1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /********************************************************************
4 * COPYRIGHT:
5 * Copyright (c) 1997-2016, International Business Machines Corporation and
6 * others. All Rights Reserved.
7 ********************************************************************/
8
9 #include "cintltst.h"
10 #include "unicode/ures.h"
11 #include "unicode/ucurr.h"
12 #include "unicode/ustring.h"
13 #include "unicode/uset.h"
14 #include "unicode/udat.h"
15 #include "unicode/uscript.h"
16 #include "unicode/ulocdata.h"
17 #include "unicode/utf16.h"
18 #include "cmemory.h"
19 #include "cstring.h"
20 #include "locmap.h"
21 #include "uresimp.h"
22
23 /*
24 returns a new UnicodeSet that is a flattened form of the original
25 UnicodeSet.
26 */
27 static USet*
createFlattenSet(USet * origSet,UErrorCode * status)28 createFlattenSet(USet *origSet, UErrorCode *status) {
29
30
31 USet *newSet = NULL;
32 int32_t origItemCount = 0;
33 int32_t idx, graphmeSize;
34 UChar32 start, end;
35 UChar graphme[64];
36 if (U_FAILURE(*status)) {
37 log_err("createFlattenSet called with %s\n", u_errorName(*status));
38 return NULL;
39 }
40 newSet = uset_open(1, 0);
41 origItemCount = uset_getItemCount(origSet);
42 for (idx = 0; idx < origItemCount; idx++) {
43 graphmeSize = uset_getItem(origSet, idx,
44 &start, &end,
45 graphme, UPRV_LENGTHOF(graphme),
46 status);
47 if (U_FAILURE(*status)) {
48 log_err("ERROR: uset_getItem returned %s\n", u_errorName(*status));
49 *status = U_ZERO_ERROR;
50 }
51 if (graphmeSize) {
52 uset_addAllCodePoints(newSet, graphme, graphmeSize);
53 }
54 else {
55 uset_addRange(newSet, start, end);
56 }
57 }
58 uset_closeOver(newSet,USET_CASE_INSENSITIVE);
59 return newSet;
60 }
61
62 static UBool
isCurrencyPreEuro(const char * currencyKey)63 isCurrencyPreEuro(const char* currencyKey){
64 if( strcmp(currencyKey, "PTE") == 0 ||
65 strcmp(currencyKey, "ESP") == 0 ||
66 strcmp(currencyKey, "LUF") == 0 ||
67 strcmp(currencyKey, "GRD") == 0 ||
68 strcmp(currencyKey, "BEF") == 0 ||
69 strcmp(currencyKey, "ITL") == 0 ||
70 strcmp(currencyKey, "EEK") == 0){
71 return TRUE;
72 }
73 return FALSE;
74 }
75 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
76 static void
TestKeyInRootRecursive(UResourceBundle * root,const char * rootName,UResourceBundle * currentBundle,const char * locale)77 TestKeyInRootRecursive(UResourceBundle *root, const char *rootName,
78 UResourceBundle *currentBundle, const char *locale) {
79 UErrorCode errorCode = U_ZERO_ERROR;
80 UResourceBundle *subRootBundle = NULL, *subBundle = NULL, *arr = NULL;
81
82 ures_resetIterator(root);
83 ures_resetIterator(currentBundle);
84 while (ures_hasNext(currentBundle)) {
85 const char *subBundleKey = NULL;
86 const char *currentBundleKey = NULL;
87
88 errorCode = U_ZERO_ERROR;
89 currentBundleKey = ures_getKey(currentBundle);
90 (void)currentBundleKey; /* Suppress set but not used warning. */
91 subBundle = ures_getNextResource(currentBundle, NULL, &errorCode);
92 if (U_FAILURE(errorCode)) {
93 log_err("Can't open a resource for lnocale %s. Error: %s\n", locale, u_errorName(errorCode));
94 continue;
95 }
96 subBundleKey = ures_getKey(subBundle);
97
98
99 subRootBundle = ures_getByKey(root, subBundleKey, NULL, &errorCode);
100 if (U_FAILURE(errorCode)) {
101 log_err("Can't open a resource with key \"%s\" in \"%s\" from %s for locale \"%s\"\n",
102 subBundleKey,
103 ures_getKey(currentBundle),
104 rootName,
105 locale);
106 ures_close(subBundle);
107 continue;
108 }
109 if (ures_getType(subRootBundle) != ures_getType(subBundle)) {
110 log_err("key \"%s\" in \"%s\" has a different type from root for locale \"%s\"\n"
111 "\troot=%d, locale=%d\n",
112 subBundleKey,
113 ures_getKey(currentBundle),
114 locale,
115 ures_getType(subRootBundle),
116 ures_getType(subBundle));
117 ures_close(subBundle);
118 continue;
119 }
120 else if (ures_getType(subBundle) == URES_INT_VECTOR) {
121 int32_t minSize;
122 int32_t subBundleSize;
123 int32_t idx;
124 UBool sameArray = TRUE;
125 const int32_t *subRootBundleArr = ures_getIntVector(subRootBundle, &minSize, &errorCode);
