1 /******************************************************************** 2 * COPYRIGHT: 3 * Copyright (c) 1997-2014, International Business Machines Corporation and 4 * others. All Rights Reserved. 5 ********************************************************************/ 6 7 #include "cintltst.h" 8 #include "unicode/ures.h" 9 #include "unicode/ucurr.h" 10 #include "unicode/ustring.h" 11 #include "unicode/uset.h" 12 #include "unicode/udat.h" 13 #include "unicode/uscript.h" 14 #include "unicode/ulocdata.h" 15 #include "cstring.h" 16 #include "locmap.h" 17 #include "uresimp.h" 18 19 /* 20 returns a new UnicodeSet that is a flattened form of the original 21 UnicodeSet. 22 */ 23 static USet* 24 createFlattenSet(USet *origSet, UErrorCode *status) { 25 26 27 USet *newSet = NULL; 28 int32_t origItemCount = 0; 29 int32_t idx, graphmeSize; 30 UChar32 start, end; 31 UChar graphme[64]; 32 if (U_FAILURE(*status)) { 33 log_err("createFlattenSet called with %s\n", u_errorName(*status)); 34 return NULL; 35 } 36 newSet = uset_open(1, 0); 37 origItemCount = uset_getItemCount(origSet); 38 for (idx = 0; idx < origItemCount; idx++) { 39 graphmeSize = uset_getItem(origSet, idx, 40 &start, &end, 41 graphme, (int32_t)(sizeof(graphme)/sizeof(graphme[0])), 42 status); 43 if (U_FAILURE(*status)) { 44 log_err("ERROR: uset_getItem returned %s\n", u_errorName(*status)); 45 *status = U_ZERO_ERROR; 46 } 47 if (graphmeSize) { 48 uset_addAllCodePoints(newSet, graphme, graphmeSize); 49 } 50 else { 51 uset_addRange(newSet, start, end); 52 } 53 } 54 uset_closeOver(newSet,USET_CASE_INSENSITIVE); 55 return newSet; 56 } 57 58 static UBool 59 isCurrencyPreEuro(const char* currencyKey){ 60 if( strcmp(currencyKey, "PTE") == 0 || 61 strcmp(currencyKey, "ESP") == 0 || 62 strcmp(currencyKey, "LUF") == 0 || 63 strcmp(currencyKey, "GRD") == 0 || 64 strcmp(currencyKey, "BEF") == 0 || 65 strcmp(currencyKey, "ITL") == 0 || 66 strcmp(currencyKey, "EEK") == 0){ 67 return TRUE; 68 } 69 return FALSE; 70 } 71 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION 72 static void 73 TestKeyInRootRecursive(UResourceBundle *root, const char *rootName, 74 UResourceBundle *currentBundle, const char *locale) { 75 UErrorCode errorCode = U_ZERO_ERROR; 76 UResourceBundle *subRootBundle = NULL, *subBundle = NULL, *arr = NULL; 77 78 ures_resetIterator(root); 79 ures_resetIterator(currentBundle); 80 while (ures_hasNext(currentBundle)) { 81 const char *subBundleKey = NULL; 82 const char *currentBundleKey = NULL; 83 84 errorCode = U_ZERO_ERROR; 85 currentBundleKey = ures_getKey(currentBundle); 86 (void)currentBundleKey; /* Suppress set but not used warning. */ 87 subBundle = ures_getNextResource(currentBundle, NULL, &errorCode); 88 if (U_FAILURE(errorCode)) { 89 log_err("Can't open a resource for lnocale %s. Error: %s\n", locale, u_errorName(errorCode)); 90 continue; 91 } 92 subBundleKey = ures_getKey(subBundle); 93 94 95 subRootBundle = ures_getByKey(root, subBundleKey, NULL, &errorCode); 96 if (U_FAILURE(errorCode)) { 97 log_err("Can't open a resource with key \"%s\" in \"%s\" from %s for locale \"%s\"\n", 98 subBundleKey, 99 ures_getKey(currentBundle), 100 rootName, 101 locale); 102 ures_close(subBundle); 103 continue; 104 } 105 if (ures_getType(subRootBundle) != ures_getType(subBundle)) { 106 log_err("key \"%s\" in \"%s\" has a different type from root for locale \"%s\"\n" 107 "\troot=%d, locale=%d\n", 108 subBundleKey, 109 ures_getKey(currentBundle), 110 locale, 111 ures_getType(subRootBundle), 112 ures_getType(subBundle)); 113 ures_close(subBundle); 114 continue; 115 } 116 else if (ures_getType(subBundle) == URES_INT_VECTOR) { 117 int32_t minSize; 118 int32_t subBundleSize; 119 int32_t idx; 120 UBool sameArray = TRUE; 121 const int32_t *subRootBundleArr = ures_getIntVector(subRootBundle, &minSize, &errorCode); 122 const int32_t *subBundleArr = ures_getIntVector(subBundle, &subBundleSize, &errorCode); 123 124 if (minSize > subBundleSize) { 125 minSize = subBundleSize; 126 log_err("Arrays are different size with key \"%s\" in \"%s\" from root for locale \"%s\"\n", 127 subBundleKey, 128 ures_getKey(currentBundle), 129 locale); 130 } 131 132 for (idx = 0; idx < minSize && sameArray; idx++) { 133 if (subRootBundleArr[idx] != subBundleArr[idx]) { 134 sameArray = FALSE; 135 } 136 if (strcmp(subBundleKey, "DateTimeElements") == 0 137 && (subBundleArr[idx] < 1 || 7 < subBundleArr[idx])) 138 { 139 log_err("Value out of range with key \"%s\" at index %d in \"%s\" for locale \"%s\"\n", 140 subBundleKey, 141 idx, 142 ures_getKey(currentBundle), 143 locale); 144 } 145 } 146 /* Special exception es_US and DateTimeElements */ 147 if (sameArray 148 && !(strcmp(locale, "es_US") == 0 && strcmp(subBundleKey, "DateTimeElements") == 0)) 149 { 150 log_err("Integer vectors are the same with key \"%s\" in \"%s\" from root for locale \"%s\"\n", 151 subBundleKey, 152 ures_getKey(currentBundle), 153 locale); 154 } 155 } 156 else if (ures_getType(subBundle) == URES_ARRAY) { 157 UResourceBundle *subSubBundle = ures_getByIndex(subBundle, 0, NULL, &errorCode); 158 UResourceBundle *subSubRootBundle = ures_getByIndex(subRootBundle, 0, NULL, &errorCode); 159 160 if (U_SUCCESS(errorCode) 161 && (ures_getType(subSubBundle) == URES_ARRAY || ures_getType(subSubRootBundle) == URES_ARRAY)) 162 { 163 /* Here is one of the recursive parts */ 164 TestKeyInRootRecursive(subRootBundle, rootName, subBundle, locale); 165 } 166 else { 167 int32_t minSize = ures_getSize(subRootBundle); 168 int32_t idx; 169 UBool sameArray = TRUE; 170 171 if (minSize > ures_getSize(subBundle)) { 172 minSize = ures_getSize(subBundle); 173 } 174 175 if ((subBundleKey == NULL 176 || (subBundleKey != NULL && strcmp(subBundleKey, "LocaleScript") != 0 && !isCurrencyPreEuro(subBundleKey))) 177 && ures_getSize(subRootBundle) != ures_getSize(subBundle)) 178 { 179 log_err("Different size array with key \"%s\" in \"%s\" from root for locale \"%s\"\n" 180 "\troot array size=%d, locale array size=%d\n", 181 subBundleKey, 182 ures_getKey(currentBundle), 183 locale, 184 ures_getSize(subRootBundle), 185 ures_getSize(subBundle)); 186 } 187 /* 188 if(isCurrencyPreEuro(subBundleKey) && ures_getSize(subBundle)!