icuSources/test/cintltst/cldrtest.c

   1 /********************************************************************
   2  * COPYRIGHT:
   3  * Copyright (c) 1997-2008, 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   Time bomb - allows temporary behavior that expires at a given
  21               release
  22  ---------------------------------------------------------------------*/
  23 static const UVersionInfo ICU_37 = {3,7,0,0};
  24
  25 /*
  26 returns a new UnicodeSet that is a flattened form of the original
  27 UnicodeSet.
  28 */
  29 static USet*
  30 createFlattenSet(USet *origSet, UErrorCode *status) {
  31
  32
  33     USet *newSet = NULL;
  34     int32_t origItemCount = 0;
  35     int32_t idx, graphmeSize;
  36     UChar32 start, end;
  37     UChar graphme[64];
  38     if (U_FAILURE(*status)) {
  39         log_err("createFlattenSet called with %s\n", u_errorName(*status));
  40         return NULL;
  41     }
  42     newSet = uset_open(1, 0);
  43     origItemCount = uset_getItemCount(origSet);
  44     for (idx = 0; idx < origItemCount; idx++) {
  45         graphmeSize = uset_getItem(origSet, idx,
  46             &start, &end,
  47             graphme, (int32_t)(sizeof(graphme)/sizeof(graphme[0])),
  48             status);
  49         if (U_FAILURE(*status)) {
  50             log_err("ERROR: uset_getItem returned %s\n", u_errorName(*status));
  51             *status = U_ZERO_ERROR;
  52         }
  53         if (graphmeSize) {
  54             uset_addAllCodePoints(newSet, graphme, graphmeSize);
  55         }
  56         else {
  57             uset_addRange(newSet, start, end);
  58         }
  59     }
  60     return newSet;
  61 }
  62 static UBool
  63 isCurrencyPreEuro(const char* currencyKey){
  64     if( strcmp(currencyKey, "PTE") == 0 ||
  65         strcmp(currencyKey, "ESP") == 0 ||
  66         strcmp(currencyKey, "LUF") == 0 ||
  67         strcmp(currencyKey, "GRD") == 0 ||
  68         strcmp(currencyKey, "BEF") == 0 ||
  69         strcmp(currencyKey, "ITL") == 0 ||
  70         strcmp(currencyKey, "EEK") == 0){
  71             return TRUE;
  72     }
  73     return FALSE;
  74 }
  75 static void
  76 TestKeyInRootRecursive(UResourceBundle *root, const char *rootName,
  77                        UResourceBundle *currentBundle, const char *locale) {
  78     UErrorCode errorCode = U_ZERO_ERROR;
  79     UResourceBundle *subRootBundle = NULL, *subBundle = NULL;
  80
  81     ures_resetIterator(root);
  82     ures_resetIterator(currentBundle);
  83     while (ures_hasNext(currentBundle)) {
  84         const char *subBundleKey = NULL;
  85         const char *currentBundleKey = NULL;
  86
  87         errorCode = U_ZERO_ERROR;
  88         currentBundleKey = ures_getKey(currentBundle);
  89         subBundle = ures_getNextResource(currentBundle, NULL, &errorCode);
  90         if (U_FAILURE(errorCode)) {
  91             log_err("Can't open a resource for locale %s. Error: %s\n", locale, u_errorName(errorCode));
  92             continue;
  93         }
  94         subBundleKey = ures_getKey(subBundle);
  95
  96
  97         subRootBundle = ures_getByKey(root, subBundleKey, NULL, &errorCode);
  98         if (U_FAILURE(errorCode)) {
  99             log_err("Can't open a resource with key \"%s\" in \"%s\" from %s for locale \"%s\"\n",
 100                     subBundleKey,
 101                     ures_getKey(currentBundle),
 102                     rootName,
 103                     locale);
 104             ures_close(subBundle);
 105             continue;
 106         }
 107         if (ures_getType(subRootBundle) != ures_getType(subBundle)) {
 108             log_err("key \"%s\" in \"%s\" has a different type from root for locale \"%s\"\n"
 109                     "\troot=%d, locale=%d\n",
 110                     subBundleKey,
 111                     ures_getKey(currentBundle),
 112                     locale,
 113                     ures_getType(subRootBundle),
 114                     ures_getType(subBundle));
 115             ures_close(subBundle);
 116             continue;
 117         }
 118         else if (ures_getType(subBundle) == URES_INT_VECTOR) {
 119             int32_t minSize;
 120             int32_t subBundleSize;
 121             int32_t idx;
 122             UBool sameArray = TRUE;
 123             const int32_t *subRootBundleArr = ures_getIntVector(subRootBundle, &minSize, &errorCode);
 124             const int32_t *subBundleArr = ures_getIntVector(subBundle, &subBundleSize, &errorCode);
 125
 126             if (minSize > subBundleSize) {
 127                 minSize = subBundleSize;
 128                 log_err("Arrays are different size with key \"%s\" in \"%s\" from root for locale \"%s\"\n",
 129                         subBundleKey,
 130                         ures_getKey(currentBundle),
 131                         locale);
 132             }
 133
 134             for (idx = 0; idx < minSize && sameArray; idx++) {
 135                 if (subRootBundleArr[idx] != subBundleArr[idx]) {
 136                     sameArray = FALSE;
 137                 }
 138                 if (strcmp(subBundleKey, "DateTimeElements") == 0
 139                     && (subBundleArr[idx] < 1 || 7 < subBundleArr[idx]))
 140                 {
 141                     log_err("Value out of range with key \"%s\" at index %d in \"%s\" for locale \"%s\"\n",
 142                             subBundleKey,
 143                             idx,
 144                             ures_getKey(currentBundle),
 145                             locale);
 146                 }
 147             }
 148             /* Special exception es_US and DateTimeElements */
 149             if (sameArray
 150                 && !(strcmp(locale, "es_US") == 0 && strcmp(subBundleKey, "DateTimeElements") == 0))
 151             {
 152                 log_err("Integer vectors are the same with key \"%s\" in \"%s\" from root for locale \"%s\"\n",
 153                         subBundleKey,
 154                         ures_getKey(currentBundle),
 155                         locale);
 156             }
 157         }
 158         else if (ures_getType(subBundle) == URES_ARRAY) {
 159             UResourceBundle *subSubBundle = ures_getByIndex(subBundle, 0, NULL, &errorCode);
 160             UResourceBundle *subSubRootBundle = ures_getByIndex(subRootBundle, 0, NULL, &errorCode);
 161
 162             if (U_SUCCESS(errorCode)
 163                 && (ures_getType(subSubBundle) == URES_ARRAY || ures_getType(subSubRootBundle) == URES_ARRAY))
 164             {
 165                 /* Here is one of the recursive parts */
 166                 TestKeyInRootRecursive(subRootBundle, rootName, subBundle, locale);
 167             }
 168             else {
 169                 int32_t minSize = ures_getSize(subRootBundle);
 170                 int32_t idx;
 171                 UBool sameArray = TRUE;
 172
 173                 if (minSize > ures_getSize(subBundle)) {
 174                     minSize = ures_getSize(subBundle);
 175                 }
 176
 177                 if ((subBundleKey == NULL
 178                     || (subBundleKey != NULL &&  strcmp(subBundleKey, "LocaleScript") != 0 && !isCurrencyPreEuro(subBundleKey)))
 179                     && ures_getSize(subRootBundle) != ures_getSize(subBundle))
 180                 {
 181                     log_err("Different size array with key \"%s\" in \"%s\" from root for locale \"%s\"\n"
 182                             "\troot array size=%d, locale array size=%d\n",
 183                             subBundleKey,
 184                             ures_getKey(currentBundle),
 185                             locale,
 186                             ures_getSize(subRootBundle),
 187                             ures_getSize(subBundle));
 188                 }
 189                 /*
 190                 if(isCurrencyPreEuro(subBundleKey) && ures_getSize(subBundle)!=3){
 191                     log_err("Different size array with key \"%s\" in \"%s\" for locale \"%s\" the expected size is 3 got size=%d\n",
