1 /*
2 *******************************************************************************
3 * Copyright (C) 2004-2014, International Business Machines
4 * Corporation and others. All Rights Reserved.
5 *******************************************************************************
6 * file name: ucol_sit.cpp
7 * encoding: US-ASCII
8 * tab size: 8 (not used)
9 * indentation:4
10 *
11 * Modification history
12 * Date Name Comments
13 * 03/12/2004 weiv Creation
14 */
15
16 #include "unicode/ustring.h"
17 #include "unicode/udata.h"
18 #include "unicode/utf16.h"
19 #include "utracimp.h"
20 #include "ucol_imp.h"
21 #include "cmemory.h"
22 #include "cstring.h"
23 #include "uresimp.h"
24 #include "unicode/coll.h"
25
26 #ifdef UCOL_TRACE_SIT
27 # include <stdio.h>
28 #endif
29
30 #if !UCONFIG_NO_COLLATION
31
32 #include "unicode/tblcoll.h"
33
34 enum OptionsList {
35 UCOL_SIT_LANGUAGE = 0,
36 UCOL_SIT_SCRIPT = 1,
37 UCOL_SIT_REGION = 2,
38 UCOL_SIT_VARIANT = 3,
39 UCOL_SIT_KEYWORD = 4,
40 UCOL_SIT_PROVIDER = 5,
41 UCOL_SIT_LOCELEMENT_MAX = UCOL_SIT_PROVIDER, /* the last element that's part of LocElements */
42
43 UCOL_SIT_BCP47,
44 UCOL_SIT_STRENGTH,
45 UCOL_SIT_CASE_LEVEL,
46 UCOL_SIT_CASE_FIRST,
47 UCOL_SIT_NUMERIC_COLLATION,
48 UCOL_SIT_ALTERNATE_HANDLING,
49 UCOL_SIT_NORMALIZATION_MODE,
50 UCOL_SIT_FRENCH_COLLATION,
51 UCOL_SIT_HIRAGANA_QUATERNARY,
52 UCOL_SIT_VARIABLE_TOP,
53 UCOL_SIT_VARIABLE_TOP_VALUE,
54 UCOL_SIT_ITEMS_COUNT
55 };
56
57 /* option starters chars. */
58 static const char alternateHArg = 'A';
59 static const char variableTopValArg = 'B';
60 static const char caseFirstArg = 'C';
61 static const char numericCollArg = 'D';
62 static const char caseLevelArg = 'E';
63 static const char frenchCollArg = 'F';
64 static const char hiraganaQArg = 'H';
65 static const char keywordArg = 'K';
66 static const char languageArg = 'L';
67 static const char normArg = 'N';
68 static const char providerArg = 'P';
69 static const char regionArg = 'R';
70 static const char strengthArg = 'S';
71 static const char variableTopArg = 'T';
72 static const char variantArg = 'V';
73 static const char RFC3066Arg = 'X';
74 static const char scriptArg = 'Z';
75
76 static const char collationKeyword[] = "@collation=";
77 static const char providerKeyword[] = "@sp=";
78
79
80 static const int32_t locElementCount = UCOL_SIT_LOCELEMENT_MAX+1;
81 static const int32_t locElementCapacity = 32;
82 static const int32_t loc3066Capacity = 256;
83 static const int32_t locProviderCapacity = 10;
84 static const int32_t internalBufferSize = 512;
85
86 /* structure containing specification of a collator. Initialized
87 * from a short string. Also used to construct a short string from a
88 * collator instance
89 */
90 struct CollatorSpec {
91 char locElements[locElementCount][locElementCapacity];
92 char locale[loc3066Capacity];
93 char provider[locProviderCapacity];
94 UColAttributeValue options[UCOL_ATTRIBUTE_COUNT];
95 uint32_t variableTopValue;
96 UChar variableTopString[locElementCapacity];
97 int32_t variableTopStringLen;
98 UBool variableTopSet;
99 struct {
100 const char *start;
101 int32_t len;
102 } entries[UCOL_SIT_ITEMS_COUNT];
103 };
104
105
106 /* structure for converting between character attribute
107 * representation and real collation attribute value.
