icuSources/test/cintltst/cldrtest.c

   1 /********************************************************************
   2  * COPYRIGHT:
   3  * Copyright (c) 1997-2013, 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/uloc.h"
  15 #include "unicode/ulocdata.h"
  16 #include "cstring.h"
  17 #include "locmap.h"
  18 #include "uresimp.h"
  19
  20 /*
  21 returns a new UnicodeSet that is a flattened form of the original
  22 UnicodeSet.
  23 */
  24 static USet*
  25 createFlattenSet(USet *origSet, UErrorCode *status) {
  26
  27
  28     USet *newSet = NULL;
  29     int32_t origItemCount = 0;
  30     int32_t idx, graphmeSize;
  31     UChar32 start, end;
  32     UChar graphme[64];
  33     if (U_FAILURE(*status)) {
  34         log_err("createFlattenSet called with %s\n", u_errorName(*status));
  35         return NULL;
  36     }
  37     newSet = uset_open(1, 0);
  38     origItemCount = uset_getItemCount(origSet);
  39     for (idx = 0; idx < origItemCount; idx++) {
  40         graphmeSize = uset_getItem(origSet, idx,
  41             &start, &end,
  42             graphme, (int32_t)(sizeof(graphme)/sizeof(graphme[0])),
  43             status);
  44         if (U_FAILURE(*status)) {
  45             log_err("ERROR: uset_getItem returned %s\n", u_errorName(*status));
  46             *status = U_ZERO_ERROR;
  47         }
  48         if (graphmeSize) {
  49             uset_addAllCodePoints(newSet, graphme, graphmeSize);
  50         }
  51         else {
  52             uset_addRange(newSet, start, end);
  53         }
  54     }
  55     uset_closeOver(newSet,USET_CASE_INSENSITIVE);
  56     return newSet;
  57 }
  58
  59 static UBool
  60 isCurrencyPreEuro(const char* currencyKey){
  61     if( strcmp(currencyKey, "PTE") == 0 ||
  62         strcmp(currencyKey, "ESP") == 0 ||
  63         strcmp(currencyKey, "LUF") == 0 ||
  64         strcmp(currencyKey, "GRD") == 0 ||
  65         strcmp(currencyKey, "BEF") == 0 ||
  66         strcmp(currencyKey, "ITL") == 0 ||
  67         strcmp(currencyKey, "EEK") == 0){
  68             return TRUE;
  69     }
  70     return FALSE;
  71 }
  72 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
  73 static void
  74 TestKeyInRootRecursive(UResourceBundle *root, const char *rootName,
  75                        UResourceBundle *currentBundle, const char *locale) {
  76     UErrorCode errorCode = U_ZERO_ERROR;
  77     UResourceBundle *subRootBundle = NULL, *subBundle = NULL, *arr = NULL;
  78
  79     ures_resetIterator(root);
  80     ures_resetIterator(currentBundle);
  81     while (ures_hasNext(currentBundle)) {
  82         const char *subBundleKey = NULL;
  83         const char *currentBundleKey = NULL;
  84
  85         errorCode = U_ZERO_ERROR;
  86         currentBundleKey = ures_getKey(currentBundle);
  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 /*                if (sameArray && strcmp(rootName, "root") == 0) {
 319                     log_err("Arrays are the same with key \"%s\" in \"%s\" from root for locale \"%s\"\n",
 320                             subBundleKey,
 321                             ures_getKey(currentBundle),
 322                             locale);
 323                 }*/
 324             }
 325             ures_close(subSubBundle);
 326             ures_close(subSubRootBundle);
 327         }
 328         else if (ures_getType(subBundle) == URES_STRING) {
 329             int32_t len = 0;
 330             const UChar *string = ures_getString(subBundle, &len, &errorCode);
 331             if (U_FAILURE(errorCode) || string == NULL) {
 332                 log_err("Can't open a string with key \"%s\" in \"%s\" for locale \"%s\"\n",
 333                         subBundleKey,
 334                         ures_getKey(currentBundle),
 335                         locale);
 336             } else if (string[0] == (UChar)0x20) {
 337                 log_err("key \"%s\" in \"%s\" starts with a space in locale \"%s\"\n",
 338                         subBundleKey,
 339                         ures_getKey(currentBundle),
 340                         locale);
 341             /* localeDisplayPattern/separator can end with a space */
 342             } else if (string[len - 1] == (UChar)0x20 && (strcmp(subBundleKey,"separator"))) {
 343                 log_err("key \"%s\" in \"%s\" ends with a space in locale \"%s\"\n",
 344                         subBundleKey,
 345                         ures_getKey(currentBundle),
 346                         locale);
 347             } else if (strcmp(subBundleKey, "localPatternChars") == 0) {
 348                 /* Note: We no longer import localPatternChars data starting
 349                  * ICU 3.8.  So it never comes into this else if block. (ticket#5597)
 350                  */
 351
 352                 /* Check well-formedness of localPatternChars.  First, the
 353                  * length must match the number of fields defined by
 354                  * DateFormat.  Second, each character in the string must
 355                  * be in the set [A-Za-z].  Finally, each character must be
 356                  * unique.
