source/common/uloc_keytype.cpp

   1 // © 2016 and later: Unicode, Inc. and others.
   2 // License & terms of use: http://www.unicode.org/copyright.html
   3 /*
   4 **********************************************************************
   5 *   Copyright (C) 2014-2016, International Business Machines
   6 *   Corporation and others.  All Rights Reserved.
   7 **********************************************************************
   8 */
   9 #include "unicode/utypes.h"
  10
  11 #include "cstring.h"
  12 #include "uassert.h"
  13 #include "ucln_cmn.h"
  14 #include "uhash.h"
  15 #include "umutex.h"
  16 #include "uresimp.h"
  17 #include "uvector.h"
  18 #include "udataswp.h" /* for InvChar functions */
  19
  20 static UHashtable* gLocExtKeyMap = NULL;
  21 static icu::UInitOnce gLocExtKeyMapInitOnce = U_INITONCE_INITIALIZER;
  22 static icu::UVector* gKeyTypeStringPool = NULL;
  23 static icu::UVector* gLocExtKeyDataEntries = NULL;
  24 static icu::UVector* gLocExtTypeEntries = NULL;
  25
  26 // bit flags for special types
  27 typedef enum {
  28     SPECIALTYPE_NONE = 0,
  29     SPECIALTYPE_CODEPOINTS = 1,
  30     SPECIALTYPE_REORDER_CODE = 2,
  31     SPECIALTYPE_RG_KEY_VALUE = 4
  32 } SpecialType;
  33
  34 typedef struct LocExtKeyData {
  35     const char*     legacyId;
  36     const char*     bcpId;
  37     UHashtable*     typeMap;
  38     uint32_t        specialTypes;
  39 } LocExtKeyData;
  40
  41 typedef struct LocExtType {
  42     const char*     legacyId;
  43     const char*     bcpId;
  44 } LocExtType;
  45
  46 U_CDECL_BEGIN
  47
  48 static UBool U_CALLCONV
  49 uloc_key_type_cleanup(void) {
  50     if (gLocExtKeyMap != NULL) {
  51         uhash_close(gLocExtKeyMap);
  52         gLocExtKeyMap = NULL;
  53     }
  54
  55     delete gLocExtKeyDataEntries;
  56     gLocExtKeyDataEntries = NULL;
  57
  58     delete gLocExtTypeEntries;
  59     gLocExtTypeEntries = NULL;
  60
  61     delete gKeyTypeStringPool;
  62     gKeyTypeStringPool = NULL;
  63
  64     gLocExtKeyMapInitOnce.reset();
  65     return TRUE;
  66 }
  67
  68 static void U_CALLCONV
  69 uloc_deleteKeyTypeStringPoolEntry(void* obj) {
  70     uprv_free(obj);
  71 }
  72
  73 static void U_CALLCONV
  74 uloc_deleteKeyDataEntry(void* obj) {
  75     LocExtKeyData* keyData = (LocExtKeyData*)obj;
  76     if (keyData->typeMap != NULL) {
  77         uhash_close(keyData->typeMap);
  78     }
  79     uprv_free(keyData);
  80 }
  81
  82 static void U_CALLCONV
  83 uloc_deleteTypeEntry(void* obj) {
  84     uprv_free(obj);
  85 }
  86
  87 U_CDECL_END
  88
  89
  90 static void U_CALLCONV
  91 initFromResourceBundle(UErrorCode& sts) {
  92     U_NAMESPACE_USE
  93     ucln_common_registerCleanup(UCLN_COMMON_LOCALE_KEY_TYPE, uloc_key_type_cleanup);
  94
  95     gLocExtKeyMap = uhash_open(uhash_hashIChars, uhash_compareIChars, NULL, &sts);
  96
  97     LocalUResourceBundlePointer keyTypeDataRes(ures_openDirect(NULL, "keyTypeData", &sts));
  98     LocalUResourceBundlePointer keyMapRes(ures_getByKey(keyTypeDataRes.getAlias(), "keyMap", NULL, &sts));
  99     LocalUResourceBundlePointer typeMapRes(ures_getByKey(keyTypeDataRes.getAlias(), "typeMap", NULL, &sts));
 100
 101     if (U_FAILURE(sts)) {
 102         return;
 103     }
 104
 105     UErrorCode tmpSts = U_ZERO_ERROR;
 106     LocalUResourceBundlePointer typeAliasRes(ures_getByKey(keyTypeDataRes.getAlias(), "typeAlias", NULL, &tmpSts));