 192                             subBundleKey,
 193                             ures_getKey(currentBundle),
 194                             locale,
 195                             ures_getSize(subBundle));
 196                 }
 197                 */
 198                 for (idx = 0; idx < minSize; idx++) {
 199                     int32_t rootStrLen, localeStrLen;
 200                     const UChar *rootStr = ures_getStringByIndex(subRootBundle,idx,&rootStrLen,&errorCode);
 201                     const UChar *localeStr = ures_getStringByIndex(subBundle,idx,&localeStrLen,&errorCode);
 202                     if (rootStr && localeStr && U_SUCCESS(errorCode)) {
 203                         if (u_strcmp(rootStr, localeStr) != 0) {
 204                             sameArray = FALSE;
 205                         }
 206                     }
 207                     else {
 208                         log_err("Got a NULL string with key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n",
 209                                 subBundleKey,
 210                                 ures_getKey(currentBundle),
 211                                 idx,
 212                                 locale);
 213                         continue;
 214                     }
 215                     if (localeStr[0] == (UChar)0x20) {
 216                         log_err("key \"%s\" at index %d in \"%s\" starts with a space in locale \"%s\"\n",
 217                                 subBundleKey,
 218                                 idx,
 219                                 ures_getKey(currentBundle),
 220                                 locale);
 221                     }
 222                     else if ((localeStr[localeStrLen - 1] == (UChar)0x20) && (strcmp(subBundleKey,"separator") != 0)) {
 223                         log_err("key \"%s\" at index %d in \"%s\" ends with a space in locale \"%s\"\n",
 224                                 subBundleKey,
 225                                 idx,
 226                                 ures_getKey(currentBundle),
 227                                 locale);
 228                     }
 229                     else if (subBundleKey != NULL
 230                         && strcmp(subBundleKey, "DateTimePatterns") == 0)
 231                     {
 232                         int32_t quoted = 0;
 233                         const UChar *localeStrItr = localeStr;
 234                         while (*localeStrItr) {
 235                             if (*localeStrItr == (UChar)0x27 /* ' */) {
 236                                 quoted++;
 237                             }
 238                             else if ((quoted % 2) == 0) {
 239                                 /* Search for unquoted characters */
 240                                 if (4 <= idx && idx <= 7
 241                                     && (*localeStrItr == (UChar)0x6B /* k */
 242                                     || *localeStrItr == (UChar)0x48 /* H */
 243                                     || *localeStrItr == (UChar)0x6D /* m */
 244                                     || *localeStrItr == (UChar)0x73 /* s */
 245                                     || *localeStrItr == (UChar)0x53 /* S */
 246                                     || *localeStrItr == (UChar)0x61 /* a */
 247                                     || *localeStrItr == (UChar)0x68 /* h */
 248                                     || *localeStrItr == (UChar)0x7A /* z */))
 249                                 {
 250                                     log_err("key \"%s\" at index %d has time pattern chars in date for locale \"%s\"\n",
 251                                             subBundleKey,
 252                                             idx,
 253                                             locale);
 254                                 }
 255                                 else if (0 <= idx && idx <= 3
 256                                     && (*localeStrItr == (UChar)0x47 /* G */
 257                                     || *localeStrItr == (UChar)0x79 /* y */
 258                                     || *localeStrItr == (UChar)0x4D /* M */
 259                                     || *localeStrItr == (UChar)0x64 /* d */
 260                                     || *localeStrItr == (UChar)0x45 /* E */
 261                                     || *localeStrItr == (UChar)0x44 /* D */
 262                                     || *localeStrItr == (UChar)0x46 /* F */
 263                                     || *localeStrItr == (UChar)0x77 /* w */
 264                                     || *localeStrItr == (UChar)0x57 /* W */))
 265                                 {
 266                                     log_err("key \"%s\" at index %d has date pattern chars in time for locale \"%s\"\n",
 267                                             subBundleKey,
 268                                             idx,
 269                                             locale);
 270                                 }
 271                             }
 272                             localeStrItr++;
 273                         }
 274                     }
 275                     else if (idx == 4 && subBundleKey != NULL
 276                         && strcmp(subBundleKey, "NumberElements") == 0
 277                         && u_charDigitValue(localeStr[0]) != 0)
 278                     {
 279                         log_err("key \"%s\" at index %d has a non-zero based number for locale \"%s\"\n",
 280                                 subBundleKey,
 281                                 idx,
 282                                 locale);
 283                     }
 284                 }
 285 /*                if (sameArray && strcmp(rootName, "root") == 0) {
 286                     log_err("Arrays are the same with key \"%s\" in \"%s\" from root for locale \"%s\"\n",
 287                             subBundleKey,
 288                             ures_getKey(currentBundle),
 289                             locale);
 290                 }*/
 291             }
 292             ures_close(subSubBundle);
 293             ures_close(subSubRootBundle);
 294         }
 295         else if (ures_getType(subBundle) == URES_STRING) {
 296             int32_t len = 0;
 297             const UChar *string = ures_getString(subBundle, &len, &errorCode);
 298             if (U_FAILURE(errorCode) || string == NULL) {
 299                 log_err("Can't open a string with key \"%s\" in \"%s\" for locale \"%s\"\n",
 300                         subBundleKey,
 301                         ures_getKey(currentBundle),
 302                         locale);
 303             } else if (string[0] == (UChar)0x20) {
 304                 log_err("key \"%s\" in \"%s\" starts with a space in locale \"%s\"\n",
 305                         subBundleKey,
 306                         ures_getKey(currentBundle),
 307                         locale);
 308             /* localeDisplayPattern/separator can end with a space */
 309             } else if (string[len - 1] == (UChar)0x20 && (strcmp(subBundleKey,"separator"))) {
 310                 log_err("key \"%s\" in \"%s\" ends with a space in locale \"%s\"\n",
 311                         subBundleKey,
 312                         ures_getKey(currentBundle),
 313                         locale);
 314             } else if (strcmp(subBundleKey, "localPatternChars") == 0) {
 315                 /* Note: We no longer import localPatternChars data starting
 316                  * ICU 3.8.  So it never comes into this else if block. (ticket#5597)
 317                  */
 318
 319                 /* Check well-formedness of localPatternChars.  First, the
 320                  * length must match the number of fields defined by
 321                  * DateFormat.  Second, each character in the string must
 322                  * be in the set [A-Za-z].  Finally, each character must be
 323                  * unique.
