]> git.saurik.com Git - apple/icu.git/blob - icuSources/common/uprops.cpp
projects / apple / icu.git / blob
? search:


dbc06ff767d32d8d9b9c902444f73afd0ec1002a
[apple/icu.git] / icuSources / common / uprops.cpp
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) 2002-2016, International Business Machines
7 * Corporation and others. All Rights Reserved.
8 *
9 *******************************************************************************
10 * file name: uprops.cpp
11 * encoding: UTF-8
12 * tab size: 8 (not used)
13 * indentation:4
14 *
15 * created on: 2002feb24
16 * created by: Markus W. Scherer
17 *
18 * Implementations for mostly non-core Unicode character properties
19 * stored in uprops.icu.
20 *
21 * With the APIs implemented here, almost all properties files and
22 * their associated implementation files are used from this file,
23 * including those for normalization and case mappings.
24 */
25
26 #include "unicode/utypes.h"
27 #include "unicode/uchar.h"
28 #include "unicode/ucptrie.h"
29 #include "unicode/udata.h"
30 #include "unicode/unorm2.h"
31 #include "unicode/uscript.h"
32 #include "unicode/ustring.h"
33 #include "cstring.h"
34 #include "mutex.h"
35 #include "normalizer2impl.h"
36 #include "umutex.h"
37 #include "ubidi_props.h"
38 #include "uprops.h"
39 #include "ucase.h"
40 #include "ucln_cmn.h"
41 #include "ulayout_props.h"
42 #include "ustr_imp.h"
43
44 U_NAMESPACE_USE
45
46 // Unicode text layout properties data -----------------------------------------
47
48 namespace {
49
50 icu::UInitOnce gLayoutInitOnce = U_INITONCE_INITIALIZER;
51 UDataMemory *gLayoutMemory = nullptr;
52
53 UCPTrie *gInpcTrie = nullptr; // Indic_Positional_Category
54 UCPTrie *gInscTrie = nullptr; // Indic_Syllabic_Category
55 UCPTrie *gVoTrie = nullptr; // Vertical_Orientation
56
57 int32_t gMaxInpcValue = 0;
58 int32_t gMaxInscValue = 0;
59 int32_t gMaxVoValue = 0;
60
61 UBool U_CALLCONV uprops_cleanup() {
62 udata_close(gLayoutMemory);
63 gLayoutMemory = nullptr;
64
65 ucptrie_close(gInpcTrie);
66 gInpcTrie = nullptr;
67 ucptrie_close(gInscTrie);
68 gInscTrie = nullptr;
69 ucptrie_close(gVoTrie);
70 gVoTrie = nullptr;
71
72 gMaxInpcValue = 0;
73 gMaxInscValue = 0;
74 gMaxVoValue = 0;
75
76 gLayoutInitOnce.reset();
77 return TRUE;
78 }
79
80 UBool U_CALLCONV
81 ulayout_isAcceptable(void * /*context*/,
82 const char * /* type */, const char * /*name*/,
83 const UDataInfo *pInfo) {
84 return pInfo->size >= 20 &&
85 pInfo->isBigEndian == U_IS_BIG_ENDIAN &&
86 pInfo->charsetFamily == U_CHARSET_FAMILY &&
87 pInfo->dataFormat[0] == ULAYOUT_FMT_0 &&
88 pInfo->dataFormat[1] == ULAYOUT_FMT_1 &&
89 pInfo->dataFormat[2] == ULAYOUT_FMT_2 &&
90 pInfo->dataFormat[3] == ULAYOUT_FMT_3 &&
91 pInfo->formatVersion[0] == 1;
92 }
93
94 // UInitOnce singleton initialization function
95 void U_CALLCONV ulayout_load(UErrorCode &errorCode) {
96 gLayoutMemory = udata_openChoice(
97 nullptr, ULAYOUT_DATA_TYPE, ULAYOUT_DATA_NAME,
98 ulayout_isAcceptable, nullptr, &errorCode);
99 if (U_FAILURE(errorCode)) { return; }
100
101 const uint8_t *inBytes = (const uint8_t *)udata_getMemory(gLayoutMemory);
102 const int32_t *inIndexes = (const int32_t *)inBytes;
103 int32_t indexesLength = inIndexes[ULAYOUT_IX_INDEXES_LENGTH];
104 if (indexesLength < 12) {
105 errorCode = U_INVALID_FORMAT_ERROR; // Not enough indexes.