 107     tmpSts = U_ZERO_ERROR;
 108     LocalUResourceBundlePointer bcpTypeAliasRes(ures_getByKey(keyTypeDataRes.getAlias(), "bcpTypeAlias", NULL, &tmpSts));
 109
 110     // initialize vectors storing dynamically allocated objects
 111     gKeyTypeStringPool = new UVector(uloc_deleteKeyTypeStringPoolEntry, NULL, sts);
 112     if (gKeyTypeStringPool == NULL) {
 113         if (U_SUCCESS(sts)) {
 114             sts = U_MEMORY_ALLOCATION_ERROR;
 115         }
 116     }
 117     if (U_FAILURE(sts)) {
 118         return;
 119     }
 120     gLocExtKeyDataEntries = new UVector(uloc_deleteKeyDataEntry, NULL, sts);
 121     if (gLocExtKeyDataEntries == NULL) {
 122         if (U_SUCCESS(sts)) {
 123             sts = U_MEMORY_ALLOCATION_ERROR;
 124         }
 125     }
 126     if (U_FAILURE(sts)) {
 127         return;
 128     }
 129     gLocExtTypeEntries = new UVector(uloc_deleteTypeEntry, NULL, sts);
 130     if (gLocExtTypeEntries == NULL) {
 131         if (U_SUCCESS(sts)) {
 132             sts = U_MEMORY_ALLOCATION_ERROR;
 133         }
 134     }
 135     if (U_FAILURE(sts)) {
 136         return;
 137     }
 138
 139     // iterate through keyMap resource
 140     LocalUResourceBundlePointer keyMapEntry;
 141
 142     while (ures_hasNext(keyMapRes.getAlias())) {
 143         keyMapEntry.adoptInstead(ures_getNextResource(keyMapRes.getAlias(), keyMapEntry.orphan(), &sts));
 144         if (U_FAILURE(sts)) {
 145             break;
 146         }
 147         const char* legacyKeyId = ures_getKey(keyMapEntry.getAlias());
 148         int32_t bcpKeyIdLen = 0;
 149         const UChar* uBcpKeyId = ures_getString(keyMapEntry.getAlias(), &bcpKeyIdLen, &sts);
 150         if (U_FAILURE(sts)) {
 151             break;
 152         }
 153
 154         // empty value indicates that BCP key is same with the legacy key.
 155         const char* bcpKeyId = legacyKeyId;
 156         if (bcpKeyIdLen > 0) {
 157             char* bcpKeyIdBuf = (char*)uprv_malloc(bcpKeyIdLen + 1);
 158             if (bcpKeyIdBuf == NULL) {
 159                 sts = U_MEMORY_ALLOCATION_ERROR;
 160                 break;
 161             }
 162             u_UCharsToChars(uBcpKeyId, bcpKeyIdBuf, bcpKeyIdLen);
 163             bcpKeyIdBuf[bcpKeyIdLen] = 0;
 164             gKeyTypeStringPool->addElement(bcpKeyIdBuf, sts);
 165             if (U_FAILURE(sts)) {
 166                 break;
 167             }
 168             bcpKeyId = bcpKeyIdBuf;
 169         }
 170
 171         UBool isTZ = uprv_strcmp(legacyKeyId, "timezone") == 0;
 172
 173         UHashtable* typeDataMap = uhash_open(uhash_hashIChars, uhash_compareIChars, NULL, &sts);
 174         if (U_FAILURE(sts)) {
 175             break;
 176         }
 177         uint32_t specialTypes = SPECIALTYPE_NONE;
 178
 179         LocalUResourceBundlePointer typeAliasResByKey;
 180         LocalUResourceBundlePointer bcpTypeAliasResByKey;
 181
 182         if (typeAliasRes.isValid()) {
 183             tmpSts = U_ZERO_ERROR;
 184             typeAliasResByKey.adoptInstead(ures_getByKey(typeAliasRes.getAlias(), legacyKeyId, NULL, &tmpSts));
 185             if (U_FAILURE(tmpSts)) {
 186                 typeAliasResByKey.orphan();
 187             }
 188         }
 189         if (bcpTypeAliasRes.isValid()) {
 190             tmpSts = U_ZERO_ERROR;
 191             bcpTypeAliasResByKey.adoptInstead(ures_getByKey(bcpTypeAliasRes.getAlias(), bcpKeyId, NULL, &tmpSts));