 324                  */
 325                 int32_t i,j;
 326 #if !UCONFIG_NO_FORMATTING
 327                 if (len != UDAT_FIELD_COUNT) {
 328                     log_err("key \"%s\" has the wrong number of characters in locale \"%s\"\n",
 329                             subBundleKey,
 330                             locale);
 331                 }
 332 #endif
 333                 /* Check char validity. */
 334                 for (i=0; i<len; ++i) {
 335                     if (!((string[i] >= 65/*'A'*/ && string[i] <= 90/*'Z'*/) ||
 336                           (string[i] >= 97/*'a'*/ && string[i] <= 122/*'z'*/))) {
 337                         log_err("key \"%s\" has illegal character '%c' in locale \"%s\"\n",
 338                                 subBundleKey,
 339                                 (char) string[i],
 340                                 locale);
 341                     }
 342                     /* Do O(n^2) check for duplicate chars. */
 343                     for (j=0; j<i; ++j) {
 344                         if (string[j] == string[i]) {
 345                             log_err("key \"%s\" has duplicate character '%c' in locale \"%s\"\n",
 346                                     subBundleKey,
 347                                     (char) string[i],
 348                                     locale);
 349                         }
 350                     }
 351                 }
 352             }
 353             /* No fallback was done. Check for duplicate data */
 354             /* The ures_* API does not do fallback of sub-resource bundles,
 355                So we can't do this now. */
 356 #if 0
 357             else if (strcmp(locale, "root") != 0 && errorCode == U_ZERO_ERROR) {
 358
 359                 const UChar *rootString = ures_getString(subRootBundle, &len, &errorCode);
 360                 if (U_FAILURE(errorCode) || rootString == NULL) {
 361                     log_err("Can't open a string with key \"%s\" in \"%s\" in root\n",
 362                             ures_getKey(subRootBundle),
 363                             ures_getKey(currentBundle));
 364                     continue;
 365                 } else if (u_strcmp(string, rootString) == 0) {
 366                     if (strcmp(locale, "de_CH") != 0 && strcmp(subBundleKey, "Countries") != 0 &&
 367                         strcmp(subBundleKey, "Version") != 0) {
 368                         log_err("Found duplicate data with key \"%s\" in \"%s\" in locale \"%s\"\n",
 369                                 ures_getKey(subRootBundle),
 370                                 ures_getKey(currentBundle),
 371                                 locale);
 372                     }
 373                     else {
 374                         /* Ignore for now. */
 375                         /* Can be fixed if fallback through de locale was done. */
 376                         log_verbose("Skipping key %s in %s\n", subBundleKey, locale);
 377                     }
 378                 }
 379             }
 380 #endif
 381         }
 382         else if (ures_getType(subBundle) == URES_TABLE) {
 383             if (strcmp(subBundleKey, "availableFormats")!=0) {
 384                 /* Here is one of the recursive parts */
 385                 TestKeyInRootRecursive(subRootBundle, rootName, subBundle, locale);
 386             }
 387             else {
 388                 log_verbose("Skipping key %s in %s\n", subBundleKey, locale);
 389             }
 390         }
 391         else if (ures_getType(subBundle) == URES_BINARY || ures_getType(subBundle) == URES_INT) {
 392             /* Can't do anything to check it */
 393             /* We'll assume it's all correct */
 394             if (strcmp(subBundleKey, "MeasurementSystem") != 0) {
 395                 log_verbose("Skipping key \"%s\" in \"%s\" for locale \"%s\"\n",
 396                         subBundleKey,
 397                         ures_getKey(currentBundle),
 398                         locale);
 399             }
 400             /* Testing for MeasurementSystem is done in VerifyTranslation */
 401         }
 402         else {
 403             log_err("Type %d for key \"%s\" in \"%s\" is unknown for locale \"%s\"\n",
 404                     ures_getType(subBundle),
 405                     subBundleKey,
 406                     ures_getKey(currentBundle),
 407                     locale);
 408         }
 409         ures_close(subRootBundle);
 410         ures_close(subBundle);
 411     }
 412 }
 413
 414
 415 static void
 416 testLCID(UResourceBundle *currentBundle,
 417          const char *localeName)
 418 {
 419     UErrorCode status = U_ZERO_ERROR;
 420     uint32_t expectedLCID;
 421     char lcidStringC[64] = {0};
 422
 423     expectedLCID = uloc_getLCID(localeName);
 424     if (expectedLCID == 0) {
 425         log_verbose("INFO:    %-5s does not have any LCID mapping\n",
 426             localeName);
 427         return;
 428     }
 429
 430     status = U_ZERO_ERROR;
 431     uprv_strcpy(lcidStringC, uprv_convertToPosix(expectedLCID, &status));
 432     if (U_FAILURE(status)) {
 433         log_err("ERROR:   %.4x does not have a POSIX mapping due to %s\n",
 434             expectedLCID, u_errorName(status));
 435     }
 436
 437     if(strcmp(localeName, lcidStringC) != 0) {
 438         char langName[1024];
 439         char langLCID[1024];
 440         uloc_getLanguage(localeName, langName, sizeof(langName), &status);
 441         uloc_getLanguage(lcidStringC, langLCID, sizeof(langLCID), &status);
 442
 443         if (strcmp(langName, langLCID) == 0) {
 444             log_verbose("WARNING: %-5s resolves to %s (0x%.4x)\n",
 445                 localeName, lcidStringC, expectedLCID);
 446         }
 447         else {
 448             log_err("ERROR:   %-5s has 0x%.4x and the number resolves wrongfully to %s\n",
 449                 localeName, expectedLCID, lcidStringC);
 450         }
 451     }
 452 }
 453
 454 static void
 455 TestLocaleStructure(void) {
 456     UResourceBundle *root, *currentLocale;
 457     int32_t locCount = uloc_countAvailable();
 458     int32_t locIndex;
 459     UErrorCode errorCode = U_ZERO_ERROR;
 460     const char *currLoc, *resolvedLoc;
 461
 462     /* TODO: Compare against parent's data too. This code can't handle fallbacks that some tools do already. */
 463 /*    char locName[ULOC_FULLNAME_CAPACITY];
 464     char *locNamePtr;
 465
 466     for (locIndex = 0; locIndex < locCount; locIndex++) {
 467         errorCode=U_ZERO_ERROR;
 468         strcpy(locName, uloc_getAvailable(locIndex));
 469         locNamePtr = strrchr(locName, '_');
 470         if (locNamePtr) {
 471             *locNamePtr = 0;
 472         }
 473         else {
 474             strcpy(locName, "root");
 475         }
 476
 477         root = ures_openDirect(NULL, locName, &errorCode);
 478         if(U_FAILURE(errorCode)) {
 479             log_err("Can't open %s\n", locName);
 480             continue;
 481         }
 482 */
 483     if (locCount <= 1) {
 484         log_data_err("At least root needs to be installed\n");
 485     }
 486
 487     root = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode);
 488     if(U_FAILURE(errorCode)) {
 489         log_data_err("Can't open structLocale\n");
 490         return;
 491     }
 492     for (locIndex = 0; locIndex < locCount; locIndex++) {
 493         errorCode=U_ZERO_ERROR;
 494         currLoc = uloc_getAvailable(locIndex);
 495         currentLocale = ures_open(NULL, currLoc, &errorCode);
 496         if(errorCode != U_ZERO_ERROR) {
 497             if(U_SUCCESS(errorCode)) {
 498                 /* It's installed, but there is no data.