126 const int32_t *subBundleArr = ures_getIntVector(subBundle, &subBundleSize, &errorCode);
127
128 if (minSize > subBundleSize) {
129 minSize = subBundleSize;
130 log_err("Arrays are different size with key \"%s\" in \"%s\" from root for locale \"%s\"\n",
131 subBundleKey,
132 ures_getKey(currentBundle),
133 locale);
134 }
135
136 for (idx = 0; idx < minSize && sameArray; idx++) {
137 if (subRootBundleArr[idx] != subBundleArr[idx]) {
138 sameArray = FALSE;
139 }
140 if (strcmp(subBundleKey, "DateTimeElements") == 0
141 && (subBundleArr[idx] < 1 || 7 < subBundleArr[idx]))
142 {
143 log_err("Value out of range with key \"%s\" at index %d in \"%s\" for locale \"%s\"\n",
144 subBundleKey,
145 idx,
146 ures_getKey(currentBundle),
147 locale);
148 }
149 }
150 /* Special exception es_US and DateTimeElements */
151 if (sameArray
152 && !(strcmp(locale, "es_US") == 0 && strcmp(subBundleKey, "DateTimeElements") == 0))
153 {
154 log_err("Integer vectors are the same with key \"%s\" in \"%s\" from root for locale \"%s\"\n",
155 subBundleKey,
156 ures_getKey(currentBundle),
157 locale);
158 }
159 }
160 else if (ures_getType(subBundle) == URES_ARRAY) {
161 UResourceBundle *subSubBundle = ures_getByIndex(subBundle, 0, NULL, &errorCode);
162 UResourceBundle *subSubRootBundle = ures_getByIndex(subRootBundle, 0, NULL, &errorCode);
163
164 if (U_SUCCESS(errorCode)
165 && (ures_getType(subSubBundle) == URES_ARRAY || ures_getType(subSubRootBundle) == URES_ARRAY))
166 {
167 /* Here is one of the recursive parts */
168 TestKeyInRootRecursive(subRootBundle, rootName, subBundle, locale);
169 }
170 else {
171 int32_t minSize = ures_getSize(subRootBundle);
172 int32_t idx;
173 UBool sameArray = TRUE;
174
175 if (minSize > ures_getSize(subBundle)) {
176 minSize = ures_getSize(subBundle);
177 }
178
179 if ((subBundleKey == NULL
180 || (subBundleKey != NULL && strcmp(subBundleKey, "LocaleScript") != 0 && !isCurrencyPreEuro(subBundleKey)))
181 && ures_getSize(subRootBundle) != ures_getSize(subBundle))
182 {
183 log_err("Different size array with key \"%s\" in \"%s\" from root for locale \"%s\"\n"
184 "\troot array size=%d, locale array size=%d\n",
185 subBundleKey,
186 ures_getKey(currentBundle),
187 locale,
188 ures_getSize(subRootBundle),
189 ures_getSize(subBundle));
190 }
191 /*
192 if(isCurrencyPreEuro(subBundleKey) && ures_getSize(subBundle)!=3){
193 log_err("Different size array with key \"%s\" in \"%s\" for locale \"%s\" the expected size is 3 got size=%d\n",
194 subBundleKey,
195 ures_getKey(currentBundle),
196 locale,
197 ures_getSize(subBundle));
198 }
199 */
200 for (idx = 0; idx < minSize; idx++) {
201 int32_t rootStrLen, localeStrLen;
202 const UChar *rootStr = ures_getStringByIndex(subRootBundle,idx,&rootStrLen,&errorCode);
203 const UChar *localeStr = ures_getStringByIndex(subBundle,idx,&localeStrLen,&errorCode);
204 if (rootStr && localeStr && U_SUCCESS(errorCode)) {
205 if (u_strcmp(rootStr, localeStr) != 0) {
206 sameArray = FALSE;
207 }
208 }
209 else {
210 if ( rootStrLen > 1 && rootStr[0] == 0x41 && rootStr[1] >= 0x30 && rootStr[1] <= 0x39 ) {
211 /* A2 or A4 in the root string indicates that the resource can optionally be an array instead of a */
212 /* string. Attempt to read it as an array. */
213 errorCode = U_ZERO_ERROR;
214 arr = ures_getByIndex(subBundle,idx,NULL,&errorCode);
215 if (U_FAILURE(errorCode)) {
216 log_err("Got a NULL string with key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n",
217 subBundleKey,
218 ures_getKey(currentBundle),
219 idx,
220 locale);
221 continue;
222 }
223 if (ures_getType(arr) != URES_ARRAY || ures_getSize(arr) != (int32_t)rootStr[1] - 0x30) {
224 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",
225 rootStr[1] - 0x30,
226 subBundleKey,
227 ures_getKey(currentBundle),
228 idx,
229 locale);
230 ures_close(arr);
231 continue;
232 }
233 localeStr = ures_getStringByIndex(arr,0,&localeStrLen,&errorCode);
234 ures_close(arr);
235 if (U_FAILURE(errorCode)) {
236 log_err("Got something other than a string or array for key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n",
237 subBundleKey,
238 ures_getKey(currentBundle),
