1 /********************************************************************
2 * COPYRIGHT:
3 * Copyright (c) 1997-2014, International Business Machines Corporation and
4 * others. All Rights Reserved.
5 ********************************************************************/
6
7 #include "cintltst.h"
8 #include "unicode/ures.h"
9 #include "unicode/ucurr.h"
10 #include "unicode/ustring.h"
11 #include "unicode/uset.h"
12 #include "unicode/udat.h"
13 #include "unicode/uscript.h"
14 #include "unicode/ulocdata.h"
15 #include "cstring.h"
16 #include "locmap.h"
17 #include "uresimp.h"
18
19 /*
20 returns a new UnicodeSet that is a flattened form of the original
21 UnicodeSet.
22 */
23 static USet*
createFlattenSet(USet * origSet,UErrorCode * status)24 createFlattenSet(USet *origSet, UErrorCode *status) {
25
26
27 USet *newSet = NULL;
28 int32_t origItemCount = 0;
29 int32_t idx, graphmeSize;
30 UChar32 start, end;
31 UChar graphme[64];
32 if (U_FAILURE(*status)) {
33 log_err("createFlattenSet called with %s\n", u_errorName(*status));
34 return NULL;
35 }
36 newSet = uset_open(1, 0);
37 origItemCount = uset_getItemCount(origSet);
38 for (idx = 0; idx < origItemCount; idx++) {
39 graphmeSize = uset_getItem(origSet, idx,
40 &start, &end,
41 graphme, (int32_t)(sizeof(graphme)/sizeof(graphme[0])),
42 status);
43 if (U_FAILURE(*status)) {
44 log_err("ERROR: uset_getItem returned %s\n", u_errorName(*status));
45 *status = U_ZERO_ERROR;
46 }
47 if (graphmeSize) {
48 uset_addAllCodePoints(newSet, graphme, graphmeSize);
49 }
50 else {
51 uset_addRange(newSet, start, end);
52 }
53 }
54 uset_closeOver(newSet,USET_CASE_INSENSITIVE);
55 return newSet;
56 }
57
58 static UBool
isCurrencyPreEuro(const char * currencyKey)59 isCurrencyPreEuro(const char* currencyKey){
60 if( strcmp(currencyKey, "PTE") == 0 ||
61 strcmp(currencyKey, "ESP") == 0 ||
62 strcmp(currencyKey, "LUF") == 0 ||
63 strcmp(currencyKey, "GRD") == 0 ||
64 strcmp(currencyKey, "BEF") == 0 ||
65 strcmp(currencyKey, "ITL") == 0 ||
66 strcmp(currencyKey, "EEK") == 0){
67 return TRUE;
68 }
69 return FALSE;
70 }
71 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
72 static void
TestKeyInRootRecursive(UResourceBundle * root,const char * rootName,UResourceBundle * currentBundle,const char * locale)73 TestKeyInRootRecursive(UResourceBundle *root, const char *rootName,
74 UResourceBundle *currentBundle, const char *locale) {
75 UErrorCode errorCode = U_ZERO_ERROR;
76 UResourceBundle *subRootBundle = NULL, *subBundle = NULL, *arr = NULL;
77
78 ures_resetIterator(root);
79 ures_resetIterator(currentBundle);
80 while (ures_hasNext(currentBundle)) {
81 const char *subBundleKey = NULL;
82 const char *currentBundleKey = NULL;
83
84 errorCode = U_ZERO_ERROR;
85 currentBundleKey = ures_getKey(currentBundle);
86 (void)currentBundleKey; /* Suppress set but not used warning. */
87 subBundle = ures_getNextResource(currentBundle, NULL, &errorCode);
88 if (U_FAILURE(errorCode)) {
89 log_err("Can't open a resource for lnocale %s. Error: %s\n", locale, u_errorName(errorCode));
90 continue;
91 }
92 subBundleKey = ures_getKey(subBundle);
93
94
95 subRootBundle = ures_getByKey(root, subBundleKey, NULL, &errorCode);
96 if (U_FAILURE(errorCode)) {
97 log_err("Can't open a resource with key \"%s\" in \"%s\" from %s for locale \"%s\"\n",
98 subBundleKey,
99 ures_getKey(currentBundle),
100 rootName,
101 locale);
102 ures_close(subBundle);
103 continue;
104 }
105 if (ures_getType(subRootBundle) != ures_getType(subBundle)) {
106 log_err("key \"%s\" in \"%s\" has a different type from root for locale \"%s\"\n"
107 "\troot=%d, locale=%d\n",
108 subBundleKey,
109 ures_getKey(currentBundle),
110 locale,
111 ures_getType(subRootBundle),
112 ures_getType(subBundle));
113 ures_close(subBundle);
114 continue;
115 }
116 else if (ures_getType(subBundle) == URES_INT_VECTOR) {
117 int32_t minSize;
118 int32_t subBundleSize;
119 int32_t idx;
120 UBool sameArray = TRUE;
121 const int32_t *subRootBundleArr = ures_getIntVector(subRootBundle, &minSize, &errorCode);
122 const int32_t *subBundleArr = ures_getIntVector(subBundle, &subBundleSize, &errorCode);
123
124 if (minSize > subBundleSize) {
125 minSize = subBundleSize;
126 log_err("Arrays are different size with key \"%s\" in \"%s\" from root for locale \"%s\"\n",
127 subBundleKey,
128 ures_getKey(currentBundle),
129 locale);
130 }
131
132 for (idx = 0; idx < minSize && sameArray; idx++) {
133 if (subRootBundleArr[idx] != subBundleArr[idx]) {
134 sameArray = FALSE;
135 }
136 if (strcmp(subBundleKey, "DateTimeElements") == 0
137 && (subBundleArr[idx] < 1 || 7 < subBundleArr[idx]))
138 {
139 log_err("Value out of range with key \"%s\" at index %d in \"%s\" for locale \"%s\"\n",
140 subBundleKey,
141 idx,
142 ures_getKey(currentBundle),
143 locale);
144 }
145 }
146 /* Special exception es_US and DateTimeElements */
147 if (sameArray
148 && !(strcmp(locale, "es_US") == 0 && strcmp(subBundleKey, "DateTimeElements") == 0))
149 {
150 log_err("Integer vectors are the same with key \"%s\" in \"%s\" from root for locale \"%s\"\n",
151 subBundleKey,
152 ures_getKey(currentBundle),
153 locale);
154 }
155 }
156 else if (ures_getType(subBundle) == URES_ARRAY) {
157 UResourceBundle *subSubBundle = ures_getByIndex(subBundle, 0, NULL, &errorCode);
158 UResourceBundle *subSubRootBundle = ures_getByIndex(subRootBundle, 0, NULL, &errorCode);
159
160 if (U_SUCCESS(errorCode)
161 && (ures_getType(subSubBundle) == URES_ARRAY || ures_getType(subSubRootBundle) == URES_ARRAY))
162 {
163 /* Here is one of the recursive parts */
164 TestKeyInRootRecursive(subRootBundle, rootName, subBundle, locale);
165 }
166 else {
167 int32_t minSize = ures_getSize(subRootBundle);
168 int32_t idx;
169 UBool sameArray = TRUE;
170
171 if (minSize > ures_getSize(subBundle)) {
172 minSize = ures_getSize(subBundle);
173 }
174
175 if ((subBundleKey == NULL
176 || (subBundleKey != NULL && strcmp(subBundleKey, "LocaleScript") != 0 && !isCurrencyPreEuro(subBundleKey)))
177 && ures_getSize(subRootBundle) != ures_getSize(subBundle))
178 {
179 log_err("Different size array with key \"%s\" in \"%s\" from root for locale \"%s\"\n"
180 "\troot array size=%d, locale array size=%d\n",