 499                    It's installed for the g18n white paper [grhoten] */
 500                 log_err("ERROR: Locale %-5s not installed, and it should be!\n",
 501                     uloc_getAvailable(locIndex));
 502             } else {
 503                 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
 504                     u_errorName(errorCode),
 505                     uloc_getAvailable(locIndex));
 506             }
 507             ures_close(currentLocale);
 508             continue;
 509         }
 510         ures_getStringByKey(currentLocale, "Version", NULL, &errorCode);
 511         if(errorCode != U_ZERO_ERROR) {
 512             log_err("No version information is available for locale %s, and it should be!\n",
 513                 currLoc);
 514         }
 515         else if (ures_getStringByKey(currentLocale, "Version", NULL, &errorCode)[0] == (UChar)(0x78)) {
 516             log_verbose("WARNING: The locale %s is experimental! It shouldn't be listed as an installed locale.\n",
 517                 currLoc);
 518         }
 519         resolvedLoc = ures_getLocaleByType(currentLocale, ULOC_ACTUAL_LOCALE, &errorCode);
 520         if (strcmp(resolvedLoc, currLoc) != 0) {
 521             /* All locales have at least a Version resource.
 522                If it's absolutely empty, then the previous test will fail too.*/
 523             log_err("Locale resolves to different locale. Is %s an alias of %s?\n",
 524                 currLoc, resolvedLoc);
 525         }
 526         TestKeyInRootRecursive(root, "root", currentLocale, currLoc);
 527
 528         testLCID(currentLocale, currLoc);
 529
 530         ures_close(currentLocale);
 531     }
 532
 533     ures_close(root);
 534 }
 535
 536 static void
 537 compareArrays(const char *keyName,
 538               UResourceBundle *fromArray, const char *fromLocale,
 539               UResourceBundle *toArray, const char *toLocale,
 540               int32_t start, int32_t end)
 541 {
 542     int32_t fromSize = ures_getSize(fromArray);
 543     int32_t toSize = ures_getSize(fromArray);
 544     int32_t idx;
 545     UErrorCode errorCode = U_ZERO_ERROR;
 546
 547     if (fromSize > toSize) {
 548         fromSize = toSize;
 549         log_err("Arrays are different size from \"%s\" to \"%s\"\n",
 550                 fromLocale,
 551                 toLocale);
 552     }
 553
 554     for (idx = start; idx <= end; idx++) {
 555         const UChar *fromBundleStr = ures_getStringByIndex(fromArray, idx, NULL, &errorCode);
 556         const UChar *toBundleStr = ures_getStringByIndex(toArray, idx, NULL, &errorCode);
 557         if (fromBundleStr && toBundleStr && u_strcmp(fromBundleStr, toBundleStr) != 0)
 558         {
 559             log_err("Difference for %s at index %d from %s= \"%s\" to %s= \"%s\"\n",
 560                     keyName,
 561                     idx,
 562                     fromLocale,
 563                     austrdup(fromBundleStr),
 564                     toLocale,
 565                     austrdup(toBundleStr));
 566         }
 567     }
 568 }
 569
 570 static void
 571 compareConsistentCountryInfo(const char *fromLocale, const char *toLocale) {
 572     UErrorCode errorCode = U_ZERO_ERROR;
 573     UResourceBundle *fromDateTimeElements, *toDateTimeElements, *fromWeekendData = NULL, *toWeekendData = NULL;
 574     UResourceBundle *fromArray, *toArray;
 575     UResourceBundle *fromLocaleBund = ures_open(NULL, fromLocale, &errorCode);
 576     UResourceBundle *toLocaleBund = ures_open(NULL, toLocale, &errorCode);
 577     UResourceBundle *toCalendar, *fromCalendar, *toGregorian, *fromGregorian;
 578
 579     if(U_FAILURE(errorCode)) {
 580         log_err("Can't open resource bundle %s or %s - %s\n", fromLocale, toLocale, u_errorName(errorCode));
 581         return;
 582     }
 583     fromCalendar = ures_getByKey(fromLocaleBund, "calendar", NULL, &errorCode);
 584     fromGregorian = ures_getByKeyWithFallback(fromCalendar, "gregorian", NULL, &errorCode);
 585     fromDateTimeElements = ures_getByKeyWithFallback(fromGregorian, "DateTimeElements", NULL, &errorCode);
 586
 587     toCalendar = ures_getByKey(toLocaleBund, "calendar", NULL, &errorCode);
 588     toGregorian = ures_getByKeyWithFallback(toCalendar, "gregorian", NULL, &errorCode);
 589     toDateTimeElements = ures_getByKeyWithFallback(toGregorian, "DateTimeElements", NULL, &errorCode);
 590
 591     if(U_FAILURE(errorCode)){
 592         log_err("Did not get DateTimeElements from the bundle %s or %s\n", fromLocale, toLocale);
 593         goto cleanup;
 594     }
 595
 596     fromWeekendData = ures_getByKeyWithFallback(fromGregorian, "weekend", NULL, &errorCode);
 597     if(U_FAILURE(errorCode)){
 598         log_err("Did not get weekend data from the bundle %s to compare against %s\n", fromLocale, toLocale);
 599         goto cleanup;
 600     }
 601     toWeekendData = ures_getByKeyWithFallback(toGregorian, "weekend", NULL, &errorCode);
 602     if(U_FAILURE(errorCode)){
 603         log_err("Did not get weekend data from the bundle %s to compare against %s\n", toLocale, fromLocale);
 604         goto cleanup;
 605     }
 606
 607     if (strcmp(fromLocale, "ar_IN") != 0)
 608     {
 609         int32_t fromSize;
 610         int32_t toSize;
 611         int32_t idx;
 612         const int32_t *fromBundleArr = ures_getIntVector(fromDateTimeElements, &fromSize, &errorCode);
 613         const int32_t *toBundleArr = ures_getIntVector(toDateTimeElements, &toSize, &errorCode);
 614
 615         if (fromSize > toSize) {
 616             fromSize = toSize;
 617             log_err("Arrays are different size with key \"DateTimeElements\" from \"%s\" to \"%s\"\n",
 618                     fromLocale,
 619                     toLocale);
 620         }
 621
 622         for (idx = 0; idx < fromSize; idx++) {
 623             if (fromBundleArr[idx] != toBundleArr[idx]) {
 624                 log_err("Difference with key \"DateTimeElements\" at index %d from \"%s\" to \"%s\"\n",
 625                         idx,
 626                         fromLocale,
 627                         toLocale);
 628             }
 629         }
 630     }
 631
 632     /* test for weekend data */
 633     {
 634         int32_t fromSize;
 635         int32_t toSize;
 636         int32_t idx;
 637         const int32_t *fromBundleArr = ures_getIntVector(fromWeekendData, &fromSize, &errorCode);
 638         const int32_t *toBundleArr = ures_getIntVector(toWeekendData, &toSize, &errorCode);
 639
 640         if (fromSize > toSize) {
 641             fromSize = toSize;
 642             log_err("Arrays are different size with key \"weekend\" data from \"%s\" to \"%s\"\n",
 643                     fromLocale,
 644                     toLocale);
 645         }
 646
 647         for (idx = 0; idx < fromSize; idx++) {
 648             if (fromBundleArr[idx] != toBundleArr[idx]) {
 649                 log_err("Difference with key \"weekend\" data at index %d from \"%s\" to \"%s\"\n",
 650                         idx,
 651                         fromLocale,
 652                         toLocale);
 653             }
 654         }
 655     }
 656
