.2 UAX 21: |
803 | * |
804 | * D1. A character C is defined to be cased |
805 | * if it meets any of the following criteria: |
806 | * |
807 | * - The general category of C is Titlecase Letter (Lt) |
808 | * - In [CoreProps], C has one of the properties Uppercase, or Lowercase |
809 | * - Given D = NFD(C), then it is not the case that: |
810 | * D = UCD_lower(D) = UCD_upper(D) = UCD_title(D) |
811 | * (This third criterion does not add any characters to the list |
812 | * for Unicode 3.2. Ignored.) |
813 | * |
814 | * D2. A character C is defined to be case-ignorable |
815 | * if it meets either of the following criteria: |
816 | * |
817 | * - The general category of C is |
818 | * Nonspacing Mark (Mn), or Enclosing Mark (Me), or Format Control (Cf), or |
819 | * Letter Modifier (Lm), or Symbol Modifier (Sk) |
820 | * - C is one of the following characters |
821 | * U+0027 APOSTROPHE |
822 | * U+00AD SOFT HYPHEN (SHY) |
823 | * U+2019 RIGHT SINGLE QUOTATION MARK |
824 | * (the preferred character for apostrophe) |
825 | * |
826 | * D3. A case-ignorable sequence is a sequence of |
827 | * zero or more case-ignorable characters. |
828 | */ |
829 | |
830 | #define is_d(c) ((c)=='d' || (c)=='D') |
831 | #define is_e(c) ((c)=='e' || (c)=='E') |
832 | #define is_i(c) ((c)=='i' || (c)=='I') |
833 | #define is_l(c) ((c)=='l' || (c)=='L') |
834 | #define is_r(c) ((c)=='r' || (c)=='R') |
835 | #define is_t(c) ((c)=='t' || (c)=='T') |
836 | #define is_u(c) ((c)=='u' || (c)=='U') |
837 | #define is_y(c) ((c)=='y' || (c)=='Y') |
838 | #define is_z(c) ((c)=='z' || (c)=='Z') |
839 | |
840 | /* separator? */ |
841 | #define is_sep(c) ((c)=='_' || (c)=='-' || (c)==0) |
842 | |
843 | /** |
844 | * Requires non-nullptr locale ID but otherwise does the equivalent of |
845 | * checking for language codes as if uloc_getLanguage() were called: |
846 | * Accepts both 2- and 3-letter codes and accepts case variants. |
847 | */ |
848 | U_CFUNC int32_t |
849 | ucase_getCaseLocale(const char *locale) { |
850 | /* |
851 | * This function used to use uloc_getLanguage(), but the current code |
852 | * removes the dependency of this low-level code on uloc implementation code |
853 | * and is faster because not the whole locale ID has to be |
854 | * examined and copied/transformed. |
855 | * |
856 | * Because this code does not want to depend on uloc, the caller must |
857 | * pass in a non-nullptr locale, i.e., may need to call uloc_getDefault(). |
858 | */ |
859 | char c=*locale++; |
860 | // Fastpath for English "en" which is often used for default (=root locale) case mappings, |
861 | // and for Chinese "zh": Very common but no special case mapping behavior. |
862 | // Then check lowercase vs. uppercase to reduce the number of comparisons |
863 | // for other locales without special behavior. |
864 | if(c=='e') { |
865 | /* el or ell? */ |
866 | c=*locale++; |
867 | if(is_l(c)) { |
868 | c=*locale++; |
869 | if(is_l(c)) { |
870 | c=*locale; |
871 | } |
872 | if(is_sep(c)) { |
873 | return UCASE_LOC_GREEK; |
874 | } |
875 | } |
876 | // en, es, ... -> root |
877 | } else if(c=='z') { |
878 | return UCASE_LOC_ROOT; |
879 | #if U_CHARSET_FAMILY==U_ASCII_FAMILY |
880 | } else if(c>='a') { // ASCII a-z = 0x61..0x7a, after A-Z |
881 | #elif U_CHARSET_FAMILY==U_EBCDIC_FAMILY |
882 | } else if(c<='z') { // EBCDIC a-z = 0x81..0xa9 with two gaps, before A-Z |
883 | #else |
884 | # error Unknown charset family! |
885 | #endif |
886 | // lowercase c |
887 | if(c=='t') { |
888 | /* tr or tur? */ |
889 | c=*locale++; |
890 | if(is_u(c)) { |
891 | c=*locale++; |
892 | } |
893 | if(is_r(c)) { |
894 | c=*locale; |
895 | if(is_sep(c)) { |
896 | return UCASE_LOC_TURKISH; |
897 | } |
898 | } |
899 | } else if(c=='a') { |
900 | /* az or aze? */ |
901 | c=*locale++; |
902 | if(is_z(c)) { |
903 | c=*locale++; |
904 | if(is_e(c)) { |
905 | c=*locale; |
906 | } |
907 | if(is_sep(c)) { |
908 | return UCASE_LOC_TURKISH; |
909 | } |
910 | } |
911 | } else if(c=='l') { |
912 | /* lt or lit? */ |
913 | c=*locale++; |
914 | if(is_i(c)) { |
915 | c=*locale++; |
916 | } |
917 | if(is_t(c)) { |
918 | c=*locale; |
919 | if(is_sep(c)) { |
920 | return UCASE_LOC_LITHUANIAN; |
921 | } |
922 | } |
923 | } else if(c=='n') { |
