1 // Copyright (C) 2016 and later: Unicode, Inc. and others. 2 // License & terms of use: http://www.unicode.org/copyright.html 3 /* 4 ******************************************************************************** 5 * Copyright (C) 1996-2015, International Business Machines 6 * Corporation and others. All Rights Reserved. 7 ******************************************************************************** 8 */ 9 10 #include "unicode/utypes.h" 11 12 #if !UCONFIG_NO_BREAK_ITERATION 13 14 #include "unicode/ubrk.h" 15 16 #include "unicode/brkiter.h" 17 #include "unicode/uloc.h" 18 #include "unicode/ustring.h" 19 #include "unicode/uchriter.h" 20 #include "unicode/rbbi.h" 21 #include "rbbirb.h" 22 #include "uassert.h" 23 24 U_NAMESPACE_USE 25 26 //------------------------------------------------------------------------------ 27 // 28 // ubrk_open Create a canned type of break iterator based on type (word, line, etc.) 29 // and locale. 30 // 31 //------------------------------------------------------------------------------ 32 U_CAPI UBreakIterator* U_EXPORT2 33 ubrk_open(UBreakIteratorType type, 34 const char *locale, 35 const UChar *text, 36 int32_t textLength, 37 UErrorCode *status) 38 { 39 40 if(U_FAILURE(*status)) return 0; 41 42 BreakIterator *result = 0; 43 44 switch(type) { 45 46 case UBRK_CHARACTER: 47 result = BreakIterator::createCharacterInstance(Locale(locale), *status); 48 break; 49 50 case UBRK_WORD: 51 result = BreakIterator::createWordInstance(Locale(locale), *status); 52 break; 53 54 case UBRK_LINE: 55 result = BreakIterator::createLineInstance(Locale(locale), *status); 56 break; 57 58 case UBRK_SENTENCE: 59 result = BreakIterator::createSentenceInstance(Locale(locale), *status); 60 break; 61 62 case UBRK_TITLE: 63 result = BreakIterator::createTitleInstance(Locale(locale), *status); 64 break; 65 66 default: 67 *status = U_ILLEGAL_ARGUMENT_ERROR; 68 } 69 70 // check for allocation error 71 if (U_FAILURE(*status)) { 72 return 0; 73 } 74 if(result == 0) { 75 *status = U_MEMORY_ALLOCATION_ERROR; 76 return 0; 77 } 78 79 80 UBreakIterator *uBI = (UBreakIterator *)result; 81 if (text != NULL) { 82 ubrk_setText(uBI, text, textLength, status); 83 } 84 return uBI; 85 } 86 87 88 89 //------------------------------------------------------------------------------ 90 // 91 // ubrk_openRules open a break iterator from a set of break rules. 92 // Invokes the rule builder. 93 // 94 //------------------------------------------------------------------------------ 95 U_CAPI UBreakIterator* U_EXPORT2 96 ubrk_openRules( const UChar *rules, 97 int32_t rulesLength, 98 const UChar *text, 99 int32_t textLength, 100 UParseError *parseErr, 101 UErrorCode *status) { 102 103 if (status == NULL || U_FAILURE(*status)){ 104 return 0; 105 } 106 107 BreakIterator *result = 0; 108 UnicodeString ruleString(rules, rulesLength); 109 result = RBBIRuleBuilder::createRuleBasedBreakIterator(ruleString, parseErr, *status); 110 if(U_FAILURE(*status)) { 111 return 0; 112 } 113 114 UBreakIterator *uBI = (UBreakIterator *)result; 115 if (text != NULL) { 116 ubrk_setText(uBI, text, textLength, status); 117 } 118 return uBI; 119 } 120 121 122 123 124 125 U_CAPI UBreakIterator * U_EXPORT2 126 ubrk_safeClone( 127 const UBreakIterator *bi, 128 void * /*stackBuffer*/, 129 int32_t *pBufferSize, 130 UErrorCode *status) 131 { 132 if (status == NULL || U_FAILURE(*status)){ 133 return NULL; 134 } 135 if (bi == NULL) { 136 *status = U_ILLEGAL_ARGUMENT_ERROR; 137 return NULL; 