108 */
109 struct AttributeConversion {
110 char letter;
111 UColAttributeValue value;
112 };
113
114 static const AttributeConversion conversions[12] = {
115 { '1', UCOL_PRIMARY },
116 { '2', UCOL_SECONDARY },
117 { '3', UCOL_TERTIARY },
118 { '4', UCOL_QUATERNARY },
119 { 'D', UCOL_DEFAULT },
120 { 'I', UCOL_IDENTICAL },
121 { 'L', UCOL_LOWER_FIRST },
122 { 'N', UCOL_NON_IGNORABLE },
123 { 'O', UCOL_ON },
124 { 'S', UCOL_SHIFTED },
125 { 'U', UCOL_UPPER_FIRST },
126 { 'X', UCOL_OFF }
127 };
128
129
130 static UColAttributeValue
ucol_sit_letterToAttributeValue(char letter,UErrorCode * status)131 ucol_sit_letterToAttributeValue(char letter, UErrorCode *status) {
132 uint32_t i = 0;
133 for(i = 0; i < sizeof(conversions)/sizeof(conversions[0]); i++) {
134 if(conversions[i].letter == letter) {
135 return conversions[i].value;
136 }
137 }
138 *status = U_ILLEGAL_ARGUMENT_ERROR;
139 #ifdef UCOL_TRACE_SIT
140 fprintf(stderr, "%s:%d: unknown letter %c: %s\n", __FILE__, __LINE__, letter, u_errorName(*status));
141 #endif
142 return UCOL_DEFAULT;
143 }
144
145 /* function prototype for functions used to parse a short string */
146 U_CDECL_BEGIN
147 typedef const char* U_CALLCONV
148 ActionFunction(CollatorSpec *spec, uint32_t value1, const char* string,
149 UErrorCode *status);
150 U_CDECL_END
151
152 U_CDECL_BEGIN
153 static const char* U_CALLCONV
_processLocaleElement(CollatorSpec * spec,uint32_t value,const char * string,UErrorCode * status)154 _processLocaleElement(CollatorSpec *spec, uint32_t value, const char* string,
155 UErrorCode *status)
156 {
157 int32_t len = 0;
158 do {
159 if(value == UCOL_SIT_LANGUAGE || value == UCOL_SIT_KEYWORD || value == UCOL_SIT_PROVIDER) {
160 spec->locElements[value][len++] = uprv_tolower(*string);
161 } else {
162 spec->locElements[value][len++] = *string;
163 }
164 } while(*(++string) != '_' && *string && len < locElementCapacity);
165 if(len >= locElementCapacity) {
166 *status = U_BUFFER_OVERFLOW_ERROR;
167 return string;
168 }
169 // don't skip the underscore at the end
170 return string;
171 }
172 U_CDECL_END
173
174 U_CDECL_BEGIN
175 static const char* U_CALLCONV
_processRFC3066Locale(CollatorSpec * spec,uint32_t,const char * string,UErrorCode * status)176 _processRFC3066Locale(CollatorSpec *spec, uint32_t, const char* string,
177 UErrorCode *status)
178 {
179 char terminator = *string;
180 string++;
181 const char *end = uprv_strchr(string+1, terminator);
182 if(end == NULL || end - string >= loc3066Capacity) {
183 *status = U_BUFFER_OVERFLOW_ERROR;
184 return string;
185 } else {
186 uprv_strncpy(spec->locale, string, end-string);
187 return end+1;
188 }
189 }
190
191 U_CDECL_END
192
193 U_CDECL_BEGIN
194 static const char* U_CALLCONV
_processCollatorOption(CollatorSpec * spec,uint32_t option,const char * string,UErrorCode * status)195 _processCollatorOption(CollatorSpec *spec, uint32_t option, const char* string,
196 UErrorCode *status)
197 {
198 spec->options[option] = ucol_sit_letterToAttributeValue(*string, status);
