icuSources/common/brkiter.cpp

   1 /*
   2 *******************************************************************************
   3 * Copyright (C) 1997-2010, International Business Machines Corporation and
   4 * others. All Rights Reserved.
   5 *******************************************************************************
   6 *
   7 * File TXTBDRY.CPP
   8 *
   9 * Modification History:
  10 *
  11 *   Date        Name        Description
  12 *   02/18/97    aliu        Converted from OpenClass.  Added DONE.
  13 *   01/13/2000  helena      Added UErrorCode parameter to createXXXInstance methods.
  14 *****************************************************************************************
  15 */
  16
  17 // *****************************************************************************
  18 // This file was generated from the java source file BreakIterator.java
  19 // *****************************************************************************
  20
  21 #include "unicode/utypes.h"
  22
  23 #if !UCONFIG_NO_BREAK_ITERATION
  24
  25 #include "unicode/rbbi.h"
  26 #include "unicode/brkiter.h"
  27 #include "unicode/udata.h"
  28 #include "unicode/ures.h"
  29 #include "unicode/ustring.h"
  30 #include "ucln_cmn.h"
  31 #include "cstring.h"
  32 #include "umutex.h"
  33 #include "servloc.h"
  34 #include "locbased.h"
  35 #include "uresimp.h"
  36 #include "uassert.h"
  37 #include "ubrkimpl.h"
  38
  39 // *****************************************************************************
  40 // class BreakIterator
  41 // This class implements methods for finding the location of boundaries in text.
  42 // Instances of BreakIterator maintain a current position and scan over text
  43 // returning the index of characters where boundaries occur.
  44 // *****************************************************************************
  45
  46 U_NAMESPACE_BEGIN
  47
  48 // -------------------------------------
  49
  50 BreakIterator*
  51 BreakIterator::buildInstance(const Locale& loc, const char *type, int32_t kind, UErrorCode &status)
  52 {
  53     char fnbuff[256];
  54     char ext[4]={'\0'};
  55     char actualLocale[ULOC_FULLNAME_CAPACITY];
  56     int32_t size;
  57     const UChar* brkfname = NULL;
  58     UResourceBundle brkRulesStack;
  59     UResourceBundle brkNameStack;
  60     UResourceBundle *brkRules = &brkRulesStack;
  61     UResourceBundle *brkName  = &brkNameStack;
  62     RuleBasedBreakIterator *result = NULL;
  63
  64     if (U_FAILURE(status))
  65         return NULL;
  66
  67     ures_initStackObject(brkRules);
  68     ures_initStackObject(brkName);
  69
  70     // Get the locale
  71     UResourceBundle *b = ures_open(U_ICUDATA_BRKITR, loc.getName(), &status);
  72     /* this is a hack for now. Should be fixed when the data is fetched from
  73         brk_index.txt */
  74     if(status==U_USING_DEFAULT_WARNING){
  75         status=U_ZERO_ERROR;
  76         ures_openFillIn(b, U_ICUDATA_BRKITR, "", &status);
  77     }
  78
  79     // Get the "boundaries" array.