924 | /* nl or nld? */ |
925 | c=*locale++; |
926 | if(is_l(c)) { |
927 | c=*locale++; |
928 | if(is_d(c)) { |
929 | c=*locale; |
930 | } |
931 | if(is_sep(c)) { |
932 | return UCASE_LOC_DUTCH; |
933 | } |
934 | } |
935 | } else if(c=='h') { |
936 | /* hy or hye? *not* hyw */ |
937 | c=*locale++; |
938 | if(is_y(c)) { |
939 | c=*locale++; |
940 | if(is_e(c)) { |
941 | c=*locale; |
942 | } |
943 | if(is_sep(c)) { |
944 | return UCASE_LOC_ARMENIAN; |
945 | } |
946 | } |
947 | } |
948 | } else { |
949 | // uppercase c |
950 | // Same code as for lowercase c but also check for 'E'. |
951 | if(c=='T') { |
952 | /* tr or tur? */ |
953 | c=*locale++; |
954 | if(is_u(c)) { |
955 | c=*locale++; |
956 | } |
957 | if(is_r(c)) { |
958 | c=*locale; |
959 | if(is_sep(c)) { |
960 | return UCASE_LOC_TURKISH; |
961 | } |
962 | } |
963 | } else if(c=='A') { |
964 | /* az or aze? */ |
965 | c=*locale++; |
966 | if(is_z(c)) { |
967 | c=*locale++; |
968 | if(is_e(c)) { |
969 | c=*locale; |
970 | } |
971 | if(is_sep(c)) { |
972 | return UCASE_LOC_TURKISH; |
973 | } |
974 | } |
975 | } else if(c=='L') { |
976 | /* lt or lit? */ |
977 | c=*locale++; |
978 | if(is_i(c)) { |
979 | c=*locale++; |
980 | } |
981 | if(is_t(c)) { |
982 | c=*locale; |
983 | if(is_sep(c)) { |
984 | return UCASE_LOC_LITHUANIAN; |
985 | } |
986 | } |
987 | } else if(c=='E') { |
988 | /* el or ell? */ |
989 | c=*locale++; |
990 | if(is_l(c)) { |
991 | c=*locale++; |
992 | if(is_l(c)) { |
993 | c=*locale; |
994 | } |
995 | if(is_sep(c)) { |
996 | return UCASE_LOC_GREEK; |
997 | } |
998 | } |
999 | } else if(c=='N') { |
1000 | /* nl or nld? */ |
1001 | c=*locale++; |
1002 | if(is_l(c)) { |
1003 | c=*locale++; |
1004 | if(is_d(c)) { |
1005 | c=*locale; |
1006 | } |
1007 | if(is_sep(c)) { |
1008 | return UCASE_LOC_DUTCH; |
1009 | } |
1010 | } |
1011 | } else if(c=='H') { |
1012 | /* hy or hye? *not* hyw */ |
1013 | c=*locale++; |
1014 | if(is_y(c)) { |
1015 | c=*locale++; |
1016 | if(is_e(c)) { |
1017 | c=*locale; |
1018 | } |
1019 | if(is_sep(c)) { |
1020 | return UCASE_LOC_ARMENIAN; |
1021 | } |
1022 | } |
1023 | } |
1024 | } |
1025 | return UCASE_LOC_ROOT; |
1026 | } |
1027 | |
1028 | /* |
1029 | * Is followed by |
1030 | * {case-ignorable}* cased |
1031 | * ? |
1032 | * (dir determines looking forward/backward) |
1033 | * If a character is case-ignorable, it is skipped regardless of whether |
1034 | * it is also cased or not. |
1035 | */ |
1036 | static UBool |
1037 | isFollowedByCasedLetter(UCaseContextIterator *iter, void *context, int8_t dir) { |
1038 | UChar32 c; |
1039 | |
1040 | if(iter==nullptr) { |
1041 | return false; |
1042 | } |
1043 | |
1044 | for(/* dir!=0 sets direction */; (c=iter(context, dir))>=0; dir=0) { |
1045 | int32_t type=ucase_getTypeOrIgnorable(c); |
1046 | if(type&4) { |
1047 | /* case-ignorable, continue with the loop */ |
1048 | } else if(type!=UCASE_NONE) { |
1049 | return true; /* followed by cased letter */ |
1050 | } else { |
1051 | return false; /* uncased and not case-ignorable */ |
1052 | } |
1053 | } |
1054 | |
1055 | return false; /* not followed by cased letter */ |
1056 | } |
1057 | |
1058 | /* Is preceded by Soft_Dotted character with no intervening cc=230 ? */ |
1059 | static UBool |
1060 | isPrecededBySoftDotted(UCaseContextIterator *iter, void *context) { |
1061 | UChar32 c; |
1062 | int32_t dotType; |
1063 | int8_t dir; |
1064 | |
1065 | if(iter==nullptr) { |
1066 | return false; |
1067 | } |
1068 | |
1069 | for(dir=-1; (c=iter(context, dir))>=0; dir=0) { |
1070 | dotType=getDotType(c); |
1071 | if(dotType==UCASE_SOFT_DOTTED) { |
1072 | return true; /* preceded by TYPE_i */ |
1073 | } else if(dotType!=UCASE_OTHER_ACCENT) { |
1074 | return false; /* preceded by different base character (not TYPE_i), or intervening cc==230 */ |
1075 | } |
1076 | } |
1077 | |
1078 | return false; /* not preceded by TYPE_i */ |
1079 | } |
1080 | |
1081 | /* |
1082 | * See Jitterbug 2344: |
1083 | * The condition After_I for Turkic-lowercasing of U+0307 combining dot above |
1084 | * is checked in ICU 2.0, 2.1, 2.6 but was not in 2.2 & 2.4 because |
1085 | * we made those releases compatible with Unicode 3.2 which had not fixed |
1086 | * a related bug in SpecialCasing.txt. |
1087 | * |