=3){ 189 log_err("Different size array with key \"%s\" in \"%s\" for locale \"%s\" the expected size is 3 got size=%d\n", 190 subBundleKey, 191 ures_getKey(currentBundle), 192 locale, 193 ures_getSize(subBundle)); 194 } 195 */ 196 for (idx = 0; idx < minSize; idx++) { 197 int32_t rootStrLen, localeStrLen; 198 const UChar *rootStr = ures_getStringByIndex(subRootBundle,idx,&rootStrLen,&errorCode); 199 const UChar *localeStr = ures_getStringByIndex(subBundle,idx,&localeStrLen,&errorCode); 200 if (rootStr && localeStr && U_SUCCESS(errorCode)) { 201 if (u_strcmp(rootStr, localeStr) != 0) { 202 sameArray = FALSE; 203 } 204 } 205 else { 206 if ( rootStrLen > 1 && rootStr[0] == 0x41 && rootStr[1] >= 0x30 && rootStr[1] <= 0x39 ) { 207 /* A2 or A4 in the root string indicates that the resource can optionally be an array instead of a */ 208 /* string. Attempt to read it as an array. */ 209 errorCode = U_ZERO_ERROR; 210 arr = ures_getByIndex(subBundle,idx,NULL,&errorCode); 211 if (U_FAILURE(errorCode)) { 212 log_err("Got a NULL string with key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n", 213 subBundleKey, 214 ures_getKey(currentBundle), 215 idx, 216 locale); 217 continue; 218 } 219 if (ures_getType(arr) != URES_ARRAY || ures_getSize(arr) != (int32_t)rootStr[1] - 0x30) { 220 log_err("Got something other than a string or array of size %d for key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n", 221 rootStr[1] - 0x30, 222 subBundleKey, 223 ures_getKey(currentBundle), 224 idx, 225 locale); 226 ures_close(arr); 227 continue; 228 } 229 localeStr = ures_getStringByIndex(arr,0,&localeStrLen,&errorCode); 230 ures_close(arr); 231 if (U_FAILURE(errorCode)) { 232 log_err("Got something other than a string or array for key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n", 233 subBundleKey, 234 ures_getKey(currentBundle), 235 idx, 236 locale); 237 continue; 238 } 239 } else { 240 log_err("Got a NULL string with key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n", 241 subBundleKey, 242 ures_getKey(currentBundle), 243 idx, 244 locale); 245 continue; 246 } 247 } 248 if (localeStr[0] == (UChar)0x20) { 249 log_err("key \"%s\" at index %d in \"%s\" starts with a space in locale \"%s\"\n", 250 subBundleKey, 251 idx, 252 ures_getKey(currentBundle), 253 locale); 254 } 255 else if ((localeStr[localeStrLen - 1] == (UChar)0x20) && (strcmp(subBundleKey,"separator") != 0)) { 256 log_err("key \"%s\" at index %d in \"%s\" ends with a space in locale \"%s\"\n", 257 subBundleKey, 258 idx, 259 ures_getKey(currentBundle), 260 locale); 261 } 262 else if (subBundleKey != NULL 263 && strcmp(subBundleKey, "DateTimePatterns") == 0) 264 { 265 int32_t quoted = 0; 266 const UChar *localeStrItr = localeStr; 267 while (*localeStrItr) { 268 if (*localeStrItr == (UChar)0x27 /* ' */) { 269 quoted++; 270 } 271 else if ((quoted % 2) == 0) { 272 /* Search for unquoted characters */ 273 if (4 <= idx && idx <= 7 274 && (*localeStrItr == (UChar)0x6B /* k */ 275 || *localeStrItr == (UChar)0x48 /* H */ 276 || *localeStrItr == (UChar)0x6D /* m */ 277 || *localeStrItr == (UChar)0x73 /* s */ 278 || *localeStrItr == (UChar)0x53 /* S */ 279 || *localeStrItr == (UChar)0x61 /* a */ 280 || *localeStrItr == (UChar)0x68 /* h */ 281 || *localeStrItr == (UChar)0x7A /* z */)) 282 { 283 log_err("key \"%s\" at index %d has time pattern chars in date for locale \"%s\"\n", 284 subBundleKey, 285 idx, 286 locale); 287 } 288 else if (0 <= idx && idx <= 3 289 && (*localeStrItr == (UChar)0x47 /* G */ 290 || *localeStrItr == (UChar)0x79 /* y */ 291 || *localeStrItr == (UChar)0x4D /* M */ 292 || *localeStrItr == (UChar)0x64 /* d */ 293 || *localeStrItr == (UChar)0x45 /* E */ 294 || *localeStrItr == (UChar)0x44 /* D */ 295 || *localeStrItr == (UChar)0x46 /* F */ 296 || *localeStrItr == (UChar)0x77 /* w */ 297 || *localeStrItr == (UChar)0x57 /* W */)) 298 { 299 log_err("key \"%s\" at index %d has date pattern chars in time for locale \"%s\"\n", 300 subBundleKey, 301 idx, 302 locale); 303 } 304 } 305 localeStrItr++; 306 } 307 } 308 else if (idx == 4 && subBundleKey != NULL 309 && strcmp(subBundleKey, "NumberElements") == 0 310 && u_charDigitValue(localeStr[0]) != 0) 311 { 312 log_err("key \"%s\" at index %d has a non-zero based number for locale \"%s\"\n", 313 subBundleKey, 314 idx, 315 locale); 316 } 317 } 318 (void)sameArray; /* Suppress set but not used warning. */ 319 /* if (sameArray && strcmp(rootName, "root") == 0) { 320 log_err("Arrays are the same with key \"%s\" in \"%s\" from root for locale \"%s\"\n", 321 subBundleKey, 322 ures_getKey(currentBundle), 323 locale); 324 }*/ 325 } 326 ures_close(subSubBundle); 327 ures_close(subSubRootBundle); 328 } 329 else if (ures_getType(subBundle) == URES_STRING) { 330 int32_t len = 0; 331 const UChar *string = ures_getString(subBundle, &len, &errorCode); 332 if (U_FAILURE(errorCode) || string == NULL) { 333 log_err("Can't open a string with key \"%s\" in \"%s\" for locale \"%s\"\n", 334 subBundleKey, 335 ures_getKey(currentBundle), 336 locale); 337 } else if (string[0] == (UChar)0x20) { 338 log_err("key \"%s\" in \"%s\" starts with a space in locale \"%s\"\n", 339 subBundleKey, 340 ures_getKey(currentBundle), 341 locale); 342 /* localeDisplayPattern/separator can end with a space */ 343 } else if (string[len - 1] == (UChar)0x20 && (strcmp(subBundleKey,"separator"))) { 344 log_err("key \"%s\" in \"%s\" ends with a space in locale \"%s\"\n", 345 subBundleKey, 346 ures_getKey(currentBundle), 347 locale); 348 } else if (strcmp(subBundleKey, "localPatternChars") == 0) { 349 /* Note: We no longer import localPatternChars data starting 350 * ICU 3.8. So it never comes into this else if block. (ticket#5597) 351 */ 352 353 /* Check well-formedness of localPatternChars. First, the 354 * length must match the number of fields defined by 355 * DateFormat. Second, each character in the string must 356 * be in the set [A-Za-z]. Finally, each character must be 357 * unique. 358 */ 359 int32_t i,j; 360 #if !UCONFIG_NO_FORMATTING 361 if (len != UDAT_FIELD_COUNT) { 362 log_err("key \"%s\" has the wrong number of characters in locale \"%s\"\n", 363 subBundleKey, 364 locale); 365 } 366 #endif 367 /* Check char validity. */ 368 for (i=0; i<len; ++i) { 369 if (!((string[i] >= 65/*'A'*/ && string[i] <= 90/*'Z'*/) || 370 (string[i] >= 97/*'a'*/ && string[i] <= 122/*'z'*/))) { 371 log_err("key \"%s\" has illegal character '%c' in locale \"%s\"\n", 372 subBundleKey, 373 (char) string[i], 374 locale); 375 } 376 /* Do O(n^2) check for duplicate chars. */ 377 for (j=0; j<i; ++j) { 378 if (string[j] == string[i]) { 379 log_err("key \"%s\" has duplicate character '%c' in locale \"%s\"\n", 380 subBundleKey, 381 (char) string[i], 382 locale); 383 } 384 } 385 } 386 } 387 /* No fallback was done. Check for duplicate data */ 388 /* The ures_* API does not do fallback of sub-resource bundles, 389 So we can't do this now. */ 390 #if 0 391 else if (strcmp(locale, "root") != 0 && errorCode == U_ZERO_ERROR) { 392 393 const UChar *rootString = ures_getString(subRootBundle, &len, &errorCode); 394 if (U_FAILURE(errorCode) || rootString == NULL) { 395 log_err("Can't open a string with key \"%s\" in \"%s\" in root\n", 396 ures_getKey(subRootBundle), 397 ures_getKey(currentBundle)); 398 continue; 399 } else if (u_strcmp(string, rootString) == 0) { 400 if (strcmp(locale, "de_CH") != 0 && strcmp(subBundleKey, "Countries") != 0 && 401 strcmp(subBundleKey, "Version") != 0) { 402 log_err("Found duplicate data with key \"%s\" in \"%s\" in locale \"%s\"\n", 403 ures_getKey(subRootBundle), 404 ures_getKey(currentBundle), 405 locale); 406 } 407 else { 408 /* Ignore for now. */ 409 /* Can be fixed if fallback through de locale was done. */ 410 log_verbose("Skipping key %s in %s\n", subBundleKey, locale); 411 } 412 } 413 } 414 #endif 415 } 416 else if (ures_getType(subBundle) == URES_TABLE) { 417 if (strcmp(subBundleKey, "availableFormats")!