199 if((*(++string) != '_' && *string) || U_FAILURE(*status)) {
200 #ifdef UCOL_TRACE_SIT
201 fprintf(stderr, "%s:%d: unknown collator option at '%s': %s\n", __FILE__, __LINE__, string, u_errorName(*status));
202 #endif
203 *status = U_ILLEGAL_ARGUMENT_ERROR;
204 }
205 return string;
206 }
207 U_CDECL_END
208
209
210 static UChar
readHexCodeUnit(const char ** string,UErrorCode * status)211 readHexCodeUnit(const char **string, UErrorCode *status)
212 {
213 UChar result = 0;
214 int32_t value = 0;
215 char c;
216 int32_t noDigits = 0;
217 while((c = **string) != 0 && noDigits < 4) {
218 if( c >= '0' && c <= '9') {
219 value = c - '0';
220 } else if ( c >= 'a' && c <= 'f') {
221 value = c - 'a' + 10;
222 } else if ( c >= 'A' && c <= 'F') {
223 value = c - 'A' + 10;
224 } else {
225 *status = U_ILLEGAL_ARGUMENT_ERROR;
226 #ifdef UCOL_TRACE_SIT
227 fprintf(stderr, "%s:%d: Bad hex char at '%s': %s\n", __FILE__, __LINE__, *string, u_errorName(*status));
228 #endif
229 return 0;
230 }
231 result = (result << 4) | (UChar)value;
232 noDigits++;
233 (*string)++;
234 }
235 // if the string was terminated before we read 4 digits, set an error
236 if(noDigits < 4) {
237 *status = U_ILLEGAL_ARGUMENT_ERROR;
238 #ifdef UCOL_TRACE_SIT
239 fprintf(stderr, "%s:%d: Short (only %d digits, wanted 4) at '%s': %s\n", __FILE__, __LINE__, noDigits,*string, u_errorName(*status));
240 #endif
241 }
242 return result;
243 }
244
245 U_CDECL_BEGIN
246 static const char* U_CALLCONV
_processVariableTop(CollatorSpec * spec,uint32_t value1,const char * string,UErrorCode * status)247 _processVariableTop(CollatorSpec *spec, uint32_t value1, const char* string, UErrorCode *status)
248 {
249 // get four digits
250 int32_t i = 0;
251 if(!value1) {
252 while(U_SUCCESS(*status) && i < locElementCapacity && *string != 0 && *string != '_') {
253 spec->variableTopString[i++] = readHexCodeUnit(&string, status);
254 }
255 spec->variableTopStringLen = i;
256 if(i == locElementCapacity && *string != 0 && *string != '_') {
257 *status = U_BUFFER_OVERFLOW_ERROR;
258 }
259 } else {
260 spec->variableTopValue = readHexCodeUnit(&string, status);
261 }
262 if(U_SUCCESS(*status)) {
263 spec->variableTopSet = TRUE;
264 }
265 return string;
266 }
267 U_CDECL_END
268
269
270 /* Table for parsing short strings */
271 struct ShortStringOptions {
272 char optionStart;
273 ActionFunction *action;
274 uint32_t attr;
275 };
276
277 static const ShortStringOptions options[UCOL_SIT_ITEMS_COUNT] =
278 {
279 /* 10 ALTERNATE_HANDLING */ {alternateHArg, _processCollatorOption, UCOL_ALTERNATE_HANDLING }, // alternate N, S, D
280 /* 15 VARIABLE_TOP_VALUE */ {variableTopValArg, _processVariableTop, 1 },
281 /* 08 CASE_FIRST */ {caseFirstArg, _processCollatorOption, UCOL_CASE_FIRST }, // case first L, U, X, D
282 /* 09 NUMERIC_COLLATION */ {numericCollArg, _processCollatorOption, UCOL_NUMERIC_COLLATION }, // codan O, X, D
283 /* 07 CASE_LEVEL */ {caseLevelArg, _processCollatorOption, UCOL_CASE_LEVEL }, // case level O, X, D