 657     fromArray = ures_getByKey(fromLocaleBund, "CurrencyElements", NULL, &errorCode);
 658     toArray = ures_getByKey(toLocaleBund, "CurrencyElements", NULL, &errorCode);
 659     if (strcmp(fromLocale, "en_CA") != 0)
 660     {
 661         /* The first one is probably localized. */
 662         compareArrays("CurrencyElements", fromArray, fromLocale, toArray, toLocale, 1, 2);
 663     }
 664     ures_close(fromArray);
 665     ures_close(toArray);
 666
 667     fromArray = ures_getByKey(fromLocaleBund, "NumberPatterns", NULL, &errorCode);
 668     toArray = ures_getByKey(toLocaleBund, "NumberPatterns", NULL, &errorCode);
 669     if (strcmp(fromLocale, "en_CA") != 0)
 670     {
 671         compareArrays("NumberPatterns", fromArray, fromLocale, toArray, toLocale, 0, 3);
 672     }
 673     ures_close(fromArray);
 674     ures_close(toArray);
 675
 676     /* Difficult to test properly */
 677 /*
 678     fromArray = ures_getByKey(fromLocaleBund, "DateTimePatterns", NULL, &errorCode);
 679     toArray = ures_getByKey(toLocaleBund, "DateTimePatterns", NULL, &errorCode);
 680     {
 681         compareArrays("DateTimePatterns", fromArray, fromLocale, toArray, toLocale);
 682     }
 683     ures_close(fromArray);
 684     ures_close(toArray);*/
 685
 686     fromArray = ures_getByKey(fromLocaleBund, "NumberElements", NULL, &errorCode);
 687     toArray = ures_getByKey(toLocaleBund, "NumberElements", NULL, &errorCode);
 688     if (strcmp(fromLocale, "en_CA") != 0)
 689     {
 690         compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 0, 3);
 691         /* Index 4 is a script based 0 */
 692         compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 5, 10);
 693     }
 694     ures_close(fromArray);
 695     ures_close(toArray);
 696
 697 cleanup:
 698     ures_close(fromDateTimeElements);
 699     ures_close(toDateTimeElements);
 700     ures_close(fromWeekendData);
 701     ures_close(toWeekendData);
 702
 703     ures_close(fromCalendar);
 704     ures_close(toCalendar);
 705     ures_close(fromGregorian);
 706     ures_close(toGregorian);
 707
 708     ures_close(fromLocaleBund);
 709     ures_close(toLocaleBund);
 710 }
 711
 712 static void
 713 TestConsistentCountryInfo(void) {
 714 /*    UResourceBundle *fromLocale, *toLocale;*/
 715     int32_t locCount = uloc_countAvailable();
 716     int32_t fromLocIndex, toLocIndex;
 717
 718     int32_t fromCountryLen, toCountryLen;
 719     char fromCountry[ULOC_FULLNAME_CAPACITY], toCountry[ULOC_FULLNAME_CAPACITY];
 720
 721     int32_t fromVariantLen, toVariantLen;
 722     char fromVariant[ULOC_FULLNAME_CAPACITY], toVariant[ULOC_FULLNAME_CAPACITY];
 723
 724     UErrorCode errorCode = U_ZERO_ERROR;
 725
 726     for (fromLocIndex = 0; fromLocIndex < locCount; fromLocIndex++) {
 727         const char *fromLocale = uloc_getAvailable(fromLocIndex);
 728
 729         errorCode=U_ZERO_ERROR;
 730         fromCountryLen = uloc_getCountry(fromLocale, fromCountry, ULOC_FULLNAME_CAPACITY, &errorCode);
 731         if (fromCountryLen <= 0) {
 732             /* Ignore countryless locales */
 733             continue;
 734         }
 735         fromVariantLen = uloc_getVariant(fromLocale, fromVariant, ULOC_FULLNAME_CAPACITY, &errorCode);
 736         if (fromVariantLen > 0) {
 737             /* Most variants are ignorable like PREEURO, or collation variants. */
 738             continue;
 739         }
 740         /* Start comparing only after the current index.
 741            Previous loop should have already compared fromLocIndex.
 742         */
 743         for (toLocIndex = fromLocIndex + 1; toLocIndex < locCount; toLocIndex++) {
 744             const char *toLocale = uloc_getAvailable(toLocIndex);
 745
 746             toCountryLen = uloc_getCountry(toLocale, toCountry, ULOC_FULLNAME_CAPACITY, &errorCode);
 747             if(U_FAILURE(errorCode)) {
 748                 log_err("Unknown failure fromLocale=%s toLocale=%s errorCode=%s\n",
 749                     fromLocale, toLocale, u_errorName(errorCode));
 750                 continue;
 751             }
 752
 753             if (toCountryLen <= 0) {
 754                 /* Ignore countryless locales */
 755                 continue;
 756             }
 757             toVariantLen = uloc_getVariant(toLocale, toVariant, ULOC_FULLNAME_CAPACITY, &errorCode);
 758             if (toVariantLen > 0) {
 759                 /* Most variants are ignorable like PREEURO, or collation variants. */
 760                 /* They're a variant for a reason. */
 761                 continue;
 762             }
 763             if (strcmp(fromCountry, toCountry) == 0) {
 764                 log_verbose("comparing fromLocale=%s toLocale=%s\n",
 765                     fromLocale, toLocale);
 766                 compareConsistentCountryInfo(fromLocale, toLocale);
 767             }
 768         }
 769     }
 770 }
 771
 772 static int32_t
 773 findStringSetMismatch(const char *currLoc, const UChar *string, int32_t langSize,
 774                       const UChar *exemplarCharacters, int32_t exemplarLen,
 775                       UBool ignoreNumbers) {
 776     UErrorCode errorCode = U_ZERO_ERROR;
 777     USet *origSet = uset_openPatternOptions(exemplarCharacters, exemplarLen, USET_CASE_INSENSITIVE, &errorCode);
 778     USet *exemplarSet = createFlattenSet(origSet, &errorCode);
 779     int32_t strIdx;
 780     uset_close(origSet);
 781     if (U_FAILURE(errorCode)) {
 782         log_err("%s: error uset_openPattern returned %s\n", currLoc, u_errorName(errorCode));
 783         return -1;
 784     }
 785
 786     for (strIdx = 0; strIdx < langSize; strIdx++) {
 787         if (!uset_contains(exemplarSet, string[strIdx])
 788             && string[strIdx] != 0x0020 && string[strIdx] != 0x00A0 && string[strIdx] != 0x002e && string[strIdx] != 0x002c && string[strIdx] != 0x002d && string[strIdx] != 0x0027
 789             && string[strIdx] != 0x200C && string[strIdx] != 0x200D) {
 790             if (!ignoreNumbers || (ignoreNumbers && (string[strIdx] < 0x30 || string[strIdx] > 0x39))) {
 791                 uset_close(exemplarSet);
 792                 return strIdx;
 793             }
 794         }
 795     }
 796     uset_close(exemplarSet);
 797     return -1;
 798 }
 799 /* include non-invariant chars */
 800 static int32_t
 801 myUCharsToChars(const UChar* us, char* cs, int32_t len){
 802     int32_t i=0;
 803     for(; i< len; i++){
 804         if(us[i] < 0x7f){
 805             cs[i] = (char)us[i];
 806         }else{
 807             return -1;
 808         }
 809     }
 810     return i;
 811 }
 812 static void
 813 findSetMatch( UScriptCode *scriptCodes, int32_t scriptsLen,
 814               USet *exemplarSet,
 815               const char  *locale){
 816     USet *scripts[10]= {0};
 817     char pattern[256] = { '[', ':', 0x000 };
 818     int32_t patternLen;
 819     UChar uPattern[256] = {0};
 820     UErrorCode status = U_ZERO_ERROR;
 821     int32_t i;
 822
 823     /* create the sets with script codes */
 824     for(i = 0; i<scriptsLen; i++){
 825         strcat(pattern, uscript_getShortName(scriptCodes[i]));
 826         strcat(pattern, ":]");
 827         patternLen = (int32_t)strlen(pattern);