181 subBundleKey,
182 ures_getKey(currentBundle),
183 locale,
184 ures_getSize(subRootBundle),
185 ures_getSize(subBundle));
186 }
187 /*
188 if(isCurrencyPreEuro(subBundleKey) && ures_getSize(subBundle)!=3){
189 log_err("Different size array with key \"%s\" in \"%s\" for locale \"%s\" the expected size is 3 got size=%d\n",
190 subBundleKey,
191 ures_getKey(currentBundle),
192 locale,
193 ures_getSize(subBundle));
194 }
195 */
196 for (idx = 0; idx < minSize; idx++) {
197 int32_t rootStrLen, localeStrLen;
198 const UChar *rootStr = ures_getStringByIndex(subRootBundle,idx,&rootStrLen,&errorCode);
199 const UChar *localeStr = ures_getStringByIndex(subBundle,idx,&localeStrLen,&errorCode);
200 if (rootStr && localeStr && U_SUCCESS(errorCode)) {
201 if (u_strcmp(rootStr, localeStr) != 0) {
202 sameArray = FALSE;
203 }
204 }
205 else {
206 if ( rootStrLen > 1 && rootStr[0] == 0x41 && rootStr[1] >= 0x30 && rootStr[1] <= 0x39 ) {
207 /* A2 or A4 in the root string indicates that the resource can optionally be an array instead of a */
208 /* string. Attempt to read it as an array. */
209 errorCode = U_ZERO_ERROR;
210 arr = ures_getByIndex(subBundle,idx,NULL,&errorCode);
211 if (U_FAILURE(errorCode)) {
212 log_err("Got a NULL string with key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n",
213 subBundleKey,
214 ures_getKey(currentBundle),
215 idx,
216 locale);
217 continue;
218 }
219 if (ures_getType(arr) != URES_ARRAY || ures_getSize(arr) != (int32_t)rootStr[1] - 0x30) {
220 log_err("Got something other than a string or array of size %d for key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n",
221 rootStr[1] - 0x30,
222 subBundleKey,
223 ures_getKey(currentBundle),
224 idx,
225 locale);
226 ures_close(arr);
227 continue;
228 }
229 localeStr = ures_getStringByIndex(arr,0,&localeStrLen,&errorCode);
230 ures_close(arr);
231 if (U_FAILURE(errorCode)) {
232 log_err("Got something other than a string or array for key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n",
233 subBundleKey,
234 ures_getKey(currentBundle),
235 idx,
236 locale);
237 continue;
238 }
239 } else {
240 log_err("Got a NULL string with key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n",
241 subBundleKey,
242 ures_getKey(currentBundle),
243 idx,
244 locale);
245 continue;
246 }
247 }
248 if (localeStr[0] == (UChar)0x20) {
249 log_err("key \"%s\" at index %d in \"%s\" starts with a space in locale \"%s\"\n",
250 subBundleKey,
251 idx,
252 ures_getKey(currentBundle),
253 locale);
254 }
255 else if ((localeStr[localeStrLen - 1] == (UChar)0x20) && (strcmp(subBundleKey,"separator") != 0)) {
256 log_err("key \"%s\" at index %d in \"%s\" ends with a space in locale \"%s\"\n",
257 subBundleKey,
258 idx,
259 ures_getKey(currentBundle),
260 locale);
261 }
262 else if (subBundleKey != NULL
263 && strcmp(subBundleKey, "DateTimePatterns") == 0)
264 {
265 int32_t quoted = 0;
266 const UChar *localeStrItr = localeStr;
267 while (*localeStrItr) {
268 if (*localeStrItr == (UChar)0x27 /* ' */) {
269 quoted++;
270 }
271 else if ((quoted % 2) == 0) {
272 /* Search for unquoted characters */
273 if (4 <= idx && idx <= 7
274 && (*localeStrItr == (UChar)0x6B /* k */
275 || *localeStrItr == (UChar)0x48 /* H */
276 || *localeStrItr == (UChar)0x6D /* m */
277 || *localeStrItr == (UChar)0x73 /* s */
278 || *localeStrItr == (UChar)0x53 /* S */
279 || *localeStrItr == (UChar)0x61 /* a */
280 || *localeStrItr == (UChar)0x68 /* h */
281 || *localeStrItr == (UChar)0x7A /* z */))
282 {
283 log_err("key \"%s\" at index %d has time pattern chars in date for locale \"%s\"\n",
284 subBundleKey,
285 idx,
286 locale);
287 }
288 else if (0 <= idx && idx <= 3
289 && (*localeStrItr == (UChar)0x47 /* G */
290 || *localeStrItr == (UChar)0x79 /* y */
291 || *localeStrItr == (UChar)0x4D /* M */
292 || *localeStrItr == (UChar)0x64 /* d */
293 || *localeStrItr == (UChar)0x45 /* E */
294 || *localeStrItr == (UChar)0x44 /* D */
295 || *localeStrItr == (UChar)0x46 /* F */
296 || *localeStrItr == (UChar)0x77 /* w */
297 || *localeStrItr == (UChar)0x57 /* W */))
298 {
299 log_err("key \"%s\" at index %d has date pattern chars in time for locale \"%s\"\n",
300 subBundleKey,
301 idx,
302 locale);
303 }
304 }
305 localeStrItr++;
306 }
307 }
308 else if (idx == 4 && subBundleKey != NULL
309 && strcmp(subBundleKey, "NumberElements") == 0
310 && u_charDigitValue(localeStr[0]) != 0)
311 {
312 log_err("key \"%s\" at index %d has a non-zero based number for locale \"%s\"\n",
313 subBundleKey,
314 idx,
315 locale);
316 }
317 }
318 (void)sameArray; /* Suppress set but not used warning. */
319 /* if (sameArray && strcmp(rootName, "root") == 0) {
320 log_err("Arrays are the same with key \"%s\" in \"%s\" from root for locale \"%s\"\n",
321 subBundleKey,
322 ures_getKey(currentBundle),
323 locale);
324 }*/
325 }
326 ures_close(subSubBundle);
327 ures_close(subSubRootBundle);
328 }
329 else if (ures_getType(subBundle) == URES_STRING) {
330 int32_t len = 0;
331 const UChar *string = ures_getString(subBundle, &len, &errorCode);
332 if (U_FAILURE(errorCode) || string == NULL) {
333 log_err("Can't open a string with key \"%s\" in \"%s\" for locale \"%s\"\n",
334 subBundleKey,
335 ures_getKey(currentBundle),
336 locale);
337 } else if (string[0] == (UChar)0x20) {
338 log_err("key \"%s\" in \"%s\" starts with a space in locale \"%s\"\n",
339 subBundleKey,
340 ures_getKey(currentBundle),
341 locale);
342 /* localeDisplayPattern/separator can end with a space */
343 } else if (string[len - 1] == (UChar)0x20 && (strcmp(subBundleKey,"separator"))) {
344 log_err("key \"%s\" in \"%s\" ends with a space in locale \"%s\"\n",
345 subBundleKey,
346 ures_getKey(currentBundle),
347 locale);
348 } else if (strcmp(subBundleKey, "localPatternChars") == 0) {
349 /* Note: We no longer import localPatternChars data starting
350 * ICU 3.8. So it never comes into this else if block. (ticket#5597)
351 */
352
353 /* Check well-formedness of localPatternChars. First, the
354 * length must match the number of fields defined by
355 * DateFormat. Second, each character in the string must
356 * be in the set [A-Za-z]. Finally, each character must be
357 * unique.
