1 /* 2 ******************************************************************** 3 * COPYRIGHT: 4 * Copyright (c) 1996-2015, International Business Machines Corporation and 5 * others. All Rights Reserved. 6 ******************************************************************** 7 * 8 * ucnv_bld.cpp: 9 * 10 * Defines functions that are used in the creation/initialization/deletion 11 * of converters and related structures. 12 * uses uconv_io.h routines to access disk information 13 * is used by ucnv.h to implement public API create/delete/flushCache routines 14 * Modification History: 15 * 16 * Date Name Description 17 * 18 * 06/20/2000 helena OS/400 port changes; mostly typecast. 19 * 06/29/2000 helena Major rewrite of the callback interface. 20 */ 21 22 #include "unicode/utypes.h" 23 24 #if !UCONFIG_NO_CONVERSION 25 26 #include "unicode/putil.h" 27 #include "unicode/udata.h" 28 #include "unicode/ucnv.h" 29 #include "unicode/uloc.h" 30 #include "mutex.h" 31 #include "putilimp.h" 32 #include "uassert.h" 33 #include "utracimp.h" 34 #include "ucnv_io.h" 35 #include "ucnv_bld.h" 36 #include "ucnvmbcs.h" 37 #include "ucnv_ext.h" 38 #include "ucnv_cnv.h" 39 #include "ucnv_imp.h" 40 #include "uhash.h" 41 #include "umutex.h" 42 #include "cstring.h" 43 #include "cmemory.h" 44 #include "ucln_cmn.h" 45 #include "ustr_cnv.h" 46 47 48 #if 0 49 #include <stdio.h> 50 extern void UCNV_DEBUG_LOG(char *what, char *who, void *p, int l); 51 #define UCNV_DEBUG_LOG(x,y,z) UCNV_DEBUG_LOG(x,y,z,__LINE__) 52 #else 53 # define UCNV_DEBUG_LOG(x,y,z) 54 #endif 55 56 static const UConverterSharedData * const 57 converterData[UCNV_NUMBER_OF_SUPPORTED_CONVERTER_TYPES]={ 58 NULL, NULL, 59 60 #if UCONFIG_NO_LEGACY_CONVERSION 61 NULL, 62 #else 63 &_MBCSData, 64 #endif 65 66 &_Latin1Data, 67 &_UTF8Data, &_UTF16BEData, &_UTF16LEData, 68 #if UCONFIG_ONLY_HTML_CONVERSION 69 NULL, NULL, 70 #else 71 &_UTF32BEData, &_UTF32LEData, 72 #endif 73 NULL, 74 75 #if UCONFIG_NO_LEGACY_CONVERSION 76 NULL, 77 #else 78 &_ISO2022Data, 79 #endif 80 81 #if UCONFIG_NO_LEGACY_CONVERSION || UCONFIG_ONLY_HTML_CONVERSION 82 NULL, NULL, NULL, NULL, NULL, NULL, 83 NULL, NULL, NULL, NULL, NULL, NULL, 84 NULL, 85 #else 86 &_LMBCSData1,&_LMBCSData2, &_LMBCSData3, &_LMBCSData4, &_LMBCSData5, &_LMBCSData6, 87 &_LMBCSData8,&_LMBCSData11,&_LMBCSData16,&_LMBCSData17,&_LMBCSData18,&_LMBCSData19, 88 &_HZData, 89 #endif 90 91 #if UCONFIG_ONLY_HTML_CONVERSION 92 NULL, 93 #else 94 &_SCSUData, 95 #endif 96 97 98 #if UCONFIG_NO_LEGACY_CONVERSION || UCONFIG_ONLY_HTML_CONVERSION 99 NULL, 100 #else 101 &_ISCIIData, 102 #endif 103 104 &_ASCIIData, 105 #if UCONFIG_ONLY_HTML_CONVERSION 106 NULL, NULL, &_UTF16Data, NULL, NULL, NULL, 107 #else 108 &_UTF7Data, &_Bocu1Data, &_UTF16Data, &_UTF32Data, &_CESU8Data, &_IMAPData, 109 #endif 110 111 #if UCONFIG_NO_LEGACY_CONVERSION || UCONFIG_ONLY_HTML_CONVERSION 112 NULL, 113 #else 114 &_CompoundTextData 115 #endif 116 }; 117 118 /* Please keep this in binary sorted order for getAlgorithmicTypeFromName. 119 Also the name should be in lower case and all spaces, dashes and underscores 120 removed 121 */ 122 static struct { 123 const char *name; 124 const UConverterType type; 125 } const cnvNameType[] = { 126 #if !UCONFIG_ONLY_HTML_CONVERSION 127 { "bocu1", UCNV_BOCU1 }, 128 { "cesu8", UCNV_CESU8 }, 129 #endif 130 #if !UCONFIG_NO_LEGACY_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION 131 { "hz",UCNV_HZ }, 132 #endif 133 #if !UCONFIG_ONLY_HTML_CONVERSION 134 { "imapmailboxname", UCNV_IMAP_MAILBOX }, 135 #endif 136 #if !UCONFIG_NO_LEGACY_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION 137 { "iscii", UCNV_ISCII }, 138 #endif 139 #if !UCONFIG_NO_LEGACY_CONVERSION 140 { "iso2022", UCNV_ISO_2022 }, 141 #endif 142 { "iso88591", UCNV_LATIN_1 }, 143 #if !UCONFIG_NO_LEGACY_CONVERSION && !UCONFIG_ONLY_HTML_CONVERSION 144 { "lmbcs1", UCNV_LMBCS_1 }, 145 { "lmbcs11",UCNV_LMBCS_11 }, 146 { "lmbcs16",UCNV_LMBCS_16 }, 147 { "lmbcs17",UCNV_LMBCS_17 }, 148 { "lmbcs18",UCNV_LMBCS_18 }, 149 { "lmbcs19",UCNV_LMBCS_19 }, 150 { "lmbcs2", UCNV_LMBCS_2 }, 151 { "lmbcs3", UCNV_LMBCS_3 }, 152 { "lmbcs4", UCNV_LMBCS_4 }, 153 { "lmbcs5", UCNV_LMBCS_5 }, 154 { "lmbcs6", UCNV_LMBCS_6 }, 155 { "lmbcs8", UCNV_LMBCS_8 }, 156 #endif 157 #if !UCONFIG_ONLY_HTML_CONVERSION 158 { "scsu", UCNV_SCSU }, 159 #endif 160 { "usascii", UCNV_US_ASCII }, 161 { "utf16", UCNV_UTF16 }, 162 { "utf16be", UCNV_UTF16_BigEndian }, 163 { "utf16le", UCNV_UTF16_LittleEndian }, 164 #if U_IS_BIG_ENDIAN 165 { "utf16oppositeendian", UCNV_UTF16_LittleEndian }, 166 { "utf16platformendian", UCNV_UTF16_BigEndian }, 167 #else 168 { "utf16oppositeendian", UCNV_UTF16_BigEndian}, 169 { "utf16platformendian", UCNV_UTF16_LittleEndian }, 170 #endif 171 #if !UCONFIG_ONLY_HTML_CONVERSION 172 { "utf32", UCNV_UTF32 }, 173 { "utf32be", UCNV_UTF32_BigEndian }, 174 { "utf32le", UCNV_UTF32_LittleEndian }, 175 #if U_IS_BIG_ENDIAN 176 { "utf32oppositeendian", UCNV_UTF32_LittleEndian }, 177 { "utf32platformendian", UCNV_UTF32_BigEndian }, 178 #else 179 { "utf32oppositeendian", UCNV_UTF32_BigEndian }, 180 { "utf32platformendian", UCNV_UTF32_LittleEndian }, 181 #endif 182 #endif 183 #if !UCONFIG_ONLY_HTML_CONVERSION 184 { "utf7", UCNV_UTF7 }, 185 #endif 186 { "utf8", UCNV_UTF8 }, 187 #if !UCONFIG_ONLY_HTML_CONVERSION 188 { "x11compoundtext", UCNV_COMPOUND_TEXT} 189 #endif 190 }; 191 192 193 /*initializes some global variables */ 194 static UHashtable *SHARED_DATA_HASHTABLE = NULL; 195 static UMutex cnvCacheMutex = U_MUTEX_INITIALIZER; /* Mutex for synchronizing cnv cache access. */ 196 /* Note: the global mutex is used for */ 197 /* reference count updates. */ 198 199 static const char **gAvailableConverters = NULL; 200 static uint16_t gAvailableConverterCount = 0; 201 static icu::UInitOnce gAvailableConvertersInitOnce = U_INITONCE_INITIALIZER; 202 203 #if !U_CHARSET_IS_UTF8 204 205 /* This contains the resolved converter name. So no further alias lookup is needed again. */ 206 static char gDefaultConverterNameBuffer[UCNV_MAX_CONVERTER_NAME_LENGTH + 1]; /* +1 for NULL */ 207 static const char *gDefaultConverterName = NULL; 208 209 /* 210 If the default converter is an algorithmic converter, this is the cached value. 211 We don't cache a full UConverter and clone it because ucnv_clone doesn't have 212 less overhead than an algorithmic open. We don't cache non-algorithmic converters 213 because ucnv_flushCache must be able to unload the default converter and its table. 214 */ 215 static const UConverterSharedData *gDefaultAlgorithmicSharedData = NULL; 216 217 /* Does gDefaultConverterName have a converter option and require extra parsing? */ 218 static UBool gDefaultConverterContainsOption; 219 220 #endif /* !U_CHARSET_IS_UTF8 */ 221 222 static const char DATA_TYPE[] = "cnv"; 223 224 /* ucnv_flushAvailableConverterCache. This is only called from ucnv_cleanup(). 225 * If it is ever to be called from elsewhere, synchronization 226 * will need to be considered. 