239 idx,
240 locale);
241 continue;
242 }
243 } else {
244 log_err("Got a NULL string with key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n",
245 subBundleKey,
246 ures_getKey(currentBundle),
247 idx,
248 locale);
249 continue;
250 }
251 }
252 if (localeStr[0] == (UChar)0x20) {
253 log_err("key \"%s\" at index %d in \"%s\" starts with a space in locale \"%s\"\n",
254 subBundleKey,
255 idx,
256 ures_getKey(currentBundle),
257 locale);
258 }
259 else if ((localeStr[localeStrLen - 1] == (UChar)0x20) && (strcmp(subBundleKey,"separator") != 0)) {
260 log_err("key \"%s\" at index %d in \"%s\" ends with a space in locale \"%s\"\n",
261 subBundleKey,
262 idx,
263 ures_getKey(currentBundle),
264 locale);
265 }
266 else if (subBundleKey != NULL
267 && strcmp(subBundleKey, "DateTimePatterns") == 0)
268 {
269 int32_t quoted = 0;
270 const UChar *localeStrItr = localeStr;
271 while (*localeStrItr) {
272 if (*localeStrItr == (UChar)0x27 /* ' */) {
273 quoted++;
274 }
275 else if ((quoted % 2) == 0) {
276 /* Search for unquoted characters */
277 if (4 <= idx && idx <= 7
278 && (*localeStrItr == (UChar)0x6B /* k */
279 || *localeStrItr == (UChar)0x48 /* H */
280 || *localeStrItr == (UChar)0x6D /* m */
281 || *localeStrItr == (UChar)0x73 /* s */
282 || *localeStrItr == (UChar)0x53 /* S */
283 || *localeStrItr == (UChar)0x61 /* a */
284 || *localeStrItr == (UChar)0x68 /* h */
285 || *localeStrItr == (UChar)0x7A /* z */))
286 {
287 log_err("key \"%s\" at index %d has time pattern chars in date for locale \"%s\"\n",
288 subBundleKey,
289 idx,
290 locale);
291 }
292 else if (0 <= idx && idx <= 3
293 && (*localeStrItr == (UChar)0x47 /* G */
294 || *localeStrItr == (UChar)0x79 /* y */
295 || *localeStrItr == (UChar)0x4D /* M */
296 || *localeStrItr == (UChar)0x64 /* d */
297 || *localeStrItr == (UChar)0x45 /* E */
298 || *localeStrItr == (UChar)0x44 /* D */
299 || *localeStrItr == (UChar)0x46 /* F */
300 || *localeStrItr == (UChar)0x77 /* w */
301 || *localeStrItr == (UChar)0x57 /* W */))
302 {
303 log_err("key \"%s\" at index %d has date pattern chars in time for locale \"%s\"\n",
304 subBundleKey,
305 idx,
306 locale);
307 }
308 }
309 localeStrItr++;
310 }
311 }
312 else if (idx == 4 && subBundleKey != NULL
313 && strcmp(subBundleKey, "NumberElements") == 0
314 && u_charDigitValue(localeStr[0]) != 0)
315 {
316 log_err("key \"%s\" at index %d has a non-zero based number for locale \"%s\"\n",
317 subBundleKey,
318 idx,
319 locale);
320 }
321 }
322 (void)sameArray; /* Suppress set but not used warning. */
323 /* if (sameArray && strcmp(rootName, "root") == 0) {
324 log_err("Arrays are the same with key \"%s\" in \"%s\" from root for locale \"%s\"\n",
325 subBundleKey,
326 ures_getKey(currentBundle),
327 locale);
328 }*/
329 }
330 ures_close(subSubBundle);
331 ures_close(subSubRootBundle);
332 }
333 else if (ures_getType(subBundle) == URES_STRING) {
334 int32_t len = 0;
335 const UChar *string = ures_getString(subBundle, &len, &errorCode);
336 if (U_FAILURE(errorCode) || string == NULL) {
337 log_err("Can't open a string with key \"%s\" in \"%s\" for locale \"%s\"\n",
338 subBundleKey,
339 ures_getKey(currentBundle),
340 locale);
341 } else if (string[0] == (UChar)0x20) {
342 log_err("key \"%s\" in \"%s\" starts with a space in locale \"%s\"\n",
343 subBundleKey,
344 ures_getKey(currentBundle),
345 locale);
346 /* localeDisplayPattern/separator can end with a space */
347 } else if (string[len - 1] == (UChar)0x20 && (strcmp(subBundleKey,"separator"))) {
348 log_err("key \"%s\" in \"%s\" ends with a space in locale \"%s\"\n",
349 subBundleKey,
350 ures_getKey(currentBundle),
351 locale);
352 } else if (strcmp(subBundleKey, "localPatternChars") == 0) {
353 /* Note: We no longer import localPatternChars data starting
354 * ICU 3.8. So it never comes into this else if block. (ticket#5597)
355 */
356
357 /* Check well-formedness of localPatternChars. First, the
358 * length must match the number of fields defined by
359 * DateFormat. Second, each character in the string must
360 * be in the set [A-Za-z]. Finally, each character must be
361 * unique.