106 return;
107 }
108 int32_t offset = indexesLength * 4;
109 int32_t top = inIndexes[ULAYOUT_IX_INPC_TRIE_TOP];
110 int32_t trieSize = top - offset;
111 if (trieSize >= 16) {
112 gInpcTrie = ucptrie_openFromBinary(
113 UCPTRIE_TYPE_ANY, UCPTRIE_VALUE_BITS_ANY,
114 inBytes + offset, trieSize, nullptr, &errorCode);
115 }
116 offset = top;
117 top = inIndexes[ULAYOUT_IX_INSC_TRIE_TOP];
118 trieSize = top - offset;
119 if (trieSize >= 16) {
120 gInscTrie = ucptrie_openFromBinary(
121 UCPTRIE_TYPE_ANY, UCPTRIE_VALUE_BITS_ANY,
122 inBytes + offset, trieSize, nullptr, &errorCode);
123 }
124 offset = top;
125 top = inIndexes[ULAYOUT_IX_VO_TRIE_TOP];
126 trieSize = top - offset;
127 if (trieSize >= 16) {
128 gVoTrie = ucptrie_openFromBinary(
129 UCPTRIE_TYPE_ANY, UCPTRIE_VALUE_BITS_ANY,
130 inBytes + offset, trieSize, nullptr, &errorCode);
131 }
132
133 uint32_t maxValues = inIndexes[ULAYOUT_IX_MAX_VALUES];
134 gMaxInpcValue = maxValues >> ULAYOUT_MAX_INPC_SHIFT;
135 gMaxInscValue = (maxValues >> ULAYOUT_MAX_INSC_SHIFT) & 0xff;
136 gMaxVoValue = (maxValues >> ULAYOUT_MAX_VO_SHIFT) & 0xff;
137
138 ucln_common_registerCleanup(UCLN_COMMON_UPROPS, uprops_cleanup);
139 }
140
141 UBool ulayout_ensureData(UErrorCode &errorCode) {
142 if (U_FAILURE(errorCode)) { return FALSE; }
143 umtx_initOnce(gLayoutInitOnce, &ulayout_load, errorCode);
144 return U_SUCCESS(errorCode);
145 }
146
147 UBool ulayout_ensureData() {
148 UErrorCode errorCode = U_ZERO_ERROR;
149 return ulayout_ensureData(errorCode);
150 }
151
152 } // namespace
153
154 /* general properties API functions ----------------------------------------- */
155
156 struct BinaryProperty;
157
158 typedef UBool BinaryPropertyContains(const BinaryProperty &prop, UChar32 c, UProperty which);
159
160 struct BinaryProperty {
161 int32_t column; // SRC_PROPSVEC column, or "source" if mask==0
162 uint32_t mask;
163 BinaryPropertyContains *contains;
164 };
165
166 static UBool defaultContains(const BinaryProperty &prop, UChar32 c, UProperty /*which*/) {
167 /* systematic, directly stored properties */
168 return (u_getUnicodeProperties(c, prop.column)&prop.mask)!=0;
169 }
170
171 static UBool caseBinaryPropertyContains(const BinaryProperty &/*prop*/, UChar32 c, UProperty which) {
172 return static_cast<UBool>(ucase_hasBinaryProperty(c, which));
173 }
174
175 static UBool isBidiControl(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
176 return ubidi_isBidiControl(c);
177 }
178
179 static UBool isMirrored(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
180 return ubidi_isMirrored(c);
181 }
182
183 static UBool isJoinControl(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
184 return ubidi_isJoinControl(c);
185 }
186
187 #if UCONFIG_NO_NORMALIZATION
188 static UBool hasFullCompositionExclusion(const BinaryProperty &, UChar32, UProperty) {
189 return FALSE;
190 }
191 #else
192 static UBool hasFullCompositionExclusion(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
193 // By definition, Full_Composition_Exclusion is the same as NFC_QC=No.
194 UErrorCode errorCode=U_ZERO_ERROR;
195 const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode);
196 return U_SUCCESS(errorCode) && impl->isCompNo(impl->getNorm16(c));
197 }
198 #endif
199
200 // UCHAR_NF*_INERT properties
201 #if UCONFIG_NO_NORMALIZATION
202 static UBool isNormInert(const BinaryProperty &, UChar32, UProperty) {
203 return FALSE;
204 }
205 #else
206 static UBool isNormInert(const BinaryProperty &/*prop*/, UChar32 c, UProperty which) {
207 UErrorCode errorCode=U_ZERO_ERROR;
208 const Normalizer2 *norm2=Normalizer2Factory::getInstance(
209 (UNormalizationMode)(which-UCHAR_NFD_INERT+UNORM_NFD), errorCode);
210 return U_SUCCESS(errorCode) && norm2->isInert(c);
211 }
212 #endif
213
214 #if UCONFIG_NO_NORMALIZATION
215 static UBool changesWhenCasefolded(const BinaryProperty &, UChar32, UProperty) {
216 return FALSE;
217 }
218 #else