 357                  */
 358                 int32_t i,j;
 359 #if !UCONFIG_NO_FORMATTING
 360                 if (len != UDAT_FIELD_COUNT) {
 361                     log_err("key \"%s\" has the wrong number of characters in locale \"%s\"\n",
 362                             subBundleKey,
 363                             locale);
 364                 }
 365 #endif
 366                 /* Check char validity. */
 367                 for (i=0; i<len; ++i) {
 368                     if (!((string[i] >= 65/*'A'*/ && string[i] <= 90/*'Z'*/) ||
 369                           (string[i] >= 97/*'a'*/ && string[i] <= 122/*'z'*/))) {
 370                         log_err("key \"%s\" has illegal character '%c' in locale \"%s\"\n",
 371                                 subBundleKey,
 372                                 (char) string[i],
 373                                 locale);
 374                     }
 375                     /* Do O(n^2) check for duplicate chars. */
 376                     for (j=0; j<i; ++j) {
 377                         if (string[j] == string[i]) {
 378                             log_err("key \"%s\" has duplicate character '%c' in locale \"%s\"\n",
 379                                     subBundleKey,
 380                                     (char) string[i],
 381                                     locale);
 382                         }
 383                     }
 384                 }
 385             }
 386             /* No fallback was done. Check for duplicate data */
 387             /* The ures_* API does not do fallback of sub-resource bundles,
 388                So we can't do this now. */
 389 #if 0
 390             else if (strcmp(locale, "root") != 0 && errorCode == U_ZERO_ERROR) {
 391
 392                 const UChar *rootString = ures_getString(subRootBundle, &len, &errorCode);
 393                 if (U_FAILURE(errorCode) || rootString == NULL) {
 394                     log_err("Can't open a string with key \"%s\" in \"%s\" in root\n",
 395                             ures_getKey(subRootBundle),
 396                             ures_getKey(currentBundle));
 397                     continue;
 398                 } else if (u_strcmp(string, rootString) == 0) {
 399                     if (strcmp(locale, "de_CH") != 0 && strcmp(subBundleKey, "Countries") != 0 &&
 400                         strcmp(subBundleKey, "Version") != 0) {
 401                         log_err("Found duplicate data with key \"%s\" in \"%s\" in locale \"%s\"\n",
 402                                 ures_getKey(subRootBundle),
 403                                 ures_getKey(currentBundle),
 404                                 locale);
 405                     }
 406                     else {
 407                         /* Ignore for now. */
 408                         /* Can be fixed if fallback through de locale was done. */
 409                         log_verbose("Skipping key %s in %s\n", subBundleKey, locale);
 410                     }
 411                 }
 412             }
 413 #endif
 414         }
 415         else if (ures_getType(subBundle) == URES_TABLE) {
 416             if (strcmp(subBundleKey, "availableFormats")!=0) {
 417                 /* Here is one of the recursive parts */
 418                 TestKeyInRootRecursive(subRootBundle, rootName, subBundle, locale);
 419             }
 420             else {
 421                 log_verbose("Skipping key %s in %s\n", subBundleKey, locale);
 422             }
 423         }
 424         else if (ures_getType(subBundle) == URES_BINARY || ures_getType(subBundle) == URES_INT) {
 425             /* Can't do anything to check it */
 426             /* We'll assume it's all correct */
 427             if (strcmp(subBundleKey, "MeasurementSystem") != 0) {
 428                 log_verbose("Skipping key \"%s\" in \"%s\" for locale \"%s\"\n",
 429                         subBundleKey,
 430                         ures_getKey(currentBundle),
 431                         locale);
 432             }
 433             /* Testing for MeasurementSystem is done in VerifyTranslation */
 434         }
 435         else {
 436             log_err("Type %d for key \"%s\" in \"%s\" is unknown for locale \"%s\"\n",
 437                     ures_getType(subBundle),
 438                     subBundleKey,
 439                     ures_getKey(currentBundle),
 440                     locale);
 441         }
 442         ures_close(subRootBundle);
 443         ures_close(subBundle);
 444     }
 445 }
 446 #endif
 447
 448 static void
 449 testLCID(UResourceBundle *currentBundle,
 450          const char *localeName)
 451 {
 452     UErrorCode status = U_ZERO_ERROR;
 453     uint32_t expectedLCID;
 454     char lcidStringC[64] = {0};
 455
 456     expectedLCID = uloc_getLCID(localeName);
 457     if (expectedLCID == 0) {
 458         log_verbose("INFO:    %-5s does not have any LCID mapping\n",
 459             localeName);
 460         return;
 461     }
 462
 463     status = U_ZERO_ERROR;
 464     uprv_strcpy(lcidStringC, uprv_convertToPosix(expectedLCID, &status));
 465     if (U_FAILURE(status)) {
 466         log_err("ERROR:   %.4x does not have a POSIX mapping due to %s\n",
 467             expectedLCID, u_errorName(status));
 468     }
 469
 470     if(strcmp(localeName, lcidStringC) != 0) {
 471         char langName[1024];
 472         char langLCID[1024];
 473         uloc_getLanguage(localeName, langName, sizeof(langName), &status);
 474         uloc_getLanguage(lcidStringC, langLCID, sizeof(langLCID), &status);
 475
 476         if (strcmp(langName, langLCID) == 0) {
 477             log_verbose("WARNING: %-5s resolves to %s (0x%.4x)\n",
 478                 localeName, lcidStringC, expectedLCID);
 479         }
 480         else {
 481             log_err("ERROR:   %-5s has 0x%.4x and the number resolves wrongfully to %s\n",
 482                 localeName, expectedLCID, lcidStringC);
 483         }
 484     }
 485 }
 486
 487 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
 488 static void
 489 TestLocaleStructure(void) {
 490     UResourceBundle *root, *currentLocale;
 491     int32_t locCount = uloc_countAvailable();
 492     int32_t locIndex;
 493     UErrorCode errorCode = U_ZERO_ERROR;
 494     const char *currLoc, *resolvedLoc;
 495
 496     /* TODO: Compare against parent's data too. This code can't handle fallbacks that some tools do already. */
 497 /*    char locName[ULOC_FULLNAME_CAPACITY];
 498     char *locNamePtr;
 499
 500     for (locIndex = 0; locIndex < locCount; locIndex++) {
 501         errorCode=U_ZERO_ERROR;
 502         strcpy(locName, uloc_getAvailable(locIndex));
 503         locNamePtr = strrchr(locName, '_');
 504         if (locNamePtr) {
 505             *locNamePtr = 0;
 506         }
 507         else {
 508             strcpy(locName, "root");
 509         }
 510
 511         root = ures_openDirect(NULL, locName, &errorCode);
 512         if(U_FAILURE(errorCode)) {
 513             log_err("Can't open %s\n", locName);
 514             continue;
 515         }
 516 */
 517     if (locCount <= 1) {
 518         log_data_err("At least root needs to be installed\n");
 519     }
 520
 521     root = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode);
 522     if(U_FAILURE(errorCode)) {
 523         log_data_err("Can't open structLocale\n");
 524         return;
 525     }
 526     for (locIndex = 0; locIndex < locCount; locIndex++) {
 527         errorCode=U_ZERO_ERROR;
 528         currLoc = uloc_getAvailable(locIndex);
 529         currentLocale = ures_open(NULL, currLoc, &errorCode);
 530         if(errorCode != U_ZERO_ERROR) {
 531             if(U_SUCCESS(errorCode)) {
 532                 /* It's installed, but there is no data.
 533                    It's installed for the g18n white paper [grhoten] */
 534                 log_err("ERROR: Locale %-5s not installed, and it should be, err %s\n",
 535                     uloc_getAvailable(locIndex), u_errorName(errorCode));
 536             } else {
 537                 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
 538                     u_errorName(errorCode),
 539                     uloc_getAvailable(locIndex));
 540             }
 541             ures_close(currentLocale);
 542             continue;
 543         }
 544         ures_getStringByKey(currentLocale, "Version", NULL, &errorCode);
 545         if(errorCode != U_ZERO_ERROR) {
 546             log_err("No version information is available for locale %s, and it should be!\n",
 547                 currLoc);
 548         }
 549         else if (ures_getStringByKey(currentLocale, "Version", NULL, &errorCode)[0] == (UChar)(0x78)) {
 550             log_verbose("WARNING: The locale %s is experimental! It shouldn't be listed as an installed locale.\n",
 551                 currLoc);
 552         }
 553         resolvedLoc = ures_getLocaleByType(currentLocale, ULOC_ACTUAL_LOCALE, &errorCode);
 554         if (strcmp(resolvedLoc, currLoc) != 0) {
 555             /* All locales have at least a Version resource.