284 /* 12 FRENCH_COLLATION */ {frenchCollArg, _processCollatorOption, UCOL_FRENCH_COLLATION }, // french O, X, D
285 /* 13 HIRAGANA_QUATERNARY] */ {hiraganaQArg, _processCollatorOption, UCOL_HIRAGANA_QUATERNARY_MODE }, // hiragana O, X, D
286 /* 04 KEYWORD */ {keywordArg, _processLocaleElement, UCOL_SIT_KEYWORD }, // keyword
287 /* 00 LANGUAGE */ {languageArg, _processLocaleElement, UCOL_SIT_LANGUAGE }, // language
288 /* 11 NORMALIZATION_MODE */ {normArg, _processCollatorOption, UCOL_NORMALIZATION_MODE }, // norm O, X, D
289 /* 02 REGION */ {regionArg, _processLocaleElement, UCOL_SIT_REGION }, // region
290 /* 06 STRENGTH */ {strengthArg, _processCollatorOption, UCOL_STRENGTH }, // strength 1, 2, 3, 4, I, D
291 /* 14 VARIABLE_TOP */ {variableTopArg, _processVariableTop, 0 },
292 /* 03 VARIANT */ {variantArg, _processLocaleElement, UCOL_SIT_VARIANT }, // variant
293 /* 05 RFC3066BIS */ {RFC3066Arg, _processRFC3066Locale, 0 }, // rfc3066bis locale name
294 /* 01 SCRIPT */ {scriptArg, _processLocaleElement, UCOL_SIT_SCRIPT }, // script
295 /* PROVIDER */ {providerArg, _processLocaleElement, UCOL_SIT_PROVIDER }
296 };
297
298
299 static
ucol_sit_readOption(const char * start,CollatorSpec * spec,UErrorCode * status)300 const char* ucol_sit_readOption(const char *start, CollatorSpec *spec,
301 UErrorCode *status)
302 {
303 int32_t i = 0;
304
305 for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
306 if(*start == options[i].optionStart) {
307 spec->entries[i].start = start;
308 const char* end = options[i].action(spec, options[i].attr, start+1, status);
309 spec->entries[i].len = (int32_t)(end - start);
310 return end;
311 }
312 }
313 *status = U_ILLEGAL_ARGUMENT_ERROR;
314 #ifdef UCOL_TRACE_SIT
315 fprintf(stderr, "%s:%d: Unknown option at '%s': %s\n", __FILE__, __LINE__, start, u_errorName(*status));
316 #endif
317 return start;
318 }
319
320 static
ucol_sit_initCollatorSpecs(CollatorSpec * spec)321 void ucol_sit_initCollatorSpecs(CollatorSpec *spec)
322 {
323 // reset everything
324 uprv_memset(spec, 0, sizeof(CollatorSpec));
325 // set collation options to default
326 int32_t i = 0;
327 for(i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) {
328 spec->options[i] = UCOL_DEFAULT;
329 }
330 }
331
332 static const char*
ucol_sit_readSpecs(CollatorSpec * s,const char * string,UParseError * parseError,UErrorCode * status)333 ucol_sit_readSpecs(CollatorSpec *s, const char *string,
334 UParseError *parseError, UErrorCode *status)
335 {
336 const char *definition = string;
337 while(U_SUCCESS(*status) && *string) {
338 string = ucol_sit_readOption(string, s, status);
339 // advance over '_'
340 while(*string && *string == '_') {
341 string++;
342 }
343 }
344 if(U_FAILURE(*status)) {
345 parseError->offset = (int32_t)(string - definition);
346 }
347 return string;
348 }
349
350 static
ucol_sit_dumpSpecs(CollatorSpec * s,char * destination,int32_t capacity,UErrorCode * status)351 int32_t ucol_sit_dumpSpecs(CollatorSpec *s, char *destination, int32_t capacity, UErrorCode *status)
