icuSources/common/brkiter.cpp

   1 /*
   2 *******************************************************************************
   3 * Copyright (C) 1997-2006, 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 "mutex.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 = 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, *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 // Gets the objectLocale display name in the default locale language.
 201 UnicodeString& U_EXPORT2
 202 BreakIterator::getDisplayName(const Locale& objectLocale,
 203                              UnicodeString& name)
 204 {
 205     return objectLocale.getDisplayName(name);
 206 }
 207
 208 // -------------------------------------
 209 // Gets the objectLocale display name in the displayLocale language.
 210 UnicodeString& U_EXPORT2
 211 BreakIterator::getDisplayName(const Locale& objectLocale,
 212                              const Locale& displayLocale,
 213                              UnicodeString& name)
 214 {
 215     return objectLocale.getDisplayName(displayLocale, name);
 216 }
 217
 218 // ------------------------------------------
 219 //
 220 // Default constructor and destructor
 221 //
 222 //-------------------------------------------
 223
 224 BreakIterator::BreakIterator()
 225 {
 226     fBufferClone = FALSE;
 227     *validLocale = *actualLocale = 0;
 228 }
 229
 230 BreakIterator::~BreakIterator()
 231 {
 232 }
 233
 234 // ------------------------------------------
 235 //
 236 // Registration
 237 //
 238 //-------------------------------------------
 239 #if !UCONFIG_NO_SERVICE
 240
 241 static ICULocaleService* gService = NULL;
 242
 243 // -------------------------------------
 244
 245 class ICUBreakIteratorFactory : public ICUResourceBundleFactory {
 246 protected:
 247     virtual UObject* handleCreate(const Locale& loc, int32_t kind, const ICUService* /*service*/, UErrorCode& status) const {
 248         return BreakIterator::makeInstance(loc, kind, status);
 249     }
 250 };
 251
 252 // -------------------------------------
 253
 254 class ICUBreakIteratorService : public ICULocaleService {
 255 public:
 256     ICUBreakIteratorService()
 257         : ICULocaleService(UNICODE_STRING("Break Iterator", 14))
 258     {
 259         UErrorCode status = U_ZERO_ERROR;
 260         registerFactory(new ICUBreakIteratorFactory(), status);
 261     }
 262
 263     virtual UObject* cloneInstance(UObject* instance) const {
 264         return ((BreakIterator*)instance)->clone();
 265     }
 266
 267     virtual UObject* handleDefault(const ICUServiceKey& key, UnicodeString* /*actualID*/, UErrorCode& status) const {
 268         LocaleKey& lkey = (LocaleKey&)key;
 269         int32_t kind = lkey.kind();
 270         Locale loc;
 271         lkey.currentLocale(loc);
 272         return BreakIterator::makeInstance(loc, kind, status);
 273     }
 274
 275     virtual UBool isDefault() const {
 276         return countFactories() == 1;
 277     }
 278 };
 279
 280 // -------------------------------------
 281
 282 U_NAMESPACE_END
 283
 284 // defined in ucln_cmn.h
 285
 286 /**
 287  * Release all static memory held by breakiterator.
 288  */
 289 U_CDECL_BEGIN
 290 static UBool U_CALLCONV breakiterator_cleanup(void) {
 291 #if !UCONFIG_NO_SERVICE
 292     if (gService) {
 293         delete gService;
 294         gService = NULL;
 295     }
 296 #endif
 297     return TRUE;
 298 }
 299 U_CDECL_END
 300 U_NAMESPACE_BEGIN
 301
 302 static ICULocaleService*
 303 getService(void)
 304 {
 305     UBool needsInit;
 306     umtx_lock(NULL);
 307     needsInit = (UBool)(gService == NULL);
 308     umtx_unlock(NULL);
 309
 310     if (needsInit) {
 311         ICULocaleService  *tService = new ICUBreakIteratorService();
 312         umtx_lock(NULL);
 313         if (gService == NULL) {
 314             gService = tService;
 315             tService = NULL;
 316             ucln_common_registerCleanup(UCLN_COMMON_BREAKITERATOR, breakiterator_cleanup);
 317         }
 318         umtx_unlock(NULL);
 319         delete tService;
 320     }
 321     return gService;
 322 }
 323
 324 // -------------------------------------
 325
 326 static UBool
 327 hasService(void)