 828         u_charsToUChars(pattern, uPattern, patternLen);
 829         scripts[i] = uset_openPattern(uPattern, patternLen, &status);
 830         if(U_FAILURE(status)){
 831             log_err("Could not create set for pattern %s. Error: %s\n", pattern, u_errorName(status));
 832             return;
 833         }
 834         pattern[2] = 0;
 835     }
 836     if (strcmp(locale, "uk") == 0 || strcmp(locale, "uk_UA") == 0) {
 837         /* Special addition. Add the modifying apostrophe, which isn't in Cyrillic. */
 838         uset_add(scripts[0], 0x2bc);
 839     }
 840     if(U_SUCCESS(status)){
 841         UBool existsInScript = FALSE;
 842         /* iterate over the exemplarSet and ascertain if all
 843          * UChars in exemplarSet belong to the scripts returned
 844          * by getScript
 845          */
 846         int32_t count = uset_getItemCount(exemplarSet);
 847
 848         for( i=0; i < count; i++){
 849             UChar32 start = 0;
 850             UChar32 end = 0;
 851             UChar *str = NULL;
 852             int32_t strCapacity = 0;
 853
 854             strCapacity = uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status);
 855             if(U_SUCCESS(status)){
 856                 int32_t j;
 857                 if(strCapacity == 0){
 858                     /* ok the item is a range */
 859                      for( j = 0; j < scriptsLen; j++){
 860                         if(uset_containsRange(scripts[j], start, end) == TRUE){
 861                             existsInScript = TRUE;
 862                         }
 863                     }
 864                     if(existsInScript == FALSE){
 865                         for( j = 0; j < scriptsLen; j++){
 866                             UChar toPattern[500]={'\0'};
 867                             char pat[500]={'\0'};
 868                             int32_t len = uset_toPattern(scripts[j], toPattern, 500, TRUE, &status);
 869                             len = myUCharsToChars(toPattern, pat, len);
 870                             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));
 871                             if(len!=-1){
 872                                 log_err("Pattern: %s\n",pat);
 873                             }
 874                         }
 875                         log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale);
 876                     }
 877                 }else{
 878                     strCapacity++; /* increment for NUL termination */
 879                     /* allocate the str and call the api again */
 880                     str = (UChar*) malloc(U_SIZEOF_UCHAR * strCapacity);
 881                     strCapacity =  uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status);
 882                     /* iterate over the scripts and figure out if the string contained is actually
 883                      * in the script set
 884                      */
 885                     for( j = 0; j < scriptsLen; j++){
 886                         if(uset_containsString(scripts[j],str, strCapacity) == TRUE){
 887                             existsInScript = TRUE;
 888                         }
 889                     }
 890                     if(existsInScript == FALSE){
 891                         log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale);
 892                     }
 893                 }
 894             }
 895         }
 896
 897     }
 898
 899     /* close the sets */
 900     for(i = 0; i<scriptsLen; i++){
 901         uset_close(scripts[i]);
 902     }
 903 }
 904
 905 static void VerifyTranslation(void) {
 906     UResourceBundle *root, *currentLocale;
 907     int32_t locCount = uloc_countAvailable();
 908     int32_t locIndex;
 909     UErrorCode errorCode = U_ZERO_ERROR;
 910     int32_t exemplarLen;
 911     const UChar *exemplarCharacters;
 912     const char *currLoc;
 913     UScriptCode scripts[USCRIPT_CODE_LIMIT];
 914     int32_t numScripts;
 915     int32_t idx;
 916     int32_t end;
 917     UResourceBundle *resArray;
 918
 919     if (locCount <= 1) {
 920         log_data_err("At least root needs to be installed\n");
 921     }
 922
 923     root = ures_openDirect(NULL, "root", &errorCode);
 924     if(U_FAILURE(errorCode)) {
 925         log_data_err("Can't open root\n");
 926         return;
 927     }
 928     for (locIndex = 0; locIndex < locCount; locIndex++) {
 929         errorCode=U_ZERO_ERROR;
 930         currLoc = uloc_getAvailable(locIndex);
 931         currentLocale = ures_open(NULL, currLoc, &errorCode);
 932         if(errorCode != U_ZERO_ERROR) {
 933             if(U_SUCCESS(errorCode)) {
 934                 /* It's installed, but there is no data.
 935                    It's installed for the g18n white paper [grhoten] */
 936                 log_err("ERROR: Locale %-5s not installed, and it should be!\n",
 937                     uloc_getAvailable(locIndex));
 938             } else {
 939                 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
 940                     u_errorName(errorCode),
 941                     uloc_getAvailable(locIndex));
 942             }
 943             ures_close(currentLocale);
 944             continue;
 945         }
 946         exemplarCharacters = ures_getStringByKey(currentLocale, "ExemplarCharacters", &exemplarLen, &errorCode);
 947         if (U_FAILURE(errorCode)) {
 948             log_err("error ures_getStringByKey returned %s\n", u_errorName(errorCode));
 949         }
 950         else if (QUICK && exemplarLen > 2048) {
 951             log_verbose("skipping test for %s\n", currLoc);
 952         }
 953         else {
 954             UChar langBuffer[128];
 955             int32_t langSize;
 956             int32_t strIdx;
 957             langSize = uloc_getDisplayLanguage(currLoc, currLoc, langBuffer, sizeof(langBuffer)/sizeof(langBuffer[0]), &errorCode);
 958             if (U_FAILURE(errorCode)) {
 959                 log_err("error uloc_getDisplayLanguage returned %s\n", u_errorName(errorCode));
 960             }
 961             else {
 962                 strIdx = findStringSetMismatch(currLoc, langBuffer, langSize, exemplarCharacters, exemplarLen, FALSE);
 963                 if (strIdx >= 0) {
 964                     log_err("getDisplayLanguage(%s) at index %d returned characters not in the exemplar characters.\n",
 965                         currLoc, strIdx);
 966                 }
 967             }
 968             langSize = uloc_getDisplayCountry(currLoc, currLoc, langBuffer, sizeof(langBuffer)/sizeof(langBuffer[0]), &errorCode);
 969             if (U_FAILURE(errorCode)) {
 970                 log_err("error uloc_getDisplayCountry returned %s\n", u_errorName(errorCode));
 971             }
 972             else {
 973               strIdx = findStringSetMismatch(currLoc, langBuffer, langSize, exemplarCharacters, exemplarLen, FALSE);
 974                 if (strIdx >= 0) {
 975                     log_err("getDisplayCountry(%s) at index %d returned characters not in the exemplar characters.\n",
 976                         currLoc, strIdx);
 977                 }
 978             }