352 {
353 int32_t i = 0, j = 0;
354 int32_t len = 0;
355 char optName;
356 if(U_SUCCESS(*status)) {
357 for(i = 0; i < UCOL_SIT_ITEMS_COUNT; i++) {
358 if(s->entries[i].start) {
359 if(len) {
360 if(len < capacity) {
361 uprv_strcat(destination, "_");
362 }
363 len++;
364 }
365 optName = *(s->entries[i].start);
366 if(optName == languageArg || optName == regionArg || optName == variantArg || optName == keywordArg) {
367 for(j = 0; j < s->entries[i].len; j++) {
368 if(len + j < capacity) {
369 destination[len+j] = uprv_toupper(*(s->entries[i].start+j));
370 }
371 }
372 len += s->entries[i].len;
373 } else {
374 len += s->entries[i].len;
375 if(len < capacity) {
376 uprv_strncat(destination,s->entries[i].start, s->entries[i].len);
377 }
378 }
379 }
380 }
381 return len;
382 } else {
383 return 0;
384 }
385 }
386
387 static void
ucol_sit_calculateWholeLocale(CollatorSpec * s)388 ucol_sit_calculateWholeLocale(CollatorSpec *s) {
389 // put the locale together, unless we have a done
390 // locale
391 if(s->locale[0] == 0) {
392 // first the language
393 uprv_strcat(s->locale, s->locElements[UCOL_SIT_LANGUAGE]);
394 // then the script, if present
395 if(*(s->locElements[UCOL_SIT_SCRIPT])) {
396 uprv_strcat(s->locale, "_");
397 uprv_strcat(s->locale, s->locElements[UCOL_SIT_SCRIPT]);
398 }
399 // then the region, if present
400 if(*(s->locElements[UCOL_SIT_REGION])) {
401 uprv_strcat(s->locale, "_");
402 uprv_strcat(s->locale, s->locElements[UCOL_SIT_REGION]);
403 } else if(*(s->locElements[UCOL_SIT_VARIANT])) { // if there is a variant, we need an underscore
404 uprv_strcat(s->locale, "_");
405 }
406 // add variant, if there
407 if(*(s->locElements[UCOL_SIT_VARIANT])) {
408 uprv_strcat(s->locale, "_");
409 uprv_strcat(s->locale, s->locElements[UCOL_SIT_VARIANT]);
410 }
411
412 // if there is a collation keyword, add that too
413 if(*(s->locElements[UCOL_SIT_KEYWORD])) {
414 uprv_strcat(s->locale, collationKeyword);
415 uprv_strcat(s->locale, s->locElements[UCOL_SIT_KEYWORD]);
416 }
417
418 // if there is a provider keyword, add that too
419 if(*(s->locElements[UCOL_SIT_PROVIDER])) {
420 uprv_strcat(s->locale, providerKeyword);
421 uprv_strcat(s->locale, s->locElements[UCOL_SIT_PROVIDER]);
422 }
423 }
424 }
425
426
427 U_CAPI void U_EXPORT2
ucol_prepareShortStringOpen(const char * definition,UBool,UParseError * parseError,UErrorCode * status)428 ucol_prepareShortStringOpen( const char *definition,
429 UBool,
430 UParseError *parseError,
431 UErrorCode *status)
432 {
433 if(U_FAILURE(*status)) return;
434
435 UParseError internalParseError;
436
437 if(!parseError) {
438 parseError = &internalParseError;
439 }
440 parseError->line = 0;
441 parseError->offset = 0;
442 parseError->preContext[0] = 0;
443 parseError->postContext[0] = 0;
444
445
446 // first we want to pick stuff out of short string.
447 // we'll end up with an UCA version, locale and a bunch of
448 // settings
449
450 // analyse the string in order to get everything we need.