219 static UBool changesWhenCasefolded(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
220 UnicodeString nfd;
221 UErrorCode errorCode=U_ZERO_ERROR;
222 const Normalizer2 *nfcNorm2=Normalizer2::getNFCInstance(errorCode);
223 if(U_FAILURE(errorCode)) {
224 return FALSE;
225 }
226 if(nfcNorm2->getDecomposition(c, nfd)) {
227 /* c has a decomposition */
228 if(nfd.length()==1) {
229 c=nfd[0]; /* single BMP code point */
230 } else if(nfd.length()<=U16_MAX_LENGTH &&
231 nfd.length()==U16_LENGTH(c=nfd.char32At(0))
232 ) {
233 /* single supplementary code point */
234 } else {
235 c=U_SENTINEL;
236 }
237 } else if(c<0) {
238 return FALSE; /* protect against bad input */
239 }
240 if(c>=0) {
241 /* single code point */
242 const UChar *resultString;
243 return (UBool)(ucase_toFullFolding(c, &resultString, U_FOLD_CASE_DEFAULT)>=0);
244 } else {
245 /* guess some large but stack-friendly capacity */
246 UChar dest[2*UCASE_MAX_STRING_LENGTH];
247 int32_t destLength;
248 destLength=u_strFoldCase(dest, UPRV_LENGTHOF(dest),
249 nfd.getBuffer(), nfd.length(),
250 U_FOLD_CASE_DEFAULT, &errorCode);
251 return (UBool)(U_SUCCESS(errorCode) &&
252 0!=u_strCompare(nfd.getBuffer(), nfd.length(),
253 dest, destLength, FALSE));
254 }
255 }
256 #endif
257
258 #if UCONFIG_NO_NORMALIZATION
259 static UBool changesWhenNFKC_Casefolded(const BinaryProperty &, UChar32, UProperty) {
260 return FALSE;
261 }
262 #else
263 static UBool changesWhenNFKC_Casefolded(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
264 UErrorCode errorCode=U_ZERO_ERROR;
265 const Normalizer2Impl *kcf=Normalizer2Factory::getNFKC_CFImpl(errorCode);
266 if(U_FAILURE(errorCode)) {
267 return FALSE;
268 }
269 UnicodeString src(c);
270 UnicodeString dest;
271 {
272 // The ReorderingBuffer must be in a block because its destructor
273 // needs to release dest's buffer before we look at its contents.
274 ReorderingBuffer buffer(*kcf, dest);
275 // Small destCapacity for NFKC_CF(c).
276 if(buffer.init(5, errorCode)) {
277 const UChar *srcArray=src.getBuffer();
278 kcf->compose(srcArray, srcArray+src.length(), FALSE,
279 TRUE, buffer, errorCode);
280 }
281 }
282 return U_SUCCESS(errorCode) && dest!=src;
283 }
284 #endif
285
286 #if UCONFIG_NO_NORMALIZATION
287 static UBool isCanonSegmentStarter(const BinaryProperty &, UChar32, UProperty) {
288 return FALSE;
289 }
290 #else
291 static UBool isCanonSegmentStarter(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
292 UErrorCode errorCode=U_ZERO_ERROR;
293 const Normalizer2Impl *impl=Normalizer2Factory::getNFCImpl(errorCode);
294 return
295 U_SUCCESS(errorCode) && impl->ensureCanonIterData(errorCode) &&
296 impl->isCanonSegmentStarter(c);
297 }
298 #endif
299
300 static UBool isPOSIX_alnum(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
301 return u_isalnumPOSIX(c);
302 }
303
304 static UBool isPOSIX_blank(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
305 return u_isblank(c);
306 }
307
308 static UBool isPOSIX_graph(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
309 return u_isgraphPOSIX(c);
310 }
311
312 static UBool isPOSIX_print(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
313 return u_isprintPOSIX(c);
314 }
315
316 static UBool isPOSIX_xdigit(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
317 return u_isxdigit(c);
318 }
319
320 static UBool isRegionalIndicator(const BinaryProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
321 // Property starts are a subset of lb=RI etc.
322 return 0x1F1E6<=c && c<=0x1F1FF;
323 }
324
325 static const BinaryProperty binProps[UCHAR_BINARY_LIMIT]={
326 /*
327 * column and mask values for binary properties from u_getUnicodeProperties().
328 * Must be in order of corresponding UProperty,
329 * and there must be exactly one entry per binary UProperty.
330 *
331 * Properties with mask==0 are handled in code.
332 * For them, column is the UPropertySource value.