227 */ 228 static void 229 ucnv_flushAvailableConverterCache() { 230 gAvailableConverterCount = 0; 231 if (gAvailableConverters) { 232 uprv_free((char **)gAvailableConverters); 233 gAvailableConverters = NULL; 234 } 235 gAvailableConvertersInitOnce.reset(); 236 } 237 238 /* ucnv_cleanup - delete all storage held by the converter cache, except any */ 239 /* in use by open converters. */ 240 /* Not thread safe. */ 241 /* Not supported API. */ 242 static UBool U_CALLCONV ucnv_cleanup(void) { 243 ucnv_flushCache(); 244 if (SHARED_DATA_HASHTABLE != NULL && uhash_count(SHARED_DATA_HASHTABLE) == 0) { 245 uhash_close(SHARED_DATA_HASHTABLE); 246 SHARED_DATA_HASHTABLE = NULL; 247 } 248 249 /* Isn't called from flushCache because other threads may have preexisting references to the table. */ 250 ucnv_flushAvailableConverterCache(); 251 252 #if !U_CHARSET_IS_UTF8 253 gDefaultConverterName = NULL; 254 gDefaultConverterNameBuffer[0] = 0; 255 gDefaultConverterContainsOption = FALSE; 256 gDefaultAlgorithmicSharedData = NULL; 257 #endif 258 259 return (SHARED_DATA_HASHTABLE == NULL); 260 } 261 262 static UBool U_CALLCONV 263 isCnvAcceptable(void * /*context*/, 264 const char * /*type*/, const char * /*name*/, 265 const UDataInfo *pInfo) { 266 return (UBool)( 267 pInfo->size>=20 && 268 pInfo->isBigEndian==U_IS_BIG_ENDIAN && 269 pInfo->charsetFamily==U_CHARSET_FAMILY && 270 pInfo->sizeofUChar==U_SIZEOF_UCHAR && 271 pInfo->dataFormat[0]==0x63 && /* dataFormat="cnvt" */ 272 pInfo->dataFormat[1]==0x6e && 273 pInfo->dataFormat[2]==0x76 && 274 pInfo->dataFormat[3]==0x74 && 275 pInfo->formatVersion[0]==6); /* Everything will be version 6 */ 276 } 277 278 /** 279 * Un flatten shared data from a UDATA.. 280 */ 281 static UConverterSharedData* 282 ucnv_data_unFlattenClone(UConverterLoadArgs *pArgs, UDataMemory *pData, UErrorCode *status) 283 { 284 /* UDataInfo info; -- necessary only if some converters have different formatVersion */ 285 const uint8_t *raw = (const uint8_t *)udata_getMemory(pData); 286 const UConverterStaticData *source = (const UConverterStaticData *) raw; 287 UConverterSharedData *data; 288 UConverterType type = (UConverterType)source->conversionType; 289 290 if(U_FAILURE(*status)) 291 return NULL; 292 293 if( (uint16_t)type >= UCNV_NUMBER_OF_SUPPORTED_CONVERTER_TYPES || 294 converterData[type] == NULL || 295 converterData[type]->referenceCounter != 1 || 296 source->structSize != sizeof(UConverterStaticData)) 297 { 298 *status = U_INVALID_TABLE_FORMAT; 299 return NULL; 300 } 301 302 data = (UConverterSharedData *)uprv_malloc(sizeof(UConverterSharedData)); 303 if(data == NULL) { 304 *status = U_MEMORY_ALLOCATION_ERROR; 305 return NULL; 306 } 307 308 /* copy initial values from the static structure for this type */ 309 uprv_memcpy(data, converterData[type], sizeof(UConverterSharedData)); 310 311 #if 0 /* made UConverterMBCSTable part of UConverterSharedData -- markus 20031107 */ 312 /* 313 * It would be much more efficient if the table were a direct member, not a pointer. 314 * However, that would add to the size of all UConverterSharedData objects 315 * even if they do not use this table (especially algorithmic ones). 316 * If this changes, then the static templates from converterData[type] 317 * need more entries. 318 * 319 * In principle, it would be cleaner if the load() function below 320 * allocated the table. 321 */ 322 data->table = (UConverterTable *)uprv_malloc(sizeof(UConverterTable)); 323 if(data->table == NULL) { 324 uprv_free(data); 325 *status = U_MEMORY_ALLOCATION_ERROR; 326 return NULL; 327 } 328 uprv_memset(data->table, 0, sizeof(UConverterTable)); 329 #endif 330 331 data->staticData = source; 332 333 data->sharedDataCached = FALSE; 334 335 /* fill in fields from the loaded data */ 336 data->dataMemory = (void*)pData; /* for future use */ 337 338 if(data->impl->load != NULL) { 339 data->impl->load(data, pArgs, raw + source->structSize, status); 340 if(U_FAILURE(*status)) { 341 uprv_free(data->table); 342 uprv_free(data); 343 return NULL; 344 } 345 } 346 return data; 347 } 348 349 /*Takes an alias name gets an actual converter file name 350 *goes to disk and opens it. 351 *allocates the memory and returns a new UConverter object 352 */ 353 static UConverterSharedData *createConverterFromFile(UConverterLoadArgs *pArgs, UErrorCode * err) 354 { 355 UDataMemory *data; 356 UConverterSharedData *sharedData; 357 358 UTRACE_ENTRY_OC(UTRACE_UCNV_LOAD); 359 360 if (U_FAILURE (*err)) { 361 UTRACE_EXIT_STATUS(*err); 362 return NULL; 363 } 364 365 UTRACE_DATA2(UTRACE_OPEN_CLOSE, "load converter %s from package %s", pArgs->name, pArgs->pkg); 366 367 data = udata_openChoice(pArgs->pkg, DATA_TYPE, pArgs->name, isCnvAcceptable, NULL, err); 368 if(U_FAILURE(*err)) 369 { 370 UTRACE_EXIT_STATUS(*err); 371 return NULL; 372 } 373 374 sharedData = ucnv_data_unFlattenClone(pArgs, data, err); 375 if(U_FAILURE(*err)) 376 { 377 udata_close(data); 378 UTRACE_EXIT_STATUS(*err); 379 return NULL; 380 } 381 382 /* 383 * TODO Store pkg in a field in the shared data so that delta-only converters 384 * can load base converters from the same package. 385 * If the pkg name is longer than the field, then either do not load the converter 386 * in the first place, or just set the pkg field to "". 387 */ 388 389 UTRACE_EXIT_PTR_STATUS(sharedData, *err); 390 return sharedData; 391 } 392 393 /*returns a converter type from a string 394 */ 395 static const UConverterSharedData * 396 getAlgorithmicTypeFromName(const char *realName) 397 { 398 uint32_t mid, start, limit; 399 uint32_t lastMid; 400 int result; 401 char strippedName[UCNV_MAX_CONVERTER_NAME_LENGTH]; 402 403 /* Lower case and remove ignoreable characters. */ 404 ucnv_io_stripForCompare(strippedName, realName); 405 406 /* do a binary search for the alias */ 407 start = 0; 408 limit = sizeof(cnvNameType)/sizeof(cnvNameType[0]); 409 mid = limit; 410 lastMid = UINT32_MAX; 411 412 for (;;) { 413 mid = (uint32_t)((start + limit) / 2); 414 if (lastMid == mid) { /* Have we moved? */ 415 break; /* We haven't moved, and it wasn't found. */ 416 } 417 lastMid = mid; 418 result = uprv_strcmp(strippedName, cnvNameType[mid].name); 419 420 if (result < 0) { 421 limit = mid; 422 } else if (result > 0) { 423 start = mid; 424 } else { 425 return converterData[cnvNameType[mid].type]; 426 } 427 } 428 429 return NULL; 430 } 431 432 /* 433 * Based on the number of known converters, this determines how many times larger 434 * the shared data hash table should be. When on small platforms, or just a couple 435 * of converters are used, this number should be 2. When memory is plentiful, or 436 * when ucnv_countAvailable is ever used with a lot of available converters, 437 * this should be 4. 438 * Larger numbers reduce the number of hash collisions, but use more memory. 