451 CollatorSpec s;
452 ucol_sit_initCollatorSpecs(&s);
453 ucol_sit_readSpecs(&s, definition, parseError, status);
454 ucol_sit_calculateWholeLocale(&s);
455
456 char buffer[internalBufferSize];
457 uprv_memset(buffer, 0, internalBufferSize);
458 uloc_canonicalize(s.locale, buffer, internalBufferSize, status);
459
460 UResourceBundle *b = ures_open(U_ICUDATA_COLL, buffer, status);
461 /* we try to find stuff from keyword */
462 UResourceBundle *collations = ures_getByKey(b, "collations", NULL, status);
463 UResourceBundle *collElem = NULL;
464 char keyBuffer[256];
465 // if there is a keyword, we pick it up and try to get elements
466 if(!uloc_getKeywordValue(buffer, "collation", keyBuffer, 256, status)) {
467 // no keyword. we try to find the default setting, which will give us the keyword value
468 UResourceBundle *defaultColl = ures_getByKeyWithFallback(collations, "default", NULL, status);
469 if(U_SUCCESS(*status)) {
470 int32_t defaultKeyLen = 0;
471 const UChar *defaultKey = ures_getString(defaultColl, &defaultKeyLen, status);
472 u_UCharsToChars(defaultKey, keyBuffer, defaultKeyLen);
473 keyBuffer[defaultKeyLen] = 0;
474 } else {
475 *status = U_INTERNAL_PROGRAM_ERROR;
476 return;
477 }
478 ures_close(defaultColl);
479 }
480 collElem = ures_getByKeyWithFallback(collations, keyBuffer, collElem, status);
481 ures_close(collElem);
482 ures_close(collations);
483 ures_close(b);
484 }
485
486
487 U_CAPI UCollator* U_EXPORT2
ucol_openFromShortString(const char * definition,UBool forceDefaults,UParseError * parseError,UErrorCode * status)488 ucol_openFromShortString( const char *definition,
489 UBool forceDefaults,
490 UParseError *parseError,
491 UErrorCode *status)
492 {
493 UTRACE_ENTRY_OC(UTRACE_UCOL_OPEN_FROM_SHORT_STRING);
494 UTRACE_DATA1(UTRACE_INFO, "short string = \"%s\"", definition);
495
496 if(U_FAILURE(*status)) return 0;
497
498 UParseError internalParseError;
499
500 if(!parseError) {
501 parseError = &internalParseError;
502 }
503 parseError->line = 0;
504 parseError->offset = 0;
505 parseError->preContext[0] = 0;
506 parseError->postContext[0] = 0;
507
508
509 // first we want to pick stuff out of short string.
510 // we'll end up with an UCA version, locale and a bunch of
511 // settings
512
513 // analyse the string in order to get everything we need.
514 const char *string = definition;
515 CollatorSpec s;
516 ucol_sit_initCollatorSpecs(&s);
517 string = ucol_sit_readSpecs(&s, definition, parseError, status);
518 ucol_sit_calculateWholeLocale(&s);
519
520 char buffer[internalBufferSize];
521 uprv_memset(buffer, 0, internalBufferSize);
522 uloc_canonicalize(s.locale, buffer, internalBufferSize, status);
523
524 UCollator *result = ucol_open(buffer, status);
525 int32_t i = 0;
526
527 for(i = 0; i < UCOL_ATTRIBUTE_COUNT; i++) {
528 if(s.options[i] != UCOL_DEFAULT) {
529 if(forceDefaults || ucol_getAttribute(result, (UColAttribute)i, status) != s.options[i]) {
530 ucol_setAttribute(result, (UColAttribute)i, s.options[i], status);
531 }
532
533 if(U_FAILURE(*status)) {
534 parseError->offset = (int32_t)(string - definition);
535 ucol_close(result);
536 return NULL;
537 }
538
539 }
540 }
541 if(s.variableTopSet) {
542 if(s.variableTopString[0]) {
543 ucol_setVariableTop(result, s.variableTopString, s.variableTopStringLen, status);
544 } else { // we set by value, using 'B'
545 ucol_restoreVariableTop(result, s.variableTopValue, status);