=0) { 418 /* Here is one of the recursive parts */ 419 TestKeyInRootRecursive(subRootBundle, rootName, subBundle, locale); 420 } 421 else { 422 log_verbose("Skipping key %s in %s\n", subBundleKey, locale); 423 } 424 } 425 else if (ures_getType(subBundle) == URES_BINARY || ures_getType(subBundle) == URES_INT) { 426 /* Can't do anything to check it */ 427 /* We'll assume it's all correct */ 428 if (strcmp(subBundleKey, "MeasurementSystem") != 0) { 429 log_verbose("Skipping key \"%s\" in \"%s\" for locale \"%s\"\n", 430 subBundleKey, 431 ures_getKey(currentBundle), 432 locale); 433 } 434 /* Testing for MeasurementSystem is done in VerifyTranslation */ 435 } 436 else { 437 log_err("Type %d for key \"%s\" in \"%s\" is unknown for locale \"%s\"\n", 438 ures_getType(subBundle), 439 subBundleKey, 440 ures_getKey(currentBundle), 441 locale); 442 } 443 ures_close(subRootBundle); 444 ures_close(subBundle); 445 } 446 } 447 #endif 448 449 static void 450 testLCID(UResourceBundle *currentBundle, 451 const char *localeName) 452 { 453 UErrorCode status = U_ZERO_ERROR; 454 uint32_t expectedLCID; 455 char lcidStringC[64] = {0}; 456 int32_t len; 457 458 expectedLCID = uloc_getLCID(localeName); 459 if (expectedLCID == 0) { 460 log_verbose("INFO: %-5s does not have any LCID mapping\n", 461 localeName); 462 return; 463 } 464 465 status = U_ZERO_ERROR; 466 len = uprv_convertToPosix(expectedLCID, lcidStringC, sizeof(lcidStringC)/sizeof(lcidStringC[0]) - 1, &status); 467 if (U_FAILURE(status)) { 468 log_err("ERROR: %.4x does not have a POSIX mapping due to %s\n", 469 expectedLCID, u_errorName(status)); 470 } 471 lcidStringC[len] = 0; 472 473 if(strcmp(localeName, lcidStringC) != 0) { 474 char langName[1024]; 475 char langLCID[1024]; 476 uloc_getLanguage(localeName, langName, sizeof(langName), &status); 477 uloc_getLanguage(lcidStringC, langLCID, sizeof(langLCID), &status); 478 479 if (strcmp(langName, langLCID) == 0) { 480 log_verbose("WARNING: %-5s resolves to %s (0x%.4x)\n", 481 localeName, lcidStringC, expectedLCID); 482 } 483 else { 484 log_err("ERROR: %-5s has 0x%.4x and the number resolves wrongfully to %s\n", 485 localeName, expectedLCID, lcidStringC); 486 } 487 } 488 } 489 490 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION 491 static void 492 TestLocaleStructure(void) { 493 // This test checks the locale structure against a key file located 494 // at source/test/testdata/structLocale.txt. When adding new data to 495 // a loale file such as en.txt, the structLocale.txt file must be changed 496 // too to include the the template of the new data. Otherwise this test 497 // will fail! 498 499 UResourceBundle *root, *currentLocale; 500 int32_t locCount = uloc_countAvailable(); 501 int32_t locIndex; 502 UErrorCode errorCode = U_ZERO_ERROR; 503 const char *currLoc, *resolvedLoc; 504 505 /* TODO: Compare against parent's data too. This code can't handle fallbacks that some tools do already. */ 506 /* char locName[ULOC_FULLNAME_CAPACITY]; 507 char *locNamePtr; 508 509 for (locIndex = 0; locIndex < locCount; locIndex++) { 510 errorCode=U_ZERO_ERROR; 511 strcpy(locName, uloc_getAvailable(locIndex)); 512 locNamePtr = strrchr(locName, '_'); 513 if (locNamePtr) { 514 *locNamePtr = 0; 515 } 516 else { 517 strcpy(locName, "root"); 518 } 519 520 root = ures_openDirect(NULL, locName, &errorCode); 521 if(U_FAILURE(errorCode)) { 522 log_err("Can't open %s\n", locName); 523 continue; 524 } 525 */ 526 if (locCount <= 1) { 527 log_data_err("At least root needs to be installed\n"); 528 } 529 530 root = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode); 531 if(U_FAILURE(errorCode)) { 532 log_data_err("Can't open structLocale\n"); 533 return; 534 } 535 for (locIndex = 0; locIndex < locCount; locIndex++) { 536 errorCode=U_ZERO_ERROR; 537 currLoc = uloc_getAvailable(locIndex); 538 currentLocale = ures_open(NULL, currLoc, &errorCode); 539 if(errorCode != U_ZERO_ERROR) { 540 if(U_SUCCESS(errorCode)) { 541 /* It's installed, but there is no data. 542 It's installed for the g18n white paper [grhoten] */ 543 log_err("ERROR: Locale %-5s not installed, and it should be, err %s\n", 544 uloc_getAvailable(locIndex), u_errorName(errorCode)); 545 } else { 546 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%", 547 u_errorName(errorCode), 548 uloc_getAvailable(locIndex)); 549 } 550 ures_close(currentLocale); 551 continue; 552 } 553 ures_getStringByKey(currentLocale, "Version", NULL, &errorCode); 554 if(errorCode != U_ZERO_ERROR) { 555 log_err("No version information is available for locale %s, and it should be!\n", 556 currLoc); 557 } 558 else if (ures_getStringByKey(currentLocale, "Version", NULL, &errorCode)[0] == (UChar)(0x78)) { 559 log_verbose("WARNING: The locale %s is experimental! It shouldn't be listed as an installed locale.\n", 560 currLoc); 561 } 562 resolvedLoc = ures_getLocaleByType(currentLocale, ULOC_ACTUAL_LOCALE, &errorCode); 563 if (strcmp(resolvedLoc, currLoc) != 0) { 564 /* All locales have at least a Version resource. 565 If it's absolutely empty, then the previous test will fail too.*/ 566 /* Google Patch: tl_PH and fil_PH are aliases of each other */ 567 if (!(strcmp(currLoc, "tl_PH")==0 && strcmp(resolvedLoc, "fil_PH")==0) && 568 !(strcmp(currLoc, "tl")==0 && strcmp(resolvedLoc, "fil")==0)) 569 log_err("Locale resolves to different locale. Is %s an alias of %s?\n", 570 currLoc, resolvedLoc); 571 } 572 TestKeyInRootRecursive(root, "root", currentLocale, currLoc); 573 574 testLCID(currentLocale, currLoc); 575 576 ures_close(currentLocale); 577 } 578 579 ures_close(root); 580 } 581 #endif 582 583 static void 584 compareArrays(const char *keyName, 585 UResourceBundle *fromArray, const char *fromLocale, 586 UResourceBundle *toArray, const char *toLocale, 587 int32_t start, int32_t end) 588 { 589 int32_t fromSize = ures_getSize(fromArray); 590 int32_t toSize = ures_getSize(fromArray); 591 int32_t idx; 592 UErrorCode errorCode = U_ZERO_ERROR; 593 594 if (fromSize > toSize) { 595 fromSize = toSize; 596 log_err("Arrays are different size from \"%s\" to \"%s\"\n", 597 fromLocale, 598 toLocale); 599 } 600 601 for (idx = start; idx <= end; idx++) { 602 const UChar *fromBundleStr = ures_getStringByIndex(fromArray, idx, NULL, &errorCode); 603 const UChar *toBundleStr = ures_getStringByIndex(toArray, idx, NULL, &errorCode); 604 if (fromBundleStr && toBundleStr && u_strcmp(fromBundleStr, toBundleStr) != 0) 605 { 606 log_err("Difference for %s at index %d from %s= \"%s\" to %s= \"%s\"\n", 607 keyName, 608 idx, 609 fromLocale, 610 austrdup(fromBundleStr), 611 toLocale, 612 austrdup(toBundleStr)); 613 } 614 } 615 } 616 617 static void 618 compareConsistentCountryInfo(const char *fromLocale, const char *toLocale) { 619 UErrorCode errorCode = U_ZERO_ERROR; 620 UResourceBundle *fromArray, *toArray; 621 UResourceBundle *fromLocaleBund = ures_open(NULL, fromLocale, &errorCode); 622 UResourceBundle *toLocaleBund = ures_open(NULL, toLocale, &errorCode); 623 UResourceBundle *toCalendar, *fromCalendar, *toGregorian, *fromGregorian; 624 625 if(U_FAILURE(errorCode)) { 626 log_err("Can't open resource bundle %s or %s - %s\n", fromLocale, toLocale, u_errorName(errorCode)); 627 return; 628 } 629 fromCalendar = ures_getByKey(fromLocaleBund, "calendar", NULL, &errorCode); 630 fromGregorian = ures_getByKeyWithFallback(fromCalendar, "gregorian", NULL, &errorCode); 631 632 toCalendar = ures_getByKey(toLocaleBund, "calendar", NULL, &errorCode); 633 toGregorian = ures_getByKeyWithFallback(toCalendar, "gregorian", NULL, &errorCode); 634 635 fromArray = ures_getByKey(fromLocaleBund, "CurrencyElements", NULL, &errorCode); 636 toArray = ures_getByKey(toLocaleBund, "CurrencyElements", NULL, &errorCode); 637 if (strcmp(fromLocale, "en_CA") != 0) 638 { 639 /* The first one is probably localized. */ 640 compareArrays("CurrencyElements", fromArray, fromLocale, toArray, toLocale, 1, 2); 641 } 642 ures_close(fromArray); 643 ures_close(toArray); 644 645 fromArray = ures_getByKey(fromLocaleBund, "NumberPatterns", NULL, &errorCode); 646 toArray = ures_getByKey(toLocaleBund, "NumberPatterns", NULL, &errorCode); 647 if (strcmp(fromLocale, "en_CA") != 0) 648 { 649 compareArrays("NumberPatterns", fromArray, fromLocale, toArray, toLocale, 0, 3); 650 } 651 ures_close(fromArray); 652 ures_close(toArray); 653 654 /* Difficult to test properly */ 655 /* 656 fromArray = ures_getByKey(fromLocaleBund, "DateTimePatterns", NULL, &errorCode); 657 toArray = ures_getByKey(toLocaleBund, "DateTimePatterns", NULL, &errorCode); 658 { 659 compareArrays("DateTimePatterns", fromArray, fromLocale, toArray, toLocale); 660 } 661 ures_close(fromArray); 662 ures_close(toArray);*/ 663 664 fromArray = ures_getByKey(fromLocaleBund, "NumberElements", NULL, &errorCode); 665 toArray = ures_getByKey(toLocaleBund, "NumberElements", NULL, &errorCode); 666 if (strcmp(fromLocale, "en_CA") != 0) 667 { 668 compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 0, 3); 669 /* Index 4 is a script based 0 */ 670 compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 5, 10); 671 } 672 ures_close(fromArray); 673 ures_close(toArray); 674 ures_close(fromCalendar); 675 ures_close(toCalendar); 676 ures_close(fromGregorian); 677 ures_close(toGregorian); 678 679 ures_close(fromLocaleBund); 680 ures_close(toLocaleBund); 681 } 682 683 static void 684 TestConsistentCountryInfo(void) { 685 /* UResourceBundle *fromLocale, *toLocale;*/ 686 int32_t locCount = uloc_countAvailable(); 687 int32_t fromLocIndex, toLocIndex; 688 689 int32_t fromCountryLen, toCountryLen; 690 char fromCountry[ULOC_FULLNAME_CAPACITY], toCountry[ULOC_FULLNAME_CAPACITY]; 691 692 int32_t fromVariantLen, toVariantLen; 693 char fromVariant[ULOC_FULLNAME_CAPACITY], toVariant[ULOC_FULLNAME_CAPACITY]; 694 695 UErrorCode errorCode = U_ZERO_ERROR; 696 697 for (fromLocIndex = 0; fromLocIndex < locCount; fromLocIndex++) { 698 const char *fromLocale = uloc_getAvailable(fromLocIndex); 699 700 errorCode=U_ZERO_ERROR; 701 fromCountryLen = uloc_getCountry(fromLocale, fromCountry, ULOC_FULLNAME_CAPACITY, &errorCode); 702 if (fromCountryLen <= 0) { 703 /* Ignore countryless locales */ 704 continue; 705 } 706 fromVariantLen = uloc_getVariant(fromLocale, fromVariant, ULOC_FULLNAME_CAPACITY, &errorCode); 707 if (fromVariantLen > 0) { 708 /* Most variants are ignorable like PREEURO, or collation variants. */ 709 continue; 710 } 711 /* Start comparing only after the current index. 712 Previous loop should have already compared fromLocIndex. 713 */ 714 for (toLocIndex = fromLocIndex + 1; toLocIndex < locCount; toLocIndex++) { 715 const char *toLocale = uloc_getAvailable(toLocIndex); 716 717 toCountryLen = uloc_getCountry(toLocale, toCountry, ULOC_FULLNAME_CAPACITY, &errorCode); 718 if(U_FAILURE(errorCode)) { 719 log_err("Unknown failure fromLocale=%s toLocale=%s errorCode=%s\n", 720 fromLocale, toLocale, u_errorName(errorCode)); 721 continue; 722 } 723 724 if (toCountryLen <= 0) { 725 /* Ignore countryless locales */ 726 continue; 727 } 728 toVariantLen = uloc_getVariant(toLocale, toVariant, ULOC_FULLNAME_CAPACITY, &errorCode); 729 if (toVariantLen > 0) { 730 /* Most variants are ignorable like PREEURO, or collation variants. */ 731 /* They're a variant for a reason. */ 732 continue; 733 } 734 if (strcmp(fromCountry, toCountry) == 0) { 735 log_verbose("comparing fromLocale=%s toLocale=%s\n", 736 fromLocale, toLocale); 737 compareConsistentCountryInfo(fromLocale, toLocale); 738 } 739 } 740 } 741 } 742 743 static int32_t 744 findStringSetMismatch(const char *currLoc, const UChar *string, int32_t langSize, 745 USet * mergedExemplarSet, 746 UBool ignoreNumbers, UChar* badCharPtr) { 747 UErrorCode errorCode = U_ZERO_ERROR; 748 USet *exemplarSet; 749 int32_t strIdx; 750 if (mergedExemplarSet == NULL) { 751 return -1; 752 } 753 exemplarSet = createFlattenSet(mergedExemplarSet, &errorCode); 754 if (U_FAILURE(errorCode)) { 755 log_err("%s: error createFlattenSet returned %s\n", currLoc, u_errorName(errorCode)); 756 return -1; 757 } 758 759 for (strIdx = 0; strIdx < langSize; strIdx++) { 760 if (!uset_contains(exemplarSet, string[strIdx]) 761 && string[strIdx] != 0x0020 && string[strIdx] != 0x00A0 && string[strIdx] != 0x002e && string[strIdx] != 0x002c && string[strIdx] != 0x002d && string[strIdx] != 0x0027 && string[strIdx] != 0x005B && string[strIdx] != 0x005D && string[strIdx] != 0x2019 && string[strIdx] != 0x0f0b 762 && string[strIdx] != 0x200C && string[strIdx] != 0x200D) { 763 if (!ignoreNumbers || (ignoreNumbers && (string[strIdx] < 0x30 || string[strIdx] > 0x39))) { 764 uset_close(exemplarSet); 765 if (badCharPtr) { 766 *badCharPtr = string[strIdx]; 767 } 768 return strIdx; 769 } 770 } 771 } 772 uset_close(exemplarSet); 773 if (badCharPtr) { 774 *badCharPtr = 0; 775 } 776 return -1; 777 } 778 /* include non-invariant chars */ 779 static int32_t 780 myUCharsToChars(const UChar* us, char* cs, int32_t len){ 781 int32_t i=0; 782 for(; i< len; i++){ 783 if(us[i] < 0x7f){ 784 cs[i] = (char)us[i]; 785 }else{ 786 return -1; 787 } 788 } 789 return i; 790 } 791 static void 792 findSetMatch( UScriptCode *scriptCodes, int32_t scriptsLen, 793 USet *exemplarSet, 794 const char *locale){ 795 USet *scripts[10]= {0}; 796 char pattern[256] = { '[', ':', 0x000 }; 797 int32_t patternLen; 798 UChar uPattern[256] = {0}; 