358 */
359 int32_t i,j;
360 #if !UCONFIG_NO_FORMATTING
361 if (len != UDAT_FIELD_COUNT) {
362 log_err("key \"%s\" has the wrong number of characters in locale \"%s\"\n",
363 subBundleKey,
364 locale);
365 }
366 #endif
367 /* Check char validity. */
368 for (i=0; i<len; ++i) {
369 if (!((string[i] >= 65/*'A'*/ && string[i] <= 90/*'Z'*/) ||
370 (string[i] >= 97/*'a'*/ && string[i] <= 122/*'z'*/))) {
371 log_err("key \"%s\" has illegal character '%c' in locale \"%s\"\n",
372 subBundleKey,
373 (char) string[i],
374 locale);
375 }
376 /* Do O(n^2) check for duplicate chars. */
377 for (j=0; j<i; ++j) {
378 if (string[j] == string[i]) {
379 log_err("key \"%s\" has duplicate character '%c' in locale \"%s\"\n",
380 subBundleKey,
381 (char) string[i],
382 locale);
383 }
384 }
385 }
386 }
387 /* No fallback was done. Check for duplicate data */
388 /* The ures_* API does not do fallback of sub-resource bundles,
389 So we can't do this now. */
390 #if 0
391 else if (strcmp(locale, "root") != 0 && errorCode == U_ZERO_ERROR) {
392
393 const UChar *rootString = ures_getString(subRootBundle, &len, &errorCode);
394 if (U_FAILURE(errorCode) || rootString == NULL) {
395 log_err("Can't open a string with key \"%s\" in \"%s\" in root\n",
396 ures_getKey(subRootBundle),
397 ures_getKey(currentBundle));
398 continue;
399 } else if (u_strcmp(string, rootString) == 0) {
400 if (strcmp(locale, "de_CH") != 0 && strcmp(subBundleKey, "Countries") != 0 &&
401 strcmp(subBundleKey, "Version") != 0) {
402 log_err("Found duplicate data with key \"%s\" in \"%s\" in locale \"%s\"\n",
403 ures_getKey(subRootBundle),
404 ures_getKey(currentBundle),
405 locale);
406 }
407 else {
408 /* Ignore for now. */
409 /* Can be fixed if fallback through de locale was done. */
410 log_verbose("Skipping key %s in %s\n", subBundleKey, locale);
411 }
412 }
413 }
414 #endif
415 }
416 else if (ures_getType(subBundle) == URES_TABLE) {
417 if (strcmp(subBundleKey, "availableFormats")!=0) {
418 /* Here is one of the recursive parts */
419 TestKeyInRootRecursive(subRootBundle, rootName, subBundle, locale);
420 }
421 else {
422 log_verbose("Skipping key %s in %s\n", subBundleKey, locale);
423 }
424 }
425 else if (ures_getType(subBundle) == URES_BINARY || ures_getType(subBundle) == URES_INT) {
426 /* Can't do anything to check it */
427 /* We'll assume it's all correct */
428 if (strcmp(subBundleKey, "MeasurementSystem") != 0) {
429 log_verbose("Skipping key \"%s\" in \"%s\" for locale \"%s\"\n",
430 subBundleKey,
431 ures_getKey(currentBundle),
432 locale);
433 }
434 /* Testing for MeasurementSystem is done in VerifyTranslation */
435 }
436 else {
437 log_err("Type %d for key \"%s\" in \"%s\" is unknown for locale \"%s\"\n",
438 ures_getType(subBundle),
439 subBundleKey,
440 ures_getKey(currentBundle),
441 locale);
442 }
443 ures_close(subRootBundle);
444 ures_close(subBundle);
445 }
446 }
447 #endif
448
449 static void
testLCID(UResourceBundle * currentBundle,const char * localeName)450 testLCID(UResourceBundle *currentBundle,
451 const char *localeName)
452 {
453 UErrorCode status = U_ZERO_ERROR;
454 uint32_t expectedLCID;
455 char lcidStringC[64] = {0};
456 int32_t len;
457
458 expectedLCID = uloc_getLCID(localeName);
459 if (expectedLCID == 0) {
460 log_verbose("INFO: %-5s does not have any LCID mapping\n",
461 localeName);
462 return;
463 }
464
465 status = U_ZERO_ERROR;
466 len = uprv_convertToPosix(expectedLCID, lcidStringC, sizeof(lcidStringC)/sizeof(lcidStringC[0]) - 1, &status);
467 if (U_FAILURE(status)) {
468 log_err("ERROR: %.4x does not have a POSIX mapping due to %s\n",
469 expectedLCID, u_errorName(status));
470 }
471 lcidStringC[len] = 0;
472
473 if(strcmp(localeName, lcidStringC) != 0) {
474 char langName[1024];
475 char langLCID[1024];
476 uloc_getLanguage(localeName, langName, sizeof(langName), &status);
477 uloc_getLanguage(lcidStringC, langLCID, sizeof(langLCID), &status);
478
479 if (strcmp(langName, langLCID) == 0) {
480 log_verbose("WARNING: %-5s resolves to %s (0x%.4x)\n",
481 localeName, lcidStringC, expectedLCID);
482 }
483 else {
484 log_err("ERROR: %-5s has 0x%.4x and the number resolves wrongfully to %s\n",
485 localeName, expectedLCID, lcidStringC);
486 }
487 }
488 }
489
490 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
491 static void
TestLocaleStructure(void)492 TestLocaleStructure(void) {
493 // This test checks the locale structure against a key file located
494 // at source/test/testdata/structLocale.txt. When adding new data to
495 // a locale file such as en.txt, the structLocale.txt file must be changed
496 // too to include the the template of the new data. Otherwise this test
497 // will fail!
498
499 UResourceBundle *root, *currentLocale;
500 int32_t locCount = uloc_countAvailable();
501 int32_t locIndex;
502 UErrorCode errorCode = U_ZERO_ERROR;
503 const char *currLoc, *resolvedLoc;
504
505 /* TODO: Compare against parent's data too. This code can't handle fallbacks that some tools do already. */
506 /* char locName[ULOC_FULLNAME_CAPACITY];
507 char *locNamePtr;
508
509 for (locIndex = 0; locIndex < locCount; locIndex++) {
510 errorCode=U_ZERO_ERROR;
511 strcpy(locName, uloc_getAvailable(locIndex));
512 locNamePtr = strrchr(locName, '_');
513 if (locNamePtr) {
514 *locNamePtr = 0;
515 }
516 else {
517 strcpy(locName, "root");
518 }
519
520 root = ures_openDirect(NULL, locName, &errorCode);
521 if(U_FAILURE(errorCode)) {
522 log_err("Can't open %s\n", locName);
523 continue;
524 }
525 */
526 if (locCount <= 1) {
527 log_data_err("At least root needs to be installed\n");
528 }
529
530 root = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode);
531 if(U_FAILURE(errorCode)) {
532 log_data_err("Can't open structLocale\n");
533 return;
534 }
535 for (locIndex = 0; locIndex < locCount; locIndex++) {
536 errorCode=U_ZERO_ERROR;
537 currLoc = uloc_getAvailable(locIndex);
538 currentLocale = ures_open(NULL, currLoc, &errorCode);
539 if(errorCode != U_ZERO_ERROR) {
540 if(U_SUCCESS(errorCode)) {
541 /* It's installed, but there is no data.
542 It's installed for the g18n white paper [grhoten] */
543 log_err("ERROR: Locale %-5s not installed, and it should be, err %s\n",
544 uloc_getAvailable(locIndex), u_errorName(errorCode));
545 } else {
546 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
547 u_errorName(errorCode),
548 uloc_getAvailable(locIndex));
549 }
550 ures_close(currentLocale);
551 continue;
552 }
553 ures_getStringByKey(currentLocale, "Version", NULL, &errorCode);
554 if(errorCode != U_ZERO_ERROR) {
555 log_err("No version information is available for locale %s, and it should be!\n",
556 currLoc);
557 }
558 else if (ures_getStringByKey(currentLocale, "Version", NULL, &errorCode)[0] == (UChar)(0x78)) {
559 log_verbose("WARNING: The locale %s is experimental! It shouldn't be listed as an installed locale.\n",
560 currLoc);
561 }
562 resolvedLoc = ures_getLocaleByType(currentLocale, ULOC_ACTUAL_LOCALE, &errorCode);
563 if (strcmp(resolvedLoc, currLoc) != 0) {
564 /* All locales have at least a Version resource.