  80     if (U_SUCCESS(status)) {
  81         brkRules = ures_getByKeyWithFallback(b, "boundaries", brkRules, &status);
  82         // Get the string object naming the rules file
  83         brkName = ures_getByKeyWithFallback(brkRules, type, brkName, &status);
  84         // Get the actual string
  85         brkfname = ures_getString(brkName, &size, &status);
  86         U_ASSERT((size_t)size<sizeof(fnbuff));
  87         if ((size_t)size>=sizeof(fnbuff)) {
  88             size=0;
  89             if (U_SUCCESS(status)) {
  90                 status = U_BUFFER_OVERFLOW_ERROR;
  91             }
  92         }
  93
  94         // Use the string if we found it
  95         if (U_SUCCESS(status) && brkfname) {
  96             uprv_strncpy(actualLocale,
  97                 ures_getLocale(brkName, &status),
  98                 sizeof(actualLocale)/sizeof(actualLocale[0]));
  99
 100             UChar* extStart=u_strchr(brkfname, 0x002e);
 101             int len = 0;
 102             if(extStart!=NULL){
 103                 len = (int)(extStart-brkfname);
 104                 u_UCharsToChars(extStart+1, ext, sizeof(ext)); // nul terminates the buff
 105                 u_UCharsToChars(brkfname, fnbuff, len);
 106             }
 107             fnbuff[len]=0; // nul terminate
 108         }
 109     }
 110
 111     ures_close(brkRules);
 112     ures_close(brkName);
 113
 114     UDataMemory* file = udata_open(U_ICUDATA_BRKITR, ext, fnbuff, &status);
 115     if (U_FAILURE(status)) {
 116         ures_close(b);
 117         return NULL;
 118     }
 119
 120     // Create a RuleBasedBreakIterator
 121     result = new RuleBasedBreakIterator(file, status);
 122
 123     // If there is a result, set the valid locale and actual locale, and the kind
 124     if (U_SUCCESS(status) && result != NULL) {
 125         U_LOCALE_BASED(locBased, *(BreakIterator*)result);
 126         locBased.setLocaleIDs(ures_getLocaleByType(b, ULOC_VALID_LOCALE, &status), actualLocale);
 127         result->setBreakType(kind);
 128     }
 129
 130     ures_close(b);
 131
 132     if (U_FAILURE(status) && result != NULL) {  // Sometimes redundant check, but simple
 133         delete result;
 134         return NULL;
 135     }
 136
 137     if (result == NULL) {
 138         udata_close(file);
 139         if (U_SUCCESS(status)) {
 140             status = U_MEMORY_ALLOCATION_ERROR;
 141         }
 142     }
 143
 144     return result;
 145 }
 146
 147 // Creates a break iterator for word breaks.
 148 BreakIterator* U_EXPORT2
 149 BreakIterator::createWordInstance(const Locale& key, UErrorCode& status)
 150 {
 151     return createInstance(key, UBRK_WORD, status);
 152 }
 153
 154 // -------------------------------------
 155
 156 // Creates a break iterator  for line breaks.
 157 BreakIterator* U_EXPORT2
 158 BreakIterator::createLineInstance(const Locale& key, UErrorCode& status)
 159 {
 160     return createInstance(key, UBRK_LINE, status);
 161 }
 162
 163 // -------------------------------------
 164
 165 // Creates a break iterator  for character breaks.
 166 BreakIterator* U_EXPORT2
 167 BreakIterator::createCharacterInstance(const Locale& key, UErrorCode& status)
 168 {
 169     return createInstance(key, UBRK_CHARACTER, status);
 170 }
 171
 172 // -------------------------------------
 173
 174 // Creates a break iterator  for sentence breaks.
 175 BreakIterator* U_EXPORT2
 176 BreakIterator::createSentenceInstance(const Locale& key, UErrorCode& status)
 177 {
 178     return createInstance(key, UBRK_SENTENCE, status);
 179 }
 180
 181 // -------------------------------------
 182
 183 // Creates a break iterator for title casing breaks.
 184 BreakIterator* U_EXPORT2
 185 BreakIterator::createTitleInstance(const Locale& key, UErrorCode& status)
 186 {
 187     return createInstance(key, UBRK_TITLE, status);
 188 }
 189
 190 // -------------------------------------
 191
 192 // Gets all the available locales that has localized text boundary data.