1088 | * From the Jitterbug 2344 text: |
1089 | * ... this bug is listed as a Unicode erratum |
1090 | * from 2002-10-31 at http://www.unicode.org/uni2errata/UnicodeErrata.html |
1091 | * <quote> |
1092 | * There are two errors in SpecialCasing.txt. |
1093 | * 1. Missing semicolons on two lines. ... [irrelevant for ICU] |
1094 | * 2. An incorrect context definition. Correct as follows: |
1095 | * < 0307; ; 0307; 0307; tr After_Soft_Dotted; # COMBINING DOT ABOVE |
1096 | * < 0307; ; 0307; 0307; az After_Soft_Dotted; # COMBINING DOT ABOVE |
1097 | * --- |
1098 | * > 0307; ; 0307; 0307; tr After_I; # COMBINING DOT ABOVE |
1099 | * > 0307; ; 0307; 0307; az After_I; # COMBINING DOT ABOVE |
1100 | * where the context After_I is defined as: |
1101 | * The last preceding base character was an uppercase I, and there is no |
1102 | * intervening combining character class 230 (ABOVE). |
1103 | * </quote> |
1104 | * |
1105 | * Note that SpecialCasing.txt even in Unicode 3.2 described the condition as: |
1106 | * |
1107 | * # When lowercasing, remove dot_above in the sequence I + dot_above, which will turn into i. |
1108 | * # This matches the behavior of the canonically equivalent I-dot_above |
1109 | * |
1110 | * See also the description in this place in older versions of uchar.c (revision 1.100). |
1111 | * |
1112 | * Markus W. Scherer 2003-feb-15 |
1113 | */ |
1114 | |
1115 | /* Is preceded by base character 'I' with no intervening cc=230 ? */ |
1116 | static UBool |
1117 | isPrecededBy_I(UCaseContextIterator *iter, void *context) { |
1118 | UChar32 c; |
1119 | int32_t dotType; |
1120 | int8_t dir; |
1121 | |
1122 | if(iter==nullptr) { |
1123 | return false; |
1124 | } |
1125 | |
1126 | for(dir=-1; (c=iter(context, dir))>=0; dir=0) { |
1127 | if(c==0x49) { |
1128 | return true; /* preceded by I */ |
1129 | } |
1130 | dotType=getDotType(c); |
1131 | if(dotType!=UCASE_OTHER_ACCENT) { |
1132 | return false; /* preceded by different base character (not I), or intervening cc==230 */ |
1133 | } |
1134 | } |
1135 | |
1136 | return false; /* not preceded by I */ |
1137 | } |
1138 | |
1139 | /* Is followed by one or more cc==230 ? */ |
1140 | static UBool |
1141 | isFollowedByMoreAbove(UCaseContextIterator *iter, void *context) { |
1142 | UChar32 c; |
1143 | int32_t dotType; |
1144 | int8_t dir; |
1145 | |
1146 | if(iter==nullptr) { |
1147 | return false; |
1148 | } |
1149 | |
1150 | for(dir=1; (c=iter(context, dir))>=0; dir=0) { |
1151 | dotType=getDotType(c); |
1152 | if(dotType==UCASE_ABOVE) { |
1153 | return true; /* at least one cc==230 following */ |
1154 | } else if(dotType!=UCASE_OTHER_ACCENT) { |
1155 | return false; /* next base character, no more cc==230 following */ |
1156 | } |
1157 | } |
1158 | |
1159 | return false; /* no more cc==230 following */ |
1160 | } |
1161 | |
1162 | /* Is followed by a dot above (without cc==230 in between) ? */ |
1163 | static UBool |
1164 | isFollowedByDotAbove(UCaseContextIterator *iter, void *context) { |
1165 | UChar32 c; |
1166 | int32_t dotType; |
1167 | int8_t dir; |
1168 | |
1169 | if(iter==nullptr) { |
1170 | return false; |
1171 | } |
1172 | |
1173 | for(dir=1; (c=iter(context, dir))>=0; dir=0) { |
1174 | if(c==0x307) { |
1175 | return true; |
1176 | } |
1177 | dotType=getDotType(c); |
1178 | if(dotType!=UCASE_OTHER_ACCENT) { |
1179 | return false; /* next base character or cc==230 in between */ |
1180 | } |
1181 | } |
1182 | |
1183 | return false; /* no dot above following */ |
1184 | } |
1185 | |
1186 | U_CAPI int32_t U_EXPORT2 |
1187 | ucase_toFullLower(UChar32 c, |
1188 | UCaseContextIterator *iter, void *context, |
1189 | const char16_t **pString, |
1190 | int32_t loc) { |
1191 | // The sign of the result has meaning, input must be non-negative so that it can be returned as is. |
1192 | U_ASSERT(c >= 0); |
1193 | UChar32 result=c; |
1194 | // Reset the output pointer in case it was uninitialized. |
1195 | *pString=nullptr; |
1196 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
1197 | if(!UCASE_HAS_EXCEPTION(props)) { |
1198 | if(UCASE_IS_UPPER_OR_TITLE(props)) { |
1199 | result=c+UCASE_GET_DELTA(props); |
1200 | } |
1201 | } else { |
1202 | const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props), *pe2; |