362 */
363 int32_t i,j;
364 #if !UCONFIG_NO_FORMATTING
365 if (len != UDAT_FIELD_COUNT) {
366 log_err("key \"%s\" has the wrong number of characters in locale \"%s\"\n",
367 subBundleKey,
368 locale);
369 }
370 #endif
371 /* Check char validity. */
372 for (i=0; i<len; ++i) {
373 if (!((string[i] >= 65/*'A'*/ && string[i] <= 90/*'Z'*/) ||
374 (string[i] >= 97/*'a'*/ && string[i] <= 122/*'z'*/))) {
375 log_err("key \"%s\" has illegal character '%c' in locale \"%s\"\n",
376 subBundleKey,
377 (char) string[i],
378 locale);
379 }
380 /* Do O(n^2) check for duplicate chars. */
381 for (j=0; j<i; ++j) {
382 if (string[j] == string[i]) {
383 log_err("key \"%s\" has duplicate character '%c' in locale \"%s\"\n",
384 subBundleKey,
385 (char) string[i],
386 locale);
387 }
388 }
389 }
390 }
391 /* No fallback was done. Check for duplicate data */
392 /* The ures_* API does not do fallback of sub-resource bundles,
393 So we can't do this now. */
394 #if 0
395 else if (strcmp(locale, "root") != 0 && errorCode == U_ZERO_ERROR) {
396
397 const UChar *rootString = ures_getString(subRootBundle, &len, &errorCode);
398 if (U_FAILURE(errorCode) || rootString == NULL) {
399 log_err("Can't open a string with key \"%s\" in \"%s\" in root\n",
400 ures_getKey(subRootBundle),
401 ures_getKey(currentBundle));
402 continue;
403 } else if (u_strcmp(string, rootString) == 0) {
404 if (strcmp(locale, "de_CH") != 0 && strcmp(subBundleKey, "Countries") != 0 &&
405 strcmp(subBundleKey, "Version") != 0) {
406 log_err("Found duplicate data with key \"%s\" in \"%s\" in locale \"%s\"\n",
407 ures_getKey(subRootBundle),
408 ures_getKey(currentBundle),
409 locale);
410 }
411 else {
412 /* Ignore for now. */
413 /* Can be fixed if fallback through de locale was done. */
414 log_verbose("Skipping key %s in %s\n", subBundleKey, locale);
415 }
416 }
417 }
418 #endif
419 }
420 else if (ures_getType(subBundle) == URES_TABLE) {
421 if (strcmp(subBundleKey, "availableFormats")!=0) {
422 /* Here is one of the recursive parts */
423 TestKeyInRootRecursive(subRootBundle, rootName, subBundle, locale);
424 }
425 else {
426 log_verbose("Skipping key %s in %s\n", subBundleKey, locale);
427 }
428 }
429 else if (ures_getType(subBundle) == URES_BINARY || ures_getType(subBundle) == URES_INT) {
430 /* Can't do anything to check it */
431 /* We'll assume it's all correct */
432 if (strcmp(subBundleKey, "MeasurementSystem") != 0) {
433 log_verbose("Skipping key \"%s\" in \"%s\" for locale \"%s\"\n",
434 subBundleKey,
435 ures_getKey(currentBundle),
436 locale);
437 }
438 /* Testing for MeasurementSystem is done in VerifyTranslation */
439 }
440 else {
441 log_err("Type %d for key \"%s\" in \"%s\" is unknown for locale \"%s\"\n",
442 ures_getType(subBundle),
443 subBundleKey,
444 ures_getKey(currentBundle),
445 locale);
446 }
447 ures_close(subRootBundle);
448 ures_close(subBundle);
449 }
450 }
451 #endif
452
453 static void
testLCID(UResourceBundle * currentBundle,const char * localeName)454 testLCID(UResourceBundle *currentBundle,
455 const char *localeName)
456 {
457 UErrorCode status = U_ZERO_ERROR;
458 uint32_t expectedLCID;
459 char lcidStringC[64] = {0};
460 int32_t len;
461
462 expectedLCID = uloc_getLCID(localeName);
463 if (expectedLCID == 0) {
464 log_verbose("INFO: %-5s does not have any LCID mapping\n",
465 localeName);
466 return;
467 }
468
469 status = U_ZERO_ERROR;
470 len = uprv_convertToPosix(expectedLCID, lcidStringC, UPRV_LENGTHOF(lcidStringC) - 1, &status);
471 if (U_FAILURE(status)) {
472 log_err("ERROR: %.4x does not have a POSIX mapping due to %s\n",
473 expectedLCID, u_errorName(status));
474 }
475 lcidStringC[len] = 0;
476
477 if(strcmp(localeName, lcidStringC) != 0) {
478 char langName[1024];
479 char langLCID[1024];
480 uloc_getLanguage(localeName, langName, sizeof(langName), &status);
481 uloc_getLanguage(lcidStringC, langLCID, sizeof(langLCID), &status);
482
483 if (strcmp(langName, langLCID) == 0) {
484 log_verbose("WARNING: %-5s resolves to %s (0x%.4x)\n",
485 localeName, lcidStringC, expectedLCID);
486 }
487 else if (!(strcmp(localeName, "ku") == 0 && log_knownIssue("20181", "ICU-20181 Fix LCID mapping for ckb vs ku"))) {
488 log_err("ERROR: %-5s has 0x%.4x and the number resolves wrongfully to %s\n",
489 localeName, expectedLCID, lcidStringC);
490 }
491 }
492 }
493
494 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
495 static void
TestLocaleStructure(void)496 TestLocaleStructure(void) {
497 // This test checks the locale structure against a key file located
498 // at source/test/testdata/structLocale.txt. When adding new data to
499 // a locale file such as en.txt, the structLocale.txt file must be changed
500 // too to include the the template of the new data. Otherwise this test
501 // will fail!