565 If it's absolutely empty, then the previous test will fail too.*/
566 log_err("Locale resolves to different locale. Is %s an alias of %s?\n",
567 currLoc, resolvedLoc);
568 }
569 TestKeyInRootRecursive(root, "root", currentLocale, currLoc);
570
571 testLCID(currentLocale, currLoc);
572
573 ures_close(currentLocale);
574 }
575
576 ures_close(root);
577 }
578 #endif
579
580 static void
compareArrays(const char * keyName,UResourceBundle * fromArray,const char * fromLocale,UResourceBundle * toArray,const char * toLocale,int32_t start,int32_t end)581 compareArrays(const char *keyName,
582 UResourceBundle *fromArray, const char *fromLocale,
583 UResourceBundle *toArray, const char *toLocale,
584 int32_t start, int32_t end)
585 {
586 int32_t fromSize = ures_getSize(fromArray);
587 int32_t toSize = ures_getSize(fromArray);
588 int32_t idx;
589 UErrorCode errorCode = U_ZERO_ERROR;
590
591 if (fromSize > toSize) {
592 fromSize = toSize;
593 log_err("Arrays are different size from \"%s\" to \"%s\"\n",
594 fromLocale,
595 toLocale);
596 }
597
598 for (idx = start; idx <= end; idx++) {
599 const UChar *fromBundleStr = ures_getStringByIndex(fromArray, idx, NULL, &errorCode);
600 const UChar *toBundleStr = ures_getStringByIndex(toArray, idx, NULL, &errorCode);
601 if (fromBundleStr && toBundleStr && u_strcmp(fromBundleStr, toBundleStr) != 0)
602 {
603 log_err("Difference for %s at index %d from %s= \"%s\" to %s= \"%s\"\n",
604 keyName,
605 idx,
606 fromLocale,
607 austrdup(fromBundleStr),
608 toLocale,
609 austrdup(toBundleStr));
610 }
611 }
612 }
613
614 static void
compareConsistentCountryInfo(const char * fromLocale,const char * toLocale)615 compareConsistentCountryInfo(const char *fromLocale, const char *toLocale) {
616 UErrorCode errorCode = U_ZERO_ERROR;
617 UResourceBundle *fromArray, *toArray;
618 UResourceBundle *fromLocaleBund = ures_open(NULL, fromLocale, &errorCode);
619 UResourceBundle *toLocaleBund = ures_open(NULL, toLocale, &errorCode);
620 UResourceBundle *toCalendar, *fromCalendar, *toGregorian, *fromGregorian;
621
622 if(U_FAILURE(errorCode)) {
623 log_err("Can't open resource bundle %s or %s - %s\n", fromLocale, toLocale, u_errorName(errorCode));
624 return;
625 }
626 fromCalendar = ures_getByKey(fromLocaleBund, "calendar", NULL, &errorCode);
627 fromGregorian = ures_getByKeyWithFallback(fromCalendar, "gregorian", NULL, &errorCode);
628
629 toCalendar = ures_getByKey(toLocaleBund, "calendar", NULL, &errorCode);
630 toGregorian = ures_getByKeyWithFallback(toCalendar, "gregorian", NULL, &errorCode);
631
632 fromArray = ures_getByKey(fromLocaleBund, "CurrencyElements", NULL, &errorCode);
633 toArray = ures_getByKey(toLocaleBund, "CurrencyElements", NULL, &errorCode);
634 if (strcmp(fromLocale, "en_CA") != 0)
635 {
636 /* The first one is probably localized. */
637 compareArrays("CurrencyElements", fromArray, fromLocale, toArray, toLocale, 1, 2);
638 }
639 ures_close(fromArray);
640 ures_close(toArray);
641
642 fromArray = ures_getByKey(fromLocaleBund, "NumberPatterns", NULL, &errorCode);
643 toArray = ures_getByKey(toLocaleBund, "NumberPatterns", NULL, &errorCode);
644 if (strcmp(fromLocale, "en_CA") != 0)
645 {
646 compareArrays("NumberPatterns", fromArray, fromLocale, toArray, toLocale, 0, 3);
647 }
648 ures_close(fromArray);
649 ures_close(toArray);
650
651 /* Difficult to test properly */
652 /*
653 fromArray = ures_getByKey(fromLocaleBund, "DateTimePatterns", NULL, &errorCode);
654 toArray = ures_getByKey(toLocaleBund, "DateTimePatterns", NULL, &errorCode);
655 {
656 compareArrays("DateTimePatterns", fromArray, fromLocale, toArray, toLocale);
657 }
658 ures_close(fromArray);
659 ures_close(toArray);*/
660
661 fromArray = ures_getByKey(fromLocaleBund, "NumberElements", NULL, &errorCode);
662 toArray = ures_getByKey(toLocaleBund, "NumberElements", NULL, &errorCode);
663 if (strcmp(fromLocale, "en_CA") != 0)
664 {
665 compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 0, 3);
666 /* Index 4 is a script based 0 */
667 compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 5, 10);
668 }
669 ures_close(fromArray);
670 ures_close(toArray);
671 ures_close(fromCalendar);
672 ures_close(toCalendar);
673 ures_close(fromGregorian);
674 ures_close(toGregorian);
675
676 ures_close(fromLocaleBund);
677 ures_close(toLocaleBund);
678 }
679
680 static void
TestConsistentCountryInfo(void)681 TestConsistentCountryInfo(void) {
682 /* UResourceBundle *fromLocale, *toLocale;*/
683 int32_t locCount = uloc_countAvailable();
684 int32_t fromLocIndex, toLocIndex;
685
686 int32_t fromCountryLen, toCountryLen;
687 char fromCountry[ULOC_FULLNAME_CAPACITY], toCountry[ULOC_FULLNAME_CAPACITY];
688
689 int32_t fromVariantLen, toVariantLen;
690 char fromVariant[ULOC_FULLNAME_CAPACITY], toVariant[ULOC_FULLNAME_CAPACITY];
691
692 UErrorCode errorCode = U_ZERO_ERROR;
693
694 for (fromLocIndex = 0; fromLocIndex < locCount; fromLocIndex++) {
695 const char *fromLocale = uloc_getAvailable(fromLocIndex);
696
697 errorCode=U_ZERO_ERROR;
698 fromCountryLen = uloc_getCountry(fromLocale, fromCountry, ULOC_FULLNAME_CAPACITY, &errorCode);
699 if (fromCountryLen <= 0) {
700 /* Ignore countryless locales */
701 continue;
702 }
703 fromVariantLen = uloc_getVariant(fromLocale, fromVariant, ULOC_FULLNAME_CAPACITY, &errorCode);
704 if (fromVariantLen > 0) {
705 /* Most variants are ignorable like PREEURO, or collation variants. */
706 continue;
707 }
708 /* Start comparing only after the current index.
709 Previous loop should have already compared fromLocIndex.
710 */
711 for (toLocIndex = fromLocIndex + 1; toLocIndex < locCount; toLocIndex++) {
712 const char *toLocale = uloc_getAvailable(toLocIndex);
713
714 toCountryLen = uloc_getCountry(toLocale, toCountry, ULOC_FULLNAME_CAPACITY, &errorCode);
715 if(U_FAILURE(errorCode)) {
716 log_err("Unknown failure fromLocale=%s toLocale=%s errorCode=%s\n",
717 fromLocale, toLocale, u_errorName(errorCode));
718 continue;
719 }
720
721 if (toCountryLen <= 0) {
722 /* Ignore countryless locales */
723 continue;
724 }
725 toVariantLen = uloc_getVariant(toLocale, toVariant, ULOC_FULLNAME_CAPACITY, &errorCode);
726 if (toVariantLen > 0) {
727 /* Most variants are ignorable like PREEURO, or collation variants. */
728 /* They're a variant for a reason. */
729 continue;
730 }
731 if (strcmp(fromCountry, toCountry) == 0) {
732 log_verbose("comparing fromLocale=%s toLocale=%s\n",
733 fromLocale, toLocale);
734 compareConsistentCountryInfo(fromLocale, toLocale);
735 }
736 }
737 }
738 }
739
740 static int32_t
findStringSetMismatch(const char * currLoc,const UChar * string,int32_t langSize,USet * mergedExemplarSet,UBool ignoreNumbers,UChar * badCharPtr)741 findStringSetMismatch(const char *currLoc, const UChar *string, int32_t langSize,
742 USet * mergedExemplarSet,
743 UBool ignoreNumbers, UChar* badCharPtr) {
744 UErrorCode errorCode = U_ZERO_ERROR;
745 USet *exemplarSet;
746 int32_t strIdx;
747 if (mergedExemplarSet == NULL) {
748 return -1;
749 }
750 exemplarSet = createFlattenSet(mergedExemplarSet, &errorCode);
751 if (U_FAILURE(errorCode)) {
752 log_err("%s: error createFlattenSet returned %s\n", currLoc, u_errorName(errorCode));