 556                If it's absolutely empty, then the previous test will fail too.*/
 557             log_err("Locale resolves to different locale. Is %s an alias of %s?\n",
 558                 currLoc, resolvedLoc);
 559         }
 560         TestKeyInRootRecursive(root, "root", currentLocale, currLoc);
 561
 562         testLCID(currentLocale, currLoc);
 563
 564         ures_close(currentLocale);
 565     }
 566
 567     ures_close(root);
 568 }
 569 #endif
 570
 571 static void
 572 compareArrays(const char *keyName,
 573               UResourceBundle *fromArray, const char *fromLocale,
 574               UResourceBundle *toArray, const char *toLocale,
 575               int32_t start, int32_t end)
 576 {
 577     int32_t fromSize = ures_getSize(fromArray);
 578     int32_t toSize = ures_getSize(fromArray);
 579     int32_t idx;
 580     UErrorCode errorCode = U_ZERO_ERROR;
 581
 582     if (fromSize > toSize) {
 583         fromSize = toSize;
 584         log_err("Arrays are different size from \"%s\" to \"%s\"\n",
 585                 fromLocale,
 586                 toLocale);
 587     }
 588
 589     for (idx = start; idx <= end; idx++) {
 590         const UChar *fromBundleStr = ures_getStringByIndex(fromArray, idx, NULL, &errorCode);
 591         const UChar *toBundleStr = ures_getStringByIndex(toArray, idx, NULL, &errorCode);
 592         if (fromBundleStr && toBundleStr && u_strcmp(fromBundleStr, toBundleStr) != 0)
 593         {
 594             log_err("Difference for %s at index %d from %s= \"%s\" to %s= \"%s\"\n",
 595                     keyName,
 596                     idx,
 597                     fromLocale,
 598                     austrdup(fromBundleStr),
 599                     toLocale,
 600                     austrdup(toBundleStr));
 601         }
 602     }
 603 }
 604
 605 static void
 606 compareConsistentCountryInfo(const char *fromLocale, const char *toLocale) {
 607     UErrorCode errorCode = U_ZERO_ERROR;
 608     UResourceBundle *fromArray, *toArray;
 609     UResourceBundle *fromLocaleBund = ures_open(NULL, fromLocale, &errorCode);
 610     UResourceBundle *toLocaleBund = ures_open(NULL, toLocale, &errorCode);
 611     UResourceBundle *toCalendar, *fromCalendar, *toGregorian, *fromGregorian;
 612
 613     if(U_FAILURE(errorCode)) {
 614         log_err("Can't open resource bundle %s or %s - %s\n", fromLocale, toLocale, u_errorName(errorCode));
 615         return;
 616     }
 617     fromCalendar = ures_getByKey(fromLocaleBund, "calendar", NULL, &errorCode);
 618     fromGregorian = ures_getByKeyWithFallback(fromCalendar, "gregorian", NULL, &errorCode);
 619
 620     toCalendar = ures_getByKey(toLocaleBund, "calendar", NULL, &errorCode);
 621     toGregorian = ures_getByKeyWithFallback(toCalendar, "gregorian", NULL, &errorCode);
 622
 623     fromArray = ures_getByKey(fromLocaleBund, "CurrencyElements", NULL, &errorCode);
 624     toArray = ures_getByKey(toLocaleBund, "CurrencyElements", NULL, &errorCode);
 625     if (strcmp(fromLocale, "en_CA") != 0)
 626     {
 627         /* The first one is probably localized. */
 628         compareArrays("CurrencyElements", fromArray, fromLocale, toArray, toLocale, 1, 2);
 629     }
 630     ures_close(fromArray);
 631     ures_close(toArray);
 632
 633     fromArray = ures_getByKey(fromLocaleBund, "NumberPatterns", NULL, &errorCode);
 634     toArray = ures_getByKey(toLocaleBund, "NumberPatterns", NULL, &errorCode);
 635     if (strcmp(fromLocale, "en_CA") != 0)
 636     {
 637         compareArrays("NumberPatterns", fromArray, fromLocale, toArray, toLocale, 0, 3);
 638     }
 639     ures_close(fromArray);
 640     ures_close(toArray);
 641
 642     /* Difficult to test properly */
 643 /*
 644     fromArray = ures_getByKey(fromLocaleBund, "DateTimePatterns", NULL, &errorCode);
 645     toArray = ures_getByKey(toLocaleBund, "DateTimePatterns", NULL, &errorCode);
 646     {
 647         compareArrays("DateTimePatterns", fromArray, fromLocale, toArray, toLocale);
 648     }
 649     ures_close(fromArray);
 650     ures_close(toArray);*/
 651
 652     fromArray = ures_getByKey(fromLocaleBund, "NumberElements", NULL, &errorCode);
 653     toArray = ures_getByKey(toLocaleBund, "NumberElements", NULL, &errorCode);
 654     if (strcmp(fromLocale, "en_CA") != 0)
 655     {
 656         compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 0, 3);
 657         /* Index 4 is a script based 0 */
 658         compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 5, 10);
 659     }
 660     ures_close(fromArray);
 661     ures_close(toArray);
 662     ures_close(fromCalendar);
 663     ures_close(toCalendar);
 664     ures_close(fromGregorian);
 665     ures_close(toGregorian);
 666
 667     ures_close(fromLocaleBund);
 668     ures_close(toLocaleBund);
 669 }
 670
 671 static void
 672 TestConsistentCountryInfo(void) {
 673 /*    UResourceBundle *fromLocale, *toLocale;*/
 674     int32_t locCount = uloc_countAvailable();
 675     int32_t fromLocIndex, toLocIndex;
 676
 677     int32_t fromCountryLen, toCountryLen;
 678     char fromCountry[ULOC_FULLNAME_CAPACITY], toCountry[ULOC_FULLNAME_CAPACITY];
 679
 680     int32_t fromVariantLen, toVariantLen;
 681     char fromVariant[ULOC_FULLNAME_CAPACITY], toVariant[ULOC_FULLNAME_CAPACITY];
 682
 683     UErrorCode errorCode = U_ZERO_ERROR;
 684
 685     for (fromLocIndex = 0; fromLocIndex < locCount; fromLocIndex++) {
 686         const char *fromLocale = uloc_getAvailable(fromLocIndex);
 687
 688         errorCode=U_ZERO_ERROR;
 689         fromCountryLen = uloc_getCountry(fromLocale, fromCountry, ULOC_FULLNAME_CAPACITY, &errorCode);
 690         if (fromCountryLen <= 0) {
 691             /* Ignore countryless locales */
 692             continue;
 693         }
 694         fromVariantLen = uloc_getVariant(fromLocale, fromVariant, ULOC_FULLNAME_CAPACITY, &errorCode);
 695         if (fromVariantLen > 0) {
 696             /* Most variants are ignorable like PREEURO, or collation variants. */
 697             continue;
 698         }
 699         /* Start comparing only after the current index.