 192             if (U_FAILURE(tmpSts)) {
 193                 bcpTypeAliasResByKey.orphan();
 194             }
 195         }
 196
 197         // look up type map for the key, and walk through the mapping data
 198         tmpSts = U_ZERO_ERROR;
 199         LocalUResourceBundlePointer typeMapResByKey(ures_getByKey(typeMapRes.getAlias(), legacyKeyId, NULL, &tmpSts));
 200         if (U_FAILURE(tmpSts)) {
 201             // type map for each key must exist
 202             U_ASSERT(FALSE);
 203         } else {
 204             LocalUResourceBundlePointer typeMapEntry;
 205
 206             while (ures_hasNext(typeMapResByKey.getAlias())) {
 207                 typeMapEntry.adoptInstead(ures_getNextResource(typeMapResByKey.getAlias(), typeMapEntry.orphan(), &sts));
 208                 if (U_FAILURE(sts)) {
 209                     break;
 210                 }
 211                 const char* legacyTypeId = ures_getKey(typeMapEntry.getAlias());
 212
 213                 // special types
 214                 if (uprv_strcmp(legacyTypeId, "CODEPOINTS") == 0) {
 215                     specialTypes |= SPECIALTYPE_CODEPOINTS;
 216                     continue;
 217                 }
 218                 if (uprv_strcmp(legacyTypeId, "REORDER_CODE") == 0) {
 219                     specialTypes |= SPECIALTYPE_REORDER_CODE;
 220                     continue;
 221                 }
 222                 if (uprv_strcmp(legacyTypeId, "RG_KEY_VALUE") == 0) {
 223                     specialTypes |= SPECIALTYPE_RG_KEY_VALUE;
 224                     continue;
 225                 }
 226
 227                 if (isTZ) {
 228                     // a timezone key uses a colon instead of a slash in the resource.
 229                     // e.g. America:Los_Angeles
 230                     if (uprv_strchr(legacyTypeId, ':') != NULL) {
 231                         int32_t legacyTypeIdLen = uprv_strlen(legacyTypeId);
 232                         char* legacyTypeIdBuf = (char*)uprv_malloc(legacyTypeIdLen + 1);
 233                         if (legacyTypeIdBuf == NULL) {
 234                             sts = U_MEMORY_ALLOCATION_ERROR;
 235                             break;
 236                         }
 237                         const char* p = legacyTypeId;
 238                         char* q = legacyTypeIdBuf;
 239                         while (*p) {
 240                             if (*p == ':') {
 241                                 *q++ = '/';
 242                             } else {
 243                                 *q++ = *p;
 244                             }
 245                             p++;
 246                         }
 247                         *q = 0;
 248
 249                         gKeyTypeStringPool->addElement(legacyTypeIdBuf, sts);
 250                         if (U_FAILURE(sts)) {
 251                             break;
 252                         }
 253                         legacyTypeId = legacyTypeIdBuf;
 254                     }
 255                 }
 256
 257                 int32_t bcpTypeIdLen = 0;
 258                 const UChar* uBcpTypeId = ures_getString(typeMapEntry.getAlias(), &bcpTypeIdLen, &sts);
 259                 if (U_FAILURE(sts)) {
 260                     break;
 261                 }
 262
 263                 // empty value indicates that BCP type is same with the legacy type.