1203 | uint16_t excWord=*pe++; |
1204 | int32_t full; |
1205 | |
1206 | pe2=pe; |
1207 | |
1208 | if(excWord&UCASE_EXC_CONDITIONAL_SPECIAL) { |
1209 | /* use hardcoded conditions and mappings */ |
1210 | |
1211 | /* |
1212 | * Test for conditional mappings first |
1213 | * (otherwise the unconditional default mappings are always taken), |
1214 | * then test for characters that have unconditional mappings in SpecialCasing.txt, |
1215 | * then get the UnicodeData.txt mappings. |
1216 | */ |
1217 | if( loc==UCASE_LOC_LITHUANIAN && |
1218 | /* base characters, find accents above */ |
1219 | (((c==0x49 || c==0x4a || c==0x12e) && |
1220 | isFollowedByMoreAbove(iter, context)) || |
1221 | /* precomposed with accent above, no need to find one */ |
1222 | (c==0xcc || c==0xcd || c==0x128)) |
1223 | ) { |
1224 | /* |
1225 | # Lithuanian |
1226 | |
1227 | # Lithuanian retains the dot in a lowercase i when followed by accents. |
1228 | |
1229 | # Introduce an explicit dot above when lowercasing capital I's and J's |
1230 | # whenever there are more accents above. |
1231 | # (of the accents used in Lithuanian: grave, acute, tilde above, and ogonek) |
1232 | |
1233 | 0049; 0069 0307; 0049; 0049; lt More_Above; # LATIN CAPITAL LETTER I |
1234 | 004A; 006A 0307; 004A; 004A; lt More_Above; # LATIN CAPITAL LETTER J |
1235 | 012E; 012F 0307; 012E; 012E; lt More_Above; # LATIN CAPITAL LETTER I WITH OGONEK |
1236 | 00CC; 0069 0307 0300; 00CC; 00CC; lt; # LATIN CAPITAL LETTER I WITH GRAVE |
1237 | 00CD; 0069 0307 0301; 00CD; 00CD; lt; # LATIN CAPITAL LETTER I WITH ACUTE |
1238 | 0128; 0069 0307 0303; 0128; 0128; lt; # LATIN CAPITAL LETTER I WITH TILDE |
1239 | */ |
1240 | switch(c) { |
1241 | case 0x49: /* LATIN CAPITAL LETTER I */ |
1242 | *pString=iDot; |
1243 | return 2; |
1244 | case 0x4a: /* LATIN CAPITAL LETTER J */ |
1245 | *pString=jDot; |
1246 | return 2; |
1247 | case 0x12e: /* LATIN CAPITAL LETTER I WITH OGONEK */ |
1248 | *pString=iOgonekDot; |
1249 | return 2; |
1250 | case 0xcc: /* LATIN CAPITAL LETTER I WITH GRAVE */ |
1251 | *pString=iDotGrave; |
1252 | return 3; |
1253 | case 0xcd: /* LATIN CAPITAL LETTER I WITH ACUTE */ |
1254 | *pString=iDotAcute; |
1255 | return 3; |
1256 | case 0x128: /* LATIN CAPITAL LETTER I WITH TILDE */ |
1257 | *pString=iDotTilde; |
1258 | return 3; |
1259 | default: |
1260 | return 0; /* will not occur */ |
1261 | } |
1262 | /* # Turkish and Azeri */ |
1263 | } else if(loc==UCASE_LOC_TURKISH && c==0x130) { |
1264 | /* |
1265 | # I and i-dotless; I-dot and i are case pairs in Turkish and Azeri |
1266 | # The following rules handle those cases. |
1267 | |
1268 | 0130; 0069; 0130; 0130; tr # LATIN CAPITAL LETTER I WITH DOT ABOVE |
1269 | 0130; 0069; 0130; 0130; az # LATIN CAPITAL LETTER I WITH DOT ABOVE |
1270 | */ |
1271 | return 0x69; |
1272 | } else if(loc==UCASE_LOC_TURKISH && c==0x307 && isPrecededBy_I(iter, context)) { |
1273 | /* |
1274 | # When lowercasing, remove dot_above in the sequence I + dot_above, which will turn into i. |
1275 | # This matches the behavior of the canonically equivalent I-dot_above |
1276 | |
1277 | 0307; ; 0307; 0307; tr After_I; # COMBINING DOT ABOVE |
1278 | 0307; ; 0307; 0307; az After_I; # COMBINING DOT ABOVE |
1279 | */ |
1280 | return 0; /* remove the dot (continue without output) */ |
1281 | } else if(loc==UCASE_LOC_TURKISH && c==0x49 && !isFollowedByDotAbove(iter, context)) { |
1282 | /* |
1283 | # When lowercasing, unless an I is before a dot_above, it turns into a dotless i. |
1284 | |
1285 | 0049; 0131; 0049; 0049; tr Not_Before_Dot; # LATIN CAPITAL LETTER I |
1286 | 0049; 0131; 0049; 0049; az Not_Before_Dot; # LATIN CAPITAL LETTER I |
1287 | */ |
1288 | return 0x131; |
1289 | } else if(c==0x130) { |
1290 | /* |
1291 | # Preserve canonical equivalence for I with dot. Turkic is handled below. |
1292 | |
1293 | 0130; 0069 0307; 0130; 0130; # LATIN CAPITAL LETTER I WITH DOT ABOVE |
1294 | */ |
1295 | *pString=iDot; |
1296 | return 2; |
1297 | } else if( c==0x3a3 && |
1298 | !isFollowedByCasedLetter(iter, context, 1) && |
1299 | isFollowedByCasedLetter(iter, context, -1) /* -1=preceded */ |
1300 | ) { |
1301 | /* greek capital sigma maps depending on surrounding cased letters (see SpecialCasing.txt) */ |