333 */
334 { 1, U_MASK(UPROPS_ALPHABETIC), defaultContains },
335 { 1, U_MASK(UPROPS_ASCII_HEX_DIGIT), defaultContains },
336 { UPROPS_SRC_BIDI, 0, isBidiControl },
337 { UPROPS_SRC_BIDI, 0, isMirrored },
338 { 1, U_MASK(UPROPS_DASH), defaultContains },
339 { 1, U_MASK(UPROPS_DEFAULT_IGNORABLE_CODE_POINT), defaultContains },
340 { 1, U_MASK(UPROPS_DEPRECATED), defaultContains },
341 { 1, U_MASK(UPROPS_DIACRITIC), defaultContains },
342 { 1, U_MASK(UPROPS_EXTENDER), defaultContains },
343 { UPROPS_SRC_NFC, 0, hasFullCompositionExclusion },
344 { 1, U_MASK(UPROPS_GRAPHEME_BASE), defaultContains },
345 { 1, U_MASK(UPROPS_GRAPHEME_EXTEND), defaultContains },
346 { 1, U_MASK(UPROPS_GRAPHEME_LINK), defaultContains },
347 { 1, U_MASK(UPROPS_HEX_DIGIT), defaultContains },
348 { 1, U_MASK(UPROPS_HYPHEN), defaultContains },
349 { 1, U_MASK(UPROPS_ID_CONTINUE), defaultContains },
350 { 1, U_MASK(UPROPS_ID_START), defaultContains },
351 { 1, U_MASK(UPROPS_IDEOGRAPHIC), defaultContains },
352 { 1, U_MASK(UPROPS_IDS_BINARY_OPERATOR), defaultContains },
353 { 1, U_MASK(UPROPS_IDS_TRINARY_OPERATOR), defaultContains },
354 { UPROPS_SRC_BIDI, 0, isJoinControl },
355 { 1, U_MASK(UPROPS_LOGICAL_ORDER_EXCEPTION), defaultContains },
356 { UPROPS_SRC_CASE, 0, caseBinaryPropertyContains }, // UCHAR_LOWERCASE
357 { 1, U_MASK(UPROPS_MATH), defaultContains },
358 { 1, U_MASK(UPROPS_NONCHARACTER_CODE_POINT), defaultContains },
359 { 1, U_MASK(UPROPS_QUOTATION_MARK), defaultContains },
360 { 1, U_MASK(UPROPS_RADICAL), defaultContains },
361 { UPROPS_SRC_CASE, 0, caseBinaryPropertyContains }, // UCHAR_SOFT_DOTTED
362 { 1, U_MASK(UPROPS_TERMINAL_PUNCTUATION), defaultContains },
363 { 1, U_MASK(UPROPS_UNIFIED_IDEOGRAPH), defaultContains },
364 { UPROPS_SRC_CASE, 0, caseBinaryPropertyContains }, // UCHAR_UPPERCASE
365 { 1, U_MASK(UPROPS_WHITE_SPACE), defaultContains },
366 { 1, U_MASK(UPROPS_XID_CONTINUE), defaultContains },
367 { 1, U_MASK(UPROPS_XID_START), defaultContains },
368 { UPROPS_SRC_CASE, 0, caseBinaryPropertyContains }, // UCHAR_CASE_SENSITIVE
369 { 1, U_MASK(UPROPS_S_TERM), defaultContains },
370 { 1, U_MASK(UPROPS_VARIATION_SELECTOR), defaultContains },
371 { UPROPS_SRC_NFC, 0, isNormInert }, // UCHAR_NFD_INERT
372 { UPROPS_SRC_NFKC, 0, isNormInert }, // UCHAR_NFKD_INERT
373 { UPROPS_SRC_NFC, 0, isNormInert }, // UCHAR_NFC_INERT
374 { UPROPS_SRC_NFKC, 0, isNormInert }, // UCHAR_NFKC_INERT
375 { UPROPS_SRC_NFC_CANON_ITER, 0, isCanonSegmentStarter },
376 { 1, U_MASK(UPROPS_PATTERN_SYNTAX), defaultContains },
377 { 1, U_MASK(UPROPS_PATTERN_WHITE_SPACE), defaultContains },
378 { UPROPS_SRC_CHAR_AND_PROPSVEC, 0, isPOSIX_alnum },
379 { UPROPS_SRC_CHAR, 0, isPOSIX_blank },
380 { UPROPS_SRC_CHAR, 0, isPOSIX_graph },
381 { UPROPS_SRC_CHAR, 0, isPOSIX_print },
382 { UPROPS_SRC_CHAR, 0, isPOSIX_xdigit },
383 { UPROPS_SRC_CASE, 0, caseBinaryPropertyContains }, // UCHAR_CASED
384 { UPROPS_SRC_CASE, 0, caseBinaryPropertyContains }, // UCHAR_CASE_IGNORABLE
385 { UPROPS_SRC_CASE, 0, caseBinaryPropertyContains }, // UCHAR_CHANGES_WHEN_LOWERCASED
386 { UPROPS_SRC_CASE, 0, caseBinaryPropertyContains }, // UCHAR_CHANGES_WHEN_UPPERCASED
387 { UPROPS_SRC_CASE, 0, caseBinaryPropertyContains }, // UCHAR_CHANGES_WHEN_TITLECASED
388 { UPROPS_SRC_CASE_AND_NORM, 0, changesWhenCasefolded },
389 { UPROPS_SRC_CASE, 0, caseBinaryPropertyContains }, // UCHAR_CHANGES_WHEN_CASEMAPPED
390 { UPROPS_SRC_NFKC_CF, 0, changesWhenNFKC_Casefolded },
391 { 2, U_MASK(UPROPS_2_EMOJI), defaultContains },
392 { 2, U_MASK(UPROPS_2_EMOJI_PRESENTATION), defaultContains },
393 { 2, U_MASK(UPROPS_2_EMOJI_MODIFIER), defaultContains },
394 { 2, U_MASK(UPROPS_2_EMOJI_MODIFIER_BASE), defaultContains },
395 { 2, U_MASK(UPROPS_2_EMOJI_COMPONENT), defaultContains },
396 { 2, 0, isRegionalIndicator },
397 { 1, U_MASK(UPROPS_PREPENDED_CONCATENATION_MARK), defaultContains },
398 { 2, U_MASK(UPROPS_2_EXTENDED_PICTOGRAPHIC), defaultContains },
399 };
400
401 U_CAPI UBool U_EXPORT2