 700            Previous loop should have already compared fromLocIndex.
 701         */
 702         for (toLocIndex = fromLocIndex + 1; toLocIndex < locCount; toLocIndex++) {
 703             const char *toLocale = uloc_getAvailable(toLocIndex);
 704
 705             toCountryLen = uloc_getCountry(toLocale, toCountry, ULOC_FULLNAME_CAPACITY, &errorCode);
 706             if(U_FAILURE(errorCode)) {
 707                 log_err("Unknown failure fromLocale=%s toLocale=%s errorCode=%s\n",
 708                     fromLocale, toLocale, u_errorName(errorCode));
 709                 continue;
 710             }
 711
 712             if (toCountryLen <= 0) {
 713                 /* Ignore countryless locales */
 714                 continue;
 715             }
 716             toVariantLen = uloc_getVariant(toLocale, toVariant, ULOC_FULLNAME_CAPACITY, &errorCode);
 717             if (toVariantLen > 0) {
 718                 /* Most variants are ignorable like PREEURO, or collation variants. */
 719                 /* They're a variant for a reason. */
 720                 continue;
 721             }
 722             if (strcmp(fromCountry, toCountry) == 0) {
 723                 log_verbose("comparing fromLocale=%s toLocale=%s\n",
 724                     fromLocale, toLocale);
 725                 compareConsistentCountryInfo(fromLocale, toLocale);
 726             }
 727         }
 728     }
 729 }
 730
 731 static int32_t
 732 findStringSetMismatch(const char *currLoc, const UChar *string, int32_t langSize,
 733                       const UChar *exemplarCharacters, int32_t exemplarLen,
 734                       UBool ignoreNumbers, UChar* badCharPtr) {
 735     UErrorCode errorCode = U_ZERO_ERROR;
 736     USet *origSet = uset_openPatternOptions(exemplarCharacters, exemplarLen, USET_CASE_INSENSITIVE, &errorCode);
 737     USet *exemplarSet = createFlattenSet(origSet, &errorCode);
 738     int32_t strIdx;
 739     uset_close(origSet);
 740     if (U_FAILURE(errorCode)) {
 741         log_err("%s: error uset_openPattern returned %s\n", currLoc, u_errorName(errorCode));
 742         return -1;
 743     }
 744
 745     for (strIdx = 0; strIdx < langSize; strIdx++) {
 746         if (!uset_contains(exemplarSet, string[strIdx])
 747             && 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
 748             && string[strIdx] != 0x200C && string[strIdx] != 0x200D) {
 749             if (!ignoreNumbers || (ignoreNumbers && (string[strIdx] < 0x30 || string[strIdx] > 0x39))) {
 750                 uset_close(exemplarSet);
 751                 if (badCharPtr) {
 752                     *badCharPtr = string[strIdx];
 753                 }
 754                 return strIdx;
 755             }
 756         }
 757     }
 758     uset_close(exemplarSet);
 759     if (badCharPtr) {
 760         *badCharPtr = 0;
 761     }
 762     return -1;
 763 }
 764 /* include non-invariant chars */
 765 static int32_t
 766 myUCharsToChars(const UChar* us, char* cs, int32_t len){
 767     int32_t i=0;
 768     for(; i< len; i++){
 769         if(us[i] < 0x7f){
 770             cs[i] = (char)us[i];
 771         }else{
 772             return -1;
 773         }
 774     }
 775     return i;
 776 }
 777 static void
 778 findSetMatch( UScriptCode *scriptCodes, int32_t scriptsLen,
 779               USet *exemplarSet,
 780               const char  *locale){
 781     USet *scripts[10]= {0};
 782     char pattern[256] = { '[', ':', 0x000 };
 783     int32_t patternLen;
 784     UChar uPattern[256] = {0};
 785     UErrorCode status = U_ZERO_ERROR;
 786     int32_t i;
 787
 788     /* create the sets with script codes */
 789     for(i = 0; i<scriptsLen; i++){
 790         strcat(pattern, uscript_getShortName(scriptCodes[i]));
 791         strcat(pattern, ":]");
 792         patternLen = (int32_t)strlen(pattern);
 793         u_charsToUChars(pattern, uPattern, patternLen);
 794         scripts[i] = uset_openPattern(uPattern, patternLen, &status);
 795         if(U_FAILURE(status)){
 796             log_err("Could not create set for pattern %s. Error: %s\n", pattern, u_errorName(status));
 797             return;
 798         }
 799         pattern[2] = 0;
 800     }
 801     if (strcmp(locale, "uk") == 0 || strcmp(locale, "uk_UA") == 0) {
 802         /* Special addition. Add the modifying apostrophe, which isn't in Cyrillic. */
 803         uset_add(scripts[0], 0x2bc);
 804     }
 805     if(U_SUCCESS(status)){
 806         UBool existsInScript = FALSE;
 807         /* iterate over the exemplarSet and ascertain if all
 808          * UChars in exemplarSet belong to the scripts returned
 809          * by getScript
 810          */
 811         int32_t count = uset_getItemCount(exemplarSet);
 812
 813         for( i=0; i < count; i++){
 814             UChar32 start = 0;
 815             UChar32 end = 0;
 816             UChar *str = NULL;
 817             int32_t strCapacity = 0;
 818
 819             strCapacity = uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status);
 820             if(U_SUCCESS(status)){
 821                 int32_t j;
 822                 if(strCapacity == 0){
 823                     /* ok the item is a range */
 824                      for( j = 0; j < scriptsLen; j++){
 825                         if(uset_containsRange(scripts[j], start, end) == TRUE){
 826                             existsInScript = TRUE;
 827                         }
 828                     }
 829                     if(existsInScript == FALSE){
 830                         for( j = 0; j < scriptsLen; j++){
 831                             UChar toPattern[500]={'\0'};
 832                             char pat[500]={'\0'};
 833                             int32_t len = uset_toPattern(scripts[j], toPattern, 500, TRUE, &status);
 834                             len = myUCharsToChars(toPattern, pat, len);
 835                             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));
 836                             if(len!=-1){
 837                                 log_err("Pattern: %s\n",pat);
 838                             }
 839                         }
 840                         log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale);
 841                     }
 842                 }else{
 843                     strCapacity++; /* increment for NUL termination */
 844                     /* allocate the str and call the api again */
 845                     str = (UChar*) malloc(U_SIZEOF_UCHAR * strCapacity);
 846                     strCapacity =  uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status);
 847                     /* iterate over the scripts and figure out if the string contained is actually
 848                      * in the script set
 849                      */
 850                     for( j = 0; j < scriptsLen; j++){
 851                         if(uset_containsString(scripts[j],str, strCapacity) == TRUE){
 852                             existsInScript = TRUE;
 853                         }
 854                     }
 855                     if(existsInScript == FALSE){
 856                         log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale);
 857                     }
 858                 }
 859             }
 860         }
 861
 862     }
 863
 864     /* close the sets */
 865     for(i = 0; i<scriptsLen; i++){
 866         uset_close(scripts[i]);
 867     }
 868 }
 869
 870 static void VerifyTranslation(void) {
 871     UResourceBundle *root, *currentLocale;
 872     int32_t locCount = uloc_countAvailable();
 873     int32_t locIndex;
 874     UErrorCode errorCode = U_ZERO_ERROR;
 875     int32_t exemplarLen;
 876     const UChar *exemplarCharacters;
 877     const char *currLoc;
 878     UScriptCode scripts[USCRIPT_CODE_LIMIT];
 879     int32_t numScripts;
 880     int32_t idx;
 881     int32_t end;
 882     UResourceBundle *resArray;
 883
 884     if (locCount <= 1) {
 885         log_data_err("At least root needs to be installed\n");
 886     }
 887
 888     root = ures_openDirect(NULL, "root", &errorCode);
 889     if(U_FAILURE(errorCode)) {
 890         log_data_err("Can't open root\n");
 891         return;
 892     }
 893     for (locIndex = 0; locIndex < locCount; locIndex++) {
 894         errorCode=U_ZERO_ERROR;
 895         currLoc = uloc_getAvailable(locIndex);
 896         currentLocale = ures_open(NULL, currLoc, &errorCode);
 897         if(errorCode != U_ZERO_ERROR) {
 898             if(U_SUCCESS(errorCode)) {
 899                 /* It's installed, but there is no data.