1302 | /* |
1303 | # Special case for final form of sigma |
1304 | |
1305 | 03A3; 03C2; 03A3; 03A3; Final_Sigma; # GREEK CAPITAL LETTER SIGMA |
1306 | */ |
1307 | return 0x3c2; /* greek small final sigma */ |
1308 | } else { |
1309 | /* no known conditional special case mapping, use a normal mapping */ |
1310 | } |
1311 | } else if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { |
1312 | GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, full); |
1313 | full&=UCASE_FULL_LOWER; |
1314 | if(full!=0) { |
1315 | /* set the output pointer to the lowercase mapping */ |
1316 | *pString=reinterpret_cast<const char16_t *>(pe+1); |
1317 | |
1318 | /* return the string length */ |
1319 | return full; |
1320 | } |
1321 | } |
1322 | |
1323 | if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_IS_UPPER_OR_TITLE(props)) { |
1324 | int32_t delta; |
1325 | GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe2, delta); |
1326 | return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; |
1327 | } |
1328 | if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { |
1329 | GET_SLOT_VALUE(excWord, UCASE_EXC_LOWER, pe2, result); |
1330 | } |
1331 | } |
1332 | |
1333 | return (result==c) ? ~result : result; |
1334 | } |
1335 | |
1336 | /* internal */ |
1337 | static int32_t |
1338 | toUpperOrTitle(UChar32 c, |
1339 | UCaseContextIterator *iter, void *context, |
1340 | const char16_t **pString, |
1341 | int32_t loc, |
1342 | UBool upperNotTitle) { |
1343 | // The sign of the result has meaning, input must be non-negative so that it can be returned as is. |
1344 | U_ASSERT(c >= 0); |
1345 | UChar32 result=c; |
1346 | // Reset the output pointer in case it was uninitialized. |
1347 | *pString=nullptr; |
1348 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
1349 | if(!UCASE_HAS_EXCEPTION(props)) { |
1350 | if(UCASE_GET_TYPE(props)==UCASE_LOWER) { |
1351 | result=c+UCASE_GET_DELTA(props); |
1352 | } |
1353 | } else { |
1354 | const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props), *pe2; |
1355 | uint16_t excWord=*pe++; |
1356 | int32_t full, idx; |
1357 | |
1358 | pe2=pe; |
1359 | |
1360 | if(excWord&UCASE_EXC_CONDITIONAL_SPECIAL) { |
1361 | /* use hardcoded conditions and mappings */ |
1362 | if(loc==UCASE_LOC_TURKISH && c==0x69) { |
1363 | /* |
1364 | # Turkish and Azeri |
1365 | |
1366 | # I and i-dotless; I-dot and i are case pairs in Turkish and Azeri |
1367 | # The following rules handle those cases. |
1368 | |
1369 | # When uppercasing, i turns into a dotted capital I |
1370 | |
1371 | 0069; 0069; 0130; 0130; tr; # LATIN SMALL LETTER I |
1372 | 0069; 0069; 0130; 0130; az; # LATIN SMALL LETTER I |
1373 | */ |
1374 | return 0x130; |
1375 | } else if(loc==UCASE_LOC_LITHUANIAN && c==0x307 && isPrecededBySoftDotted(iter, context)) { |
1376 | /* |
1377 | # Lithuanian |
1378 | |
1379 | # Lithuanian retains the dot in a lowercase i when followed by accents. |
1380 | |
1381 | # Remove DOT ABOVE after "i" with upper or titlecase |
1382 | |
1383 | 0307; 0307; ; ; lt After_Soft_Dotted; # COMBINING DOT ABOVE |
1384 | */ |
1385 | return 0; /* remove the dot (continue without output) */ |
1386 | } else if(c==0x0587) { |
1387 | // See ICU-13416: |
1388 | // և ligature ech-yiwn |
1389 | // uppercases to ԵՒ=ech+yiwn by default and in Western Armenian, |
1390 | // but to ԵՎ=ech+vew in Eastern Armenian. |
1391 | if(loc==UCASE_LOC_ARMENIAN) { |
1392 | *pString=upperNotTitle ? u"ԵՎ" : u"Եվ" ; |
1393 | } else { |
1394 | *pString=upperNotTitle ? u"ԵՒ" : u"Եւ" ; |
1395 | } |
1396 | return 2; |
1397 | } else { |
1398 | /* no known conditional special case mapping, use a normal mapping */ |
1399 | } |
1400 | } else if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { |
1401 | GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, full); |
1402 | |
1403 | /* start of full case mapping strings */ |
1404 | ++pe; |
1405 | |
1406 | /* skip the lowercase and case-folding result strings */ |
1407 | pe+=full&UCASE_FULL_LOWER; |
1408 | full>>=4; |
1409 | pe+=full&0xf; |
1410 | full>>=4; |
1411 | |
1412 | if(upperNotTitle) { |
1413 | full&=0xf; |
1414 | } else { |
1415 | /* skip the uppercase result string */ |
1416 | pe+=full&0xf; |
1417 | full=(full>>4)&0xf; |
1418 | } |