 264                 const char* bcpTypeId = legacyTypeId;
 265                 if (bcpTypeIdLen > 0) {
 266                     char* bcpTypeIdBuf = (char*)uprv_malloc(bcpTypeIdLen + 1);
 267                     if (bcpTypeIdBuf == NULL) {
 268                         sts = U_MEMORY_ALLOCATION_ERROR;
 269                         break;
 270                     }
 271                     u_UCharsToChars(uBcpTypeId, bcpTypeIdBuf, bcpTypeIdLen);
 272                     bcpTypeIdBuf[bcpTypeIdLen] = 0;
 273                     gKeyTypeStringPool->addElement(bcpTypeIdBuf, sts);
 274                     if (U_FAILURE(sts)) {
 275                         break;
 276                     }
 277                     bcpTypeId = bcpTypeIdBuf;
 278                 }
 279
 280                 // Note: legacy type value should never be
 281                 // equivalent to bcp type value of a different
 282                 // type under the same key. So we use a single
 283                 // map for lookup.
 284                 LocExtType* t = (LocExtType*)uprv_malloc(sizeof(LocExtType));
 285                 if (t == NULL) {
 286                     sts = U_MEMORY_ALLOCATION_ERROR;
 287                     break;
 288                 }
 289                 t->bcpId = bcpTypeId;
 290                 t->legacyId = legacyTypeId;
 291                 gLocExtTypeEntries->addElement((void*)t, sts);
 292                 if (U_FAILURE(sts)) {
 293                     break;
 294                 }
 295
 296                 uhash_put(typeDataMap, (void*)legacyTypeId, t, &sts);
 297                 if (bcpTypeId != legacyTypeId) {
 298                     // different type value
 299                     uhash_put(typeDataMap, (void*)bcpTypeId, t, &sts);
 300                 }
 301                 if (U_FAILURE(sts)) {
 302                     break;
 303                 }
 304
 305                 // also put aliases in the map
 306                 if (typeAliasResByKey.isValid()) {
 307                     LocalUResourceBundlePointer typeAliasDataEntry;
 308
 309                     ures_resetIterator(typeAliasResByKey.getAlias());
 310                     while (ures_hasNext(typeAliasResByKey.getAlias()) && U_SUCCESS(sts)) {
 311                         int32_t toLen;
 312                         typeAliasDataEntry.adoptInstead(ures_getNextResource(typeAliasResByKey.getAlias(), typeAliasDataEntry.orphan(), &sts));
 313                         const UChar* to = ures_getString(typeAliasDataEntry.getAlias(), &toLen, &sts);
 314                         if (U_FAILURE(sts)) {
 315                             break;
 316                         }
 317                         // check if this is an alias of canoncal legacy type
 318                         if (uprv_compareInvWithUChar(NULL, legacyTypeId, -1, to, toLen) == 0) {
 319                             const char* from = ures_getKey(typeAliasDataEntry.getAlias());
 320                             if (isTZ) {
 321                                 // replace colon with slash if necessary
 322                                 if (uprv_strchr(from, ':') != NULL) {
 323                                     int32_t fromLen = uprv_strlen(from);
 324                                     char* fromBuf = (char*)uprv_malloc(fromLen + 1);
 325                                     if (fromBuf == NULL) {
 326                                         sts = U_MEMORY_ALLOCATION_ERROR;
 327                                         break;
 328                                     }
 329                                     const char* p = from;
 330                                     char* q = fromBuf;
 331                                     while (*p) {
 332                                         if (*p == ':') {
 333                                             *q++ = '/';
 334                                         } else {
 335                                             *q++ = *p;
 336                                         }
 337                                         p++;
 338                                     }
 339                                     *q = 0;
 340
 341                                     gKeyTypeStringPool->addElement(fromBuf, sts);
 342                                     if (U_FAILURE(sts)) {