753 return -1;
754 }
755
756 for (strIdx = 0; strIdx < langSize; strIdx++) {
757 if (!uset_contains(exemplarSet, string[strIdx])
758 && 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
759 && string[strIdx] != 0x200C && string[strIdx] != 0x200D) {
760 if (!ignoreNumbers || (ignoreNumbers && (string[strIdx] < 0x30 || string[strIdx] > 0x39))) {
761 uset_close(exemplarSet);
762 if (badCharPtr) {
763 *badCharPtr = string[strIdx];
764 }
765 return strIdx;
766 }
767 }
768 }
769 uset_close(exemplarSet);
770 if (badCharPtr) {
771 *badCharPtr = 0;
772 }
773 return -1;
774 }
775 /* include non-invariant chars */
776 static int32_t
myUCharsToChars(const UChar * us,char * cs,int32_t len)777 myUCharsToChars(const UChar* us, char* cs, int32_t len){
778 int32_t i=0;
779 for(; i< len; i++){
780 if(us[i] < 0x7f){
781 cs[i] = (char)us[i];
782 }else{
783 return -1;
784 }
785 }
786 return i;
787 }
788 static void
findSetMatch(UScriptCode * scriptCodes,int32_t scriptsLen,USet * exemplarSet,const char * locale)789 findSetMatch( UScriptCode *scriptCodes, int32_t scriptsLen,
790 USet *exemplarSet,
791 const char *locale){
792 USet *scripts[10]= {0};
793 char pattern[256] = { '[', ':', 0x000 };
794 int32_t patternLen;
795 UChar uPattern[256] = {0};
796 UErrorCode status = U_ZERO_ERROR;
797 int32_t i;
798
799 /* create the sets with script codes */
800 for(i = 0; i<scriptsLen; i++){
801 strcat(pattern, uscript_getShortName(scriptCodes[i]));
802 strcat(pattern, ":]");
803 patternLen = (int32_t)strlen(pattern);
804 u_charsToUChars(pattern, uPattern, patternLen);
805 scripts[i] = uset_openPattern(uPattern, patternLen, &status);
806 if(U_FAILURE(status)){
807 log_err("Could not create set for pattern %s. Error: %s\n", pattern, u_errorName(status));
808 return;
809 }
810 pattern[2] = 0;
811 }
812 if (strcmp(locale, "uk") == 0 || strcmp(locale, "uk_UA") == 0) {
813 /* Special addition. Add the modifying apostrophe, which isn't in Cyrillic. */
814 uset_add(scripts[0], 0x2bc);
815 }
816 if(U_SUCCESS(status)){
817 UBool existsInScript = FALSE;
818 /* iterate over the exemplarSet and ascertain if all
819 * UChars in exemplarSet belong to the scripts returned
820 * by getScript
821 */
822 int32_t count = uset_getItemCount(exemplarSet);
823
824 for( i=0; i < count; i++){
825 UChar32 start = 0;
826 UChar32 end = 0;
827 UChar *str = NULL;
828 int32_t strCapacity = 0;
829
830 strCapacity = uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status);
831 if(U_SUCCESS(status)){
832 int32_t j;
833 if(strCapacity == 0){
834 /* ok the item is a range */
835 for( j = 0; j < scriptsLen; j++){
836 if(uset_containsRange(scripts[j], start, end) == TRUE){
837 existsInScript = TRUE;
838 }
839 }
840 if(existsInScript == FALSE){
841 for( j = 0; j < scriptsLen; j++){
842 UChar toPattern[500]={'\0'};
843 char pat[500]={'\0'};
844 int32_t len = uset_toPattern(scripts[j], toPattern, 500, TRUE, &status);
845 len = myUCharsToChars(toPattern, pat, len);
846 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));
847 if(len!=-1){
848 log_err("Pattern: %s\n",pat);
849 }
850 }
851 log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale);
852 }
853 }else{
854 strCapacity++; /* increment for NUL termination */
855 /* allocate the str and call the api again */
856 str = (UChar*) malloc(U_SIZEOF_UCHAR * strCapacity);
857 strCapacity = uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status);
858 /* iterate over the scripts and figure out if the string contained is actually
859 * in the script set
860 */
861 for( j = 0; j < scriptsLen; j++){
862 if(uset_containsString(scripts[j],str, strCapacity) == TRUE){
863 existsInScript = TRUE;
864 }
865 }
866 if(existsInScript == FALSE){
867 log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale);
868 }
869 }
870 }
871 }
872
873 }
874
875 /* close the sets */
876 for(i = 0; i<scriptsLen; i++){
877 uset_close(scripts[i]);
878 }
879 }
880
VerifyTranslation(void)881 static void VerifyTranslation(void) {
882 UResourceBundle *root, *currentLocale;
883 int32_t locCount = uloc_countAvailable();
884 int32_t locIndex;
885 UErrorCode errorCode = U_ZERO_ERROR;
886 const char *currLoc;
887 UScriptCode scripts[USCRIPT_CODE_LIMIT];
888 int32_t numScripts;
889 int32_t idx;
890 int32_t end;
891 UResourceBundle *resArray;
892
893 if (locCount <= 1) {
894 log_data_err("At least root needs to be installed\n");
895 }
896
897 root = ures_openDirect(NULL, "root", &errorCode);
898 if(U_FAILURE(errorCode)) {
899 log_data_err("Can't open root\n");
900 return;
901 }
902 for (locIndex = 0; locIndex < locCount; locIndex++) {
903 USet * mergedExemplarSet = NULL;
904 errorCode=U_ZERO_ERROR;
905 currLoc = uloc_getAvailable(locIndex);
906 currentLocale = ures_open(NULL, currLoc, &errorCode);
907 if(errorCode != U_ZERO_ERROR) {
908 if(U_SUCCESS(errorCode)) {
909 /* It's installed, but there is no data.
910 It's installed for the g18n white paper [grhoten] */
911 log_err("ERROR: Locale %-5s not installed, and it should be!\n",
912 uloc_getAvailable(locIndex));
913 } else {
914 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
915 u_errorName(errorCode),
916 uloc_getAvailable(locIndex));
917 }
918 ures_close(currentLocale);
919 continue;
920 }
921 {
922 UErrorCode exemplarStatus = U_ZERO_ERROR;
923 ULocaleData * uld = ulocdata_open(currLoc, &exemplarStatus);
924 if (U_SUCCESS(exemplarStatus)) {
925 USet * exemplarSet = ulocdata_getExemplarSet(uld, NULL, USET_ADD_CASE_MAPPINGS, ULOCDATA_ES_STANDARD, &exemplarStatus);
926 if (U_SUCCESS(exemplarStatus)) {
927 mergedExemplarSet = uset_cloneAsThawed(exemplarSet);
928 uset_close(exemplarSet);
929 exemplarSet = ulocdata_getExemplarSet(uld, NULL, USET_ADD_CASE_MAPPINGS, ULOCDATA_ES_AUXILIARY, &exemplarStatus);
930 if (U_SUCCESS(exemplarStatus)) {
931 uset_addAll(mergedExemplarSet, exemplarSet);
932 uset_close(exemplarSet);
933 }
934 exemplarStatus = U_ZERO_ERROR;
935 exemplarSet = ulocdata_getExemplarSet(uld, NULL, 0, ULOCDATA_ES_PUNCTUATION, &exemplarStatus);
936 if (U_SUCCESS(exemplarStatus)) {
937 uset_addAll(mergedExemplarSet, exemplarSet);
938 uset_close(exemplarSet);
939 }
940 } else {
941 log_err("error ulocdata_getExemplarSet (main) for locale %s returned %s\n", currLoc, u_errorName(errorCode));
942 }
943 ulocdata_close(uld);
944 } else {
945 log_err("error ulocdata_open for locale %s returned %s\n", currLoc, u_errorName(errorCode));
946 }
947 }
948 if (mergedExemplarSet == NULL /*|| (getTestOption(QUICK_OPTION) && uset_size() > 2048)*/) {
949 log_verbose("skipping test for %s\n", currLoc);
950 }
951 //else if (uprv_strncmp(currLoc,"bem",3) == 0 || uprv_strncmp(currLoc,"mgo",3) == 0 || uprv_strncmp(currLoc,"nl",2) == 0) {
952 // log_verbose("skipping test for %s, some month and country names known to use aux exemplars\n", currLoc);
953 //}
954 else {
955 UChar langBuffer[128];
956 int32_t langSize;
957 int32_t strIdx;
958 UChar badChar;
959 langSize = uloc_getDisplayLanguage(currLoc, currLoc, langBuffer, sizeof(langBuffer)/sizeof(langBuffer[0]), &errorCode);
960 if (U_FAILURE(errorCode)) {
961 log_err("error uloc_getDisplayLanguage returned %s\n", u_errorName(errorCode));
962 }