402 u_hasBinaryProperty(UChar32 c, UProperty which) {
403 /* c is range-checked in the functions that are called from here */
404 if(which<UCHAR_BINARY_START || UCHAR_BINARY_LIMIT<=which) {
405 /* not a known binary property */
406 return FALSE;
407 } else {
408 const BinaryProperty &prop=binProps[which];
409 return prop.contains(prop, c, which);
410 }
411 }
412
413 // Apple-only specific version of the above
414 U_CAPI UBool U_EXPORT2
415 u_isEmoji(UChar32 c) {
416 const BinaryProperty &prop=binProps[UCHAR_EMOJI];
417 return prop.contains(prop, c, UCHAR_EMOJI);
418 }
419
420 struct IntProperty;
421
422 typedef int32_t IntPropertyGetValue(const IntProperty &prop, UChar32 c, UProperty which);
423 typedef int32_t IntPropertyGetMaxValue(const IntProperty &prop, UProperty which);
424
425 struct IntProperty {
426 int32_t column; // SRC_PROPSVEC column, or "source" if mask==0
427 uint32_t mask;
428 int32_t shift; // =maxValue if getMaxValueFromShift() is used
429 IntPropertyGetValue *getValue;
430 IntPropertyGetMaxValue *getMaxValue;
431 };
432
433 static int32_t defaultGetValue(const IntProperty &prop, UChar32 c, UProperty /*which*/) {
434 /* systematic, directly stored properties */
435 return (int32_t)(u_getUnicodeProperties(c, prop.column)&prop.mask)>>prop.shift;
436 }
437
438 static int32_t defaultGetMaxValue(const IntProperty &prop, UProperty /*which*/) {
439 return (uprv_getMaxValues(prop.column)&prop.mask)>>prop.shift;
440 }
441
442 static int32_t getMaxValueFromShift(const IntProperty &prop, UProperty /*which*/) {
443 return prop.shift;
444 }
445
446 static int32_t getBiDiClass(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
447 return (int32_t)u_charDirection(c);
448 }
449
450 static int32_t getBiDiPairedBracketType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
451 return (int32_t)ubidi_getPairedBracketType(c);
452 }
453
454 static int32_t biDiGetMaxValue(const IntProperty &/*prop*/, UProperty which) {
455 return ubidi_getMaxValue(which);
456 }
457
458 #if UCONFIG_NO_NORMALIZATION
459 static int32_t getCombiningClass(const IntProperty &, UChar32, UProperty) {
460 return 0;
461 }
462 #else
463 static int32_t getCombiningClass(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
464 return u_getCombiningClass(c);
465 }
466 #endif
467
468 static int32_t getGeneralCategory(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
469 return (int32_t)u_charType(c);
470 }
471
472 static int32_t getJoiningGroup(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
473 return ubidi_getJoiningGroup(c);
474 }
475
476 static int32_t getJoiningType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
477 return ubidi_getJoiningType(c);
478 }
479
480 static int32_t getNumericType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
481 int32_t ntv=(int32_t)GET_NUMERIC_TYPE_VALUE(u_getMainProperties(c));
482 return UPROPS_NTV_GET_TYPE(ntv);
483 }
484
485 static int32_t getScript(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
486 UErrorCode errorCode=U_ZERO_ERROR;
487 return (int32_t)uscript_getScript(c, &errorCode);
488 }
489
490 static int32_t scriptGetMaxValue(const IntProperty &/*prop*/, UProperty /*which*/) {
491 uint32_t scriptX=uprv_getMaxValues(0)&UPROPS_SCRIPT_X_MASK;
492 return uprops_mergeScriptCodeOrIndex(scriptX);
493 }
494
495 /*
496 * Map some of the Grapheme Cluster Break values to Hangul Syllable Types.
497 * Hangul_Syllable_Type is fully redundant with a subset of Grapheme_Cluster_Break.
498 */
499 static const UHangulSyllableType gcbToHst[]={
500 U_HST_NOT_APPLICABLE, /* U_GCB_OTHER */
501 U_HST_NOT_APPLICABLE, /* U_GCB_CONTROL */
502 U_HST_NOT_APPLICABLE, /* U_GCB_CR */
503 U_HST_NOT_APPLICABLE, /* U_GCB_EXTEND */
504 U_HST_LEADING_JAMO, /* U_GCB_L */
505 U_HST_NOT_APPLICABLE, /* U_GCB_LF */
506 U_HST_LV_SYLLABLE, /* U_GCB_LV */
507 U_HST_LVT_SYLLABLE, /* U_GCB_LVT */
508 U_HST_TRAILING_JAMO, /* U_GCB_T */
509 U_HST_VOWEL_JAMO /* U_GCB_V */
510 /*
511 * Omit GCB values beyond what we need for hst.