502
503 UResourceBundle *root, *currentLocale;
504 int32_t locCount = uloc_countAvailable();
505 int32_t locIndex;
506 UErrorCode errorCode = U_ZERO_ERROR;
507 const char *currLoc, *resolvedLoc;
508
509 /* TODO: Compare against parent's data too. This code can't handle fallbacks that some tools do already. */
510 /* char locName[ULOC_FULLNAME_CAPACITY];
511 char *locNamePtr;
512
513 for (locIndex = 0; locIndex < locCount; locIndex++) {
514 errorCode=U_ZERO_ERROR;
515 strcpy(locName, uloc_getAvailable(locIndex));
516 locNamePtr = strrchr(locName, '_');
517 if (locNamePtr) {
518 *locNamePtr = 0;
519 }
520 else {
521 strcpy(locName, "root");
522 }
523
524 root = ures_openDirect(NULL, locName, &errorCode);
525 if(U_FAILURE(errorCode)) {
526 log_err("Can't open %s\n", locName);
527 continue;
528 }
529 */
530 if (locCount <= 1) {
531 log_data_err("At least root needs to be installed\n");
532 }
533
534 root = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode);
535 if(U_FAILURE(errorCode)) {
536 log_data_err("Can't open structLocale\n");
537 return;
538 }
539 for (locIndex = 0; locIndex < locCount; locIndex++) {
540 errorCode=U_ZERO_ERROR;
541 currLoc = uloc_getAvailable(locIndex);
542 currentLocale = ures_open(NULL, currLoc, &errorCode);
543 if(errorCode != U_ZERO_ERROR) {
544 if(U_SUCCESS(errorCode)) {
545 /* It's installed, but there is no data.
546 It's installed for the g18n white paper [grhoten] */
547 log_err("ERROR: Locale %-5s not installed, and it should be, err %s\n",
548 uloc_getAvailable(locIndex), u_errorName(errorCode));
549 } else {
550 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
551 u_errorName(errorCode),
552 uloc_getAvailable(locIndex));
553 }
554 ures_close(currentLocale);
555 continue;
556 }
557 ures_getStringByKey(currentLocale, "Version", NULL, &errorCode);
558 if(errorCode != U_ZERO_ERROR) {
559 log_err("No version information is available for locale %s, and it should be!\n",
560 currLoc);
561 }
562 else if (ures_getStringByKey(currentLocale, "Version", NULL, &errorCode)[0] == (UChar)(0x78)) {
563 log_verbose("WARNING: The locale %s is experimental! It shouldn't be listed as an installed locale.\n",
564 currLoc);
565 }
566 resolvedLoc = ures_getLocaleByType(currentLocale, ULOC_ACTUAL_LOCALE, &errorCode);
567 if (strcmp(resolvedLoc, currLoc) != 0) {
568 /* All locales have at least a Version resource.