546 }
547 }
548
549
550 if(U_FAILURE(*status)) { // here it can only be a bogus value
551 ucol_close(result);
552 result = NULL;
553 }
554
555 UTRACE_EXIT_PTR_STATUS(result, *status);
556 return result;
557 }
558
559
560 U_CAPI int32_t U_EXPORT2
ucol_getShortDefinitionString(const UCollator * coll,const char * locale,char * dst,int32_t capacity,UErrorCode * status)561 ucol_getShortDefinitionString(const UCollator *coll,
562 const char *locale,
563 char *dst,
564 int32_t capacity,
565 UErrorCode *status)
566 {
567 if(U_FAILURE(*status)) return 0;
568 if(coll == NULL) {
569 *status = U_ILLEGAL_ARGUMENT_ERROR;
570 return 0;
571 }
572 return ((icu::Collator*)coll)->internalGetShortDefinitionString(locale,dst,capacity,*status);
573 }
574
575 U_CAPI int32_t U_EXPORT2
ucol_normalizeShortDefinitionString(const char * definition,char * destination,int32_t capacity,UParseError * parseError,UErrorCode * status)576 ucol_normalizeShortDefinitionString(const char *definition,
577 char *destination,
578 int32_t capacity,
579 UParseError *parseError,
580 UErrorCode *status)
581 {
582
583 if(U_FAILURE(*status)) {
584 return 0;
585 }
586
587 if(destination) {
588 uprv_memset(destination, 0, capacity*sizeof(char));
589 }
590
591 UParseError pe;
592 if(!parseError) {
593 parseError = &pe;
594 }
595
596 // validate
597 CollatorSpec s;
598 ucol_sit_initCollatorSpecs(&s);
599 ucol_sit_readSpecs(&s, definition, parseError, status);
600 return ucol_sit_dumpSpecs(&s, destination, capacity, status);
601 }
602
603 /**
604 * Get a set containing the contractions defined by the collator. The set includes
605 * both the UCA contractions and the contractions defined by the collator
606 * @param coll collator
607 * @param conts the set to hold the result
608 * @param status to hold the error code
609 * @return the size of the contraction set
610 */
611 U_CAPI int32_t U_EXPORT2
ucol_getContractions(const UCollator * coll,USet * contractions,UErrorCode * status)612 ucol_getContractions( const UCollator *coll,
613 USet *contractions,
614 UErrorCode *status)
615 {
616 ucol_getContractionsAndExpansions(coll, contractions, NULL, FALSE, status);
617 return uset_getItemCount(contractions);
618 }
619
620 /**
621 * Get a set containing the expansions defined by the collator. The set includes
622 * both the UCA expansions and the expansions defined by the tailoring
623 * @param coll collator
624 * @param conts the set to hold the result
625 * @param addPrefixes add the prefix contextual elements to contractions
626 * @param status to hold the error code
627 *
628 * @draft ICU 3.4
629 */
630 U_CAPI void U_EXPORT2
ucol_getContractionsAndExpansions(const UCollator * coll,USet * contractions,USet * expansions,UBool addPrefixes,UErrorCode * status)631 ucol_getContractionsAndExpansions( const UCollator *coll,
632 USet *contractions,
633 USet *expansions,
634 UBool addPrefixes,
635 UErrorCode *status)
636 {
637 if(U_FAILURE(*status)) {
638 return;
639 }
640 if(coll == NULL) {
641 *status = U_ILLEGAL_ARGUMENT_ERROR;
642 return;
643 }
644 const icu::RuleBasedCollator *rbc = icu::RuleBasedCollator::rbcFromUCollator(coll);
645 if(rbc == NULL) {
646 *status = U_UNSUPPORTED_ERROR;
647 return;
648 }
649 rbc->internalGetContractionsAndExpansions(
650 icu::UnicodeSet::fromUSet(contractions),
651 icu::UnicodeSet::fromUSet(expansions),
652 addPrefixes, *status);
653 }
654 #endif
655