439 */ 440 #define UCNV_CACHE_LOAD_FACTOR 2 441 442 /* Puts the shared data in the static hashtable SHARED_DATA_HASHTABLE */ 443 /* Will always be called with the cnvCacheMutex alrady being held */ 444 /* by the calling function. */ 445 /* Stores the shared data in the SHARED_DATA_HASHTABLE 446 * @param data The shared data 447 */ 448 static void 449 ucnv_shareConverterData(UConverterSharedData * data) 450 { 451 UErrorCode err = U_ZERO_ERROR; 452 /*Lazy evaluates the Hashtable itself */ 453 /*void *sanity = NULL;*/ 454 455 if (SHARED_DATA_HASHTABLE == NULL) 456 { 457 SHARED_DATA_HASHTABLE = uhash_openSize(uhash_hashChars, uhash_compareChars, NULL, 458 ucnv_io_countKnownConverters(&err)*UCNV_CACHE_LOAD_FACTOR, 459 &err); 460 ucln_common_registerCleanup(UCLN_COMMON_UCNV, ucnv_cleanup); 461 462 if (U_FAILURE(err)) 463 return; 464 } 465 466 /* ### check to see if the element is not already there! */ 467 468 /* 469 sanity = ucnv_getSharedConverterData (data->staticData->name); 470 if(sanity != NULL) 471 { 472 UCNV_DEBUG_LOG("put:overwrite!",data->staticData->name,sanity); 473 } 474 UCNV_DEBUG_LOG("put:chk",data->staticData->name,sanity); 475 */ 476 477 /* Mark it shared */ 478 data->sharedDataCached = TRUE; 479 480 uhash_put(SHARED_DATA_HASHTABLE, 481 (void*) data->staticData->name, /* Okay to cast away const as long as 482 keyDeleter == NULL */ 483 data, 484 &err); 485 UCNV_DEBUG_LOG("put", data->staticData->name,data); 486 487 } 488 489 /* Look up a converter name in the shared data cache. */ 490 /* cnvCacheMutex must be held by the caller to protect the hash table. */ 491 /* gets the shared data from the SHARED_DATA_HASHTABLE (might return NULL if it isn't there) 492 * @param name The name of the shared data 493 * @return the shared data from the SHARED_DATA_HASHTABLE 494 */ 495 static UConverterSharedData * 496 ucnv_getSharedConverterData(const char *name) 497 { 498 /*special case when no Table has yet been created we return NULL */ 499 if (SHARED_DATA_HASHTABLE == NULL) 500 { 501 return NULL; 502 } 503 else 504 { 505 UConverterSharedData *rc; 506 507 rc = (UConverterSharedData*)uhash_get(SHARED_DATA_HASHTABLE, name); 508 UCNV_DEBUG_LOG("get",name,rc); 509 return rc; 510 } 511 } 512 513 /*frees the string of memory blocks associates with a sharedConverter 514 *if and only if the referenceCounter == 0 515 */ 516 /* Deletes (frees) the Shared data it's passed. first it checks the referenceCounter to 517 * see if anyone is using it, if not it frees all the memory stemming from sharedConverterData and 518 * returns TRUE, 519 * otherwise returns FALSE 520 * @param sharedConverterData The shared data 521 * @return if not it frees all the memory stemming from sharedConverterData and 522 * returns TRUE, otherwise returns FALSE 523 */ 524 static UBool 525 ucnv_deleteSharedConverterData(UConverterSharedData * deadSharedData) 526 { 527 UTRACE_ENTRY_OC(UTRACE_UCNV_UNLOAD); 528 UTRACE_DATA2(UTRACE_OPEN_CLOSE, "unload converter %s shared data %p", deadSharedData->staticData->name, deadSharedData); 529 530 if (deadSharedData->referenceCounter > 0) { 531 UTRACE_EXIT_VALUE((int32_t)FALSE); 532 return FALSE; 533 } 534 535 if (deadSharedData->impl->unload != NULL) { 536 deadSharedData->impl->unload(deadSharedData); 537 } 538 539 if(deadSharedData->dataMemory != NULL) 540 { 541 UDataMemory *data = (UDataMemory*)deadSharedData->dataMemory; 542 udata_close(data); 543 } 544 545 if(deadSharedData->table != NULL) 546 { 547 uprv_free(deadSharedData->table); 548 } 549 550 #if 0 551 /* if the static data is actually owned by the shared data */ 552 /* enable if we ever have this situation. */ 553 if(deadSharedData->staticDataOwned == TRUE) /* see ucnv_bld.h */ 554 { 555 uprv_free((void*)deadSharedData->staticData); 556 } 557 #endif 558 559 #if 0 560 /* Zap it ! */ 561 uprv_memset(deadSharedData->0, sizeof(*deadSharedData)); 562 #endif 563 564 uprv_free(deadSharedData); 565 566 UTRACE_EXIT_VALUE((int32_t)TRUE); 567 return TRUE; 568 } 569 570 /** 571 * Load a non-algorithmic converter. 572 * If pkg==NULL, then this function must be called inside umtx_lock(&cnvCacheMutex). 573 */ 574 UConverterSharedData * 575 ucnv_load(UConverterLoadArgs *pArgs, UErrorCode *err) { 576 UConverterSharedData *mySharedConverterData; 577 578 if(err == NULL || U_FAILURE(*err)) { 579 return NULL; 580 } 581 582 if(pArgs->pkg != NULL && *pArgs->pkg != 0) { 583 /* application-provided converters are not currently cached */ 584 return createConverterFromFile(pArgs, err); 585 } 586 587 mySharedConverterData = ucnv_getSharedConverterData(pArgs->name); 588 if (mySharedConverterData == NULL) 589 { 590 /*Not cached, we need to stream it in from file */ 591 mySharedConverterData = createConverterFromFile(pArgs, err); 592 if (U_FAILURE (*err) || (mySharedConverterData == NULL)) 593 { 594 return NULL; 595 } 596 else if (!pArgs->onlyTestIsLoadable) 597 { 598 /* share it with other library clients */ 599 ucnv_shareConverterData(mySharedConverterData); 600 } 601 } 602 else 603 { 604 /* The data for this converter was already in the cache. */ 605 /* Update the reference counter on the shared data: one more client */ 606 mySharedConverterData->referenceCounter++; 607 } 608 609 return mySharedConverterData; 610 } 611 612 /** 613 * Unload a non-algorithmic converter. 614 * It must be sharedData->referenceCounter != ~0 615 * and this function must be called inside umtx_lock(&cnvCacheMutex). 616 */ 617 U_CAPI void 618 ucnv_unload(UConverterSharedData *sharedData) { 619 if(sharedData != NULL) { 620 if (sharedData->referenceCounter > 0) { 621 sharedData->referenceCounter--; 622 } 623 624 if((sharedData->referenceCounter <= 0)&&(sharedData->sharedDataCached == FALSE)) { 625 ucnv_deleteSharedConverterData(sharedData); 626 } 627 } 628 } 629 630 U_CFUNC void 631 ucnv_unloadSharedDataIfReady(UConverterSharedData *sharedData) 632 { 633 /* 634 Checking whether it's an algorithic converter is okay 635 in multithreaded applications because the value never changes. 636 Don't check referenceCounter for any other value. 637 */ 638 if(sharedData != NULL && sharedData->referenceCounter != (uint32_t)~0) { 639 umtx_lock(&cnvCacheMutex); 640 ucnv_unload(sharedData); 641 umtx_unlock(&cnvCacheMutex); 642 } 643 } 644 645 U_CFUNC void 646 ucnv_incrementRefCount(UConverterSharedData *sharedData) 647 { 648 /* 649 Checking whether it's an algorithic converter is okay 650 in multithreaded applications because the value never changes. 651 Don't check referenceCounter for any other value. 652 */ 653 if(sharedData != NULL && sharedData->referenceCounter != (uint32_t)~0) { 654 umtx_lock(&cnvCacheMutex); 655 sharedData->referenceCounter++; 656 umtx_unlock(&cnvCacheMutex); 657 } 658 } 659 660 /* 661 * *pPieces must be initialized. 662 * The name without options will be copied to pPieces->cnvName. 663 * The locale and options will be copied to pPieces only if present in inName, 664 * otherwise the existing values in pPieces remain. 665 * *pArgs will be set to the pPieces values. 666 */ 667 static void 668 parseConverterOptions(const char *inName, 669 UConverterNamePieces *pPieces, 670 UConverterLoadArgs *pArgs, 671 UErrorCode *err) 672 { 673 char *cnvName = pPieces->cnvName; 674 char c; 675 int32_t len = 0; 676 677 pArgs->name=inName; 678 pArgs->locale=pPieces->locale; 679 pArgs->options=pPieces->options; 680 681 /* copy the converter name itself to cnvName */ 682 while((c=*inName)!=0 && c!=UCNV_OPTION_SEP_CHAR) { 683 if (++len>=UCNV_MAX_CONVERTER_NAME_LENGTH) { 684 *err = U_ILLEGAL_ARGUMENT_ERROR; /* bad name */ 685 pPieces->cnvName[0]=0; 686 return; 687 } 688 *cnvName++=c; 689 inName++; 690 } 691 *cnvName=0; 692 pArgs->name=pPieces->cnvName; 693 694 /* parse options. No more name copying should occur. */ 695 while((c=*inName)!