138 } 139 if (pBufferSize != NULL) { 140 int32_t inputSize = *pBufferSize; 141 *pBufferSize = 1; 142 if (inputSize == 0) { 143 return NULL; // preflighting for deprecated functionality 144 } 145 } 146 BreakIterator *newBI = ((BreakIterator *)bi)->clone(); 147 if (newBI == NULL) { 148 *status = U_MEMORY_ALLOCATION_ERROR; 149 } else { 150 *status = U_SAFECLONE_ALLOCATED_WARNING; 151 } 152 return (UBreakIterator *)newBI; 153 } 154 155 156 157 U_CAPI void U_EXPORT2 158 ubrk_close(UBreakIterator *bi) 159 { 160 delete (BreakIterator *)bi; 161 } 162 163 U_CAPI void U_EXPORT2 164 ubrk_setText(UBreakIterator* bi, 165 const UChar* text, 166 int32_t textLength, 167 UErrorCode* status) 168 { 169 UText ut = UTEXT_INITIALIZER; 170 utext_openUChars(&ut, text, textLength, status); 171 ((BreakIterator*)bi)->setText(&ut, *status); 172 // A stack allocated UText wrapping a UChar * string 173 // can be dumped without explicitly closing it. 174 } 175 176 177 178 U_CAPI void U_EXPORT2 179 ubrk_setUText(UBreakIterator *bi, 180 UText *text, 181 UErrorCode *status) 182 { 183 ((BreakIterator*)bi)->setText(text, *status); 184 } 185 186 187 188 189 190 U_CAPI int32_t U_EXPORT2 191 ubrk_current(const UBreakIterator *bi) 192 { 193 194 return ((BreakIterator*)bi)->current(); 195 } 196 197 U_CAPI int32_t U_EXPORT2 198 ubrk_next(UBreakIterator *bi) 199 { 200 201 return ((BreakIterator*)bi)->next(); 202 } 203 204 U_CAPI int32_t U_EXPORT2 205 ubrk_previous(UBreakIterator *bi) 206 { 207 208 return ((BreakIterator*)bi)->previous(); 209 } 210 211 U_CAPI int32_t U_EXPORT2 212 ubrk_first(UBreakIterator *bi) 213 { 214 215 return ((BreakIterator*)bi)->first(); 216 } 217 218 U_CAPI int32_t U_EXPORT2 219 ubrk_last(UBreakIterator *bi) 220 { 221 222 return ((BreakIterator*)bi)->last(); 223 } 224 225 U_CAPI int32_t U_EXPORT2 226 ubrk_preceding(UBreakIterator *bi, 227 int32_t offset) 228 { 229 230 return ((BreakIterator*)bi)->preceding(offset); 231 } 232 233 U_CAPI int32_t U_EXPORT2 234 ubrk_following(UBreakIterator *bi, 235 int32_t offset) 236 { 237 238 return ((BreakIterator*)bi)->following(offset); 239 } 240 241 U_CAPI const char* U_EXPORT2 242 ubrk_getAvailable(int32_t index) 243 { 244 245 return uloc_getAvailable(index); 246 } 247 248 U_CAPI int32_t U_EXPORT2 249 ubrk_countAvailable() 250 { 251 252 return uloc_countAvailable(); 253 } 254 255 256 U_CAPI UBool U_EXPORT2 257 ubrk_isBoundary(UBreakIterator *bi, int32_t offset) 258 { 259 return ((BreakIterator*)bi)->isBoundary(offset); 260 } 261 262 263 U_CAPI int32_t U_EXPORT2 264 ubrk_getRuleStatus(UBreakIterator *bi) 265 { 266 return ((BreakIterator*)bi)->getRuleStatus(); 267 } 268 269 U_CAPI int32_t U_EXPORT2 270 ubrk_getRuleStatusVec(UBreakIterator *bi, int32_t *fillInVec, int32_t capacity, UErrorCode *status) 271 { 272 return ((BreakIterator*)bi)->getRuleStatusVec(fillInVec, capacity, *status); 273 } 274 275 276 U_CAPI const char* U_EXPORT2 277 ubrk_getLocaleByType(const UBreakIterator *bi, 278 ULocDataLocaleType type, 279 UErrorCode* status) 280 { 281 if (bi == NULL) { 282 if (U_SUCCESS(*status)) { 283 *status = U_ILLEGAL_ARGUMENT_ERROR; 284 } 285 return NULL; 286 } 287 return ((BreakIterator*)bi)->getLocaleID(type, *status); 288 } 289 290 291 void ubrk_refreshUText(UBreakIterator *bi, 292 UText *text, 293 UErrorCode *status) 294 { 295 BreakIterator *bii = reinterpret_cast<BreakIterator *>(bi); 296 bii->refreshInputText(text, *status); 297 } 298 299 300 301 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */ 302