963 else {
964 strIdx = findStringSetMismatch(currLoc, langBuffer, langSize, mergedExemplarSet, FALSE, &badChar);
965 if (strIdx >= 0) {
966 log_err("getDisplayLanguage(%s) at index %d returned characters not in the exemplar characters: %04X.\n",
967 currLoc, strIdx, badChar);
968 }
969 }
970 langSize = uloc_getDisplayCountry(currLoc, currLoc, langBuffer, sizeof(langBuffer)/sizeof(langBuffer[0]), &errorCode);
971 if (U_FAILURE(errorCode)) {
972 log_err("error uloc_getDisplayCountry returned %s\n", u_errorName(errorCode));
973 }
974 {
975 UResourceBundle* cal = ures_getByKey(currentLocale, "calendar", NULL, &errorCode);
976 UResourceBundle* greg = ures_getByKeyWithFallback(cal, "gregorian", NULL, &errorCode);
977 UResourceBundle* names = ures_getByKeyWithFallback(greg, "dayNames", NULL, &errorCode);
978 UResourceBundle* format = ures_getByKeyWithFallback(names, "format", NULL, &errorCode);
979 resArray = ures_getByKeyWithFallback(format, "wide", NULL, &errorCode);
980
981 if (U_FAILURE(errorCode)) {
982 log_err("error ures_getByKey returned %s\n", u_errorName(errorCode));
983 }
984 if (getTestOption(QUICK_OPTION)) {
985 end = 1;
986 }
987 else {
988 end = ures_getSize(resArray);
989 }
990
991
992 for (idx = 0; idx < end; idx++) {
993 const UChar *fromBundleStr = ures_getStringByIndex(resArray, idx, &langSize, &errorCode);
994 if (U_FAILURE(errorCode)) {
995 log_err("error ures_getStringByIndex(%d) returned %s\n", idx, u_errorName(errorCode));
996 continue;
997 }
998 strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, mergedExemplarSet, TRUE, &badChar);
999 if (strIdx >= 0) {
1000 log_err("getDayNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n",
1001 currLoc, idx, strIdx, badChar);
1002 }
1003 }
1004 ures_close(resArray);
1005 ures_close(format);
1006 ures_close(names);
1007
1008 names = ures_getByKeyWithFallback(greg, "monthNames", NULL, &errorCode);
1009 format = ures_getByKeyWithFallback(names,"format", NULL, &errorCode);
1010 resArray = ures_getByKeyWithFallback(format, "wide", NULL, &errorCode);
1011 if (U_FAILURE(errorCode)) {
1012 log_err("error ures_getByKey returned %s\n", u_errorName(errorCode));
1013 }
1014 if (getTestOption(QUICK_OPTION)) {
1015 end = 1;
1016 }
1017 else {
1018 end = ures_getSize(resArray);
1019 }
1020
1021 for (idx = 0; idx < end; idx++) {
1022 const UChar *fromBundleStr = ures_getStringByIndex(resArray, idx, &langSize, &errorCode);
1023 if (U_FAILURE(errorCode)) {
1024 log_err("error ures_getStringByIndex(%d) returned %s\n", idx, u_errorName(errorCode));
1025 continue;
1026 }
1027 strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, mergedExemplarSet, TRUE, &badChar);
1028 if (strIdx >= 0) {
1029 log_err("getMonthNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n",
1030 currLoc, idx, strIdx, badChar);
1031 }
1032 }
1033 ures_close(resArray);
1034 ures_close(format);
1035 ures_close(names);
1036 ures_close(greg);
1037 ures_close(cal);
1038 }
1039 errorCode = U_ZERO_ERROR;
1040 numScripts = uscript_getCode(currLoc, scripts, sizeof(scripts)/sizeof(scripts[0]), &errorCode);
1041 if (strcmp(currLoc, "yi") == 0 && numScripts > 0 && log_knownIssue("11217", "Fix result of uscript_getCode for yi: USCRIPT_YI -> USCRIPT_HEBREW")) {
1042 scripts[0] = USCRIPT_HEBREW;
1043 }
1044 if (numScripts == 0) {
1045 log_err("uscript_getCode(%s) doesn't work.\n", currLoc);
1046 }else if(scripts[0] == USCRIPT_COMMON){
1047 log_err("uscript_getCode(%s) returned USCRIPT_COMMON.\n", currLoc);
1048 }
1049
1050 /* test that the scripts are a superset of exemplar characters. */
1051 {
1052 ULocaleData *uld = ulocdata_open(currLoc,&errorCode);
1053 USet *exemplarSet = ulocdata_getExemplarSet(uld, NULL, 0, ULOCDATA_ES_STANDARD, &errorCode);
1054 /* test if exemplar characters are part of script code */
1055 findSetMatch(scripts, numScripts, exemplarSet, currLoc);
1056 uset_close(exemplarSet);
1057 ulocdata_close(uld);
1058 }
1059
1060 /* test that the paperSize API works */
1061 {
1062 int32_t height=0, width=0;
1063 ulocdata_getPaperSize(currLoc, &height, &width, &errorCode);
1064 if(U_FAILURE(errorCode)){
1065 log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1066 }
1067 if(strstr(currLoc, "_US")!=NULL && height != 279 && width != 216 ){
1068 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc);
1069 }
1070 }
1071 /* test that the MeasurementSystem API works */
1072 {
1073 char fullLoc[ULOC_FULLNAME_CAPACITY];
1074 UMeasurementSystem measurementSystem;
1075 int32_t height = 0, width = 0;
1076
1077 uloc_addLikelySubtags(currLoc, fullLoc, ULOC_FULLNAME_CAPACITY, &errorCode);
1078
1079 errorCode = U_ZERO_ERROR;
1080 measurementSystem = ulocdata_getMeasurementSystem(currLoc, &errorCode);
1081 if (U_FAILURE(errorCode)) {
1082 log_err("ulocdata_getMeasurementSystem failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1083 } else {
1084 if ( strstr(fullLoc, "_US")!=NULL || strstr(fullLoc, "_MM")!=NULL || strstr(fullLoc, "_LR")!=NULL ) {
1085 if(measurementSystem != UMS_US){
1086 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1087 }
1088 } else if ( strstr(fullLoc, "_GB")!=NULL ) {
1089 if(measurementSystem != UMS_UK){
1090 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1091 }
1092 } else if (measurementSystem != UMS_SI) {
1093 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1094 }
1095 }
1096
1097 errorCode = U_ZERO_ERROR;
1098 ulocdata_getPaperSize(currLoc, &height, &width, &errorCode);
1099 if (U_FAILURE(errorCode)) {
1100 log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1101 } else {
1102 if ( strstr(fullLoc, "_US")!=NULL || strstr(fullLoc, "_BZ")!=NULL || strstr(fullLoc, "_CA")!=NULL || strstr(fullLoc, "_CL")!=NULL ||
1103 strstr(fullLoc, "_CO")!=NULL || strstr(fullLoc, "_CR")!=NULL || strstr(fullLoc, "_GT")!=NULL || strstr(fullLoc, "_MX")!=NULL ||
1104 strstr(fullLoc, "_NI")!=NULL || strstr(fullLoc, "_PA")!=NULL || strstr(fullLoc, "_PH")!=NULL || strstr(fullLoc, "_PR")!=NULL ||
1105 strstr(fullLoc, "_SV")!=NULL || strstr(fullLoc, "_VE")!=NULL ) {
1106 if (height != 279 || width != 216) {
1107 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc);
1108 }
1109 } else if (height != 297 || width != 210) {
1110 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc);
1111 }
1112 }
1113 }
1114 }
1115 if (mergedExemplarSet != NULL) {
1116 uset_close(mergedExemplarSet);
1117 }
1118 ures_close(currentLocale);
1119 }
1120
1121 ures_close(root);
1122 }
1123
1124 /* adjust this limit as appropriate */
1125 #define MAX_SCRIPTS_PER_LOCALE 8
1126
TestExemplarSet(void)1127 static void TestExemplarSet(void){
1128 int32_t i, j, k, m, n;
1129 int32_t equalCount = 0;
1130 UErrorCode ec = U_ZERO_ERROR;
1131 UEnumeration* avail;
1132 USet* exemplarSets[2];
1133 USet* unassignedSet;
1134 UScriptCode code[MAX_SCRIPTS_PER_LOCALE];
1135 USet* codeSets[MAX_SCRIPTS_PER_LOCALE];
1136 int32_t codeLen;
1137 char cbuf[32]; /* 9 should be enough */
1138 UChar ubuf[64]; /* adjust as needed */
1139 UBool existsInScript;
1140 int32_t itemCount;
1141 int32_t strLen;
1142 UChar32 start, end;
1143
1144 unassignedSet = NULL;
1145 exemplarSets[0] = NULL;
1146 exemplarSets[1] = NULL;