512 * The code below checks for the array length.
513 */
514 };
515
516 static int32_t getHangulSyllableType(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
517 /* see comments on gcbToHst[] above */
518 int32_t gcb=(int32_t)(u_getUnicodeProperties(c, 2)&UPROPS_GCB_MASK)>>UPROPS_GCB_SHIFT;
519 if(gcb<UPRV_LENGTHOF(gcbToHst)) {
520 return gcbToHst[gcb];
521 } else {
522 return U_HST_NOT_APPLICABLE;
523 }
524 }
525
526 #if UCONFIG_NO_NORMALIZATION
527 static int32_t getNormQuickCheck(const IntProperty &, UChar32, UProperty) {
528 return 0;
529 }
530 #else
531 static int32_t getNormQuickCheck(const IntProperty &/*prop*/, UChar32 c, UProperty which) {
532 return (int32_t)unorm_getQuickCheck(c, (UNormalizationMode)(which-UCHAR_NFD_QUICK_CHECK+UNORM_NFD));
533 }
534 #endif
535
536 #if UCONFIG_NO_NORMALIZATION
537 static int32_t getLeadCombiningClass(const IntProperty &, UChar32, UProperty) {
538 return 0;
539 }
540 #else
541 static int32_t getLeadCombiningClass(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
542 return unorm_getFCD16(c)>>8;
543 }
544 #endif
545
546 #if UCONFIG_NO_NORMALIZATION
547 static int32_t getTrailCombiningClass(const IntProperty &, UChar32, UProperty) {
548 return 0;
549 }
550 #else
551 static int32_t getTrailCombiningClass(const IntProperty &/*prop*/, UChar32 c, UProperty /*which*/) {
552 return unorm_getFCD16(c)&0xff;
553 }
554 #endif
555
556 static int32_t getInPC(const IntProperty &, UChar32 c, UProperty) {
557 return ulayout_ensureData() && gInpcTrie != nullptr ? ucptrie_get(gInpcTrie, c) : 0;
558 }
559
560 static int32_t getInSC(const IntProperty &, UChar32 c, UProperty) {
561 return ulayout_ensureData() && gInscTrie != nullptr ? ucptrie_get(gInscTrie, c) : 0;
562 }
563
564 static int32_t getVo(const IntProperty &, UChar32 c, UProperty) {
565 return ulayout_ensureData() && gVoTrie != nullptr ? ucptrie_get(gVoTrie, c) : 0;
566 }
567
568 static int32_t layoutGetMaxValue(const IntProperty &/*prop*/, UProperty which) {
569 if (!ulayout_ensureData()) { return 0; }
570 switch (which) {
571 case UCHAR_INDIC_POSITIONAL_CATEGORY:
572 return gMaxInpcValue;
573 case UCHAR_INDIC_SYLLABIC_CATEGORY:
574 return gMaxInscValue;
575 case UCHAR_VERTICAL_ORIENTATION:
576 return gMaxVoValue;
577 default:
578 return 0;
579 }
580 }
581
582 static const IntProperty intProps[UCHAR_INT_LIMIT-UCHAR_INT_START]={
583 /*
584 * column, mask and shift values for int-value properties from u_getUnicodeProperties().
585 * Must be in order of corresponding UProperty,
586 * and there must be exactly one entry per int UProperty.
587 *
588 * Properties with mask==0 are handled in code.
589 * For them, column is the UPropertySource value.
590 */
591 { UPROPS_SRC_BIDI, 0, 0, getBiDiClass, biDiGetMaxValue },
592 { 0, UPROPS_BLOCK_MASK, UPROPS_BLOCK_SHIFT, defaultGetValue, defaultGetMaxValue },
593 { UPROPS_SRC_NFC, 0, 0xff, getCombiningClass, getMaxValueFromShift },
594 { 2, UPROPS_DT_MASK, 0, defaultGetValue, defaultGetMaxValue },
595 { 0, UPROPS_EA_MASK, UPROPS_EA_SHIFT, defaultGetValue, defaultGetMaxValue },
596 { UPROPS_SRC_CHAR, 0, (int32_t)U_CHAR_CATEGORY_COUNT-1,getGeneralCategory, getMaxValueFromShift },
597 { UPROPS_SRC_BIDI, 0, 0, getJoiningGroup, biDiGetMaxValue },
598 { UPROPS_SRC_BIDI, 0, 0, getJoiningType, biDiGetMaxValue },
599 { 2, UPROPS_LB_MASK, UPROPS_LB_SHIFT, defaultGetValue, defaultGetMaxValue },
600 { UPROPS_SRC_CHAR, 0, (int32_t)U_NT_COUNT-1, getNumericType, getMaxValueFromShift },
601 { UPROPS_SRC_PROPSVEC, 0, 0, getScript, scriptGetMaxValue },