 193 const Locale* U_EXPORT2
 194 BreakIterator::getAvailableLocales(int32_t& count)
 195 {
 196     return Locale::getAvailableLocales(count);
 197 }
 198
 199 // ------------------------------------------
 200 //
 201 // Default constructor and destructor
 202 //
 203 //-------------------------------------------
 204
 205 BreakIterator::BreakIterator()
 206 {
 207     fBufferClone = FALSE;
 208     *validLocale = *actualLocale = 0;
 209 }
 210
 211 BreakIterator::~BreakIterator()
 212 {
 213 }
 214
 215 // ------------------------------------------
 216 //
 217 // Registration
 218 //
 219 //-------------------------------------------
 220 #if !UCONFIG_NO_SERVICE
 221
 222 // -------------------------------------
 223
 224 class ICUBreakIteratorFactory : public ICUResourceBundleFactory {
 225 protected:
 226     virtual UObject* handleCreate(const Locale& loc, int32_t kind, const ICUService* /*service*/, UErrorCode& status) const {
 227         return BreakIterator::makeInstance(loc, kind, status);
 228     }
 229 };
 230
 231 // -------------------------------------
 232
 233 class ICUBreakIteratorService : public ICULocaleService {
 234 public:
 235     ICUBreakIteratorService()
 236         : ICULocaleService(UNICODE_STRING("Break Iterator", 14))
 237     {
 238         UErrorCode status = U_ZERO_ERROR;
 239         registerFactory(new ICUBreakIteratorFactory(), status);
 240     }
 241
 242     virtual UObject* cloneInstance(UObject* instance) const {
 243         return ((BreakIterator*)instance)->clone();
 244     }
 245
 246     virtual UObject* handleDefault(const ICUServiceKey& key, UnicodeString* /*actualID*/, UErrorCode& status) const {
 247         LocaleKey& lkey = (LocaleKey&)key;
 248         int32_t kind = lkey.kind();
 249         Locale loc;
 250         lkey.currentLocale(loc);
 251         return BreakIterator::makeInstance(loc, kind, status);
 252     }
 253
 254     virtual UBool isDefault() const {
 255         return countFactories() == 1;
 256     }
 257 };
 258
 259 // -------------------------------------
 260
 261 U_NAMESPACE_END
 262
 263 // defined in ucln_cmn.h
 264
 265 static U_NAMESPACE_QUALIFIER ICULocaleService* gService = NULL;
 266
 267 /**
 268  * Release all static memory held by breakiterator.
 269  */
 270 U_CDECL_BEGIN
 271 static UBool U_CALLCONV breakiterator_cleanup(void) {
 272 #if !UCONFIG_NO_SERVICE
 273     if (gService) {
 274         delete gService;
 275         gService = NULL;
 276     }
 277 #endif
 278     return TRUE;
 279 }
 280 U_CDECL_END
 281 U_NAMESPACE_BEGIN
 282
 283 static ICULocaleService*
 284 getService(void)
 285 {
 286     UBool needsInit;
 287     UMTX_CHECK(NULL, (UBool)(gService == NULL), needsInit);
 288
 289     if (needsInit) {
 290         ICULocaleService  *tService = new ICUBreakIteratorService();
 291         umtx_lock(NULL);
 292         if (gService == NULL) {
 293             gService = tService;
 294             tService = NULL;
 295             ucln_common_registerCleanup(UCLN_COMMON_BREAKITERATOR, breakiterator_cleanup);
 296         }
 297         umtx_unlock(NULL);
 298         delete tService;
 299     }
 300     return gService;
 301 }
 302
 303 // -------------------------------------
 304
 305 static inline UBool
 306 hasService(void)
 307 {
 308     UBool retVal;
 309     UMTX_CHECK(NULL, gService != NULL, retVal);
 310     return retVal;
 311 }
 312
 313 // -------------------------------------
 314
 315 URegistryKey U_EXPORT2
 316 BreakIterator::registerInstance(BreakIterator* toAdopt, const Locale& locale, UBreakIteratorType kind, UErrorCode& status)
 317 {
 318     ICULocaleService *service = getService();
 319     if (service == NULL) {
 320         status = U_MEMORY_ALLOCATION_ERROR;
 321         return NULL;
 322     }
 323     return service->registerInstance(toAdopt, locale, kind, status);
 324 }
 325
 326 // -------------------------------------