=0) { 696 if(c==UCNV_OPTION_SEP_CHAR) { 697 ++inName; 698 } 699 700 /* inName is behind an option separator */ 701 if(uprv_strncmp(inName, "locale=", 7)==0) { 702 /* do not modify locale itself in case we have multiple locale options */ 703 char *dest=pPieces->locale; 704 705 /* copy the locale option value */ 706 inName+=7; 707 len=0; 708 while((c=*inName)!=0 && c!=UCNV_OPTION_SEP_CHAR) { 709 ++inName; 710 711 if(++len>=ULOC_FULLNAME_CAPACITY) { 712 *err=U_ILLEGAL_ARGUMENT_ERROR; /* bad name */ 713 pPieces->locale[0]=0; 714 return; 715 } 716 717 *dest++=c; 718 } 719 *dest=0; 720 } else if(uprv_strncmp(inName, "version=", 8)==0) { 721 /* copy the version option value into bits 3..0 of pPieces->options */ 722 inName+=8; 723 c=*inName; 724 if(c==0) { 725 pArgs->options=(pPieces->options&=~UCNV_OPTION_VERSION); 726 return; 727 } else if((uint8_t)(c-'0')<10) { 728 pArgs->options=pPieces->options=(pPieces->options&~UCNV_OPTION_VERSION)|(uint32_t)(c-'0'); 729 ++inName; 730 } 731 } else if(uprv_strncmp(inName, "swaplfnl", 8)==0) { 732 inName+=8; 733 pArgs->options=(pPieces->options|=UCNV_OPTION_SWAP_LFNL); 734 /* add processing for new options here with another } else if(uprv_strncmp(inName, "option-name=", XX)==0) { */ 735 } else { 736 /* ignore any other options until we define some */ 737 while(((c = *inName++) != 0) && (c != UCNV_OPTION_SEP_CHAR)) { 738 } 739 if(c==0) { 740 return; 741 } 742 } 743 } 744 } 745 746 /*Logic determines if the converter is Algorithmic AND/OR cached 747 *depending on that: 748 * -we either go to get data from disk and cache it (Data=TRUE, Cached=False) 749 * -Get it from a Hashtable (Data=X, Cached=TRUE) 750 * -Call dataConverter initializer (Data=TRUE, Cached=TRUE) 751 * -Call AlgorithmicConverter initializer (Data=FALSE, Cached=TRUE) 752 */ 753 U_CFUNC UConverterSharedData * 754 ucnv_loadSharedData(const char *converterName, 755 UConverterNamePieces *pPieces, 756 UConverterLoadArgs *pArgs, 757 UErrorCode * err) { 758 UConverterNamePieces stackPieces; 759 UConverterLoadArgs stackArgs; 760 UConverterSharedData *mySharedConverterData = NULL; 761 UErrorCode internalErrorCode = U_ZERO_ERROR; 762 UBool mayContainOption = TRUE; 763 UBool checkForAlgorithmic = TRUE; 764 765 if (U_FAILURE (*err)) { 766 return NULL; 767 } 768 769 if(pPieces == NULL) { 770 if(pArgs != NULL) { 771 /* 772 * Bad: We may set pArgs pointers to stackPieces fields 773 * which will be invalid after this function returns. 774 */ 775 *err = U_INTERNAL_PROGRAM_ERROR; 776 return NULL; 777 } 778 pPieces = &stackPieces; 779 } 780 if(pArgs == NULL) { 781 uprv_memset(&stackArgs, 0, sizeof(stackArgs)); 782 stackArgs.size = (int32_t)sizeof(stackArgs); 783 pArgs = &stackArgs; 784 } 785 786 pPieces->cnvName[0] = 0; 787 pPieces->locale[0] = 0; 788 pPieces->options = 0; 789 790 pArgs->name = converterName; 791 pArgs->locale = pPieces->locale; 792 pArgs->options = pPieces->options; 793 794 /* In case "name" is NULL we want to open the default converter. */ 795 if (converterName == NULL) { 796 #if U_CHARSET_IS_UTF8 797 pArgs->name = "UTF-8"; 798 return (UConverterSharedData *)converterData[UCNV_UTF8]; 799 #else 800 /* Call ucnv_getDefaultName first to query the name from the OS. */ 801 pArgs->name = ucnv_getDefaultName(); 802 if (pArgs->name == NULL) { 803 *err = U_MISSING_RESOURCE_ERROR; 804 return NULL; 805 } 806 mySharedConverterData = (UConverterSharedData *)gDefaultAlgorithmicSharedData; 807 checkForAlgorithmic = FALSE; 808 mayContainOption = gDefaultConverterContainsOption; 809 /* the default converter name is already canonical */ 810 #endif 811 } 812 else if(UCNV_FAST_IS_UTF8(converterName)) { 813 /* fastpath for UTF-8 */ 814 pArgs->name = "UTF-8"; 815 return (UConverterSharedData *)converterData[UCNV_UTF8]; 816 } 817 else { 818 /* separate the converter name from the options */ 819 parseConverterOptions(converterName, pPieces, pArgs, err); 820 if (U_FAILURE(*err)) { 821 /* Very bad name used. */ 822 return NULL; 823 } 824 825 /* get the canonical converter name */ 826 pArgs->name = ucnv_io_getConverterName(pArgs->name, &mayContainOption, &internalErrorCode); 827 if (U_FAILURE(internalErrorCode) || pArgs->name == NULL) { 828 /* 829 * set the input name in case the converter was added 830 * without updating the alias table, or when there is no alias table 831 */ 832 pArgs->name = pPieces->cnvName; 833 } else if (internalErrorCode == U_AMBIGUOUS_ALIAS_WARNING) { 834 *err = U_AMBIGUOUS_ALIAS_WARNING; 835 } 836 } 837 838 /* separate the converter name from the options */ 839 if(mayContainOption && pArgs->name != pPieces->cnvName) { 840 parseConverterOptions(pArgs->name, pPieces, pArgs, err); 841 } 842 843 /* get the shared data for an algorithmic converter, if it is one */ 844 if (checkForAlgorithmic) { 845 mySharedConverterData = (UConverterSharedData *)getAlgorithmicTypeFromName(pArgs->name); 846 } 847 if (mySharedConverterData == NULL) 848 { 849 /* it is a data-based converter, get its shared data. */ 850 /* Hold the cnvCacheMutex through the whole process of checking the */ 851 /* converter data cache, and adding new entries to the cache */ 852 /* to prevent other threads from modifying the cache during the */ 853 /* process. */ 854 pArgs->nestedLoads=1; 855 pArgs->pkg=NULL; 856 857 umtx_lock(&cnvCacheMutex); 858 mySharedConverterData = ucnv_load(pArgs, err); 859 umtx_unlock(&cnvCacheMutex); 860 if (U_FAILURE (*err) || (mySharedConverterData == NULL)) 861 { 862 return NULL; 863 } 864 } 865 866 return mySharedConverterData; 867 } 868 869 U_CAPI UConverter * 870 ucnv_createConverter(UConverter *myUConverter, const char *converterName, UErrorCode * err) 871 { 872 UConverterNamePieces stackPieces; 873 UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; 874 UConverterSharedData *mySharedConverterData; 875 876 UTRACE_ENTRY_OC(UTRACE_UCNV_OPEN); 877 878 if(U_SUCCESS(*err)) { 879 UTRACE_DATA1(UTRACE_OPEN_CLOSE, "open converter %s", converterName); 880 881 mySharedConverterData = ucnv_loadSharedData(converterName, &stackPieces, &stackArgs, err); 882 883 myUConverter = ucnv_createConverterFromSharedData( 884 myUConverter, mySharedConverterData, 885 &stackArgs, 886 err); 887 888 if(U_SUCCESS(*err)) { 889 UTRACE_EXIT_PTR_STATUS(myUConverter, *err); 890 return myUConverter; 891 } 892 } 893 894 /* exit with error */ 895 UTRACE_EXIT_STATUS(*err); 896 return NULL; 897 } 898 899 U_CFUNC UBool 900 ucnv_canCreateConverter(const char *converterName, UErrorCode *err) { 901 UConverter myUConverter; 902 UConverterNamePieces stackPieces; 903 UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; 904 UConverterSharedData *mySharedConverterData; 905 906 UTRACE_ENTRY_OC(UTRACE_UCNV_OPEN); 907 908 if(U_SUCCESS(*err)) { 909 UTRACE_DATA1(UTRACE_OPEN_CLOSE, "test if can open converter %s", converterName); 910 911 stackArgs.onlyTestIsLoadable=TRUE; 912 mySharedConverterData = ucnv_loadSharedData(converterName, &stackPieces, &stackArgs, err); 913 ucnv_createConverterFromSharedData( 914 &myUConverter, mySharedConverterData, 915 &stackArgs, 916 err); 917 ucnv_unloadSharedDataIfReady(mySharedConverterData); 918 } 919 920 UTRACE_EXIT_STATUS(*err); 921 return U_SUCCESS(*err); 922 } 923 924 UConverter * 925 ucnv_createAlgorithmicConverter(UConverter *myUConverter, 926 UConverterType type, 927 const char *locale, uint32_t options, 928 UErrorCode *err) { 929 UConverter *cnv; 930 const UConverterSharedData *sharedData; 931 UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; 932 933 UTRACE_ENTRY_OC(UTRACE_UCNV_OPEN_ALGORITHMIC); 934 UTRACE_DATA1(UTRACE_OPEN_CLOSE, "open algorithmic converter type %d", (int32_t)type); 935 936 if(type<0 || UCNV_NUMBER_OF_SUPPORTED_CONVERTER_TYPES<=type) { 937 *err = U_ILLEGAL_ARGUMENT_ERROR; 938 UTRACE_EXIT_STATUS(U_ILLEGAL_ARGUMENT_ERROR); 939 return NULL; 940 } 941 942 sharedData = converterData[type]; 943 /* 944 Checking whether it's an algorithic converter is okay 945 in multithreaded applications because the value never changes. 