 343                                         break;
 344                                     }
 345                                     from = fromBuf;
 346                                 }
 347                             }
 348                             uhash_put(typeDataMap, (void*)from, t, &sts);
 349                         }
 350                     }
 351                     if (U_FAILURE(sts)) {
 352                         break;
 353                     }
 354                 }
 355
 356                 if (bcpTypeAliasResByKey.isValid()) {
 357                     LocalUResourceBundlePointer bcpTypeAliasDataEntry;
 358
 359                     ures_resetIterator(bcpTypeAliasResByKey.getAlias());
 360                     while (ures_hasNext(bcpTypeAliasResByKey.getAlias()) && U_SUCCESS(sts)) {
 361                         int32_t toLen;
 362                         bcpTypeAliasDataEntry.adoptInstead(ures_getNextResource(bcpTypeAliasResByKey.getAlias(), bcpTypeAliasDataEntry.orphan(), &sts));
 363                         const UChar* to = ures_getString(bcpTypeAliasDataEntry.getAlias(), &toLen, &sts);
 364                         if (U_FAILURE(sts)) {
 365                             break;
 366                         }
 367                         // check if this is an alias of bcp type
 368                         if (uprv_compareInvWithUChar(NULL, bcpTypeId, -1, to, toLen) == 0) {
 369                             const char* from = ures_getKey(bcpTypeAliasDataEntry.getAlias());
 370                             uhash_put(typeDataMap, (void*)from, t, &sts);
 371                         }
 372                     }
 373                     if (U_FAILURE(sts)) {
 374                         break;
 375                     }
 376                 }
 377             }
 378         }
 379         if (U_FAILURE(sts)) {
 380             break;
 381         }
 382
 383         LocExtKeyData* keyData = (LocExtKeyData*)uprv_malloc(sizeof(LocExtKeyData));
 384         if (keyData == NULL) {
 385             sts = U_MEMORY_ALLOCATION_ERROR;
 386             break;
 387         }
 388         keyData->bcpId = bcpKeyId;
 389         keyData->legacyId = legacyKeyId;
 390         keyData->specialTypes = specialTypes;
 391         keyData->typeMap = typeDataMap;
 392
 393         gLocExtKeyDataEntries->addElement((void*)keyData, sts);
 394         if (U_FAILURE(sts)) {
 395             break;
 396         }
 397
 398         uhash_put(gLocExtKeyMap, (void*)legacyKeyId, keyData, &sts);
 399         if (legacyKeyId != bcpKeyId) {
 400             // different key value
 401             uhash_put(gLocExtKeyMap, (void*)bcpKeyId, keyData, &sts);
 402         }
 403         if (U_FAILURE(sts)) {
 404             break;
 405         }
 406     }
 407 }
 408
 409 static UBool
 410 init() {
 411     UErrorCode sts = U_ZERO_ERROR;
 412     umtx_initOnce(gLocExtKeyMapInitOnce, &initFromResourceBundle, sts);
 413     if (U_FAILURE(sts)) {
 414         return FALSE;
 415     }
 416     return TRUE;
 417 }
 418
 419 static UBool
 420 isSpecialTypeCodepoints(const char* val) {
 421     int32_t subtagLen = 0;
 422     const char* p = val;
 423     while (*p) {
 424         if (*p == '-') {
 425             if (subtagLen < 4 || subtagLen > 6) {
 426                 return FALSE;
 427             }
 428             subtagLen = 0;
 429         } else if ((*p >= '0' && *p <= '9') ||
 430                     (*p >= 'A' && *p <= 'F') || // A-F/a-f are contiguous
 431                     (*p >= 'a' && *p <= 'f')) { // also in EBCDIC
 432             subtagLen++;
 433         } else {
 434             return FALSE;
 435         }
 436         p++;
 437     }
 438     return (subtagLen >= 4 && subtagLen <= 6);
 439 }
 440
 441 static UBool
 442 isSpecialTypeReorderCode(const char* val) {
 443     int32_t subtagLen = 0;
 444     const char* p = val;
 445     while (*p) {
 446         if (*p == '-') {
 447             if (subtagLen < 3 || subtagLen > 8) {
 448                 return FALSE;
 449             }
 450             subtagLen = 0;
 451         } else if (uprv_isASCIILetter(*p)) {
 452             subtagLen++;
 453         } else {
 454             return FALSE;
 455         }
 456         p++;
 457     }
 458     return (subtagLen >=3 && subtagLen <=8);