 900                    It's installed for the g18n white paper [grhoten] */
 901                 log_err("ERROR: Locale %-5s not installed, and it should be!\n",
 902                     uloc_getAvailable(locIndex));
 903             } else {
 904                 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
 905                     u_errorName(errorCode),
 906                     uloc_getAvailable(locIndex));
 907             }
 908             ures_close(currentLocale);
 909             continue;
 910         }
 911         exemplarCharacters = ures_getStringByKey(currentLocale, "ExemplarCharacters", &exemplarLen, &errorCode);
 912         if (U_FAILURE(errorCode)) {
 913             log_err("error ures_getStringByKey returned %s\n", u_errorName(errorCode));
 914         }
 915         else if (getTestOption(QUICK_OPTION) && exemplarLen > 2048) {
 916             log_verbose("skipping test for %s\n", currLoc);
 917         }
 918         else if (uprv_strncmp(currLoc,"bem",3) == 0 || uprv_strncmp(currLoc,"mgo",3) == 0 || uprv_strncmp(currLoc,"nl",2) == 0 || uprv_strncmp(currLoc,"ky",2) == 0) {
 919             log_verbose("skipping test for %s, some month and country names known to use aux exemplars\n", currLoc);
 920         }
 921         else {
 922             UChar langBuffer[128];
 923             int32_t langSize;
 924             int32_t strIdx;
 925             UChar badChar;
 926             langSize = uloc_getDisplayLanguage(currLoc, currLoc, langBuffer, sizeof(langBuffer)/sizeof(langBuffer[0]), &errorCode);
 927             if (U_FAILURE(errorCode)) {
 928                 log_err("error uloc_getDisplayLanguage returned %s\n", u_errorName(errorCode));
 929             }
 930             else {
 931                 strIdx = findStringSetMismatch(currLoc, langBuffer, langSize, exemplarCharacters, exemplarLen, FALSE, &badChar);
 932                 if (strIdx >= 0) {
 933                     log_err("getDisplayLanguage(%s) at index %d returned characters not in the exemplar characters: %04X.\n",
 934                         currLoc, strIdx, badChar);
 935                 }
 936             }
 937             langSize = uloc_getDisplayCountry(currLoc, currLoc, langBuffer, sizeof(langBuffer)/sizeof(langBuffer[0]), &errorCode);
 938             if (U_FAILURE(errorCode)) {
 939                 log_err("error uloc_getDisplayCountry returned %s\n", u_errorName(errorCode));
 940             }
 941             else if (uprv_strstr(currLoc, "ti_") != currLoc || isICUVersionAtLeast(52, 0, 1)) { /* TODO: FIX or REMOVE this test!  Was: restore DisplayCountry test for ti_* when cldrbug 3058 is fixed) - but CldrBug:3058 is wontfix */
 942               strIdx = findStringSetMismatch(currLoc, langBuffer, langSize, exemplarCharacters, exemplarLen, FALSE, &badChar);
 943                 if (strIdx >= 0) {
 944                     log_err("getDisplayCountry(%s) at index %d returned characters not in the exemplar characters: %04X.\n",
 945                         currLoc, strIdx, badChar);
 946                 }
 947             }
 948             {
 949                 UResourceBundle* cal = ures_getByKey(currentLocale, "calendar", NULL, &errorCode);
 950                 UResourceBundle* greg = ures_getByKeyWithFallback(cal, "gregorian", NULL, &errorCode);
 951                 UResourceBundle* names = ures_getByKeyWithFallback(greg,  "dayNames", NULL, &errorCode);
 952                 UResourceBundle* format = ures_getByKeyWithFallback(names,  "format", NULL, &errorCode);
 953                 resArray = ures_getByKeyWithFallback(format,  "wide", NULL, &errorCode);
 954
 955                 if (U_FAILURE(errorCode)) {
 956                     log_err("error ures_getByKey returned %s\n", u_errorName(errorCode));
 957                 }
 958                 if (getTestOption(QUICK_OPTION)) {
 959                     end = 1;
 960                 }
 961                 else {
 962                     end = ures_getSize(resArray);
 963                 }
 964
 965
 966                 for (idx = 0; idx < end; idx++) {
 967                     const UChar *fromBundleStr = ures_getStringByIndex(resArray, idx, &langSize, &errorCode);
 968                     if (U_FAILURE(errorCode)) {
 969                         log_err("error ures_getStringByIndex(%d) returned %s\n", idx, u_errorName(errorCode));
 970                         continue;
 971                     }
 972                     if (uprv_strstr(currLoc, "uz_Arab") != currLoc || isICUVersionAtLeast(52, 0, 1)) { /* TODO: FIX or REMOVE this test! */
 973                         strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, exemplarCharacters, exemplarLen, TRUE, &badChar);
 974                         if (strIdx >= 0) {
 975                             log_err("getDayNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n",
 976                                 currLoc, idx, strIdx, badChar);
 977                         }
 978                     }
 979                 }
 980                 ures_close(resArray);
 981                 ures_close(format);
 982                 ures_close(names);
 983
 984                 names = ures_getByKeyWithFallback(greg, "monthNames", NULL, &errorCode);
 985                 format = ures_getByKeyWithFallback(names,"format", NULL, &errorCode);
 986                 resArray = ures_getByKeyWithFallback(format, "wide", NULL, &errorCode);
 987                 if (U_FAILURE(errorCode)) {
 988                     log_err("error ures_getByKey returned %s\n", u_errorName(errorCode));
 989                 }
 990                 if (getTestOption(QUICK_OPTION)) {
 991                     end = 1;
 992                 }
 993                 else {
 994                     end = ures_getSize(resArray);
 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                     if (uprv_strstr(currLoc, "uz_Arab") != currLoc || isICUVersionAtLeast(52, 0, 1)) { /* TODO: FIX or REMOVE this test! */
1004                         strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, exemplarCharacters, exemplarLen, TRUE, &badChar);
1005                         if (strIdx >= 0) {
1006                             log_err("getMonthNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n",