 328 {
 329     Mutex mutex;
 330     return gService != NULL;
 331 }
 332
 333 // -------------------------------------
 334
 335 URegistryKey U_EXPORT2
 336 BreakIterator::registerInstance(BreakIterator* toAdopt, const Locale& locale, UBreakIteratorType kind, UErrorCode& status)
 337 {
 338     return getService()->registerInstance(toAdopt, locale, kind, status);
 339 }
 340
 341 // -------------------------------------
 342
 343 UBool U_EXPORT2
 344 BreakIterator::unregister(URegistryKey key, UErrorCode& status)
 345 {
 346     if (U_SUCCESS(status)) {
 347         if (hasService()) {
 348             return gService->unregister(key, status);
 349         }
 350         status = U_ILLEGAL_ARGUMENT_ERROR;
 351     }
 352     return FALSE;
 353 }
 354
 355 // -------------------------------------
 356
 357 StringEnumeration* U_EXPORT2
 358 BreakIterator::getAvailableLocales(void)
 359 {
 360   return getService()->getAvailableLocales();
 361 }
 362 #endif /* UCONFIG_NO_SERVICE */
 363
 364 // -------------------------------------
 365
 366 BreakIterator*
 367 BreakIterator::createInstance(const Locale& loc, int32_t kind, UErrorCode& status)
 368 {
 369     if (U_FAILURE(status)) {
 370         return NULL;
 371     }
 372
 373     u_init(&status);
 374 #if !UCONFIG_NO_SERVICE
 375     if (hasService()) {
 376         Locale actualLoc("");
 377         BreakIterator *result = (BreakIterator*)gService->get(loc, kind, &actualLoc, status);
 378         // TODO: The way the service code works in ICU 2.8 is that if
 379         // there is a real registered break iterator, the actualLoc
 380         // will be populated, but if the handleDefault path is taken
 381         // (because nothing is registered that can handle the
 382         // requested locale) then the actualLoc comes back empty.  In
 383         // that case, the returned object already has its actual/valid
 384         // locale data populated (by makeInstance, which is what
 385         // handleDefault calls), so we don't touch it.  YES, A COMMENT
 386         // THIS LONG is a sign of bad code -- so the action item is to
 387         // revisit this in ICU 3.0 and clean it up/fix it/remove it.
 388         if (U_SUCCESS(status) && (result != NULL) && *actualLoc.getName() != 0) {
 389             U_LOCALE_BASED(locBased, *result);
 390             locBased.setLocaleIDs(actualLoc.getName(), actualLoc.getName());
 391         }
 392         return result;
 393     }
 394     else
 395 #endif
 396     {
 397         return makeInstance(loc, kind, status);
 398     }
 399 }
 400
 401 // -------------------------------------
 402
 403 BreakIterator*
 404 BreakIterator::makeInstance(const Locale& loc, int32_t kind, UErrorCode& status)
 405 {
 406
 407     if (U_FAILURE(status)) {
 408         return NULL;
 409     }
 410
 411     BreakIterator *result = NULL;
 412     switch (kind) {
 413     case UBRK_CHARACTER:
 414         result = BreakIterator::buildInstance(loc, "grapheme", kind, status);
 415         break;
 416     case UBRK_WORD:
 417         result = BreakIterator::buildInstance(loc, "word", kind, status);
 418         break;
 419     case UBRK_LINE:
 420         result = BreakIterator::buildInstance(loc, "line", kind, status);
 421         break;
 422     case UBRK_SENTENCE:
 423         result = BreakIterator::buildInstance(loc, "sentence", kind, status);
 424         break;
 425     case UBRK_TITLE:
 426         result = BreakIterator::buildInstance(loc, "title", kind, status);
 427         break;
 428     default:
 429         status = U_ILLEGAL_ARGUMENT_ERROR;
 430     }
 431
 432     if (U_FAILURE(status)) {
 433         return NULL;
 434     }
 435
 436     return result;
 437 }
 438
 439 Locale
 440 BreakIterator::getLocale(ULocDataLocaleType type, UErrorCode& status) const {
 441     U_LOCALE_BASED(locBased, *this);
 442     return locBased.getLocale(type, status);
 443 }
 444
 445 const char *
 446 BreakIterator::getLocaleID(ULocDataLocaleType type, UErrorCode& status) const {
 447     U_LOCALE_BASED(locBased, *this);
 448     return locBased.getLocaleID(type, status);
 449 }
 450
 451 U_NAMESPACE_END
 452
 453 #endif /* #if !UCONFIG_NO_BREAK_ITERATION */
 454
 455 //eof