 979             {
 980                 UResourceBundle* cal = ures_getByKey(currentLocale, "calendar", NULL, &errorCode);
 981                 UResourceBundle* greg = ures_getByKeyWithFallback(cal, "gregorian", NULL, &errorCode);
 982                 UResourceBundle* names = ures_getByKeyWithFallback(greg,  "dayNames", NULL, &errorCode);
 983                 UResourceBundle* format = ures_getByKeyWithFallback(names,  "format", NULL, &errorCode);
 984                 resArray = ures_getByKeyWithFallback(format,  "wide", NULL, &errorCode);
 985
 986                 if (U_FAILURE(errorCode)) {
 987                     log_err("error ures_getByKey returned %s\n", u_errorName(errorCode));
 988                 }
 989                 if (QUICK) {
 990                     end = 1;
 991                 }
 992                 else {
 993                     end = ures_getSize(resArray);
 994                 }
 995
 996
 997                 for (idx = 0; idx < end; idx++) {
 998                     const UChar *fromBundleStr = ures_getStringByIndex(resArray, idx, &langSize, &errorCode);
 999                     if (U_FAILURE(errorCode)) {
1000                         log_err("error ures_getStringByIndex(%d) returned %s\n", idx, u_errorName(errorCode));
1001                         continue;
1002                     }
1003                     strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, exemplarCharacters, exemplarLen, TRUE);
1004                     if (strIdx >= 0) {
1005                         log_err("getDayNames(%s, %d) at index %d returned characters not in the exemplar characters.\n",
1006                             currLoc, idx, strIdx);
1007                     }
1008                 }
1009                 ures_close(resArray);
1010                 ures_close(format);
1011                 ures_close(names);
1012
1013                 names = ures_getByKeyWithFallback(greg, "monthNames", NULL, &errorCode);
1014                 format = ures_getByKeyWithFallback(names,"format", NULL, &errorCode);
1015                 resArray = ures_getByKeyWithFallback(format, "wide", NULL, &errorCode);
1016                 if (U_FAILURE(errorCode)) {
1017                     log_err("error ures_getByKey returned %s\n", u_errorName(errorCode));
1018                 }
1019                 if (QUICK) {
1020                     end = 1;
1021                 }
1022                 else {
1023                     end = ures_getSize(resArray);
1024                 }
1025
1026                 for (idx = 0; idx < end; idx++) {
1027                     const UChar *fromBundleStr = ures_getStringByIndex(resArray, idx, &langSize, &errorCode);
1028                     if (U_FAILURE(errorCode)) {
1029                         log_err("error ures_getStringByIndex(%d) returned %s\n", idx, u_errorName(errorCode));
1030                         continue;
1031                     }
1032                     strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, exemplarCharacters, exemplarLen, TRUE);
1033                     if (strIdx >= 0) {
1034                         log_err("getMonthNames(%s, %d) at index %d returned characters not in the exemplar characters.\n",
1035                             currLoc, idx, strIdx);
1036                     }
1037                 }
1038                 ures_close(resArray);
1039                 ures_close(format);
1040                 ures_close(names);
1041                 ures_close(greg);
1042                 ures_close(cal);
1043             }
1044             errorCode = U_ZERO_ERROR;
1045             numScripts = uscript_getCode(currLoc, scripts, sizeof(scripts)/sizeof(scripts[0]), &errorCode);
1046             if (numScripts == 0) {
1047                 log_err("uscript_getCode(%s) doesn't work.\n", currLoc);
1048             }else if(scripts[0] == USCRIPT_COMMON){
1049                 log_err("uscript_getCode(%s) returned USCRIPT_COMMON.\n", currLoc);
1050             }
1051
1052             /* test that the scripts are a superset of exemplar characters. */
1053            {
1054                 ULocaleData *uld = ulocdata_open(currLoc,&errorCode);
1055                 USet *exemplarSet =  ulocdata_getExemplarSet(uld, NULL, 0, ULOCDATA_ES_STANDARD, &errorCode);
1056                 /* test if exemplar characters are part of script code */
1057                 findSetMatch(scripts, numScripts, exemplarSet, currLoc);
1058                 uset_close(exemplarSet);
1059                 ulocdata_close(uld);
1060             }
1061
1062            /* test that the paperSize API works */
1063            {
1064                int32_t height=0, width=0;
1065                ulocdata_getPaperSize(currLoc, &height, &width, &errorCode);
1066                if(U_FAILURE(errorCode)){
1067                    log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1068                }
1069                if(strstr(currLoc, "_US")!=NULL && height != 279 && width != 216 ){
1070                    log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc);
1071                }
1072            }
1073             /* test that the MeasurementSystem works API works */
1074            {
1075                UMeasurementSystem measurementSystem = ulocdata_getMeasurementSystem(currLoc, &errorCode);
1076                if(U_FAILURE(errorCode)){
1077                    log_err("ulocdata_getMeasurementSystem failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1078                }
1079                if(strstr(currLoc, "_US")!=NULL){
1080                    if(measurementSystem != UMS_US){
1081                         log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1082                    }
1083                }else if(measurementSystem != UMS_SI){
1084                    log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1085                }
1086            }
1087         }
1088         ures_close(currentLocale);
1089     }
1090
1091     ures_close(root);
1092 }
1093
1094 /* adjust this limit as appropriate */
1095 #define MAX_SCRIPTS_PER_LOCALE 8
1096
1097 static void TestExemplarSet(void){
1098     int32_t i, j, k, m, n;
1099     int32_t equalCount = 0;
1100     UErrorCode ec = U_ZERO_ERROR;
1101     UEnumeration* avail;
1102     USet* exemplarSets[2];
1103     USet* unassignedSet;
1104     UScriptCode code[MAX_SCRIPTS_PER_LOCALE];
1105     USet* codeSets[MAX_SCRIPTS_PER_LOCALE];
1106     int32_t codeLen;
1107     char cbuf[32]; /* 9 should be enough */
1108     UChar ubuf[64]; /* adjust as needed */
1109     UBool existsInScript;
1110     int32_t itemCount;
1111     int32_t strLen;
1112     UChar32 start, end;
1113
1114     unassignedSet = NULL;
1115     exemplarSets[0] = NULL;
1116     exemplarSets[1] = NULL;
1117     for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) {
1118         codeSets[i] = NULL;
1119     }
1120
1121     avail = ures_openAvailableLocales(NULL, &ec);
1122     if (!assertSuccess("ures_openAvailableLocales", &ec)) goto END;
1123     n = uenum_count(avail, &ec);
1124     if (!assertSuccess("uenum_count", &ec)) goto END;
1125
1126     u_uastrcpy(ubuf, "[:unassigned:]");
1127     unassignedSet = uset_openPattern(ubuf, -1, &ec);