1147 for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) {
1148 codeSets[i] = NULL;
1149 }
1150
1151 avail = ures_openAvailableLocales(NULL, &ec);
1152 if (!assertSuccess("ures_openAvailableLocales", &ec)) goto END;
1153 n = uenum_count(avail, &ec);
1154 if (!assertSuccess("uenum_count", &ec)) goto END;
1155
1156 u_uastrcpy(ubuf, "[:unassigned:]");
1157 unassignedSet = uset_openPattern(ubuf, -1, &ec);
1158 if (!assertSuccess("uset_openPattern", &ec)) goto END;
1159
1160 for(i=0; i<n; i++){
1161 const char* locale = uenum_next(avail, NULL, &ec);
1162 if (!assertSuccess("uenum_next", &ec)) goto END;
1163 log_verbose("%s\n", locale);
1164 for (k=0; k<2; ++k) {
1165 uint32_t option = (k==0) ? 0 : USET_CASE_INSENSITIVE;
1166 ULocaleData *uld = ulocdata_open(locale,&ec);
1167 USet* exemplarSet = ulocdata_getExemplarSet(uld,NULL, option, ULOCDATA_ES_STANDARD, &ec);
1168 uset_close(exemplarSets[k]);
1169 ulocdata_close(uld);
1170 exemplarSets[k] = exemplarSet;
1171 if (!assertSuccess("ulocaledata_getExemplarSet", &ec)) goto END;
1172
1173 if (uset_containsSome(exemplarSet, unassignedSet)) {
1174 log_err("ExemplarSet contains unassigned characters for locale : %s\n", locale);
1175 }
1176 codeLen = uscript_getCode(locale, code, 8, &ec);
1177 if (strcmp(locale, "yi") == 0 && codeLen > 0 && log_knownIssue("11217", "Fix result of uscript_getCode for yi: USCRIPT_YI -> USCRIPT_HEBREW")) {
1178 code[0] = USCRIPT_HEBREW;
1179 }
1180 if (!assertSuccess("uscript_getCode", &ec)) goto END;
1181
1182 for (j=0; j<MAX_SCRIPTS_PER_LOCALE; ++j) {
1183 uset_close(codeSets[j]);
1184 codeSets[j] = NULL;
1185 }
1186 for (j=0; j<codeLen; ++j) {
1187 uprv_strcpy(cbuf, "[:");
1188 if(code[j]==-1){
1189 log_err("USCRIPT_INVALID_CODE returned for locale: %s\n", locale);
1190 continue;
1191 }
1192 uprv_strcat(cbuf, uscript_getShortName(code[j]));
1193 uprv_strcat(cbuf, ":]");
1194 u_uastrcpy(ubuf, cbuf);
1195 codeSets[j] = uset_openPattern(ubuf, -1, &ec);
1196 }
1197 if (!assertSuccess("uset_openPattern", &ec)) goto END;
1198
1199 existsInScript = FALSE;
1200 itemCount = uset_getItemCount(exemplarSet);
1201 for (m=0; m<itemCount && !existsInScript; ++m) {
1202 strLen = uset_getItem(exemplarSet, m, &start, &end, ubuf,
1203 sizeof(ubuf)/sizeof(ubuf[0]), &ec);
1204 /* failure here might mean str[] needs to be larger */
1205 if (!assertSuccess("uset_getItem", &ec)) goto END;
1206 if (strLen == 0) {
1207 for (j=0; j<codeLen; ++j) {
1208 if (codeSets[j]!=NULL && uset_containsRange(codeSets[j], start, end)) {
1209 existsInScript = TRUE;
1210 break;
1211 }
1212 }
1213 } else {
1214 for (j=0; j<codeLen; ++j) {
1215 if (codeSets[j]!=NULL && uset_containsString(codeSets[j], ubuf, strLen)) {
1216 existsInScript = TRUE;
1217 break;
1218 }
1219 }
1220 }
1221 }
1222
1223 if (existsInScript == FALSE){
1224 log_err("ExemplarSet containment failed for locale : %s\n", locale);
1225 }
1226 }
1227 assertTrue("case-folded is a superset",
1228 uset_containsAll(exemplarSets[1], exemplarSets[0]));
1229 if (uset_equals(exemplarSets[1], exemplarSets[0])) {
1230 ++equalCount;
1231 }
1232 }
1233 /* Note: The case-folded set should sometimes be a strict superset
1234 and sometimes be equal. */
1235 assertTrue("case-folded is sometimes a strict superset, and sometimes equal",
1236 equalCount > 0 && equalCount < n);
1237
1238 END:
1239 uenum_close(avail);
1240 uset_close(exemplarSets[0]);
1241 uset_close(exemplarSets[1]);
1242 uset_close(unassignedSet);
1243 for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) {
1244 uset_close(codeSets[i]);
1245 }
1246 }
1247
1248 enum { kUBufMax = 32 };
TestLocaleDisplayPattern(void)1249 static void TestLocaleDisplayPattern(void){
1250 UErrorCode status;
1251 UChar pattern[kUBufMax] = {0,};
1252 UChar separator[kUBufMax] = {0,};
1253 ULocaleData *uld;
1254 static const UChar enExpectPat[] = { 0x007B,0x0030,0x007D,0x0020,0x0028,0x007B,0x0031,0x007D,0x0029,0 }; /* "{0} ({1})" */
1255 static const UChar enExpectSep[] = { 0x002C,0x0020,0 }; /* ", " */
1256 static const UChar zhExpectPat[] = { 0x007B,0x0030,0x007D,0xFF08,0x007B,0x0031,0x007D,0xFF09,0 };
1257 static const UChar zhExpectSep[] = { 0x3001,0 };
1258
1259 status = U_ZERO_ERROR;
1260 uld = ulocdata_open("en", &status);
1261 if(U_FAILURE(status)){
1262 log_data_err("ulocdata_open en error %s", u_errorName(status));
1263 } else {
1264 ulocdata_getLocaleDisplayPattern(uld, pattern, kUBufMax, &status);
1265 if (U_FAILURE(status)){
1266 log_err("ulocdata_getLocaleDisplayPattern en error %s", u_errorName(status));
1267 } else if (u_strcmp(pattern, enExpectPat) != 0) {
1268 log_err("ulocdata_getLocaleDisplayPattern en returns unexpected pattern");
1269 }
1270 status = U_ZERO_ERROR;
1271 ulocdata_getLocaleSeparator(uld, separator, kUBufMax, &status);
1272 if (U_FAILURE(status)){
1273 log_err("ulocdata_getLocaleSeparator en error %s", u_errorName(status));
1274 } else if (u_strcmp(separator, enExpectSep) != 0) {
1275 log_err("ulocdata_getLocaleSeparator en returns unexpected string ");
1276 }
1277 ulocdata_close(uld);
1278 }
1279
1280 status = U_ZERO_ERROR;
1281 uld = ulocdata_open("zh", &status);
1282 if(U_FAILURE(status)){
1283 log_data_err("ulocdata_open zh error %s", u_errorName(status));
1284 } else {
1285 ulocdata_getLocaleDisplayPattern(uld, pattern, kUBufMax, &status);
1286 if (U_FAILURE(status)){
1287 log_err("ulocdata_getLocaleDisplayPattern zh error %s", u_errorName(status));
1288 } else if (u_strcmp(pattern, zhExpectPat) != 0) {
1289 log_err("ulocdata_getLocaleDisplayPattern zh returns unexpected pattern");
1290 }
1291 status = U_ZERO_ERROR;
1292 ulocdata_getLocaleSeparator(uld, separator, kUBufMax, &status);
1293 if (U_FAILURE(status)){
1294 log_err("ulocdata_getLocaleSeparator zh error %s", u_errorName(status));
1295 } else if (u_strcmp(separator, zhExpectSep) != 0) {
1296 log_err("ulocdata_getLocaleSeparator zh returns unexpected string ");
1297 }
1298 ulocdata_close(uld);
1299 }
1300 }
1301
TestCoverage(void)1302 static void TestCoverage(void){
1303 ULocaleDataDelimiterType types[] = {
1304 ULOCDATA_QUOTATION_START, /* Quotation start */
1305 ULOCDATA_QUOTATION_END, /* Quotation end */
1306 ULOCDATA_ALT_QUOTATION_START, /* Alternate quotation start */
1307 ULOCDATA_ALT_QUOTATION_END, /* Alternate quotation end */
1308 ULOCDATA_DELIMITER_COUNT
1309 };
1310 int i;
1311 UBool sub;
1312 UErrorCode status = U_ZERO_ERROR;
1313 ULocaleData *uld = ulocdata_open(uloc_getDefault(), &status);
1314
1315 if(U_FAILURE(status)){
1316 log_data_err("ulocdata_open error");
1317 return;
1318 }
1319
1320
1321 for(i = 0; i < ULOCDATA_DELIMITER_COUNT; i++){
1322 UChar result[32] = {0,};
1323 status = U_ZERO_ERROR;
1324 ulocdata_getDelimiter(uld, types[i], result, 32, &status);
1325 if (U_FAILURE(status)){
1326 log_err("ulocdata_getgetDelimiter error with type %d", types[i]);
1327 }
1328 }
1329
1330 sub = ulocdata_getNoSubstitute(uld);
1331 ulocdata_setNoSubstitute(uld,sub);
1332 ulocdata_close(uld);
1333 }
1334
TestIndexChars(void)1335 static void TestIndexChars(void) {
1336 /* Very basic test of ULOCDATA_ES_INDEX.
1337 * No comprehensive test of data, just basic check that the code path is alive.