602 { UPROPS_SRC_PROPSVEC, 0, (int32_t)U_HST_COUNT-1, getHangulSyllableType, getMaxValueFromShift },
603 // UCHAR_NFD_QUICK_CHECK: max=1=YES -- never "maybe", only "no" or "yes"
604 { UPROPS_SRC_NFC, 0, (int32_t)UNORM_YES, getNormQuickCheck, getMaxValueFromShift },
605 // UCHAR_NFKD_QUICK_CHECK: max=1=YES -- never "maybe", only "no" or "yes"
606 { UPROPS_SRC_NFKC, 0, (int32_t)UNORM_YES, getNormQuickCheck, getMaxValueFromShift },
607 // UCHAR_NFC_QUICK_CHECK: max=2=MAYBE
608 { UPROPS_SRC_NFC, 0, (int32_t)UNORM_MAYBE, getNormQuickCheck, getMaxValueFromShift },
609 // UCHAR_NFKC_QUICK_CHECK: max=2=MAYBE
610 { UPROPS_SRC_NFKC, 0, (int32_t)UNORM_MAYBE, getNormQuickCheck, getMaxValueFromShift },
611 { UPROPS_SRC_NFC, 0, 0xff, getLeadCombiningClass, getMaxValueFromShift },
612 { UPROPS_SRC_NFC, 0, 0xff, getTrailCombiningClass, getMaxValueFromShift },
613 { 2, UPROPS_GCB_MASK, UPROPS_GCB_SHIFT, defaultGetValue, defaultGetMaxValue },
614 { 2, UPROPS_SB_MASK, UPROPS_SB_SHIFT, defaultGetValue, defaultGetMaxValue },
615 { 2, UPROPS_WB_MASK, UPROPS_WB_SHIFT, defaultGetValue, defaultGetMaxValue },
616 { UPROPS_SRC_BIDI, 0, 0, getBiDiPairedBracketType, biDiGetMaxValue },
617 { UPROPS_SRC_INPC, 0, 0, getInPC, layoutGetMaxValue },
618 { UPROPS_SRC_INSC, 0, 0, getInSC, layoutGetMaxValue },
619 { UPROPS_SRC_VO, 0, 0, getVo, layoutGetMaxValue },
620 };
621
622 U_CAPI int32_t U_EXPORT2
623 u_getIntPropertyValue(UChar32 c, UProperty which) {
624 if(which<UCHAR_INT_START) {
625 if(UCHAR_BINARY_START<=which && which<UCHAR_BINARY_LIMIT) {
626 const BinaryProperty &prop=binProps[which];
627 return prop.contains(prop, c, which);
628 }
629 } else if(which<UCHAR_INT_LIMIT) {
630 const IntProperty &prop=intProps[which-UCHAR_INT_START];
631 return prop.getValue(prop, c, which);
632 } else if(which==UCHAR_GENERAL_CATEGORY_MASK) {
633 return U_MASK(u_charType(c));
634 }
635 return 0; // undefined
636 }
637
638 U_CAPI int32_t U_EXPORT2
639 u_getIntPropertyMinValue(UProperty /*which*/) {
640 return 0; /* all binary/enum/int properties have a minimum value of 0 */
641 }
642
643 U_CAPI int32_t U_EXPORT2
644 u_getIntPropertyMaxValue(UProperty which) {
645 if(which<UCHAR_INT_START) {
646 if(UCHAR_BINARY_START<=which && which<UCHAR_BINARY_LIMIT) {
647 return 1; // maximum TRUE for all binary properties
648 }
649 } else if(which<UCHAR_INT_LIMIT) {
650 const IntProperty &prop=intProps[which-UCHAR_INT_START];
651 return prop.getMaxValue(prop, which);
652 }
653 return -1; // undefined
654 }
655
656 U_CFUNC UPropertySource U_EXPORT2
657 uprops_getSource(UProperty which) {
658 if(which<UCHAR_BINARY_START) {
659 return UPROPS_SRC_NONE; /* undefined */
660 } else if(which<UCHAR_BINARY_LIMIT) {
661 const BinaryProperty &prop=binProps[which];
662 if(prop.mask!=0) {
663 return UPROPS_SRC_PROPSVEC;
664 } else {
665 return (UPropertySource)prop.column;
666 }
667 } else if(which<UCHAR_INT_START) {
668 return UPROPS_SRC_NONE; /* undefined */
669 } else if(which<UCHAR_INT_LIMIT) {
670 const IntProperty &prop=intProps[which-UCHAR_INT_START];
671 if(prop.mask!=0) {
672 return UPROPS_SRC_PROPSVEC;
673 } else {
674 return (UPropertySource)prop.column;
675 }
676 } else if(which<UCHAR_STRING_START) {
677 switch(which) {
678 case UCHAR_GENERAL_CATEGORY_MASK:
679 case UCHAR_NUMERIC_VALUE:
680 return UPROPS_SRC_CHAR;
681
682 default:
683 return UPROPS_SRC_NONE;
684 }
685 } else if(which<UCHAR_STRING_LIMIT) {
686 switch(which) {
687 case UCHAR_AGE:
688 return UPROPS_SRC_PROPSVEC;
689
690 case UCHAR_BIDI_MIRRORING_GLYPH:
691 return UPROPS_SRC_BIDI;
692
693 case UCHAR_CASE_FOLDING:
694 case UCHAR_LOWERCASE_MAPPING:
695 case UCHAR_SIMPLE_CASE_FOLDING:
696 case UCHAR_SIMPLE_LOWERCASE_MAPPING:
697 case UCHAR_SIMPLE_TITLECASE_MAPPING:
698 case UCHAR_SIMPLE_UPPERCASE_MAPPING:
699 case UCHAR_TITLECASE_MAPPING:
700 case UCHAR_UPPERCASE_MAPPING:
701 return UPROPS_SRC_CASE;
702
703 case UCHAR_ISO_COMMENT:
704 case UCHAR_NAME:
705 case UCHAR_UNICODE_1_NAME:
706 return UPROPS_SRC_NAMES;
707
708 default:
709 return UPROPS_SRC_NONE;
710 }
711 } else {
712 switch(which) {
713 case UCHAR_SCRIPT_EXTENSIONS:
714 return UPROPS_SRC_PROPSVEC;
715 default:
716 return UPROPS_SRC_NONE; /* undefined */
717 }
718 }
719 }
720
721 U_CFUNC void U_EXPORT2
722 uprops_addPropertyStarts(UPropertySource src, const USetAdder *sa, UErrorCode *pErrorCode) {
723 if (!ulayout_ensureData(*pErrorCode)) { return; }
724 const UCPTrie *trie;
725 switch (src) {
726 case UPROPS_SRC_INPC:
727 trie = gInpcTrie;
728 break;
729 case UPROPS_SRC_INSC:
730 trie = gInscTrie;
731 break;
732 case UPROPS_SRC_VO:
733 trie = gVoTrie;
734 break;
735 default:
736 *pErrorCode = U_ILLEGAL_ARGUMENT_ERROR;
737 return;
738 }
739
740 if (trie == nullptr) {
741 *pErrorCode = U_MISSING_RESOURCE_ERROR;
742 return;
743 }
744
745 // Add the start code point of each same-value range of the trie.
746 UChar32 start = 0, end;
747 while ((end = ucptrie_getRange(trie, start, UCPMAP_RANGE_NORMAL, 0,
748 nullptr, nullptr, nullptr)) >= 0) {
749 sa->add(sa->set, start);
750 start = end + 1;
751 }
752 }
753
754 #if !UCONFIG_NO_NORMALIZATION
755
756 U_CAPI int32_t U_EXPORT2
757 u_getFC_NFKC_Closure(UChar32 c, UChar *dest, int32_t destCapacity, UErrorCode *pErrorCode) {
758 if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) {
759 return 0;
760 }
761 if(destCapacity<0 || (dest==NULL && destCapacity>0)) {
762 *pErrorCode=U_ILLEGAL_ARGUMENT_ERROR;
763 return 0;
764 }
765 // Compute the FC_NFKC_Closure on the fly:
766 // We have the API for complete coverage of Unicode properties, although
767 // this value by itself is not useful via API.
768 // (What could be useful is a custom normalization table that combines
769 // case folding and NFKC.)
770 // For the derivation, see Unicode's DerivedNormalizationProps.txt.
771 const Normalizer2 *nfkc=Normalizer2::getNFKCInstance(*pErrorCode);
772 if(U_FAILURE(*pErrorCode)) {
773 return 0;
774 }
775 // first: b = NFKC(Fold(a))
776 UnicodeString folded1String;
777 const UChar *folded1;
778 int32_t folded1Length=ucase_toFullFolding(c, &folded1, U_FOLD_CASE_DEFAULT);
779 if(folded1Length<0) {
780 const Normalizer2Impl *nfkcImpl=Normalizer2Factory::getImpl(nfkc);
781 if(nfkcImpl->getCompQuickCheck(nfkcImpl->getNorm16(c))!=UNORM_NO) {
782 return u_terminateUChars(dest, destCapacity, 0, pErrorCode); // c does not change at all under CaseFolding+NFKC
783 }
784 folded1String.setTo(c);
785 } else {
786 if(folded1Length>UCASE_MAX_STRING_LENGTH) {
787 folded1String.setTo(folded1Length);
788 } else {
789 folded1String.setTo(FALSE, folded1, folded1Length);
790 }
791 }
792 UnicodeString kc1=nfkc->normalize(folded1String, *pErrorCode);
793 // second: c = NFKC(Fold(b))
794 UnicodeString folded2String(kc1);
795 UnicodeString kc2=nfkc->normalize(folded2String.foldCase(), *pErrorCode);
796 // if (c != b) add the mapping from a to c
797 if(U_FAILURE(*pErrorCode) || kc1==kc2) {
798 return u_terminateUChars(dest, destCapacity, 0, pErrorCode);
799 } else {
800 return kc2.extract(dest, destCapacity, *pErrorCode);
801 }
802 }
803
804 #endif
mirror of ICU