569 If it's absolutely empty, then the previous test will fail too.*/
570 log_err("Locale resolves to different locale. Is %s an alias of %s?\n",
571 currLoc, resolvedLoc);
572 }
573 TestKeyInRootRecursive(root, "root", currentLocale, currLoc);
574
575 testLCID(currentLocale, currLoc);
576
577 ures_close(currentLocale);
578 }
579
580 ures_close(root);
581 }
582 #endif
583
584 static void
compareArrays(const char * keyName,UResourceBundle * fromArray,const char * fromLocale,UResourceBundle * toArray,const char * toLocale,int32_t start,int32_t end)585 compareArrays(const char *keyName,
586 UResourceBundle *fromArray, const char *fromLocale,
587 UResourceBundle *toArray, const char *toLocale,
588 int32_t start, int32_t end)
589 {
590 int32_t fromSize = ures_getSize(fromArray);
591 int32_t toSize = ures_getSize(fromArray);
592 int32_t idx;
593 UErrorCode errorCode = U_ZERO_ERROR;
594
595 if (fromSize > toSize) {
596 fromSize = toSize;
597 log_err("Arrays are different size from \"%s\" to \"%s\"\n",
598 fromLocale,
599 toLocale);
600 }
601
602 for (idx = start; idx <= end; idx++) {
603 const UChar *fromBundleStr = ures_getStringByIndex(fromArray, idx, NULL, &errorCode);
604 const UChar *toBundleStr = ures_getStringByIndex(toArray, idx, NULL, &errorCode);
605 if (fromBundleStr && toBundleStr && u_strcmp(fromBundleStr, toBundleStr) != 0)
606 {
607 log_err("Difference for %s at index %d from %s= \"%s\" to %s= \"%s\"\n",
608 keyName,
609 idx,
610 fromLocale,
611 austrdup(fromBundleStr),
612 toLocale,
613 austrdup(toBundleStr));
614 }
615 }
616 }
617
618 static void
compareConsistentCountryInfo(const char * fromLocale,const char * toLocale)619 compareConsistentCountryInfo(const char *fromLocale, const char *toLocale) {
620 UErrorCode errorCode = U_ZERO_ERROR;
621 UResourceBundle *fromArray, *toArray;
622 UResourceBundle *fromLocaleBund = ures_open(NULL, fromLocale, &errorCode);
623 UResourceBundle *toLocaleBund = ures_open(NULL, toLocale, &errorCode);
624 UResourceBundle *toCalendar, *fromCalendar, *toGregorian, *fromGregorian;
625
626 if(U_FAILURE(errorCode)) {
627 log_err("Can't open resource bundle %s or %s - %s\n", fromLocale, toLocale, u_errorName(errorCode));
628 return;
629 }
630 fromCalendar = ures_getByKey(fromLocaleBund, "calendar", NULL, &errorCode);
631 fromGregorian = ures_getByKeyWithFallback(fromCalendar, "gregorian", NULL, &errorCode);
632
633 toCalendar = ures_getByKey(toLocaleBund, "calendar", NULL, &errorCode);
634 toGregorian = ures_getByKeyWithFallback(toCalendar, "gregorian", NULL, &errorCode);
635
636 fromArray = ures_getByKey(fromLocaleBund, "CurrencyElements", NULL, &errorCode);
637 toArray = ures_getByKey(toLocaleBund, "CurrencyElements", NULL, &errorCode);
638 if (strcmp(fromLocale, "en_CA") != 0)
639 {
640 /* The first one is probably localized. */
641 compareArrays("CurrencyElements", fromArray, fromLocale, toArray, toLocale, 1, 2);
642 }
643 ures_close(fromArray);
644 ures_close(toArray);
645
646 fromArray = ures_getByKey(fromLocaleBund, "NumberPatterns", NULL, &errorCode);
647 toArray = ures_getByKey(toLocaleBund, "NumberPatterns", NULL, &errorCode);
648 if (strcmp(fromLocale, "en_CA") != 0)
649 {
650 compareArrays("NumberPatterns", fromArray, fromLocale, toArray, toLocale, 0, 3);
651 }
652 ures_close(fromArray);
653 ures_close(toArray);
654
655 /* Difficult to test properly */
656 /*
657 fromArray = ures_getByKey(fromLocaleBund, "DateTimePatterns", NULL, &errorCode);
658 toArray = ures_getByKey(toLocaleBund, "DateTimePatterns", NULL, &errorCode);
659 {
660 compareArrays("DateTimePatterns", fromArray, fromLocale, toArray, toLocale);
661 }
662 ures_close(fromArray);
663 ures_close(toArray);*/
664
665 fromArray = ures_getByKey(fromLocaleBund, "NumberElements", NULL, &errorCode);
666 toArray = ures_getByKey(toLocaleBund, "NumberElements", NULL, &errorCode);
667 if (strcmp(fromLocale, "en_CA") != 0)
668 {
669 compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 0, 3);
670 /* Index 4 is a script based 0 */
671 compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 5, 10);
672 }
673 ures_close(fromArray);
674 ures_close(toArray);
675 ures_close(fromCalendar);
676 ures_close(toCalendar);
677 ures_close(fromGregorian);
678 ures_close(toGregorian);
679
680 ures_close(fromLocaleBund);
681 ures_close(toLocaleBund);
682 }
683
684 static void
TestConsistentCountryInfo(void)685 TestConsistentCountryInfo(void) {
686 /* UResourceBundle *fromLocale, *toLocale;*/
687 int32_t locCount = uloc_countAvailable();
688 int32_t fromLocIndex, toLocIndex;
689
690 int32_t fromCountryLen, toCountryLen;
691 char fromCountry[ULOC_FULLNAME_CAPACITY], toCountry[ULOC_FULLNAME_CAPACITY];
692
693 int32_t fromVariantLen, toVariantLen;
694 char fromVariant[ULOC_FULLNAME_CAPACITY], toVariant[ULOC_FULLNAME_CAPACITY];
695
696 UErrorCode errorCode = U_ZERO_ERROR;
697
698 for (fromLocIndex = 0; fromLocIndex < locCount; fromLocIndex++) {
699 const char *fromLocale = uloc_getAvailable(fromLocIndex);
700
701 errorCode=U_ZERO_ERROR;
702 fromCountryLen = uloc_getCountry(fromLocale, fromCountry, ULOC_FULLNAME_CAPACITY, &errorCode);
703 if (fromCountryLen <= 0) {
704 /* Ignore countryless locales */
705 continue;
706 }
707 fromVariantLen = uloc_getVariant(fromLocale, fromVariant, ULOC_FULLNAME_CAPACITY, &errorCode);
708 if (fromVariantLen > 0) {
709 /* Most variants are ignorable like PREEURO, or collation variants. */
710 continue;
711 }
712 /* Start comparing only after the current index.