799 UErrorCode status = U_ZERO_ERROR; 800 int32_t i; 801 802 /* create the sets with script codes */ 803 for(i = 0; i<scriptsLen; i++){ 804 strcat(pattern, uscript_getShortName(scriptCodes[i])); 805 strcat(pattern, ":]"); 806 patternLen = (int32_t)strlen(pattern); 807 u_charsToUChars(pattern, uPattern, patternLen); 808 scripts[i] = uset_openPattern(uPattern, patternLen, &status); 809 if(U_FAILURE(status)){ 810 log_err("Could not create set for pattern %s. Error: %s\n", pattern, u_errorName(status)); 811 return; 812 } 813 pattern[2] = 0; 814 } 815 if (strcmp(locale, "uk") == 0 || strcmp(locale, "uk_UA") == 0) { 816 /* Special addition. Add the modifying apostrophe, which isn't in Cyrillic. */ 817 uset_add(scripts[0], 0x2bc); 818 } 819 if(U_SUCCESS(status)){ 820 UBool existsInScript = FALSE; 821 /* iterate over the exemplarSet and ascertain if all 822 * UChars in exemplarSet belong to the scripts returned 823 * by getScript 824 */ 825 int32_t count = uset_getItemCount(exemplarSet); 826 827 for( i=0; i < count; i++){ 828 UChar32 start = 0; 829 UChar32 end = 0; 830 UChar *str = NULL; 831 int32_t strCapacity = 0; 832 833 strCapacity = uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status); 834 if(U_SUCCESS(status)){ 835 int32_t j; 836 if(strCapacity == 0){ 837 /* ok the item is a range */ 838 for( j = 0; j < scriptsLen; j++){ 839 if(uset_containsRange(scripts[j], start, end) == TRUE){ 840 existsInScript = TRUE; 841 } 842 } 843 if(existsInScript == FALSE){ 844 for( j = 0; j < scriptsLen; j++){ 845 UChar toPattern[500]={'\0'}; 846 char pat[500]={'\0'}; 847 int32_t len = uset_toPattern(scripts[j], toPattern, 500, TRUE, &status); 848 len = myUCharsToChars(toPattern, pat, len); 849 log_err("uset_indexOf(\\u%04X)=%i uset_indexOf(\\u%04X)=%i\n", start, uset_indexOf(scripts[0], start), end, uset_indexOf(scripts[0], end)); 850 if(len!=-1){ 851 log_err("Pattern: %s\n",pat); 852 } 853 } 854 log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale); 855 } 856 }else{ 857 strCapacity++; /* increment for NUL termination */ 858 /* allocate the str and call the api again */ 859 str = (UChar*) malloc(U_SIZEOF_UCHAR * strCapacity); 860 strCapacity = uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status); 861 /* iterate over the scripts and figure out if the string contained is actually 862 * in the script set 863 */ 864 for( j = 0; j < scriptsLen; j++){ 865 if(uset_containsString(scripts[j],str, strCapacity) == TRUE){ 866 existsInScript = TRUE; 867 } 868 } 869 if(existsInScript == FALSE){ 870 log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale); 871 } 872 } 873 } 874 } 875 876 } 877 878 /* close the sets */ 879 for(i = 0; i<scriptsLen; i++){ 880 uset_close(scripts[i]); 881 } 882 } 883 884 static void VerifyTranslation(void) { 885 UResourceBundle *root, *currentLocale; 886 int32_t locCount = uloc_countAvailable(); 887 int32_t locIndex; 888 UErrorCode errorCode = U_ZERO_ERROR; 889 const char *currLoc; 890 UScriptCode scripts[USCRIPT_CODE_LIMIT]; 891 int32_t numScripts; 892 int32_t idx; 893 int32_t end; 894 UResourceBundle *resArray; 895 896 if (locCount <= 1) { 897 log_data_err("At least root needs to be installed\n"); 898 } 899 900 root = ures_openDirect(NULL, "root", &errorCode); 901 if(U_FAILURE(errorCode)) { 902 log_data_err("Can't open root\n"); 903 return; 904 } 905 for (locIndex = 0; locIndex < locCount; locIndex++) { 906 USet * mergedExemplarSet = NULL; 907 errorCode=U_ZERO_ERROR; 908 currLoc = uloc_getAvailable(locIndex); 909 currentLocale = ures_open(NULL, currLoc, &errorCode); 910 if(errorCode != U_ZERO_ERROR) { 911 if(U_SUCCESS(errorCode)) { 912 /* It's installed, but there is no data. 913 It's installed for the g18n white paper [grhoten] */ 914 log_err("ERROR: Locale %-5s not installed, and it should be!\n", 915 uloc_getAvailable(locIndex)); 916 } else { 917 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%", 918 u_errorName(errorCode), 919 uloc_getAvailable(locIndex)); 920 } 921 ures_close(currentLocale); 922 continue; 923 } 924 { 925 UErrorCode exemplarStatus = U_ZERO_ERROR; 926 ULocaleData * uld = ulocdata_open(currLoc, &exemplarStatus); 927 if (U_SUCCESS(exemplarStatus)) { 928 USet * exemplarSet = ulocdata_getExemplarSet(uld, NULL, USET_ADD_CASE_MAPPINGS, ULOCDATA_ES_STANDARD, &exemplarStatus); 929 if (U_SUCCESS(exemplarStatus)) { 930 mergedExemplarSet = uset_cloneAsThawed(exemplarSet); 931 uset_close(exemplarSet); 932 exemplarSet = ulocdata_getExemplarSet(uld, NULL, USET_ADD_CASE_MAPPINGS, ULOCDATA_ES_AUXILIARY, &exemplarStatus); 933 if (U_SUCCESS(exemplarStatus)) { 934 uset_addAll(mergedExemplarSet, exemplarSet); 935 uset_close(exemplarSet); 936 } 937 exemplarStatus = U_ZERO_ERROR; 938 exemplarSet = ulocdata_getExemplarSet(uld, NULL, 0, ULOCDATA_ES_PUNCTUATION, &exemplarStatus); 939 if (U_SUCCESS(exemplarStatus)) { 940 uset_addAll(mergedExemplarSet, exemplarSet); 941 uset_close(exemplarSet); 942 } 943 } else { 944 log_err("error ulocdata_getExemplarSet (main) for locale %s returned %s\n", currLoc, u_errorName(errorCode)); 945 } 946 ulocdata_close(uld); 947 } else { 948 log_err("error ulocdata_open for locale %s returned %s\n", currLoc, u_errorName(errorCode)); 949 } 950 } 951 if (mergedExemplarSet == NULL /*|| (getTestOption(QUICK_OPTION) && uset_size() > 2048)*/) { 952 log_verbose("skipping test for %s\n", currLoc); 953 } 954 //else if (uprv_strncmp(currLoc,"bem",3) == 0 || uprv_strncmp(currLoc,"mgo",3) == 0 || uprv_strncmp(currLoc,"nl",2) == 0) { 955 // log_verbose("skipping test for %s, some month and country names known to use aux exemplars\n", currLoc); 956 //} 957 else { 958 UChar langBuffer[128]; 959 int32_t langSize; 960 int32_t strIdx; 961 UChar badChar; 962 langSize = uloc_getDisplayLanguage(currLoc, currLoc, langBuffer, sizeof(langBuffer)/sizeof(langBuffer[0]), &errorCode); 963 if (U_FAILURE(errorCode)) { 964 log_err("error uloc_getDisplayLanguage returned %s\n", u_errorName(errorCode)); 965 } 966 else { 967 strIdx = findStringSetMismatch(currLoc, langBuffer, langSize, mergedExemplarSet, FALSE, &badChar); 968 if (strIdx >= 0) { 969 log_err("getDisplayLanguage(%s) at index %d returned characters not in the exemplar characters: %04X.