1419 | |
1420 | if(full!=0) { |
1421 | /* set the output pointer to the result string */ |
1422 | *pString=reinterpret_cast<const char16_t *>(pe); |
1423 | |
1424 | /* return the string length */ |
1425 | return full; |
1426 | } |
1427 | } |
1428 | |
1429 | if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_GET_TYPE(props)==UCASE_LOWER) { |
1430 | int32_t delta; |
1431 | GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe2, delta); |
1432 | return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; |
1433 | } |
1434 | if(!upperNotTitle && HAS_SLOT(excWord, UCASE_EXC_TITLE)) { |
1435 | idx=UCASE_EXC_TITLE; |
1436 | } else if(HAS_SLOT(excWord, UCASE_EXC_UPPER)) { |
1437 | /* here, titlecase is same as uppercase */ |
1438 | idx=UCASE_EXC_UPPER; |
1439 | } else { |
1440 | return ~c; |
1441 | } |
1442 | GET_SLOT_VALUE(excWord, idx, pe2, result); |
1443 | } |
1444 | |
1445 | return (result==c) ? ~result : result; |
1446 | } |
1447 | |
1448 | U_CAPI int32_t U_EXPORT2 |
1449 | ucase_toFullUpper(UChar32 c, |
1450 | UCaseContextIterator *iter, void *context, |
1451 | const char16_t **pString, |
1452 | int32_t caseLocale) { |
1453 | return toUpperOrTitle(c, iter, context, pString, caseLocale, true); |
1454 | } |
1455 | |
1456 | U_CAPI int32_t U_EXPORT2 |
1457 | ucase_toFullTitle(UChar32 c, |
1458 | UCaseContextIterator *iter, void *context, |
1459 | const char16_t **pString, |
1460 | int32_t caseLocale) { |
1461 | return toUpperOrTitle(c, iter, context, pString, caseLocale, false); |
1462 | } |
1463 | |
1464 | /* case folding ------------------------------------------------------------- */ |
1465 | |
1466 | /* |
1467 | * Case folding is similar to lowercasing. |
1468 | * The result may be a simple mapping, i.e., a single code point, or |
1469 | * a full mapping, i.e., a string. |
1470 | * If the case folding for a code point is the same as its simple (1:1) lowercase mapping, |
1471 | * then only the lowercase mapping is stored. |
1472 | * |
1473 | * Some special cases are hardcoded because their conditions cannot be |
1474 | * parsed and processed from CaseFolding.txt. |
1475 | * |
1476 | * Unicode 3.2 CaseFolding.txt specifies for its status field: |
1477 | |
1478 | # C: common case folding, common mappings shared by both simple and full mappings. |
1479 | # F: full case folding, mappings that cause strings to grow in length. Multiple characters are separated by spaces. |
1480 | # S: simple case folding, mappings to single characters where different from F. |
1481 | # T: special case for uppercase I and dotted uppercase I |
1482 | # - For non-Turkic languages, this mapping is normally not used. |
1483 | # - For Turkic languages (tr, az), this mapping can be used instead of the normal mapping for these characters. |
1484 | # |
1485 | # Usage: |
1486 | # A. To do a simple case folding, use the mappings with status C + S. |
1487 | # B. To do a full case folding, use the mappings with status C + F. |
1488 | # |
1489 | # The mappings with status T can be used or omitted depending on the desired case-folding |
1490 | # behavior. (The default option is to exclude them.) |
1491 | |
1492 | * Unicode 3.2 has 'T' mappings as follows: |
1493 | |
1494 | 0049; T; 0131; # LATIN CAPITAL LETTER I |
1495 | 0130; T; 0069; # LATIN CAPITAL LETTER I WITH DOT ABOVE |
1496 | |
1497 | * while the default mappings for these code points are: |
1498 | |
1499 | 0049; C; 0069; # LATIN CAPITAL LETTER I |
1500 | 0130; F; 0069 0307; # LATIN CAPITAL LETTER I WITH DOT ABOVE |
1501 | |
1502 | * U+0130 has no simple case folding (simple-case-folds to itself). |
1503 | */ |
1504 | |
1505 | /* return the simple case folding mapping for c */ |
1506 | U_CAPI UChar32 U_EXPORT2 |
1507 | ucase_fold(UChar32 c, uint32_t options) { |
1508 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
1509 | if(!UCASE_HAS_EXCEPTION(props)) { |
1510 | if(UCASE_IS_UPPER_OR_TITLE(props)) { |
1511 | c+=UCASE_GET_DELTA(props); |
1512 | } |
1513 | } else { |
1514 | const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props); |
1515 | uint16_t excWord=*pe++; |
1516 | int32_t idx; |
1517 | if(excWord&UCASE_EXC_CONDITIONAL_FOLD) { |
1518 | /* special case folding mappings, hardcoded */ |