1128     if (!assertSuccess("uset_openPattern", &ec)) goto END;
1129
1130     for(i=0; i<n; i++){
1131         const char* locale = uenum_next(avail, NULL, &ec);
1132         if (!assertSuccess("uenum_next", &ec)) goto END;
1133         log_verbose("%s\n", locale);
1134         for (k=0; k<2; ++k) {
1135             uint32_t option = (k==0) ? 0 : USET_CASE_INSENSITIVE;
1136             ULocaleData *uld = ulocdata_open(locale,&ec);
1137             USet* exemplarSet = ulocdata_getExemplarSet(uld,NULL, option, ULOCDATA_ES_STANDARD, &ec);
1138             uset_close(exemplarSets[k]);
1139             ulocdata_close(uld);
1140             exemplarSets[k] = exemplarSet;
1141             if (!assertSuccess("ulocaledata_getExemplarSet", &ec)) goto END;
1142
1143             if (uset_containsSome(exemplarSet, unassignedSet)) {
1144                 log_err("ExemplarSet contains unassigned characters for locale : %s\n", locale);
1145             }
1146             codeLen = uscript_getCode(locale, code, 8, &ec);
1147             if (!assertSuccess("uscript_getCode", &ec)) goto END;
1148
1149             for (j=0; j<MAX_SCRIPTS_PER_LOCALE; ++j) {
1150                 uset_close(codeSets[j]);
1151                 codeSets[j] = NULL;
1152             }
1153             for (j=0; j<codeLen; ++j) {
1154                 uprv_strcpy(cbuf, "[:");
1155                 if(code[j]==-1){
1156                     log_err("USCRIPT_INVALID_CODE returned for locale: %s\n", locale);
1157                     continue;
1158                 }
1159                 uprv_strcat(cbuf, uscript_getShortName(code[j]));
1160                 uprv_strcat(cbuf, ":]");
1161                 u_uastrcpy(ubuf, cbuf);
1162                 codeSets[j] = uset_openPattern(ubuf, -1, &ec);
1163             }
1164             if (!assertSuccess("uset_openPattern", &ec)) goto END;
1165
1166             existsInScript = FALSE;
1167             itemCount = uset_getItemCount(exemplarSet);
1168             for (m=0; m<itemCount && !existsInScript; ++m) {
1169                 strLen = uset_getItem(exemplarSet, m, &start, &end, ubuf,
1170                                       sizeof(ubuf)/sizeof(ubuf[0]), &ec);
1171                 /* failure here might mean str[] needs to be larger */
1172                 if (!assertSuccess("uset_getItem", &ec)) goto END;
1173                 if (strLen == 0) {
1174                     for (j=0; j<codeLen; ++j) {
1175                         if (codeSets[j]!=NULL && uset_containsRange(codeSets[j], start, end)) {
1176                             existsInScript = TRUE;
1177                             break;
1178                         }
1179                     }
1180                 } else {
1181                     for (j=0; j<codeLen; ++j) {
1182                         if (codeSets[j]!=NULL && uset_containsString(codeSets[j], ubuf, strLen)) {
1183                             existsInScript = TRUE;
1184                             break;
1185                         }
1186                     }
1187                 }
1188             }
1189
1190             if (existsInScript == FALSE){
1191                 log_err("ExemplarSet containment failed for locale : %s\n", locale);
1192             }
1193         }
1194         assertTrue("case-folded is a superset",
1195                    uset_containsAll(exemplarSets[1], exemplarSets[0]));
1196         if (uset_equals(exemplarSets[1], exemplarSets[0])) {
1197             ++equalCount;
1198         }
1199     }
1200     /* Note: The case-folded set should sometimes be a strict superset
1201        and sometimes be equal. */
1202     assertTrue("case-folded is sometimes a strict superset, and sometimes equal",
1203                equalCount > 0 && equalCount < n);
1204
1205  END:
1206     uenum_close(avail);
1207     uset_close(exemplarSets[0]);
1208     uset_close(exemplarSets[1]);
1209     uset_close(unassignedSet);
1210     for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) {
1211         uset_close(codeSets[i]);
1212     }
1213 }
1214
1215 static void TestCoverage(void){
1216     ULocaleDataDelimiterType types[] = {
1217      ULOCDATA_QUOTATION_START,     /* Quotation start */
1218      ULOCDATA_QUOTATION_END,       /* Quotation end */
1219      ULOCDATA_ALT_QUOTATION_START, /* Alternate quotation start */
1220      ULOCDATA_ALT_QUOTATION_END,   /* Alternate quotation end */
1221      ULOCDATA_DELIMITER_COUNT
1222     };
1223     int i;
1224     UBool sub;
1225     UErrorCode status = U_ZERO_ERROR;
1226     ULocaleData *uld = ulocdata_open(uloc_getDefault(), &status);
1227
1228     if(U_FAILURE(status)){
1229         log_err("ulocdata_open error");
1230         return;
1231     }
1232
1233
1234     for(i = 0; i < ULOCDATA_DELIMITER_COUNT; i++){
1235         UChar result[32] = {0,};
1236         status = U_ZERO_ERROR;
1237         ulocdata_getDelimiter(uld, types[i], result, 32, &status);
1238         if (U_FAILURE(status)){
1239             log_err("ulocdata_getgetDelimiter error with type %d", types[i]);
1240         }
1241     }
1242
1243     sub = ulocdata_getNoSubstitute(uld);
1244     ulocdata_setNoSubstitute(uld,sub);
1245     ulocdata_close(uld);
1246 }
1247
1248 static void TestCurrencyList(void){
1249 #if !UCONFIG_NO_FORMATTING
1250     UErrorCode errorCode = U_ZERO_ERROR;
1251     int32_t structLocaleCount, currencyCount;
1252     UEnumeration *en = ucurr_openISOCurrencies(UCURR_ALL, &errorCode);
1253     const char *isoCode, *structISOCode;
1254     UResourceBundle *subBundle;
1255     UResourceBundle *currencies = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode);
1256     if(U_FAILURE(errorCode)) {
1257         log_data_err("Can't open structLocale\n");
1258         return;
1259     }
1260     currencies = ures_getByKey(currencies, "Currencies", currencies, &errorCode);
1261     currencyCount = uenum_count(en, &errorCode);
1262     structLocaleCount = ures_getSize(currencies);
1263     if (currencyCount != structLocaleCount) {
1264         log_err("structLocale(%d) and ISO4217(%d) currency list are out of sync.\n", structLocaleCount, currencyCount);
1265 #if U_CHARSET_FAMILY == U_ASCII_FAMILY
1266         ures_resetIterator(currencies);
1267         while ((isoCode = uenum_next(en, NULL, &errorCode)) != NULL && ures_hasNext(currencies)) {
1268             subBundle = ures_getNextResource(currencies, NULL, &errorCode);
1269             structISOCode = ures_getKey(subBundle);
1270             ures_close(subBundle);
1271             if (strcmp(structISOCode, isoCode) != 0) {
1272                 log_err("First difference found at structLocale(%s) and ISO4217(%s).\n", structISOCode, isoCode);
1273                 break;
1274             }
1275         }
1276 #endif
1277     }
1278     ures_close(currencies);
1279     uenum_close(en);
1280 #endif
1281 }
1282
1283 #define TESTCASE(name) addTest(root, &name, "tsutil/cldrtest/" #name)
1284
1285 void addCLDRTest(TestNode** root);
1286
1287 void addCLDRTest(TestNode** root)
1288 {
1289     TESTCASE(TestLocaleStructure);
1290     TESTCASE(TestConsistentCountryInfo);
1291     TESTCASE(VerifyTranslation);
1292     TESTCASE(TestExemplarSet);
1293     TESTCASE(TestCurrencyList);
1294     TESTCASE(TestCoverage);
1295 }