\n", 970 currLoc, strIdx, badChar); 971 } 972 } 973 langSize = uloc_getDisplayCountry(currLoc, currLoc, langBuffer, sizeof(langBuffer)/sizeof(langBuffer[0]), &errorCode); 974 if (U_FAILURE(errorCode)) { 975 log_err("error uloc_getDisplayCountry returned %s\n", u_errorName(errorCode)); 976 } 977 { 978 UResourceBundle* cal = ures_getByKey(currentLocale, "calendar", NULL, &errorCode); 979 UResourceBundle* greg = ures_getByKeyWithFallback(cal, "gregorian", NULL, &errorCode); 980 UResourceBundle* names = ures_getByKeyWithFallback(greg, "dayNames", NULL, &errorCode); 981 UResourceBundle* format = ures_getByKeyWithFallback(names, "format", NULL, &errorCode); 982 resArray = ures_getByKeyWithFallback(format, "wide", NULL, &errorCode); 983 984 if (U_FAILURE(errorCode)) { 985 log_err("error ures_getByKey returned %s\n", u_errorName(errorCode)); 986 } 987 if (getTestOption(QUICK_OPTION)) { 988 end = 1; 989 } 990 else { 991 end = ures_getSize(resArray); 992 } 993 994 995 for (idx = 0; idx < end; idx++) { 996 const UChar *fromBundleStr = ures_getStringByIndex(resArray, idx, &langSize, &errorCode); 997 if (U_FAILURE(errorCode)) { 998 log_err("error ures_getStringByIndex(%d) returned %s\n", idx, u_errorName(errorCode)); 999 continue; 1000 } 1001 strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, mergedExemplarSet, TRUE, &badChar); 1002 if (strIdx >= 0) { 1003 log_err("getDayNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n", 1004 currLoc, idx, strIdx, badChar); 1005 } 1006 } 1007 ures_close(resArray); 1008 ures_close(format); 1009 ures_close(names); 1010 1011 names = ures_getByKeyWithFallback(greg, "monthNames", NULL, &errorCode); 1012 format = ures_getByKeyWithFallback(names,"format", NULL, &errorCode); 1013 resArray = ures_getByKeyWithFallback(format, "wide", NULL, &errorCode); 1014 if (U_FAILURE(errorCode)) { 1015 log_err("error ures_getByKey returned %s\n", u_errorName(errorCode)); 1016 } 1017 if (getTestOption(QUICK_OPTION)) { 1018 end = 1; 1019 } 1020 else { 1021 end = ures_getSize(resArray); 1022 } 1023 1024 for (idx = 0; idx < end; idx++) { 1025 const UChar *fromBundleStr = ures_getStringByIndex(resArray, idx, &langSize, &errorCode); 1026 if (U_FAILURE(errorCode)) { 1027 log_err("error ures_getStringByIndex(%d) returned %s\n", idx, u_errorName(errorCode)); 1028 continue; 1029 } 1030 strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, mergedExemplarSet, TRUE, &badChar); 1031 if (strIdx >= 0) { 1032 log_err("getMonthNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n", 1033 currLoc, idx, strIdx, badChar); 1034 } 1035 } 1036 ures_close(resArray); 1037 ures_close(format); 1038 ures_close(names); 1039 ures_close(greg); 1040 ures_close(cal); 1041 } 1042 errorCode = U_ZERO_ERROR; 1043 numScripts = uscript_getCode(currLoc, scripts, sizeof(scripts)/sizeof(scripts[0]), &errorCode); 1044 if (numScripts == 0) { 1045 log_err("uscript_getCode(%s) doesn't work.\n", currLoc); 1046 }else if(scripts[0] == USCRIPT_COMMON){ 1047 log_err("uscript_getCode(%s) returned USCRIPT_COMMON.\n", currLoc); 1048 } 1049 1050 /* test that the scripts are a superset of exemplar characters. */ 1051 { 1052 ULocaleData *uld = ulocdata_open(currLoc,&errorCode); 1053 USet *exemplarSet = ulocdata_getExemplarSet(uld, NULL, 0, ULOCDATA_ES_STANDARD, &errorCode); 1054 /* test if exemplar characters are part of script code */ 1055 findSetMatch(scripts, numScripts, exemplarSet, currLoc); 1056 uset_close(exemplarSet); 1057 ulocdata_close(uld); 1058 } 1059 1060 /* test that the paperSize API works */ 1061 { 1062 int32_t height=0, width=0; 1063 ulocdata_getPaperSize(currLoc, &height, &width, &errorCode); 1064 if(U_FAILURE(errorCode)){ 1065 log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode)); 1066 } 1067 if(strstr(currLoc, "_US")!=NULL && height != 279 && width != 216 ){ 1068 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc); 1069 } 1070 } 1071 /* test that the MeasurementSystem API works */ 1072 { 1073 char fullLoc[ULOC_FULLNAME_CAPACITY]; 1074 UMeasurementSystem measurementSystem; 1075 int32_t height = 0, width = 0; 1076 1077 uloc_addLikelySubtags(currLoc, fullLoc, ULOC_FULLNAME_CAPACITY, &errorCode); 1078 1079 errorCode = U_ZERO_ERROR; 1080 measurementSystem = ulocdata_getMeasurementSystem(currLoc, &errorCode); 1081 if (U_FAILURE(errorCode)) { 1082 log_err("ulocdata_getMeasurementSystem failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode)); 1083 } else { 1084 if ( strstr(fullLoc, "_US")!=NULL || strstr(fullLoc, "_MM")!=NULL || strstr(fullLoc, "_LR")!=NULL ) { 1085 if(measurementSystem != UMS_US){ 1086 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc); 1087 } 1088 } else if (measurementSystem != UMS_SI) { 1089 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc); 1090 } 1091 } 1092 1093 errorCode = U_ZERO_ERROR; 1094 ulocdata_getPaperSize(currLoc, &height, &width, &errorCode); 1095 if (U_FAILURE(errorCode)) { 1096 log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode)); 1097 } else { 1098 if ( strstr(fullLoc, "_US")!=NULL || strstr(fullLoc, "_BZ")!=NULL || strstr(fullLoc, "_CA")!=NULL || strstr(fullLoc, "_CL")!=NULL || 1099 strstr(fullLoc, "_CO")!=NULL || strstr(fullLoc, "_CR")!=NULL || strstr(fullLoc, "_GT")!=NULL || strstr(fullLoc, "_MX")!=NULL || 1100 strstr(fullLoc, "_NI")!=NULL || strstr(fullLoc, "_PA")!=NULL || strstr(fullLoc, "_PH")!=NULL || strstr(fullLoc, "_PR")!=NULL || 1101 strstr(fullLoc, "_SV")!=NULL || strstr(fullLoc, "_VE")!=NULL ) { 1102 if (height != 279 || width != 216) { 1103 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc); 1104 } 1105 } else if (height != 297 || width != 210) { 1106 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc); 1107 } 1108 } 1109 } 1110 } 1111 if (mergedExemplarSet != NULL) { 1112 uset_close(mergedExemplarSet); 1113 } 1114 ures_close(currentLocale); 1115 } 1116 1117 ures_close(root); 1118 } 1119 1120 /* adjust this limit as appropriate */ 1121 #define MAX_SCRIPTS_PER_LOCALE 8 1122 1123 static void TestExemplarSet(void){ 1124 int32_t i, j, k, m, n; 1125 int32_t equalCount = 0; 1126 UErrorCode ec = U_ZERO_ERROR; 1127 UEnumeration* avail; 1128 USet* exemplarSets[2]; 1129 USet* unassignedSet; 1130 UScriptCode code[MAX_SCRIPTS_PER_LOCALE]; 1131 USet* codeSets[MAX_SCRIPTS_PER_LOCALE]; 1132 int32_t codeLen; 1133 char cbuf[32]; /* 9 should be enough */ 1134 UChar ubuf[64]; /* adjust as needed */ 1135 UBool existsInScript; 1136 int32_t itemCount; 1137 int32_t strLen; 1138 UChar32 start, end; 1139 1140 unassignedSet = NULL; 1141 exemplarSets[0] = NULL; 1142 exemplarSets[1] = NULL; 1143 for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) { 1144 codeSets[i] = NULL; 1145 } 1146 1147 avail = ures_openAvailableLocales(NULL, &ec); 1148 if (!assertSuccess("ures_openAvailableLocales", &ec)) goto END; 1149 n = uenum_count(avail, &ec); 1150 if (!assertSuccess("uenum_count", &ec)) goto END; 1151 1152 u_uastrcpy(ubuf, "[:unassigned:]"); 1153 unassignedSet = uset_openPattern(ubuf, -1, &ec); 1154 if (!assertSuccess("uset_openPattern", &ec)) goto END; 1155 1156 for(i=0; i<n; i++){ 1157 const char* locale = uenum_next(avail, NULL, &ec); 1158 if (!assertSuccess("uenum_next", &ec)) goto END; 1159 log_verbose("%s\n", locale); 1160 for (k=0; k<2; ++k) { 1161 uint32_t option = (k==0) ? 