1519 | if((options&_FOLD_CASE_OPTIONS_MASK)==U_FOLD_CASE_DEFAULT) { |
1520 | /* default mappings */ |
1521 | if(c==0x49) { |
1522 | /* 0049; C; 0069; # LATIN CAPITAL LETTER I */ |
1523 | return 0x69; |
1524 | } else if(c==0x130) { |
1525 | /* no simple case folding for U+0130 */ |
1526 | return c; |
1527 | } |
1528 | } else { |
1529 | /* Turkic mappings */ |
1530 | if(c==0x49) { |
1531 | /* 0049; T; 0131; # LATIN CAPITAL LETTER I */ |
1532 | return 0x131; |
1533 | } else if(c==0x130) { |
1534 | /* 0130; T; 0069; # LATIN CAPITAL LETTER I WITH DOT ABOVE */ |
1535 | return 0x69; |
1536 | } |
1537 | } |
1538 | } |
1539 | if((excWord&UCASE_EXC_NO_SIMPLE_CASE_FOLDING)!=0) { |
1540 | return c; |
1541 | } |
1542 | if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_IS_UPPER_OR_TITLE(props)) { |
1543 | int32_t delta; |
1544 | GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe, delta); |
1545 | return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; |
1546 | } |
1547 | if(HAS_SLOT(excWord, UCASE_EXC_FOLD)) { |
1548 | idx=UCASE_EXC_FOLD; |
1549 | } else if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { |
1550 | idx=UCASE_EXC_LOWER; |
1551 | } else { |
1552 | return c; |
1553 | } |
1554 | GET_SLOT_VALUE(excWord, idx, pe, c); |
1555 | } |
1556 | return c; |
1557 | } |
1558 | |
1559 | /* |
1560 | * Issue for canonical caseless match (UAX #21): |
1561 | * Turkic casefolding (using "T" mappings in CaseFolding.txt) does not preserve |
1562 | * canonical equivalence, unlike default-option casefolding. |
1563 | * For example, I-grave and I + grave fold to strings that are not canonically |
1564 | * equivalent. |
1565 | * For more details, see the comment in unorm_compare() in unorm.cpp |
1566 | * and the intermediate prototype changes for Jitterbug 2021. |
1567 | * (For example, revision 1.104 of uchar.c and 1.4 of CaseFolding.txt.) |
1568 | * |
1569 | * This did not get fixed because it appears that it is not possible to fix |
1570 | * it for uppercase and lowercase characters (I-grave vs. i-grave) |
1571 | * together in a way that they still fold to common result strings. |
1572 | */ |
1573 | |
1574 | U_CAPI int32_t U_EXPORT2 |
1575 | ucase_toFullFolding(UChar32 c, |
1576 | const char16_t **pString, |
1577 | uint32_t options) { |
1578 | // The sign of the result has meaning, input must be non-negative so that it can be returned as is. |
1579 | U_ASSERT(c >= 0); |
1580 | UChar32 result=c; |
1581 | // Reset the output pointer in case it was uninitialized. |
1582 | *pString=nullptr; |
1583 | uint16_t props=UTRIE2_GET16(&ucase_props_singleton.trie, c); |
1584 | if(!UCASE_HAS_EXCEPTION(props)) { |
1585 | if(UCASE_IS_UPPER_OR_TITLE(props)) { |
1586 | result=c+UCASE_GET_DELTA(props); |
1587 | } |
1588 | } else { |
1589 | const uint16_t *pe=GET_EXCEPTIONS(&ucase_props_singleton, props), *pe2; |
1590 | uint16_t excWord=*pe++; |
1591 | int32_t full, idx; |
1592 | |
1593 | pe2=pe; |
1594 | |
1595 | if(excWord&UCASE_EXC_CONDITIONAL_FOLD) { |
1596 | /* use hardcoded conditions and mappings */ |
1597 | if((options&_FOLD_CASE_OPTIONS_MASK)==U_FOLD_CASE_DEFAULT) { |
1598 | /* default mappings */ |
1599 | if(c==0x49) { |
1600 | /* 0049; C; 0069; # LATIN CAPITAL LETTER I */ |
1601 | return 0x69; |
1602 | } else if(c==0x130) { |
1603 | /* 0130; F; 0069 0307; # LATIN CAPITAL LETTER I WITH DOT ABOVE */ |
1604 | *pString=iDot; |
1605 | return 2; |
1606 | } |
1607 | } else { |
1608 | /* Turkic mappings */ |
1609 | if(c==0x49) { |
1610 | /* 0049; T; 0131; # LATIN CAPITAL LETTER I */ |
1611 | return 0x131; |
1612 | } else if(c==0x130) { |
1613 | /* 0130; T; 0069; # LATIN CAPITAL LETTER I WITH DOT ABOVE */ |
1614 | return 0x69; |
1615 | } |
1616 | } |
1617 | } else if(HAS_SLOT(excWord, UCASE_EXC_FULL_MAPPINGS)) { |
1618 | GET_SLOT_VALUE(excWord, UCASE_EXC_FULL_MAPPINGS, pe, full); |
1619 | |
1620 | /* start of full case mapping strings */ |
1621 | ++pe; |
1622 | |
1623 | /* skip the lowercase result string */ |
1624 | pe+=full&UCASE_FULL_LOWER; |
1625 | full=(full>>4)&0xf; |
1626 | |
1627 | if(full!=0) { |
1628 | /* set the output pointer to the result string */ |
1629 | *pString=reinterpret_cast<const char16_t *>(pe); |
1630 | |
1631 | /* return the string length */ |
1632 | return full; |
1633 | } |