946 Don't check referenceCounter for any other value. 947 */ 948 if(sharedData == NULL || sharedData->referenceCounter != (uint32_t)~0) { 949 /* not a valid type, or not an algorithmic converter */ 950 *err = U_ILLEGAL_ARGUMENT_ERROR; 951 UTRACE_EXIT_STATUS(U_ILLEGAL_ARGUMENT_ERROR); 952 return NULL; 953 } 954 955 stackArgs.name = ""; 956 stackArgs.options = options; 957 stackArgs.locale=locale; 958 cnv = ucnv_createConverterFromSharedData( 959 myUConverter, (UConverterSharedData *)sharedData, 960 &stackArgs, err); 961 962 UTRACE_EXIT_PTR_STATUS(cnv, *err); 963 return cnv; 964 } 965 966 U_CFUNC UConverter* 967 ucnv_createConverterFromPackage(const char *packageName, const char *converterName, UErrorCode * err) 968 { 969 UConverter *myUConverter; 970 UConverterSharedData *mySharedConverterData; 971 UConverterNamePieces stackPieces; 972 UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; 973 974 UTRACE_ENTRY_OC(UTRACE_UCNV_OPEN_PACKAGE); 975 976 if(U_FAILURE(*err)) { 977 UTRACE_EXIT_STATUS(*err); 978 return NULL; 979 } 980 981 UTRACE_DATA2(UTRACE_OPEN_CLOSE, "open converter %s from package %s", converterName, packageName); 982 983 /* first, get the options out of the converterName string */ 984 stackPieces.cnvName[0] = 0; 985 stackPieces.locale[0] = 0; 986 stackPieces.options = 0; 987 parseConverterOptions(converterName, &stackPieces, &stackArgs, err); 988 if (U_FAILURE(*err)) { 989 /* Very bad name used. */ 990 UTRACE_EXIT_STATUS(*err); 991 return NULL; 992 } 993 stackArgs.nestedLoads=1; 994 stackArgs.pkg=packageName; 995 996 /* open the data, unflatten the shared structure */ 997 mySharedConverterData = createConverterFromFile(&stackArgs, err); 998 999 if (U_FAILURE(*err)) { 1000 UTRACE_EXIT_STATUS(*err); 1001 return NULL; 1002 } 1003 1004 /* create the actual converter */ 1005 myUConverter = ucnv_createConverterFromSharedData(NULL, mySharedConverterData, &stackArgs, err); 1006 1007 if (U_FAILURE(*err)) { 1008 ucnv_close(myUConverter); 1009 UTRACE_EXIT_STATUS(*err); 1010 return NULL; 1011 } 1012 1013 UTRACE_EXIT_PTR_STATUS(myUConverter, *err); 1014 return myUConverter; 1015 } 1016 1017 1018 U_CFUNC UConverter* 1019 ucnv_createConverterFromSharedData(UConverter *myUConverter, 1020 UConverterSharedData *mySharedConverterData, 1021 UConverterLoadArgs *pArgs, 1022 UErrorCode *err) 1023 { 1024 UBool isCopyLocal; 1025 1026 if(U_FAILURE(*err)) { 1027 ucnv_unloadSharedDataIfReady(mySharedConverterData); 1028 return myUConverter; 1029 } 1030 if(myUConverter == NULL) 1031 { 1032 myUConverter = (UConverter *) uprv_malloc (sizeof (UConverter)); 1033 if(myUConverter == NULL) 1034 { 1035 *err = U_MEMORY_ALLOCATION_ERROR; 1036 ucnv_unloadSharedDataIfReady(mySharedConverterData); 1037 return NULL; 1038 } 1039 isCopyLocal = FALSE; 1040 } else { 1041 isCopyLocal = TRUE; 1042 } 1043 1044 /* initialize the converter */ 1045 uprv_memset(myUConverter, 0, sizeof(UConverter)); 1046 myUConverter->isCopyLocal = isCopyLocal; 1047 /*myUConverter->isExtraLocal = FALSE;*/ /* Set by the memset call */ 1048 myUConverter->sharedData = mySharedConverterData; 1049 myUConverter->options = pArgs->options; 1050 if(!pArgs->onlyTestIsLoadable) { 1051 myUConverter->preFromUFirstCP = U_SENTINEL; 1052 myUConverter->fromCharErrorBehaviour = UCNV_TO_U_DEFAULT_CALLBACK; 1053 myUConverter->fromUCharErrorBehaviour = UCNV_FROM_U_DEFAULT_CALLBACK; 1054 myUConverter->toUnicodeStatus = mySharedConverterData->toUnicodeStatus; 1055 myUConverter->maxBytesPerUChar = mySharedConverterData->staticData->maxBytesPerChar; 1056 myUConverter->subChar1 = mySharedConverterData->staticData->subChar1; 1057 myUConverter->subCharLen = mySharedConverterData->staticData->subCharLen; 1058 myUConverter->subChars = (uint8_t *)myUConverter->subUChars; 1059 uprv_memcpy(myUConverter->subChars, mySharedConverterData->staticData->subChar, myUConverter->subCharLen); 1060 myUConverter->toUCallbackReason = UCNV_ILLEGAL; /* default reason to invoke (*fromCharErrorBehaviour) */ 1061 } 1062 1063 if(mySharedConverterData->impl->open != NULL) { 1064 mySharedConverterData->impl->open(myUConverter, pArgs, err); 1065 if(U_FAILURE(*err) && !pArgs->onlyTestIsLoadable) { 1066 /* don't ucnv_close() if onlyTestIsLoadable because not fully initialized */ 1067 ucnv_close(myUConverter); 1068 return NULL; 1069 } 1070 } 1071 1072 return myUConverter; 1073 } 1074 1075 /*Frees all shared immutable objects that aren't referred to (reference count = 0) 1076 */ 1077 U_CAPI int32_t U_EXPORT2 1078 ucnv_flushCache () 1079 { 1080 UConverterSharedData *mySharedData = NULL; 1081 int32_t pos; 1082 int32_t tableDeletedNum = 0; 1083 const UHashElement *e; 1084 /*UErrorCode status = U_ILLEGAL_ARGUMENT_ERROR;*/ 1085 int32_t i, remaining; 1086 1087 UTRACE_ENTRY_OC(UTRACE_UCNV_FLUSH_CACHE); 1088 1089 /* Close the default converter without creating a new one so that everything will be flushed. */ 1090 u_flushDefaultConverter(); 1091 1092 /*if shared data hasn't even been lazy evaluated yet 1093 * return 0 1094 */ 1095 if (SHARED_DATA_HASHTABLE == NULL) { 1096 UTRACE_EXIT_VALUE((int32_t)0); 1097 return 0; 1098 } 1099 1100 /*creates an enumeration to iterate through every element in the 1101 * table 1102 * 1103 * Synchronization: holding cnvCacheMutex will prevent any other thread from 1104 * accessing or modifying the hash table during the iteration. 1105 * The reference count of an entry may be decremented by 1106 * ucnv_close while the iteration is in process, but this is 1107 * benign. It can't be incremented (in ucnv_createConverter()) 1108 * because the sequence of looking up in the cache + incrementing 1109 * is protected by cnvCacheMutex. 1110 */ 1111 umtx_lock(&cnvCacheMutex); 1112 /* 1113 * double loop: A delta/extension-only converter has a pointer to its base table's 1114 * shared data; the first iteration of the outer loop may see the delta converter 1115 * before the base converter, and unloading the delta converter may get the base 1116 * converter's reference counter down to 0. 