 327
 328 UBool U_EXPORT2
 329 BreakIterator::unregister(URegistryKey key, UErrorCode& status)
 330 {
 331     if (U_SUCCESS(status)) {
 332         if (hasService()) {
 333             return gService->unregister(key, status);
 334         }
 335         status = U_MEMORY_ALLOCATION_ERROR;
 336     }
 337     return FALSE;
 338 }
 339
 340 // -------------------------------------
 341
 342 StringEnumeration* U_EXPORT2
 343 BreakIterator::getAvailableLocales(void)
 344 {
 345     ICULocaleService *service = getService();
 346     if (service == NULL) {
 347         return NULL;
 348     }
 349     return service->getAvailableLocales();
 350 }
 351 #endif /* UCONFIG_NO_SERVICE */
 352
 353 // -------------------------------------
 354
 355 BreakIterator*
 356 BreakIterator::createInstance(const Locale& loc, int32_t kind, UErrorCode& status)
 357 {
 358     if (U_FAILURE(status)) {
 359         return NULL;
 360     }
 361
 362 #if !UCONFIG_NO_SERVICE
 363     if (hasService()) {
 364         Locale actualLoc("");
 365         BreakIterator *result = (BreakIterator*)gService->get(loc, kind, &actualLoc, status);
 366         // TODO: The way the service code works in ICU 2.8 is that if
 367         // there is a real registered break iterator, the actualLoc
 368         // will be populated, but if the handleDefault path is taken
 369         // (because nothing is registered that can handle the
 370         // requested locale) then the actualLoc comes back empty.  In
 371         // that case, the returned object already has its actual/valid
 372         // locale data populated (by makeInstance, which is what
 373         // handleDefault calls), so we don't touch it.  YES, A COMMENT
 374         // THIS LONG is a sign of bad code -- so the action item is to
 375         // revisit this in ICU 3.0 and clean it up/fix it/remove it.
 376         if (U_SUCCESS(status) && (result != NULL) && *actualLoc.getName() != 0) {
 377             U_LOCALE_BASED(locBased, *result);
 378             locBased.setLocaleIDs(actualLoc.getName(), actualLoc.getName());
 379         }
 380         return result;
 381     }
 382     else
 383 #endif
 384     {
 385         return makeInstance(loc, kind, status);
 386     }
 387 }
 388
 389 // -------------------------------------
 390
 391 BreakIterator*
 392 BreakIterator::makeInstance(const Locale& loc, int32_t kind, UErrorCode& status)
 393 {
 394
 395     if (U_FAILURE(status)) {
 396         return NULL;
 397     }
 398
 399     BreakIterator *result = NULL;
 400     switch (kind) {
 401     case UBRK_CHARACTER:
 402         result = BreakIterator::buildInstance(loc, "grapheme", kind, status);
 403         break;
 404     case UBRK_WORD:
 405         result = BreakIterator::buildInstance(loc, "word", kind, status);
 406         break;
 407     case UBRK_LINE:
 408         result = BreakIterator::buildInstance(loc, "line", kind, status);
 409         break;
 410     case UBRK_SENTENCE:
 411         result = BreakIterator::buildInstance(loc, "sentence", kind, status);
 412         break;
 413     case UBRK_TITLE:
 414         result = BreakIterator::buildInstance(loc, "title", kind, status);
 415         break;
 416     default:
 417         status = U_ILLEGAL_ARGUMENT_ERROR;
 418     }
 419
 420     if (U_FAILURE(status)) {
 421         return NULL;
 422     }
 423
 424     return result;
 425 }
 426
 427 Locale
 428 BreakIterator::getLocale(ULocDataLocaleType type, UErrorCode& status) const {
 429     U_LOCALE_BASED(locBased, *this);
 430     return locBased.getLocale(type, status);
 431 }
 432
 433 const char *
 434 BreakIterator::getLocaleID(ULocDataLocaleType type, UErrorCode& status) const {
 435     U_LOCALE_BASED(locBased, *this);
 436     return locBased.getLocaleID(type, status);
 437 }
 438
 439 U_NAMESPACE_END
 440
 441 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */
 442
 443 //eof