1007                                 currLoc, idx, strIdx, badChar);
1008                         }
1009                     }
1010                 }
1011                 ures_close(resArray);
1012                 ures_close(format);
1013                 ures_close(names);
1014                 ures_close(greg);
1015                 ures_close(cal);
1016             }
1017             errorCode = U_ZERO_ERROR;
1018             numScripts = uscript_getCode(currLoc, scripts, sizeof(scripts)/sizeof(scripts[0]), &errorCode);
1019             if (numScripts == 0) {
1020                 log_err("uscript_getCode(%s) doesn't work.\n", currLoc);
1021             }else if(scripts[0] == USCRIPT_COMMON){
1022                 log_err("uscript_getCode(%s) returned USCRIPT_COMMON.\n", currLoc);
1023             }
1024
1025             /* test that the scripts are a superset of exemplar characters. */
1026            {
1027                 ULocaleData *uld = ulocdata_open(currLoc,&errorCode);
1028                 USet *exemplarSet =  ulocdata_getExemplarSet(uld, NULL, 0, ULOCDATA_ES_STANDARD, &errorCode);
1029                 /* test if exemplar characters are part of script code */
1030                 findSetMatch(scripts, numScripts, exemplarSet, currLoc);
1031                 uset_close(exemplarSet);
1032                 ulocdata_close(uld);
1033             }
1034
1035            /* test that the paperSize API works */
1036            {
1037                int32_t height=0, width=0;
1038                ulocdata_getPaperSize(currLoc, &height, &width, &errorCode);
1039                if(U_FAILURE(errorCode)){
1040                    log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1041                }
1042                if(strstr(currLoc, "_US")!=NULL && height != 279 && width != 216 ){
1043                    log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc);
1044                }
1045            }
1046             /* test that the MeasurementSystem works API works */
1047            {
1048                char fullLoc[ULOC_FULLNAME_CAPACITY];
1049                UMeasurementSystem measurementSystem = ulocdata_getMeasurementSystem(currLoc, &errorCode);
1050                if(U_FAILURE(errorCode)){
1051                    log_err("ulocdata_getMeasurementSystem failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1052                }
1053                uloc_addLikelySubtags(currLoc, fullLoc, ULOC_FULLNAME_CAPACITY, &errorCode);
1054                if(strstr(fullLoc, "_US")!=NULL || strstr(fullLoc, "_MM")!=NULL || strstr(fullLoc, "_LR")!=NULL){
1055                    if(measurementSystem != UMS_US){
1056                         log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1057                    }
1058                }else if(measurementSystem != UMS_SI){
1059                    log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1060                }
1061            }
1062         }
1063         ures_close(currentLocale);
1064     }
1065
1066     ures_close(root);
1067 }
1068
1069 /* adjust this limit as appropriate */
1070 #define MAX_SCRIPTS_PER_LOCALE 8
1071
1072 static void TestExemplarSet(void){
1073     int32_t i, j, k, m, n;
1074     int32_t equalCount = 0;
1075     UErrorCode ec = U_ZERO_ERROR;
1076     UEnumeration* avail;
1077     USet* exemplarSets[2];
1078     USet* unassignedSet;
1079     UScriptCode code[MAX_SCRIPTS_PER_LOCALE];
1080     USet* codeSets[MAX_SCRIPTS_PER_LOCALE];
1081     int32_t codeLen;
1082     char cbuf[32]; /* 9 should be enough */
1083     UChar ubuf[64]; /* adjust as needed */
1084     UBool existsInScript;
1085     int32_t itemCount;
1086     int32_t strLen;
1087     UChar32 start, end;
1088
1089     unassignedSet = NULL;
1090     exemplarSets[0] = NULL;
1091     exemplarSets[1] = NULL;
1092     for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) {
1093         codeSets[i] = NULL;
1094     }
1095
1096     avail = ures_openAvailableLocales(NULL, &ec);
1097     if (!assertSuccess("ures_openAvailableLocales", &ec)) goto END;
1098     n = uenum_count(avail, &ec);
1099     if (!assertSuccess("uenum_count", &ec)) goto END;
1100
1101     u_uastrcpy(ubuf, "[:unassigned:]");
1102     unassignedSet = uset_openPattern(ubuf, -1, &ec);
1103     if (!assertSuccess("uset_openPattern", &ec)) goto END;
1104
1105     for(i=0; i<n; i++){
1106         const char* locale = uenum_next(avail, NULL, &ec);
1107         if (!assertSuccess("uenum_next", &ec)) goto END;
1108         log_verbose("%s\n", locale);
1109         for (k=0; k<2; ++k) {
1110             uint32_t option = (k==0) ? 0 : USET_CASE_INSENSITIVE;
1111             ULocaleData *uld = ulocdata_open(locale,&ec);
1112             USet* exemplarSet = ulocdata_getExemplarSet(uld,NULL, option, ULOCDATA_ES_STANDARD, &ec);
1113             uset_close(exemplarSets[k]);
1114             ulocdata_close(uld);
1115             exemplarSets[k] = exemplarSet;
1116             if (!assertSuccess("ulocaledata_getExemplarSet", &ec)) goto END;
1117
1118             if (uset_containsSome(exemplarSet, unassignedSet)) {
1119                 log_err("ExemplarSet contains unassigned characters for locale : %s\n", locale);
1120             }
1121             codeLen = uscript_getCode(locale, code, 8, &ec);
1122             if (!assertSuccess("uscript_getCode", &ec)) goto END;
1123
1124             for (j=0; j<MAX_SCRIPTS_PER_LOCALE; ++j) {
1125                 uset_close(codeSets[j]);
1126                 codeSets[j] = NULL;
1127             }
1128             for (j=0; j<codeLen; ++j) {
1129                 uprv_strcpy(cbuf, "[:");
1130                 if(code[j]==-1){
1131                     log_err("USCRIPT_INVALID_CODE returned for locale: %s\n", locale);
1132                     continue;
1133                 }
1134                 uprv_strcat(cbuf, uscript_getShortName(code[j]));
1135                 uprv_strcat(cbuf, ":]");
1136                 u_uastrcpy(ubuf, cbuf);
1137                 codeSets[j] = uset_openPattern(ubuf, -1, &ec);
1138             }
1139             if (!assertSuccess("uset_openPattern", &ec)) goto END;
1140
1141             existsInScript = FALSE;
1142             itemCount = uset_getItemCount(exemplarSet);
1143             for (m=0; m<itemCount && !existsInScript; ++m) {