1117 */ 1118 i = 0; 1119 do { 1120 remaining = 0; 1121 pos = UHASH_FIRST; 1122 while ((e = uhash_nextElement (SHARED_DATA_HASHTABLE, &pos)) != NULL) 1123 { 1124 mySharedData = (UConverterSharedData *) e->value.pointer; 1125 /*deletes only if reference counter == 0 */ 1126 if (mySharedData->referenceCounter == 0) 1127 { 1128 tableDeletedNum++; 1129 1130 UCNV_DEBUG_LOG("del",mySharedData->staticData->name,mySharedData); 1131 1132 uhash_removeElement(SHARED_DATA_HASHTABLE, e); 1133 mySharedData->sharedDataCached = FALSE; 1134 ucnv_deleteSharedConverterData (mySharedData); 1135 } else { 1136 ++remaining; 1137 } 1138 } 1139 } while(++i == 1 && remaining > 0); 1140 umtx_unlock(&cnvCacheMutex); 1141 1142 UTRACE_DATA1(UTRACE_INFO, "ucnv_flushCache() exits with %d converters remaining", remaining); 1143 1144 UTRACE_EXIT_VALUE(tableDeletedNum); 1145 return tableDeletedNum; 1146 } 1147 1148 /* available converters list --------------------------------------------------- */ 1149 1150 static void U_CALLCONV initAvailableConvertersList(UErrorCode &errCode) { 1151 U_ASSERT(gAvailableConverterCount == 0); 1152 U_ASSERT(gAvailableConverters == NULL); 1153 1154 ucln_common_registerCleanup(UCLN_COMMON_UCNV, ucnv_cleanup); 1155 UEnumeration *allConvEnum = ucnv_openAllNames(&errCode); 1156 int32_t allConverterCount = uenum_count(allConvEnum, &errCode); 1157 if (U_FAILURE(errCode)) { 1158 return; 1159 } 1160 1161 /* We can't have more than "*converterTable" converters to open */ 1162 gAvailableConverters = (const char **) uprv_malloc(allConverterCount * sizeof(char*)); 1163 if (!gAvailableConverters) { 1164 errCode = U_MEMORY_ALLOCATION_ERROR; 1165 return; 1166 } 1167 1168 /* Open the default converter to make sure that it has first dibs in the hash table. */ 1169 UErrorCode localStatus = U_ZERO_ERROR; 1170 UConverter tempConverter; 1171 ucnv_close(ucnv_createConverter(&tempConverter, NULL, &localStatus)); 1172 1173 gAvailableConverterCount = 0; 1174 1175 for (int32_t idx = 0; idx < allConverterCount; idx++) { 1176 localStatus = U_ZERO_ERROR; 1177 const char *converterName = uenum_next(allConvEnum, NULL, &localStatus); 1178 if (ucnv_canCreateConverter(converterName, &localStatus)) { 1179 gAvailableConverters[gAvailableConverterCount++] = converterName; 1180 } 1181 } 1182 1183 uenum_close(allConvEnum); 1184 } 1185 1186 1187 static UBool haveAvailableConverterList(UErrorCode *pErrorCode) { 1188 umtx_initOnce(gAvailableConvertersInitOnce, &initAvailableConvertersList, *pErrorCode); 1189 return U_SUCCESS(*pErrorCode); 1190 } 1191 1192 U_CFUNC uint16_t 1193 ucnv_bld_countAvailableConverters(UErrorCode *pErrorCode) { 1194 if (haveAvailableConverterList(pErrorCode)) { 1195 return gAvailableConverterCount; 1196 } 1197 return 0; 1198 } 1199 1200 U_CFUNC const char * 1201 ucnv_bld_getAvailableConverter(uint16_t n, UErrorCode *pErrorCode) { 1202 if (haveAvailableConverterList(pErrorCode)) { 1203 if (n < gAvailableConverterCount) { 1204 return gAvailableConverters[n]; 1205 } 1206 *pErrorCode = U_INDEX_OUTOFBOUNDS_ERROR; 1207 } 1208 return NULL; 1209 } 1210 1211 /* default converter name --------------------------------------------------- */ 1212 1213 #if !U_CHARSET_IS_UTF8 1214 /* 1215 Copy the canonical converter name. 1216 ucnv_getDefaultName must be thread safe, which can call this function. 1217 1218 ucnv_setDefaultName calls this function and it doesn't have to be 1219 thread safe because there is no reliable/safe way to reset the 1220 converter in use in all threads. If you did reset the converter, you 1221 would not be sure that retrieving a default converter for one string 1222 would be the same type of default converter for a successive string. 1223 Since the name is a returned via ucnv_getDefaultName without copying, 1224 you shouldn't be modifying or deleting the string from a separate thread. 1225 */ 1226 static inline void 1227 internalSetName(const char *name, UErrorCode *status) { 1228 UConverterNamePieces stackPieces; 1229 UConverterLoadArgs stackArgs=UCNV_LOAD_ARGS_INITIALIZER; 1230 int32_t length=(int32_t)(uprv_strlen(name)); 1231 UBool containsOption = (UBool)(uprv_strchr(name, UCNV_OPTION_SEP_CHAR) != NULL); 1232 const UConverterSharedData *algorithmicSharedData; 1233 1234 stackArgs.name = name; 1235 if(containsOption) { 1236 stackPieces.cnvName[0] = 0; 1237 stackPieces.locale[0] = 0; 1238 stackPieces.options = 0; 1239 parseConverterOptions(name, &stackPieces, &stackArgs, status); 1240 if(U_FAILURE(*status)) { 1241 return; 1242 } 1243 } 1244 algorithmicSharedData = getAlgorithmicTypeFromName(stackArgs.name); 1245 1246 umtx_lock(&cnvCacheMutex); 1247 1248 gDefaultAlgorithmicSharedData = algorithmicSharedData; 1249 gDefaultConverterContainsOption = containsOption; 1250 uprv_memcpy(gDefaultConverterNameBuffer, name, length); 1251 gDefaultConverterNameBuffer[length]=0; 1252 1253 /* gDefaultConverterName MUST be the last global var set by this function. */ 1254 /* It is the variable checked in ucnv_getDefaultName() to see if initialization is required. */ 1255 // But there is nothing here preventing that from being reordered, either by the compiler 1256 // or hardware. I'm adding the mutex to ucnv_getDefaultName for now. UMTX_CHECK is not enough. 1257 // -- Andy 1258 gDefaultConverterName = gDefaultConverterNameBuffer; 1259 1260 ucln_common_registerCleanup(UCLN_COMMON_UCNV, ucnv_cleanup); 1261 1262 umtx_unlock(&cnvCacheMutex); 1263 } 1264 #endif 1265 1266 /* 1267 * In order to be really thread-safe, the get function would have to take 1268 * a buffer parameter and copy the current string inside a mutex block. 1269 * This implementation only tries to be really thread-safe while 1270 * setting the name. 1271 * It assumes that setting a pointer is atomic. 1272 */ 1273 1274 U_CAPI const char* U_EXPORT2 1275 ucnv_getDefaultName() { 1276 #if U_CHARSET_IS_UTF8 1277 return "UTF-8"; 1278 #else 1279 /* local variable to be thread-safe */ 1280 const char *name; 1281 1282 /* 1283 Concurrent calls to ucnv_getDefaultName must be thread safe, 1284 but ucnv_setDefaultName is not thread safe. 1285 */ 1286 { 1287 icu::Mutex lock(&cnvCacheMutex); 1288 name = gDefaultConverterName; 1289 } 1290 if(name==NULL) { 1291 UErrorCode errorCode = U_ZERO_ERROR; 1292 UConverter *cnv = NULL; 1293 1294 name = uprv_getDefaultCodepage(); 1295 1296 /* if the name is there, test it out and get the canonical name with options */ 1297 if(name != NULL) { 1298 cnv = ucnv_open(name, &errorCode); 1299 if(U_SUCCESS(errorCode) && cnv != NULL) { 1300 name = ucnv_getName(cnv, &errorCode); 1301 } 1302 } 1303 1304 if(name == NULL || name[0] == 0 1305 || U_FAILURE(errorCode) || cnv == NULL 1306 || uprv_strlen(name)>=sizeof(gDefaultConverterNameBuffer)) 1307 { 1308 /* Panic time, let's use a fallback. */ 1309 #if (U_CHARSET_FAMILY == U_ASCII_FAMILY) 1310 name = "US-ASCII"; 1311 /* there is no 'algorithmic' converter for EBCDIC */ 1312 #elif U_PLATFORM == U_PF_OS390 1313 name = "ibm-1047_P100-1995" UCNV_SWAP_LFNL_OPTION_STRING; 1314 #else 1315 name = "ibm-37_P100-1995"; 1316 #endif 1317 } 1318 1319 internalSetName(name, &errorCode); 1320 1321 /* The close may make the current name go away. */ 1322 ucnv_close(cnv); 1323 } 1324 1325 return name; 1326 #endif 1327 } 1328 1329 #if U_CHARSET_IS_UTF8 1330 U_CAPI void U_EXPORT2 ucnv_setDefaultName(const char *) {} 1331 #else 1332 /* 1333 This function is not thread safe, and it can't be thread safe. 1334 See internalSetName or the API reference for details. 