1338 */
1339 UErrorCode status = U_ZERO_ERROR;
1340 ULocaleData *uld;
1341 USet *exemplarChars;
1342 USet *indexChars;
1343
1344 uld = ulocdata_open("en", &status);
1345 exemplarChars = uset_openEmpty();
1346 indexChars = uset_openEmpty();
1347 ulocdata_getExemplarSet(uld, exemplarChars, 0, ULOCDATA_ES_STANDARD, &status);
1348 ulocdata_getExemplarSet(uld, indexChars, 0, ULOCDATA_ES_INDEX, &status);
1349 if (U_FAILURE(status)) {
1350 log_data_err("File %s, line %d, Failure opening exemplar chars: %s", __FILE__, __LINE__, u_errorName(status));
1351 goto close_sets;
1352 }
1353 /* en data, standard exemplars are [a-z], lower case. */
1354 /* en data, index characters are [A-Z], upper case. */
1355 if ((uset_contains(exemplarChars, (UChar32)0x41) || uset_contains(indexChars, (UChar32)0x61))) {
1356 log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__, __LINE__ );
1357 goto close_sets;
1358 }
1359 if (!(uset_contains(exemplarChars, (UChar32)0x61) && uset_contains(indexChars, (UChar32)0x41) )) {
1360 log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__, __LINE__ );
1361 goto close_sets;
1362 }
1363
1364 close_sets:
1365 uset_close(exemplarChars);
1366 uset_close(indexChars);
1367 ulocdata_close(uld);
1368 }
1369
1370
1371
1372 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
TestCurrencyList(void)1373 static void TestCurrencyList(void){
1374 #if !UCONFIG_NO_FORMATTING
1375 UErrorCode errorCode = U_ZERO_ERROR;
1376 int32_t structLocaleCount, currencyCount;
1377 UEnumeration *en = ucurr_openISOCurrencies(UCURR_ALL, &errorCode);
1378 const char *isoCode, *structISOCode;
1379 UResourceBundle *subBundle;
1380 UResourceBundle *currencies = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode);
1381 if(U_FAILURE(errorCode)) {
1382 log_data_err("Can't open structLocale\n");
1383 return;
1384 }
1385 currencies = ures_getByKey(currencies, "Currencies", currencies, &errorCode);
1386 currencyCount = uenum_count(en, &errorCode);
1387 structLocaleCount = ures_getSize(currencies);
1388 if (currencyCount != structLocaleCount) {
1389 log_err("structLocale(%d) and ISO4217(%d) currency list are out of sync.\n", structLocaleCount, currencyCount);
1390 #if U_CHARSET_FAMILY == U_ASCII_FAMILY
1391 ures_resetIterator(currencies);
1392 while ((isoCode = uenum_next(en, NULL, &errorCode)) != NULL && ures_hasNext(currencies)) {
1393 subBundle = ures_getNextResource(currencies, NULL, &errorCode);
1394 structISOCode = ures_getKey(subBundle);
1395 ures_close(subBundle);
1396 if (strcmp(structISOCode, isoCode) != 0) {
1397 log_err("First difference found at structLocale(%s) and ISO4217(%s).\n", structISOCode, isoCode);
1398 break;
1399 }
1400 }
1401 #endif
1402 }
1403 ures_close(currencies);
1404 uenum_close(en);
1405 #endif
1406 }
1407 #endif
1408
TestAvailableIsoCodes(void)1409 static void TestAvailableIsoCodes(void){
1410 #if !UCONFIG_NO_FORMATTING
1411 UErrorCode errorCode = U_ZERO_ERROR;
1412 const char* eurCode = "EUR";
1413 const char* usdCode = "USD";
1414 const char* lastCode = "RHD";
1415 const char* zzzCode = "ZZZ";
1416 UDate date1950 = (UDate)-630720000000.0;/* year 1950 */
1417 UDate date1970 = (UDate)0.0; /* year 1970 */
1418 UDate date1975 = (UDate)173448000000.0; /* year 1975 */
1419 UDate date1978 = (UDate)260172000000.0; /* year 1978 */
1420 UDate date1981 = (UDate)346896000000.0; /* year 1981 */
1421 UDate date1992 = (UDate)693792000000.0; /* year 1992 */
1422 UChar* isoCode = (UChar*)malloc(sizeof(UChar) * (uprv_strlen(usdCode) + 1));
1423
1424 /* testing available codes with no time ranges */
1425 u_charsToUChars(eurCode, isoCode, uprv_strlen(usdCode) + 1);
1426 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) {
1427 log_data_err("FAIL: ISO code (%s) is not found.\n", eurCode);
1428 }
1429
1430 u_charsToUChars(usdCode, isoCode, uprv_strlen(zzzCode) + 1);
1431 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) {
1432 log_data_err("FAIL: ISO code (%s) is not found.\n", usdCode);
1433 }
1434
1435 u_charsToUChars(zzzCode, isoCode, uprv_strlen(zzzCode) + 1);
1436 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == TRUE) {
1437 log_err("FAIL: ISO code (%s) is reported as available, but it doesn't exist.\n", zzzCode);
1438 }
1439
1440 u_charsToUChars(lastCode, isoCode, uprv_strlen(zzzCode) + 1);
1441 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) {
1442 log_data_err("FAIL: ISO code (%s) is not found.\n", lastCode);
1443 }
1444
1445 /* RHD was used from 1970-02-17 to 1980-04-18*/
1446
1447 /* to = null */
1448 if (ucurr_isAvailable(isoCode, date1970, U_DATE_MAX, &errorCode) == FALSE) {
1449 log_data_err("FAIL: ISO code (%s) was available in time range >1970-01-01.\n", lastCode);
1450 }
1451
1452 if (ucurr_isAvailable(isoCode, date1975, U_DATE_MAX, &errorCode) == FALSE) {
1453 log_data_err("FAIL: ISO code (%s) was available in time range >1975.\n", lastCode);
1454 }
1455
1456 if (ucurr_isAvailable(isoCode, date1981, U_DATE_MAX, &errorCode) == TRUE) {
1457 log_err("FAIL: ISO code (%s) was not available in time range >1981.\n", lastCode);
1458 }
1459
1460 /* from = null */
1461 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1970, &errorCode) == TRUE) {
1462 log_err("FAIL: ISO code (%s) was not available in time range <1970.\n", lastCode);
1463 }
1464
1465 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1975, &errorCode) == FALSE) {
1466 log_data_err("FAIL: ISO code (%s) was available in time range <1975.\n", lastCode);
1467 }
1468
1469 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1981, &errorCode) == FALSE) {
1470 log_data_err("FAIL: ISO code (%s) was available in time range <1981.\n", lastCode);
1471 }
1472
1473 /* full ranges */
1474 if (ucurr_isAvailable(isoCode, date1975, date1978, &errorCode) == FALSE) {
1475 log_data_err("FAIL: ISO code (%s) was available in time range 1975-1978.\n", lastCode);
1476 }
1477
1478 if (ucurr_isAvailable(isoCode, date1970, date1975, &errorCode) == FALSE) {
1479 log_data_err("FAIL: ISO code (%s) was available in time range 1970-1975.\n", lastCode);
1480 }
1481
1482 if (ucurr_isAvailable(isoCode, date1975, date1981, &errorCode) == FALSE) {
1483 log_data_err("FAIL: ISO code (%s) was available in time range 1975-1981.\n", lastCode);
1484 }
1485
1486 if (ucurr_isAvailable(isoCode, date1970, date1981, &errorCode) == FALSE) {
1487 log_data_err("FAIL: ISO code (%s) was available in time range 1970-1981.\n", lastCode);
1488 }
1489
1490 if (ucurr_isAvailable(isoCode, date1981, date1992, &errorCode) == TRUE) {
1491 log_err("FAIL: ISO code (%s) was not available in time range 1981-1992.\n", lastCode);
1492 }
1493
1494 if (ucurr_isAvailable(isoCode, date1950, date1970, &errorCode) == TRUE) {
1495 log_err("FAIL: ISO code (%s) was not available in time range 1950-1970.\n", lastCode);
1496 }
1497
1498 /* wrong range - from > to*/
1499 if (ucurr_isAvailable(isoCode, date1975, date1970, &errorCode) == TRUE) {
1500 log_err("FAIL: Wrong range 1975-1970 for ISO code (%s) was not reported.\n", lastCode);
1501 } else if (errorCode != U_ILLEGAL_ARGUMENT_ERROR) {
1502 log_data_err("FAIL: Error code not reported for wrong range 1975-1970 for ISO code (%s).\n", lastCode);
1503 }
1504
1505 free(isoCode);
1506 #endif
1507 }
1508
1509 #define TESTCASE(name) addTest(root, &name, "tsutil/cldrtest/" #name)
1510
1511 void addCLDRTest(TestNode** root);
1512
addCLDRTest(TestNode ** root)1513 void addCLDRTest(TestNode** root)
1514 {
1515 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
1516 TESTCASE(TestLocaleStructure);
1517 TESTCASE(TestCurrencyList);
1518 #endif
1519 TESTCASE(TestConsistentCountryInfo);
1520 TESTCASE(VerifyTranslation);
1521 TESTCASE(TestExemplarSet);
1522 TESTCASE(TestLocaleDisplayPattern);
1523 TESTCASE(TestCoverage);
1524 TESTCASE(TestIndexChars);
1525 TESTCASE(TestAvailableIsoCodes);
1526 }
1527
1528