713 Previous loop should have already compared fromLocIndex.
714 */
715 for (toLocIndex = fromLocIndex + 1; toLocIndex < locCount; toLocIndex++) {
716 const char *toLocale = uloc_getAvailable(toLocIndex);
717
718 toCountryLen = uloc_getCountry(toLocale, toCountry, ULOC_FULLNAME_CAPACITY, &errorCode);
719 if(U_FAILURE(errorCode)) {
720 log_err("Unknown failure fromLocale=%s toLocale=%s errorCode=%s\n",
721 fromLocale, toLocale, u_errorName(errorCode));
722 continue;
723 }
724
725 if (toCountryLen <= 0) {
726 /* Ignore countryless locales */
727 continue;
728 }
729 toVariantLen = uloc_getVariant(toLocale, toVariant, ULOC_FULLNAME_CAPACITY, &errorCode);
730 if (toVariantLen > 0) {
731 /* Most variants are ignorable like PREEURO, or collation variants. */
732 /* They're a variant for a reason. */
733 continue;
734 }
735 if (strcmp(fromCountry, toCountry) == 0) {
736 log_verbose("comparing fromLocale=%s toLocale=%s\n",
737 fromLocale, toLocale);
738 compareConsistentCountryInfo(fromLocale, toLocale);
739 }
740 }
741 }
742 }
743
744 static int32_t
findStringSetMismatch(const char * currLoc,const UChar * string,int32_t langSize,USet * mergedExemplarSet,UBool ignoreNumbers,UChar32 * badCharPtr)745 findStringSetMismatch(const char *currLoc, const UChar *string, int32_t langSize,
746 USet * mergedExemplarSet,
747 UBool ignoreNumbers, UChar32* badCharPtr) {
748 UErrorCode errorCode = U_ZERO_ERROR;
749 USet *exemplarSet;
750 int32_t strIdx;
751 if (mergedExemplarSet == NULL) {
752 return -1;
753 }
754 exemplarSet = createFlattenSet(mergedExemplarSet, &errorCode);
755 if (U_FAILURE(errorCode)) {
756 log_err("%s: error createFlattenSet returned %s\n", currLoc, u_errorName(errorCode));
757 return -1;
758 }
759
760 for (strIdx = 0; strIdx < langSize;) {
761 UChar32 testChar;
762 U16_NEXT(string, strIdx, langSize, testChar);
763 if (!uset_contains(exemplarSet, testChar)
764 && testChar != 0x0020 && testChar != 0x00A0 && testChar != 0x002e && testChar != 0x002c && testChar != 0x002d && testChar != 0x0027
765 && testChar != 0x005B && testChar != 0x005D && testChar != 0x2019 && testChar != 0x0f0b && testChar != 0x200C && testChar != 0x200D) {
766 if (!ignoreNumbers || (ignoreNumbers && (testChar < 0x30 || testChar > 0x39))) {
767 uset_close(exemplarSet);
768 if (badCharPtr) {
769 *badCharPtr = testChar;
770 }
771 return strIdx;
772 }
773 }
774 }
775 uset_close(exemplarSet);
776 if (badCharPtr) {
777 *badCharPtr = 0;
778 }
779 return -1;
780 }
781 /* include non-invariant chars */
782 static int32_t
myUCharsToChars(const UChar * us,char * cs,int32_t len)783 myUCharsToChars(const UChar* us, char* cs, int32_t len){
784 int32_t i=0;
785 for(; i< len; i++){
786 if(us[i] < 0x7f){
787 cs[i] = (char)us[i];
788 }else{
789 return -1;
790 }
791 }
792 return i;
793 }
794 static void
findSetMatch(UScriptCode * scriptCodes,int32_t scriptsLen,USet * exemplarSet,const char * locale)795 findSetMatch( UScriptCode *scriptCodes, int32_t scriptsLen,
796 USet *exemplarSet,
797 const char *locale){
798 USet *scripts[10]= {0};
799 char pattern[256] = { '[', ':', 0x000 };
800 int32_t patternLen;
801 UChar uPattern[256] = {0};
802 UErrorCode status = U_ZERO_ERROR;
803 int32_t i;
804
805 /* create the sets with script codes */
806 for(i = 0; i<scriptsLen; i++){
807 strcat(pattern, uscript_getShortName(scriptCodes[i]));
808 strcat(pattern, ":]");
809 patternLen = (int32_t)strlen(pattern);
810 u_charsToUChars(pattern, uPattern, patternLen);
811 scripts[i] = uset_openPattern(uPattern, patternLen, &status);
812 if(U_FAILURE(status)){