1335 */ 1336 U_CAPI void U_EXPORT2 1337 ucnv_setDefaultName(const char *converterName) { 1338 if(converterName==NULL) { 1339 /* reset to the default codepage */ 1340 gDefaultConverterName=NULL; 1341 } else { 1342 UErrorCode errorCode = U_ZERO_ERROR; 1343 UConverter *cnv = NULL; 1344 const char *name = NULL; 1345 1346 /* if the name is there, test it out and get the canonical name with options */ 1347 cnv = ucnv_open(converterName, &errorCode); 1348 if(U_SUCCESS(errorCode) && cnv != NULL) { 1349 name = ucnv_getName(cnv, &errorCode); 1350 } 1351 1352 if(U_SUCCESS(errorCode) && name!=NULL) { 1353 internalSetName(name, &errorCode); 1354 } 1355 /* else this converter is bad to use. Don't change it to a bad value. */ 1356 1357 /* The close may make the current name go away. */ 1358 ucnv_close(cnv); 1359 1360 /* reset the converter cache */ 1361 u_flushDefaultConverter(); 1362 } 1363 } 1364 #endif 1365 1366 /* data swapping ------------------------------------------------------------ */ 1367 1368 /* most of this might belong more properly into ucnvmbcs.c, but that is so large */ 1369 1370 #if !UCONFIG_NO_LEGACY_CONVERSION 1371 1372 U_CAPI int32_t U_EXPORT2 1373 ucnv_swap(const UDataSwapper *ds, 1374 const void *inData, int32_t length, void *outData, 1375 UErrorCode *pErrorCode) { 1376 const UDataInfo *pInfo; 1377 int32_t headerSize; 1378 1379 const uint8_t *inBytes; 1380 uint8_t *outBytes; 1381 1382 uint32_t offset, count, staticDataSize; 1383 int32_t size; 1384 1385 const UConverterStaticData *inStaticData; 1386 UConverterStaticData *outStaticData; 1387 1388 const _MBCSHeader *inMBCSHeader; 1389 _MBCSHeader *outMBCSHeader; 1390 _MBCSHeader mbcsHeader; 1391 uint32_t mbcsHeaderLength; 1392 UBool noFromU=FALSE; 1393 1394 uint8_t outputType; 1395 1396 int32_t maxFastUChar, mbcsIndexLength; 1397 1398 const int32_t *inExtIndexes; 1399 int32_t extOffset; 1400 1401 /* udata_swapDataHeader checks the arguments */ 1402 headerSize=udata_swapDataHeader(ds, inData, length, outData, pErrorCode); 1403 if(pErrorCode==NULL || U_FAILURE(*pErrorCode)) { 1404 return 0; 1405 } 1406 1407 /* check data format and format version */ 1408 pInfo=(const UDataInfo *)((const char *)inData+4); 1409 if(!( 1410 pInfo->dataFormat[0]==0x63 && /* dataFormat="cnvt" */ 1411 pInfo->dataFormat[1]==0x6e && 1412 pInfo->dataFormat[2]==0x76 && 1413 pInfo->dataFormat[3]==0x74 && 1414 pInfo->formatVersion[0]==6 && 1415 pInfo->formatVersion[1]>=2 1416 )) { 1417 udata_printError(ds, "ucnv_swap(): data format %02x.%02x.%02x.%02x (format version %02x.%02x) is not recognized as an ICU .cnv conversion table\n", 1418 pInfo->dataFormat[0], pInfo->dataFormat[1], 1419 pInfo->dataFormat[2], pInfo->dataFormat[3], 1420 pInfo->formatVersion[0], pInfo->formatVersion[1]); 1421 *pErrorCode=U_UNSUPPORTED_ERROR; 1422 return 0; 1423 } 1424 1425 inBytes=(const uint8_t *)inData+headerSize; 1426 outBytes=(uint8_t *)outData+headerSize; 1427 1428 /* read the initial UConverterStaticData structure after the UDataInfo header */ 1429 inStaticData=(const UConverterStaticData *)inBytes; 1430 outStaticData=(UConverterStaticData *)outBytes; 1431 1432 if(length<0) { 1433 staticDataSize=ds->readUInt32(inStaticData->structSize); 1434 } else { 1435 length-=headerSize; 1436 if( length<(int32_t)sizeof(UConverterStaticData) || 1437 (uint32_t)length<(staticDataSize=ds->readUInt32(inStaticData->structSize)) 1438 ) { 1439 udata_printError(ds, "ucnv_swap(): too few bytes (%d after header) for an ICU .cnv conversion table\n", 1440 length); 1441 *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; 1442 return 0; 1443 } 1444 } 1445 1446 if(length>=0) { 1447 /* swap the static data */ 1448 if(inStaticData!=outStaticData) { 1449 uprv_memcpy(outStaticData, inStaticData, staticDataSize); 1450 } 1451 1452 ds->swapArray32(ds, &inStaticData->structSize, 4, 1453 &outStaticData->structSize, pErrorCode); 1454 ds->swapArray32(ds, &inStaticData->codepage, 4, 1455 &outStaticData->codepage, pErrorCode); 1456 1457 ds->swapInvChars(ds, inStaticData->name, (int32_t)uprv_strlen(inStaticData->name), 1458 outStaticData->name, pErrorCode); 1459 if(U_FAILURE(*pErrorCode)) { 1460 udata_printError(ds, "ucnv_swap(): error swapping converter name\n"); 1461 return 0; 1462 } 1463 } 1464 1465 inBytes+=staticDataSize; 1466 outBytes+=staticDataSize; 1467 if(length>=0) { 1468 length-=(int32_t)staticDataSize; 1469 } 1470 1471 /* check for supported conversionType values */ 1472 if(inStaticData->conversionType==UCNV_MBCS) { 1473 /* swap MBCS data */ 1474 inMBCSHeader=(const _MBCSHeader *)inBytes; 1475 outMBCSHeader=(_MBCSHeader *)outBytes; 1476 1477 if(0<=length && length<(int32_t)sizeof(_MBCSHeader)) { 1478 udata_printError(ds, "ucnv_swap(): too few bytes (%d after headers) for an ICU MBCS .cnv conversion table\n", 1479 length); 1480 *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; 1481 return 0; 1482 } 1483 if(inMBCSHeader->version[0]==4 && inMBCSHeader->version[1]>=1) { 1484 mbcsHeaderLength=MBCS_HEADER_V4_LENGTH; 1485 } else if(inMBCSHeader->version[0]==5 && inMBCSHeader->version[1]>=3 && 1486 ((mbcsHeader.options=ds->readUInt32(inMBCSHeader->options))& 1487 MBCS_OPT_UNKNOWN_INCOMPATIBLE_MASK)==0 1488 ) { 1489 mbcsHeaderLength=mbcsHeader.options&MBCS_OPT_LENGTH_MASK; 1490 noFromU=(UBool)((mbcsHeader.options&MBCS_OPT_NO_FROM_U)!=0); 1491 } else { 1492 udata_printError(ds, "ucnv_swap(): unsupported _MBCSHeader.version %d.%d\n", 1493 inMBCSHeader->version[0], inMBCSHeader->version[1]); 1494 *pErrorCode=U_UNSUPPORTED_ERROR; 1495 return 0; 1496 } 1497 1498 uprv_memcpy(mbcsHeader.version, inMBCSHeader->version, 4); 1499 mbcsHeader.countStates= ds->readUInt32(inMBCSHeader->countStates); 1500 mbcsHeader.countToUFallbacks= ds->readUInt32(inMBCSHeader->countToUFallbacks); 1501 mbcsHeader.offsetToUCodeUnits= ds->readUInt32(inMBCSHeader->offsetToUCodeUnits); 1502 mbcsHeader.offsetFromUTable= ds->readUInt32(inMBCSHeader->offsetFromUTable); 1503 mbcsHeader.offsetFromUBytes= ds->readUInt32(inMBCSHeader->offsetFromUBytes); 1504 mbcsHeader.flags= ds->readUInt32(inMBCSHeader->flags); 1505 mbcsHeader.fromUBytesLength= ds->readUInt32(inMBCSHeader->fromUBytesLength); 1506 /* mbcsHeader.options have been read above */ 1507 1508 extOffset=(int32_t)(mbcsHeader.flags>>8); 1509 outputType=(uint8_t)mbcsHeader.flags; 1510 if(noFromU && outputType==MBCS_OUTPUT_1) { 1511 udata_printError(ds, "ucnv_swap(): unsupported combination of makeconv --small with SBCS\n"); 1512 *pErrorCode=U_UNSUPPORTED_ERROR; 1513 return 0; 1514 } 1515 1516 /* make sure that the output type is known */ 1517 switch(outputType) { 1518 case MBCS_OUTPUT_1: 1519 case MBCS_OUTPUT_2: 1520 case MBCS_OUTPUT_3: 1521 case MBCS_OUTPUT_4: 1522 case MBCS_OUTPUT_3_EUC: 1523 case MBCS_OUTPUT_4_EUC: 1524 case MBCS_OUTPUT_2_SISO: 1525 case MBCS_OUTPUT_EXT_ONLY: 1526 /* OK */ 1527 break; 1528 default: 1529 udata_printError(ds, "ucnv_swap(): unsupported MBCS output type 0x%x\n", 1530 outputType); 1531 *pErrorCode=U_UNSUPPORTED_ERROR; 1532 return 0; 1533 } 1534 1535 /* calculate the length of the MBCS data */ 1536 1537 /* 1538 * utf8Friendly MBCS files (mbcsHeader.version 4.3) 1539 * contain an additional mbcsIndex table: 1540 * uint16_t[(maxFastUChar+1)>>6]; 1541 * where maxFastUChar=((mbcsHeader.version[2]<<8)|0xff). 