1634 | } |
1635 | |
1636 | if((excWord&UCASE_EXC_NO_SIMPLE_CASE_FOLDING)!=0) { |
1637 | return ~c; |
1638 | } |
1639 | if(HAS_SLOT(excWord, UCASE_EXC_DELTA) && UCASE_IS_UPPER_OR_TITLE(props)) { |
1640 | int32_t delta; |
1641 | GET_SLOT_VALUE(excWord, UCASE_EXC_DELTA, pe2, delta); |
1642 | return (excWord&UCASE_EXC_DELTA_IS_NEGATIVE)==0 ? c+delta : c-delta; |
1643 | } |
1644 | if(HAS_SLOT(excWord, UCASE_EXC_FOLD)) { |
1645 | idx=UCASE_EXC_FOLD; |
1646 | } else if(HAS_SLOT(excWord, UCASE_EXC_LOWER)) { |
1647 | idx=UCASE_EXC_LOWER; |
1648 | } else { |
1649 | return ~c; |
1650 | } |
1651 | GET_SLOT_VALUE(excWord, idx, pe2, result); |
1652 | } |
1653 | |
1654 | return (result==c) ? ~result : result; |
1655 | } |
1656 | |
1657 | /* case mapping properties API ---------------------------------------------- */ |
1658 | |
1659 | /* public API (see uchar.h) */ |
1660 | |
1661 | U_CAPI UBool U_EXPORT2 |
1662 | u_isULowercase(UChar32 c) { |
1663 | return (UBool)(UCASE_LOWER==ucase_getType(c)); |
1664 | } |
1665 | |
1666 | U_CAPI UBool U_EXPORT2 |
1667 | u_isUUppercase(UChar32 c) { |
1668 | return (UBool)(UCASE_UPPER==ucase_getType(c)); |
1669 | } |
1670 | |
1671 | /* Transforms the Unicode character to its lower case equivalent.*/ |
1672 | U_CAPI UChar32 U_EXPORT2 |
1673 | u_tolower(UChar32 c) { |
1674 | return ucase_tolower(c); |
1675 | } |
1676 | |
1677 | /* Transforms the Unicode character to its upper case equivalent.*/ |
1678 | U_CAPI UChar32 U_EXPORT2 |
1679 | u_toupper(UChar32 c) { |
1680 | return ucase_toupper(c); |
1681 | } |
1682 | |
1683 | /* Transforms the Unicode character to its title case equivalent.*/ |
1684 | U_CAPI UChar32 U_EXPORT2 |
1685 | u_totitle(UChar32 c) { |
1686 | return ucase_totitle(c); |
1687 | } |
1688 | |
1689 | /* return the simple case folding mapping for c */ |
1690 | U_CAPI UChar32 U_EXPORT2 |
1691 | u_foldCase(UChar32 c, uint32_t options) { |
1692 | return ucase_fold(c, options); |
1693 | } |
1694 | |
1695 | U_CFUNC int32_t U_EXPORT2 |
1696 | ucase_hasBinaryProperty(UChar32 c, UProperty which) { |
1697 | /* case mapping properties */ |
1698 | const char16_t *resultString; |
1699 | switch(which) { |
1700 | case UCHAR_LOWERCASE: |
1701 | return (UBool)(UCASE_LOWER==ucase_getType(c)); |
1702 | case UCHAR_UPPERCASE: |
1703 | return (UBool)(UCASE_UPPER==ucase_getType(c)); |
1704 | case UCHAR_SOFT_DOTTED: |
1705 | return ucase_isSoftDotted(c); |
1706 | case UCHAR_CASE_SENSITIVE: |
1707 | return ucase_isCaseSensitive(c); |
1708 | case UCHAR_CASED: |
1709 | return (UBool)(UCASE_NONE!=ucase_getType(c)); |
1710 | case UCHAR_CASE_IGNORABLE: |
1711 | return (UBool)(ucase_getTypeOrIgnorable(c)>>2); |
1712 | /* |
1713 | * Note: The following Changes_When_Xyz are defined as testing whether |
1714 | * the NFD form of the input changes when Xyz-case-mapped. |
1715 | * However, this simpler implementation of these properties, |
1716 | * ignoring NFD, passes the tests. |
1717 | * The implementation needs to be changed if the tests start failing. |
1718 | * When that happens, optimizations should be used to work with the |
1719 | * per-single-code point ucase_toFullXyz() functions unless |
1720 | * the NFD form has more than one code point, |
1721 | * and the property starts set needs to be the union of the |
1722 | * start sets for normalization and case mappings. |
1723 | */ |
1724 | case UCHAR_CHANGES_WHEN_LOWERCASED: |
1725 | return (UBool)(ucase_toFullLower(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0); |
1726 | case UCHAR_CHANGES_WHEN_UPPERCASED: |
1727 | return (UBool)(ucase_toFullUpper(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0); |
1728 | case UCHAR_CHANGES_WHEN_TITLECASED: |
1729 | return (UBool)(ucase_toFullTitle(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0); |
1730 | /* case UCHAR_CHANGES_WHEN_CASEFOLDED: -- in uprops.c */ |
1731 | case UCHAR_CHANGES_WHEN_CASEMAPPED: |
1732 | return (UBool)( |
1733 | ucase_toFullLower(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0 || |
1734 | ucase_toFullUpper(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0 || |
1735 | ucase_toFullTitle(c, nullptr, nullptr, &resultString, UCASE_LOC_ROOT)>=0); |
1736 | default: |
1737 | return false; |
1738 | } |
1739 | } |
1740 | |