0 : USET_CASE_INSENSITIVE; 1162 ULocaleData *uld = ulocdata_open(locale,&ec); 1163 USet* exemplarSet = ulocdata_getExemplarSet(uld,NULL, option, ULOCDATA_ES_STANDARD, &ec); 1164 uset_close(exemplarSets[k]); 1165 ulocdata_close(uld); 1166 exemplarSets[k] = exemplarSet; 1167 if (!assertSuccess("ulocaledata_getExemplarSet", &ec)) goto END; 1168 1169 if (uset_containsSome(exemplarSet, unassignedSet)) { 1170 log_err("ExemplarSet contains unassigned characters for locale : %s\n", locale); 1171 } 1172 codeLen = uscript_getCode(locale, code, 8, &ec); 1173 if (!assertSuccess("uscript_getCode", &ec)) goto END; 1174 1175 for (j=0; j<MAX_SCRIPTS_PER_LOCALE; ++j) { 1176 uset_close(codeSets[j]); 1177 codeSets[j] = NULL; 1178 } 1179 for (j=0; j<codeLen; ++j) { 1180 uprv_strcpy(cbuf, "[:"); 1181 if(code[j]==-1){ 1182 log_err("USCRIPT_INVALID_CODE returned for locale: %s\n", locale); 1183 continue; 1184 } 1185 uprv_strcat(cbuf, uscript_getShortName(code[j])); 1186 uprv_strcat(cbuf, ":]"); 1187 u_uastrcpy(ubuf, cbuf); 1188 codeSets[j] = uset_openPattern(ubuf, -1, &ec); 1189 } 1190 if (!assertSuccess("uset_openPattern", &ec)) goto END; 1191 1192 existsInScript = FALSE; 1193 itemCount = uset_getItemCount(exemplarSet); 1194 for (m=0; m<itemCount && !existsInScript; ++m) { 1195 strLen = uset_getItem(exemplarSet, m, &start, &end, ubuf, 1196 sizeof(ubuf)/sizeof(ubuf[0]), &ec); 1197 /* failure here might mean str[] needs to be larger */ 1198 if (!assertSuccess("uset_getItem", &ec)) goto END; 1199 if (strLen == 0) { 1200 for (j=0; j<codeLen; ++j) { 1201 if (codeSets[j]!=NULL && uset_containsRange(codeSets[j], start, end)) { 1202 existsInScript = TRUE; 1203 break; 1204 } 1205 } 1206 } else { 1207 for (j=0; j<codeLen; ++j) { 1208 if (codeSets[j]!=NULL && uset_containsString(codeSets[j], ubuf, strLen)) { 1209 existsInScript = TRUE; 1210 break; 1211 } 1212 } 1213 } 1214 } 1215 1216 if (existsInScript == FALSE){ 1217 log_err("ExemplarSet containment failed for locale : %s\n", locale); 1218 } 1219 } 1220 assertTrue("case-folded is a superset", 1221 uset_containsAll(exemplarSets[1], exemplarSets[0])); 1222 if (uset_equals(exemplarSets[1], exemplarSets[0])) { 1223 ++equalCount; 1224 } 1225 } 1226 /* Note: The case-folded set should sometimes be a strict superset 1227 and sometimes be equal. */ 1228 assertTrue("case-folded is sometimes a strict superset, and sometimes equal", 1229 equalCount > 0 && equalCount < n); 1230 1231 END: 1232 uenum_close(avail); 1233 uset_close(exemplarSets[0]); 1234 uset_close(exemplarSets[1]); 1235 uset_close(unassignedSet); 1236 for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) { 1237 uset_close(codeSets[i]); 1238 } 1239 } 1240 1241 enum { kUBufMax = 32 }; 1242 static void TestLocaleDisplayPattern(void){ 1243 UErrorCode status; 1244 UChar pattern[kUBufMax] = {0,}; 1245 UChar separator[kUBufMax] = {0,}; 1246 ULocaleData *uld; 1247 static const UChar enExpectPat[] = { 0x007B,0x0030,0x007D,0x0020,0x0028,0x007B,0x0031,0x007D,0x0029,0 }; /* "{0} ({1})" */ 1248 static const UChar enExpectSep[] = { 0x002C,0x0020,0 }; /* ", " */ 1249 static const UChar zhExpectPat[] = { 0x007B,0x0030,0x007D,0xFF08,0x007B,0x0031,0x007D,0xFF09,0 }; 1250 static const UChar zhExpectSep[] = { 0x3001,0 }; 1251 1252 status = U_ZERO_ERROR; 1253 uld = ulocdata_open("en", &status); 1254 if(U_FAILURE(status)){ 1255 log_data_err("ulocdata_open en error %s", u_errorName(status)); 1256 } else { 1257 ulocdata_getLocaleDisplayPattern(uld, pattern, kUBufMax, &status); 1258 if (U_FAILURE(status)){ 1259 log_err("ulocdata_getLocaleDisplayPattern en error %s", u_errorName(status)); 1260 } else if (u_strcmp(pattern, enExpectPat) != 0) { 1261 log_err("ulocdata_getLocaleDisplayPattern en returns unexpected pattern"); 1262 } 1263 status = U_ZERO_ERROR; 1264 ulocdata_getLocaleSeparator(uld, separator, kUBufMax, &status); 1265 if (U_FAILURE(status)){ 1266 log_err("ulocdata_getLocaleSeparator en error %s", u_errorName(status)); 1267 } else if (u_strcmp(separator, enExpectSep) != 0) { 1268 log_err("ulocdata_getLocaleSeparator en returns unexpected string "); 1269 } 1270 ulocdata_close(uld); 1271 } 1272 1273 status = U_ZERO_ERROR; 1274 uld = ulocdata_open("zh", &status); 1275 if(U_FAILURE(status)){ 1276 log_data_err("ulocdata_open zh error %s", u_errorName(status)); 1277 } else { 1278 ulocdata_getLocaleDisplayPattern(uld, pattern, kUBufMax, &status); 1279 if (U_FAILURE(status)){ 1280 log_err("ulocdata_getLocaleDisplayPattern zh error %s", u_errorName(status)); 1281 } else if (u_strcmp(pattern, zhExpectPat) != 0) { 1282 log_err("ulocdata_getLocaleDisplayPattern zh returns unexpected pattern"); 1283 } 1284 status = U_ZERO_ERROR; 1285 ulocdata_getLocaleSeparator(uld, separator, kUBufMax, &status); 1286 if (U_FAILURE(status)){ 1287 log_err("ulocdata_getLocaleSeparator zh error %s", u_errorName(status)); 1288 } else if (u_strcmp(separator, zhExpectSep) != 0) { 1289 log_err("ulocdata_getLocaleSeparator zh returns unexpected string "); 1290 } 1291 ulocdata_close(uld); 1292 } 1293 } 1294 1295 static void TestCoverage(void){ 1296 ULocaleDataDelimiterType types[] = { 1297 ULOCDATA_QUOTATION_START, /* Quotation start */ 1298 ULOCDATA_QUOTATION_END, /* Quotation end */ 1299 ULOCDATA_ALT_QUOTATION_START, /* Alternate quotation start */ 1300 ULOCDATA_ALT_QUOTATION_END, /* Alternate quotation end */ 1301 ULOCDATA_DELIMITER_COUNT 1302 }; 1303 int i; 1304 UBool sub; 1305 UErrorCode status = U_ZERO_ERROR; 1306 ULocaleData *uld = ulocdata_open(uloc_getDefault(), &status); 1307 1308 if(U_FAILURE(status)){ 1309 log_data_err("ulocdata_open error"); 1310 return; 1311 } 1312 1313 1314 for(i = 0; i < ULOCDATA_DELIMITER_COUNT; i++){ 1315 UChar result[32] = {0,}; 1316 status = U_ZERO_ERROR; 1317 ulocdata_getDelimiter(uld, types[i], result, 32, &status); 1318 if (U_FAILURE(status)){ 1319 log_err("ulocdata_getgetDelimiter error with type %d", types[i]); 1320 } 1321 } 1322 1323 sub = ulocdata_getNoSubstitute(uld); 1324 ulocdata_setNoSubstitute(uld,sub); 1325 ulocdata_close(uld); 1326 } 1327 1328 static void TestIndexChars(void) { 1329 /* Very basic test of ULOCDATA_ES_INDEX. 1330 * No comprehensive test of data, just basic check that the code path is alive. 1331 */ 1332 UErrorCode status = U_ZERO_ERROR; 1333 ULocaleData *uld; 1334 USet *exemplarChars; 1335 USet *indexChars; 1336 1337 uld = ulocdata_open("en", &status); 1338 exemplarChars = uset_openEmpty(); 1339 indexChars = uset_openEmpty(); 1340 ulocdata_getExemplarSet(uld, exemplarChars, 0, ULOCDATA_ES_STANDARD, &status); 1341 ulocdata_getExemplarSet(uld, indexChars, 0, ULOCDATA_ES_INDEX, &status); 1342 if (U_FAILURE(status)) { 1343 log_data_err("File %s, line %d, Failure opening exemplar chars: %s", __FILE__, __LINE__, u_errorName(status)); 1344 goto close_sets; 1345 } 1346 /* en data, standard exemplars are [a-z], lower case. */ 1347 /* en data, index characters are [A-Z], upper case. */ 1348 if ((uset_contains(exemplarChars, (UChar32)0x41) || uset_contains(indexChars, (UChar32)0x61))) { 1349 log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__, __LINE__ ); 1350 goto close_sets; 1351 } 1352 if (!