1144                 strLen = uset_getItem(exemplarSet, m, &start, &end, ubuf,
1145                                       sizeof(ubuf)/sizeof(ubuf[0]), &ec);
1146                 /* failure here might mean str[] needs to be larger */
1147                 if (!assertSuccess("uset_getItem", &ec)) goto END;
1148                 if (strLen == 0) {
1149                     for (j=0; j<codeLen; ++j) {
1150                         if (codeSets[j]!=NULL && uset_containsRange(codeSets[j], start, end)) {
1151                             existsInScript = TRUE;
1152                             break;
1153                         }
1154                     }
1155                 } else {
1156                     for (j=0; j<codeLen; ++j) {
1157                         if (codeSets[j]!=NULL && uset_containsString(codeSets[j], ubuf, strLen)) {
1158                             existsInScript = TRUE;
1159                             break;
1160                         }
1161                     }
1162                 }
1163             }
1164
1165             if (existsInScript == FALSE){
1166                 log_err("ExemplarSet containment failed for locale : %s\n", locale);
1167             }
1168         }
1169         assertTrue("case-folded is a superset",
1170                    uset_containsAll(exemplarSets[1], exemplarSets[0]));
1171         if (uset_equals(exemplarSets[1], exemplarSets[0])) {
1172             ++equalCount;
1173         }
1174     }
1175     /* Note: The case-folded set should sometimes be a strict superset
1176        and sometimes be equal. */
1177     assertTrue("case-folded is sometimes a strict superset, and sometimes equal",
1178                equalCount > 0 && equalCount < n);
1179
1180  END:
1181     uenum_close(avail);
1182     uset_close(exemplarSets[0]);
1183     uset_close(exemplarSets[1]);
1184     uset_close(unassignedSet);
1185     for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) {
1186         uset_close(codeSets[i]);
1187     }
1188 }
1189
1190 static void TestLocaleDisplayPattern(void){
1191     UErrorCode status = U_ZERO_ERROR;
1192     UChar pattern[32] = {0,};
1193     UChar separator[32] = {0,};
1194     ULocaleData *uld = ulocdata_open(uloc_getDefault(), &status);
1195
1196     if(U_FAILURE(status)){
1197         log_data_err("ulocdata_open error");
1198         return;
1199     }
1200     ulocdata_getLocaleDisplayPattern(uld, pattern, 32, &status);
1201     if (U_FAILURE(status)){
1202         log_err("ulocdata_getLocaleDisplayPattern error!");
1203     }
1204     status = U_ZERO_ERROR;
1205     ulocdata_getLocaleSeparator(uld, separator, 32, &status);
1206     if (U_FAILURE(status)){
1207         log_err("ulocdata_getLocaleSeparator error!");
1208     }
1209     ulocdata_close(uld);
1210 }
1211
1212 static void TestCoverage(void){
1213     ULocaleDataDelimiterType types[] = {
1214      ULOCDATA_QUOTATION_START,     /* Quotation start */
1215      ULOCDATA_QUOTATION_END,       /* Quotation end */
1216      ULOCDATA_ALT_QUOTATION_START, /* Alternate quotation start */
1217      ULOCDATA_ALT_QUOTATION_END,   /* Alternate quotation end */
1218      ULOCDATA_DELIMITER_COUNT
1219     };
1220     int i;
1221     UBool sub;
1222     UErrorCode status = U_ZERO_ERROR;
1223     ULocaleData *uld = ulocdata_open(uloc_getDefault(), &status);
1224
1225     if(U_FAILURE(status)){
1226         log_data_err("ulocdata_open error");
1227         return;
1228     }
1229
1230
1231     for(i = 0; i < ULOCDATA_DELIMITER_COUNT; i++){
1232         UChar result[32] = {0,};
1233         status = U_ZERO_ERROR;
1234         ulocdata_getDelimiter(uld, types[i], result, 32, &status);
1235         if (U_FAILURE(status)){
1236             log_err("ulocdata_getgetDelimiter error with type %d", types[i]);
1237         }
1238     }
1239
1240     sub = ulocdata_getNoSubstitute(uld);
1241     ulocdata_setNoSubstitute(uld,sub);
1242     ulocdata_close(uld);
1243 }
1244
1245 static void TestIndexChars(void) {
1246     /* Very basic test of ULOCDATA_ES_INDEX.
1247      * No comprehensive test of data, just basic check that the code path is alive.
1248      */
1249     UErrorCode status = U_ZERO_ERROR;
1250     ULocaleData  *uld;
1251     USet *exemplarChars;
1252     USet *indexChars;
1253
1254     uld = ulocdata_open("en", &status);
1255     exemplarChars = uset_openEmpty();
1256     indexChars = uset_openEmpty();
1257     ulocdata_getExemplarSet(uld, exemplarChars, 0, ULOCDATA_ES_STANDARD, &status);
1258     ulocdata_getExemplarSet(uld, indexChars, 0, ULOCDATA_ES_INDEX, &status);
1259     if (U_FAILURE(status)) {
1260         log_data_err("File %s, line %d, Failure opening exemplar chars: %s", __FILE__, __LINE__, u_errorName(status));
1261         goto close_sets;
1262     }
1263     /* en data, standard exemplars are [a-z], lower case. */
1264     /* en data, index characters are [A-Z], upper case. */
1265     if ((uset_contains(exemplarChars, (UChar32)0x41) || uset_contains(indexChars, (UChar32)0x61))) {
1266         log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__, __LINE__ );
1267         goto close_sets;
1268     }
1269     if (!(uset_contains(exemplarChars, (UChar32)0x61) && uset_contains(indexChars, (UChar32)0x41) )) {
1270         log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__, __LINE__ );
1271         goto close_sets;
1272     }
1273
1274   close_sets:
1275     uset_close(exemplarChars);
1276     uset_close(indexChars);
1277     ulocdata_close(uld);
1278 }
1279
1280
1281
1282 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
1283 static void TestCurrencyList(void){
1284 #if !UCONFIG_NO_FORMATTING
1285     UErrorCode errorCode = U_ZERO_ERROR;
1286     int32_t structLocaleCount, currencyCount;
1287     UEnumeration *en = ucurr_openISOCurrencies(UCURR_ALL, &errorCode);
1288     const char *isoCode, *structISOCode;
1289     UResourceBundle *subBundle;
1290     UResourceBundle *currencies = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode);
1291     if(U_FAILURE(errorCode)) {
1292         log_data_err("Can't open structLocale\n");
1293         return;
1294     }
1295     currencies = ures_getByKey(currencies, "Currencies", currencies, &errorCode);
1296     currencyCount = uenum_count(en, &errorCode);
1297     structLocaleCount = ures_getSize(currencies);