 459 }
 460
 461 static UBool
 462 isSpecialTypeRgKeyValue(const char* val) {
 463     int32_t subtagLen = 0;
 464     const char* p = val;
 465     while (*p) {
 466         if ( (subtagLen < 2 && uprv_isASCIILetter(*p)) ||
 467                     (subtagLen >= 2 && (*p == 'Z' || *p == 'z')) ) {
 468             subtagLen++;
 469         } else {
 470             return FALSE;
 471         }
 472         p++;
 473     }
 474     return (subtagLen == 6);
 475     return TRUE;
 476 }
 477
 478 U_CFUNC const char*
 479 ulocimp_toBcpKey(const char* key) {
 480     if (!init()) {
 481         return NULL;
 482     }
 483
 484     LocExtKeyData* keyData = (LocExtKeyData*)uhash_get(gLocExtKeyMap, key);
 485     if (keyData != NULL) {
 486         return keyData->bcpId;
 487     }
 488     return NULL;
 489 }
 490
 491 U_CFUNC const char*
 492 ulocimp_toLegacyKey(const char* key) {
 493     if (!init()) {
 494         return NULL;
 495     }
 496
 497     LocExtKeyData* keyData = (LocExtKeyData*)uhash_get(gLocExtKeyMap, key);
 498     if (keyData != NULL) {
 499         return keyData->legacyId;
 500     }
 501     return NULL;
 502 }
 503
 504 U_CFUNC const char*
 505 ulocimp_toBcpType(const char* key, const char* type, UBool* isKnownKey, UBool* isSpecialType) {
 506     if (isKnownKey != NULL) {
 507         *isKnownKey = FALSE;
 508     }
 509     if (isSpecialType != NULL) {
 510         *isSpecialType = FALSE;
 511     }
 512
 513     if (!init()) {
 514         return NULL;
 515     }
 516
 517     LocExtKeyData* keyData = (LocExtKeyData*)uhash_get(gLocExtKeyMap, key);
 518     if (keyData != NULL) {
 519         if (isKnownKey != NULL) {
 520             *isKnownKey = TRUE;
 521         }
 522         LocExtType* t = (LocExtType*)uhash_get(keyData->typeMap, type);
 523         if (t != NULL) {
 524             return t->bcpId;
 525         }
 526         if (keyData->specialTypes != SPECIALTYPE_NONE) {
 527             UBool matched = FALSE;
 528             if (keyData->specialTypes & SPECIALTYPE_CODEPOINTS) {
 529                 matched = isSpecialTypeCodepoints(type);
 530             }
 531             if (!matched && keyData->specialTypes & SPECIALTYPE_REORDER_CODE) {
 532                 matched = isSpecialTypeReorderCode(type);
 533             }
 534             if (!matched && keyData->specialTypes & SPECIALTYPE_RG_KEY_VALUE) {
 535                 matched = isSpecialTypeRgKeyValue(type);
 536             }
 537             if (matched) {
 538                 if (isSpecialType != NULL) {
 539                     *isSpecialType = TRUE;
 540                 }
 541                 return type;
 542             }
 543         }
 544     }
 545     return NULL;
 546 }
 547
 548
 549 U_CFUNC const char*
 550 ulocimp_toLegacyType(const char* key, const char* type, UBool* isKnownKey, UBool* isSpecialType) {
 551     if (isKnownKey != NULL) {
 552         *isKnownKey = FALSE;
 553     }
 554     if (isSpecialType != NULL) {
 555         *isSpecialType = FALSE;
 556     }
 557
 558     if (!init()) {
 559         return NULL;
 560     }
 561
 562     LocExtKeyData* keyData = (LocExtKeyData*)uhash_get(gLocExtKeyMap, key);
 563     if (keyData != NULL) {
 564         if (isKnownKey != NULL) {
 565             *isKnownKey = TRUE;
 566         }
 567         LocExtType* t = (LocExtType*)uhash_get(keyData->typeMap, type);
 568         if (t != NULL) {
 569             return t->legacyId;
 570         }
 571         if (keyData->specialTypes != SPECIALTYPE_NONE) {
 572             UBool matched = FALSE;
 573             if (keyData->specialTypes & SPECIALTYPE_CODEPOINTS) {
 574                 matched = isSpecialTypeCodepoints(type);
 575             }
 576             if (!matched && keyData->specialTypes & SPECIALTYPE_REORDER_CODE) {
 577                 matched = isSpecialTypeReorderCode(type);
 578             }
 579             if (!matched && keyData->specialTypes & SPECIALTYPE_RG_KEY_VALUE) {
 580                 matched = isSpecialTypeRgKeyValue(type);
 581             }
 582             if (matched) {
 583                 if (isSpecialType != NULL) {
 584                     *isSpecialType = TRUE;
 585                 }
 586                 return type;
 587             }
 588         }
 589     }
 590     return NULL;
 591 }
 592