(uset_contains(exemplarChars, (UChar32)0x61) && uset_contains(indexChars, (UChar32)0x41) )) { 1353 log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__, __LINE__ ); 1354 goto close_sets; 1355 } 1356 1357 close_sets: 1358 uset_close(exemplarChars); 1359 uset_close(indexChars); 1360 ulocdata_close(uld); 1361 } 1362 1363 1364 1365 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION 1366 static void TestCurrencyList(void){ 1367 #if !UCONFIG_NO_FORMATTING 1368 UErrorCode errorCode = U_ZERO_ERROR; 1369 int32_t structLocaleCount, currencyCount; 1370 UEnumeration *en = ucurr_openISOCurrencies(UCURR_ALL, &errorCode); 1371 const char *isoCode, *structISOCode; 1372 UResourceBundle *subBundle; 1373 UResourceBundle *currencies = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode); 1374 if(U_FAILURE(errorCode)) { 1375 log_data_err("Can't open structLocale\n"); 1376 return; 1377 } 1378 currencies = ures_getByKey(currencies, "Currencies", currencies, &errorCode); 1379 currencyCount = uenum_count(en, &errorCode); 1380 structLocaleCount = ures_getSize(currencies); 1381 if (currencyCount != structLocaleCount) { 1382 log_err("structLocale(%d) and ISO4217(%d) currency list are out of sync.\n", structLocaleCount, currencyCount); 1383 #if U_CHARSET_FAMILY == U_ASCII_FAMILY 1384 ures_resetIterator(currencies); 1385 while ((isoCode = uenum_next(en, NULL, &errorCode)) != NULL && ures_hasNext(currencies)) { 1386 subBundle = ures_getNextResource(currencies, NULL, &errorCode); 1387 structISOCode = ures_getKey(subBundle); 1388 ures_close(subBundle); 1389 if (strcmp(structISOCode, isoCode) != 0) { 1390 log_err("First difference found at structLocale(%s) and ISO4217(%s).\n", structISOCode, isoCode); 1391 break; 1392 } 1393 } 1394 #endif 1395 } 1396 ures_close(currencies); 1397 uenum_close(en); 1398 #endif 1399 } 1400 #endif 1401 1402 static void TestAvailableIsoCodes(void){ 1403 #if !UCONFIG_NO_FORMATTING 1404 UErrorCode errorCode = U_ZERO_ERROR; 1405 const char* eurCode = "EUR"; 1406 const char* usdCode = "USD"; 1407 const char* lastCode = "RHD"; 1408 const char* zzzCode = "ZZZ"; 1409 UDate date1950 = (UDate)-630720000000.0;/* year 1950 */ 1410 UDate date1970 = (UDate)0.0; /* year 1970 */ 1411 UDate date1975 = (UDate)173448000000.0; /* year 1975 */ 1412 UDate date1978 = (UDate)260172000000.0; /* year 1978 */ 1413 UDate date1981 = (UDate)346896000000.0; /* year 1981 */ 1414 UDate date1992 = (UDate)693792000000.0; /* year 1992 */ 1415 UChar* isoCode = (UChar*)malloc(sizeof(UChar) * (uprv_strlen(usdCode) + 1)); 1416 1417 /* testing available codes with no time ranges */ 1418 u_charsToUChars(eurCode, isoCode, uprv_strlen(usdCode) + 1); 1419 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) { 1420 log_data_err("FAIL: ISO code (%s) is not found.\n", eurCode); 1421 } 1422 1423 u_charsToUChars(usdCode, isoCode, uprv_strlen(zzzCode) + 1); 1424 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) { 1425 log_data_err("FAIL: ISO code (%s) is not found.\n", usdCode); 1426 } 1427 1428 u_charsToUChars(zzzCode, isoCode, uprv_strlen(zzzCode) + 1); 1429 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == TRUE) { 1430 log_err("FAIL: ISO code (%s) is reported as available, but it doesn't exist.\n", zzzCode); 1431 } 1432 1433 u_charsToUChars(lastCode, isoCode, uprv_strlen(zzzCode) + 1); 1434 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) { 1435 log_data_err("FAIL: ISO code (%s) is not found.\n", lastCode); 1436 } 1437 1438 /* RHD was used from 1970-02-17 to 1980-04-18*/ 1439 1440 /* to = null */ 1441 if (ucurr_isAvailable(isoCode, date1970, U_DATE_MAX, &errorCode) == FALSE) { 1442 log_data_err("FAIL: ISO code (%s) was available in time range >1970-01-01.\n", lastCode); 1443 } 1444 1445 if (ucurr_isAvailable(isoCode, date1975, U_DATE_MAX, &errorCode) == FALSE) { 1446 log_data_err("FAIL: ISO code (%s) was available in time range >1975.\n", lastCode); 1447 } 1448 1449 if (ucurr_isAvailable(isoCode, date1981, U_DATE_MAX, &errorCode) == TRUE) { 1450 log_err("FAIL: ISO code (%s) was not available in time range >1981.\n", lastCode); 1451 } 1452 1453 /* from = null */ 1454 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1970, &errorCode) == TRUE) { 1455 log_err("FAIL: ISO code (%s) was not available in time range <1970.\n", lastCode); 1456 } 1457 1458 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1975, &errorCode) == FALSE) { 1459 log_data_err("FAIL: ISO code (%s) was available in time range <1975.\n", lastCode); 1460 } 1461 1462 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1981, &errorCode) == FALSE) { 1463 log_data_err("FAIL: ISO code (%s) was available in time range <1981.\n", lastCode); 1464 } 1465 1466 /* full ranges */ 1467 if (ucurr_isAvailable(isoCode, date1975, date1978, &errorCode) == FALSE) { 1468 log_data_err("FAIL: ISO code (%s) was available in time range 1975-1978.\n", lastCode); 1469 } 1470 1471 if (ucurr_isAvailable(isoCode, date1970, date1975, &errorCode) == FALSE) { 1472 log_data_err("FAIL: ISO code (%s) was available in time range 1970-1975.\n", lastCode); 1473 } 1474 1475 if (ucurr_isAvailable(isoCode, date1975, date1981, &errorCode) == FALSE) { 1476 log_data_err("FAIL: ISO code (%s) was available in time range 1975-1981.\n", lastCode); 1477 } 1478 1479 if (ucurr_isAvailable(isoCode, date1970, date1981, &errorCode) == FALSE) { 1480 log_data_err("FAIL: ISO code (%s) was available in time range 1970-1981.\n", lastCode); 1481 } 1482 1483 if (ucurr_isAvailable(isoCode, date1981, date1992, &errorCode) == TRUE) { 1484 log_err("FAIL: ISO code (%s) was not available in time range 1981-1992.\n", lastCode); 1485 } 1486 1487 if (ucurr_isAvailable(isoCode, date1950, date1970, &errorCode) == TRUE) { 1488 log_err("FAIL: ISO code (%s) was not available in time range 1950-1970.\n", lastCode); 1489 } 1490 1491 /* wrong range - from > to*/ 1492 if (ucurr_isAvailable(isoCode, date1975, date1970, &errorCode) == TRUE) { 1493 log_err("FAIL: Wrong range 1975-1970 for ISO code (%s) was not reported.\n", lastCode); 1494 } else if (errorCode != U_ILLEGAL_ARGUMENT_ERROR) { 1495 log_data_err("FAIL: Error code not reported for wrong range 1975-1970 for ISO code (%s).\n", lastCode); 1496 } 1497 1498 free(isoCode); 1499 #endif 1500 } 1501 1502 #define TESTCASE(name) addTest(root, &name, "tsutil/cldrtest/" #name) 1503 1504 void addCLDRTest(TestNode** root); 1505 1506 void addCLDRTest(TestNode** root) 1507 { 1508 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION 1509 TESTCASE(TestLocaleStructure); 1510 TESTCASE(TestCurrencyList); 1511 #endif 1512 TESTCASE(TestConsistentCountryInfo); 1513 TESTCASE(VerifyTranslation); 1514 TESTCASE(TestExemplarSet); 1515 TESTCASE(TestLocaleDisplayPattern); 1516 TESTCASE(TestCoverage); 1517 TESTCASE(TestIndexChars); 1518 TESTCASE(TestAvailableIsoCodes); 1519 } 1520 1521