1542 */ 1543 maxFastUChar=0; 1544 mbcsIndexLength=0; 1545 if( outputType!=MBCS_OUTPUT_EXT_ONLY && outputType!=MBCS_OUTPUT_1 && 1546 mbcsHeader.version[1]>=3 && (maxFastUChar=mbcsHeader.version[2])!=0 1547 ) { 1548 maxFastUChar=(maxFastUChar<<8)|0xff; 1549 mbcsIndexLength=((maxFastUChar+1)>>6)*2; /* number of bytes */ 1550 } 1551 1552 if(extOffset==0) { 1553 size=(int32_t)(mbcsHeader.offsetFromUBytes+mbcsIndexLength); 1554 if(!noFromU) { 1555 size+=(int32_t)mbcsHeader.fromUBytesLength; 1556 } 1557 1558 /* avoid compiler warnings - not otherwise necessary, and the value does not matter */ 1559 inExtIndexes=NULL; 1560 } else { 1561 /* there is extension data after the base data, see ucnv_ext.h */ 1562 if(length>=0 && length<(extOffset+UCNV_EXT_INDEXES_MIN_LENGTH*4)) { 1563 udata_printError(ds, "ucnv_swap(): too few bytes (%d after headers) for an ICU MBCS .cnv conversion table with extension data\n", 1564 length); 1565 *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; 1566 return 0; 1567 } 1568 1569 inExtIndexes=(const int32_t *)(inBytes+extOffset); 1570 size=extOffset+udata_readInt32(ds, inExtIndexes[UCNV_EXT_SIZE]); 1571 } 1572 1573 if(length>=0) { 1574 if(length<size) { 1575 udata_printError(ds, "ucnv_swap(): too few bytes (%d after headers) for an ICU MBCS .cnv conversion table\n", 1576 length); 1577 *pErrorCode=U_INDEX_OUTOFBOUNDS_ERROR; 1578 return 0; 1579 } 1580 1581 /* copy the data for inaccessible bytes */ 1582 if(inBytes!=outBytes) { 1583 uprv_memcpy(outBytes, inBytes, size); 1584 } 1585 1586 /* swap the MBCSHeader, except for the version field */ 1587 count=mbcsHeaderLength*4; 1588 ds->swapArray32(ds, &inMBCSHeader->countStates, count-4, 1589 &outMBCSHeader->countStates, pErrorCode); 1590 1591 if(outputType==MBCS_OUTPUT_EXT_ONLY) { 1592 /* 1593 * extension-only file, 1594 * contains a base name instead of normal base table data 1595 */ 1596 1597 /* swap the base name, between the header and the extension data */ 1598 const char *inBaseName=(const char *)inBytes+count; 1599 char *outBaseName=(char *)outBytes+count; 1600 ds->swapInvChars(ds, inBaseName, (int32_t)uprv_strlen(inBaseName), 1601 outBaseName, pErrorCode); 1602 } else { 1603 /* normal file with base table data */ 1604 1605 /* swap the state table, 1kB per state */ 1606 offset=count; 1607 count=mbcsHeader.countStates*1024; 1608 ds->swapArray32(ds, inBytes+offset, (int32_t)count, 1609 outBytes+offset, pErrorCode); 1610 1611 /* swap the toUFallbacks[] */ 1612 offset+=count; 1613 count=mbcsHeader.countToUFallbacks*8; 1614 ds->swapArray32(ds, inBytes+offset, (int32_t)count, 1615 outBytes+offset, pErrorCode); 1616 1617 /* swap the unicodeCodeUnits[] */ 1618 offset=mbcsHeader.offsetToUCodeUnits; 1619 count=mbcsHeader.offsetFromUTable-offset; 1620 ds->swapArray16(ds, inBytes+offset, (int32_t)count, 1621 outBytes+offset, pErrorCode); 1622 1623 /* offset to the stage 1 table, independent of the outputType */ 1624 offset=mbcsHeader.offsetFromUTable; 1625 1626 if(outputType==MBCS_OUTPUT_1) { 1627 /* SBCS: swap the fromU tables, all 16 bits wide */ 1628 count=(mbcsHeader.offsetFromUBytes-offset)+mbcsHeader.fromUBytesLength; 1629 ds->swapArray16(ds, inBytes+offset, (int32_t)count, 1630 outBytes+offset, pErrorCode); 1631 } else { 1632 /* otherwise: swap the stage tables separately */ 1633 1634 /* stage 1 table: uint16_t[0x440 or 0x40] */ 1635 if(inStaticData->unicodeMask&UCNV_HAS_SUPPLEMENTARY) { 1636 count=0x440*2; /* for all of Unicode */ 1637 } else { 1638 count=0x40*2; /* only BMP */ 1639 } 1640 ds->swapArray16(ds, inBytes+offset, (int32_t)count, 1641 outBytes+offset, pErrorCode); 1642 1643 /* stage 2 table: uint32_t[] */ 1644 offset+=count; 1645 count=mbcsHeader.offsetFromUBytes-offset; 1646 ds->swapArray32(ds, inBytes+offset, (int32_t)count, 1647 outBytes+offset, pErrorCode); 1648 1649 /* stage 3/result bytes: sometimes uint16_t[] or uint32_t[] */ 1650 offset=mbcsHeader.offsetFromUBytes; 1651 count= noFromU ? 0 : mbcsHeader.fromUBytesLength; 1652 switch(outputType) { 1653 case MBCS_OUTPUT_2: 1654 case MBCS_OUTPUT_3_EUC: 1655 case MBCS_OUTPUT_2_SISO: 1656 ds->swapArray16(ds, inBytes+offset, (int32_t)count, 1657 outBytes+offset, pErrorCode); 1658 break; 1659 case MBCS_OUTPUT_4: 1660 ds->swapArray32(ds, inBytes+offset, (int32_t)count, 1661 outBytes+offset, pErrorCode); 1662 break; 1663 default: 1664 /* just uint8_t[], nothing to swap */ 1665 break; 1666 } 1667 1668 if(mbcsIndexLength!=0) { 1669 offset+=count; 1670 count=mbcsIndexLength; 1671 ds->swapArray16(ds, inBytes+offset, (int32_t)count, 1672 outBytes+offset, pErrorCode); 1673 } 1674 } 1675 } 1676 1677 if(extOffset!=0) { 1678 /* swap the extension data */ 1679 inBytes+=extOffset; 1680 outBytes+=extOffset; 1681 1682 /* swap toUTable[] */ 1683 offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_TO_U_INDEX]); 1684 length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_TO_U_LENGTH]); 1685 ds->swapArray32(ds, inBytes+offset, length*4, outBytes+offset, pErrorCode); 1686 1687 /* swap toUUChars[] */ 1688 offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_TO_U_UCHARS_INDEX]); 1689 length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_TO_U_UCHARS_LENGTH]); 1690 ds->swapArray16(ds, inBytes+offset, length*2, outBytes+offset, pErrorCode); 1691 1692 /* swap fromUTableUChars[] */ 1693 offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_UCHARS_INDEX]); 1694 length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_LENGTH]); 1695 ds->swapArray16(ds, inBytes+offset, length*2, outBytes+offset, pErrorCode); 1696 1697 /* swap fromUTableValues[] */ 1698 offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_VALUES_INDEX]); 1699 /* same length as for fromUTableUChars[] */ 1700 ds->swapArray32(ds, inBytes+offset, length*4, outBytes+offset, pErrorCode); 1701 1702 /* no need to swap fromUBytes[] */ 1703 1704 /* swap fromUStage12[] */ 1705 offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_12_INDEX]); 1706 length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_12_LENGTH]); 1707 ds->swapArray16(ds, inBytes+offset, length*2, outBytes+offset, pErrorCode); 1708 1709 /* swap fromUStage3[] */ 1710 offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_3_INDEX]); 1711 length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_3_LENGTH]); 1712 ds->swapArray16(ds, inBytes+offset, length*2, outBytes+offset, pErrorCode); 1713 1714 /* swap fromUStage3b[] */ 1715 offset=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_3B_INDEX]); 1716 length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_FROM_U_STAGE_3B_LENGTH]); 1717 ds->swapArray32(ds, inBytes+offset, length*4, outBytes+offset, pErrorCode); 1718 1719 /* swap indexes[] */ 1720 length=udata_readInt32(ds, inExtIndexes[UCNV_EXT_INDEXES_LENGTH]); 1721 ds->swapArray32(ds, inBytes, length*4, outBytes, pErrorCode); 1722 } 1723 } 1724 } else { 1725 udata_printError(ds, "ucnv_swap(): unknown conversionType=%d!=UCNV_MBCS\n", 1726 inStaticData->conversionType); 1727 *pErrorCode=U_UNSUPPORTED_ERROR; 1728 return 0; 1729 } 1730 1731 return headerSize+(int32_t)staticDataSize+size; 1732 } 1733 1734 #endif /* #if !UCONFIG_NO_LEGACY_CONVERSION */ 1735 1736 #endif 1737