1298     if (currencyCount != structLocaleCount) {
1299         log_err("structLocale(%d) and ISO4217(%d) currency list are out of sync.\n", structLocaleCount, currencyCount);
1300 #if U_CHARSET_FAMILY == U_ASCII_FAMILY
1301         ures_resetIterator(currencies);
1302         while ((isoCode = uenum_next(en, NULL, &errorCode)) != NULL && ures_hasNext(currencies)) {
1303             subBundle = ures_getNextResource(currencies, NULL, &errorCode);
1304             structISOCode = ures_getKey(subBundle);
1305             ures_close(subBundle);
1306             if (strcmp(structISOCode, isoCode) != 0) {
1307                 log_err("First difference found at structLocale(%s) and ISO4217(%s).\n", structISOCode, isoCode);
1308                 break;
1309             }
1310         }
1311 #endif
1312     }
1313     ures_close(currencies);
1314     uenum_close(en);
1315 #endif
1316 }
1317 #endif
1318
1319 static void TestAvailableIsoCodes(void){
1320 #if !UCONFIG_NO_FORMATTING
1321     UErrorCode errorCode = U_ZERO_ERROR;
1322     const char* eurCode = "EUR";
1323     const char* usdCode = "USD";
1324     const char* lastCode = "RHD";
1325     const char* zzzCode = "ZZZ";
1326     UDate date1950 = (UDate)-630720000000.0;/* year 1950 */
1327     UDate date1970 = (UDate)0.0;            /* year 1970 */
1328     UDate date1975 = (UDate)173448000000.0; /* year 1975 */
1329     UDate date1978 = (UDate)260172000000.0; /* year 1978 */
1330     UDate date1981 = (UDate)346896000000.0; /* year 1981 */
1331     UDate date1992 = (UDate)693792000000.0; /* year 1992 */
1332     UChar* isoCode = (UChar*)malloc(sizeof(UChar) * (uprv_strlen(usdCode) + 1));
1333
1334     /* testing available codes with no time ranges */
1335     u_charsToUChars(eurCode, isoCode, uprv_strlen(usdCode) + 1);
1336     if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) {
1337        log_data_err("FAIL: ISO code (%s) is not found.\n", eurCode);
1338     }
1339
1340     u_charsToUChars(usdCode, isoCode, uprv_strlen(zzzCode) + 1);
1341     if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) {
1342        log_data_err("FAIL: ISO code (%s) is not found.\n", usdCode);
1343     }
1344
1345     u_charsToUChars(zzzCode, isoCode, uprv_strlen(zzzCode) + 1);
1346     if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == TRUE) {
1347        log_err("FAIL: ISO code (%s) is reported as available, but it doesn't exist.\n", zzzCode);
1348     }
1349
1350     u_charsToUChars(lastCode, isoCode, uprv_strlen(zzzCode) + 1);
1351     if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) {
1352        log_data_err("FAIL: ISO code (%s) is not found.\n", lastCode);
1353     }
1354
1355     /* RHD was used from 1970-02-17  to 1980-04-18*/
1356
1357     /* to = null */
1358     if (ucurr_isAvailable(isoCode, date1970, U_DATE_MAX, &errorCode) == FALSE) {
1359        log_data_err("FAIL: ISO code (%s) was available in time range >1970-01-01.\n", lastCode);
1360     }
1361
1362     if (ucurr_isAvailable(isoCode, date1975, U_DATE_MAX, &errorCode) == FALSE) {
1363        log_data_err("FAIL: ISO code (%s) was available in time range >1975.\n", lastCode);
1364     }
1365
1366     if (ucurr_isAvailable(isoCode, date1981, U_DATE_MAX, &errorCode) == TRUE) {
1367        log_err("FAIL: ISO code (%s) was not available in time range >1981.\n", lastCode);
1368     }
1369
1370     /* from = null */
1371     if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1970, &errorCode) == TRUE) {
1372        log_err("FAIL: ISO code (%s) was not available in time range <1970.\n", lastCode);
1373     }
1374
1375     if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1975, &errorCode) == FALSE) {
1376        log_data_err("FAIL: ISO code (%s) was available in time range <1975.\n", lastCode);
1377     }
1378
1379     if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1981, &errorCode) == FALSE) {
1380        log_data_err("FAIL: ISO code (%s) was available in time range <1981.\n", lastCode);
1381     }
1382
1383     /* full ranges */
1384     if (ucurr_isAvailable(isoCode, date1975, date1978, &errorCode) == FALSE) {
1385        log_data_err("FAIL: ISO code (%s) was available in time range 1975-1978.\n", lastCode);
1386     }
1387
1388     if (ucurr_isAvailable(isoCode, date1970, date1975, &errorCode) == FALSE) {
1389        log_data_err("FAIL: ISO code (%s) was available in time range 1970-1975.\n", lastCode);
1390     }
1391
1392     if (ucurr_isAvailable(isoCode, date1975, date1981, &errorCode) == FALSE) {
1393        log_data_err("FAIL: ISO code (%s) was available in time range 1975-1981.\n", lastCode);
1394     }
1395
1396     if (ucurr_isAvailable(isoCode, date1970,  date1981, &errorCode) == FALSE) {
1397        log_data_err("FAIL: ISO code (%s) was available in time range 1970-1981.\n", lastCode);
1398     }
1399
1400     if (ucurr_isAvailable(isoCode, date1981,  date1992, &errorCode) == TRUE) {
1401        log_err("FAIL: ISO code (%s) was not available in time range 1981-1992.\n", lastCode);
1402     }
1403
1404     if (ucurr_isAvailable(isoCode, date1950,  date1970, &errorCode) == TRUE) {
1405        log_err("FAIL: ISO code (%s) was not available in time range 1950-1970.\n", lastCode);
1406     }
1407
1408     /* wrong range - from > to*/
1409     if (ucurr_isAvailable(isoCode, date1975,  date1970, &errorCode) == TRUE) {
1410        log_err("FAIL: Wrong range 1975-1970 for ISO code (%s) was not reported.\n", lastCode);
1411     } else if (errorCode != U_ILLEGAL_ARGUMENT_ERROR) {
1412        log_data_err("FAIL: Error code not reported for wrong range 1975-1970 for ISO code (%s).\n", lastCode);
1413     }
1414
1415     free(isoCode);
1416 #endif
1417 }
1418
1419 #define TESTCASE(name) addTest(root, &name, "tsutil/cldrtest/" #name)
1420
1421 void addCLDRTest(TestNode** root);
1422
1423 void addCLDRTest(TestNode** root)
1424 {
1425 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
1426     TESTCASE(TestLocaleStructure);
1427     TESTCASE(TestCurrencyList);
1428 #endif
1429     TESTCASE(TestConsistentCountryInfo);
1430     TESTCASE(VerifyTranslation);
1431     TESTCASE(TestExemplarSet);
1432     TESTCASE(TestLocaleDisplayPattern);
1433     TESTCASE(TestCoverage);
1434     TESTCASE(TestIndexChars);
1435     TESTCASE(TestAvailableIsoCodes);
1436 }
1437