813 log_err("Could not create set for pattern %s. Error: %s\n", pattern, u_errorName(status));
814 return;
815 }
816 pattern[2] = 0;
817 }
818 if (strcmp(locale, "uk") == 0 || strcmp(locale, "uk_UA") == 0) {
819 /* Special addition. Add the modifying apostrophe, which isn't in Cyrillic. */
820 uset_add(scripts[0], 0x2bc);
821 }
822 if(U_SUCCESS(status)){
823 UBool existsInScript = FALSE;
824 /* iterate over the exemplarSet and ascertain if all
825 * UChars in exemplarSet belong to the scripts returned
826 * by getScript
827 */
828 int32_t count = uset_getItemCount(exemplarSet);
829
830 for( i=0; i < count; i++){
831 UChar32 start = 0;
832 UChar32 end = 0;
833 UChar *str = NULL;
834 int32_t strCapacity = 0;
835
836 strCapacity = uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status);
837 if(U_SUCCESS(status)){
838 int32_t j;
839 if(strCapacity == 0){
840 /* ok the item is a range */
841 for( j = 0; j < scriptsLen; j++){
842 if(uset_containsRange(scripts[j], start, end) == TRUE){
843 existsInScript = TRUE;
844 }
845 }
846 if(existsInScript == FALSE){
847 for( j = 0; j < scriptsLen; j++){
848 UChar toPattern[500]={'\0'};
849 char pat[500]={'\0'};
850 int32_t len = uset_toPattern(scripts[j], toPattern, 500, TRUE, &status);
851 len = myUCharsToChars(toPattern, pat, len);
852 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));
853 if(len!=-1){
854 log_err("Pattern: %s\n",pat);
855 }
856 }
857 log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale);
858 }
859 }else{
860 strCapacity++; /* increment for NUL termination */
861 /* allocate the str and call the api again */
862 str = (UChar*) malloc(U_SIZEOF_UCHAR * strCapacity);
863 strCapacity = uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status);
864 /* iterate over the scripts and figure out if the string contained is actually
865 * in the script set
866 */
867 for( j = 0; j < scriptsLen; j++){
868 if(uset_containsString(scripts[j],str, strCapacity) == TRUE){
869 existsInScript = TRUE;
870 }
871 }
872 if(existsInScript == FALSE){
873 log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale);
874 }
875 }
876 }
877 }
878
879 }
880
881 /* close the sets */
882 for(i = 0; i<scriptsLen; i++){
883 uset_close(scripts[i]);
884 }
885 }
886
VerifyTranslation(void)887 static void VerifyTranslation(void) {
888 UResourceBundle *root, *currentLocale;
889 int32_t locCount = uloc_countAvailable();
890 int32_t locIndex;
891 UErrorCode errorCode = U_ZERO_ERROR;
892 const char *currLoc;
893 UScriptCode scripts[USCRIPT_CODE_LIMIT];
894 int32_t numScripts;
895 int32_t idx;
896 int32_t end;
897 UResourceBundle *resArray;
898
899 if (locCount <= 1) {
900 log_data_err("At least root needs to be installed\n");
901 }
902
903 root = ures_openDirect(NULL, "root", &errorCode);
904 if(U_FAILURE(errorCode)) {
905 log_data_err("Can't open root\n");
906 return;
907 }
908 for (locIndex = 0; locIndex < locCount; locIndex++) {
909 USet * mergedExemplarSet = NULL;
910 errorCode=U_ZERO_ERROR;
911 currLoc = uloc_getAvailable(locIndex);
912 currentLocale = ures_open(NULL, currLoc, &errorCode);
913 if(errorCode != U_ZERO_ERROR) {
914 if(U_SUCCESS(errorCode)) {
915 /* It's installed, but there is no data.
916 It's installed for the g18n white paper [grhoten] */
917 log_err("ERROR: Locale %-5s not installed, and it should be!\n",
918 uloc_getAvailable(locIndex));
919 } else {
920 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
921 u_errorName(errorCode),
922 uloc_getAvailable(locIndex));
923 }
924 ures_close(currentLocale);
925 continue;
926 }
927 {
928 UErrorCode exemplarStatus = U_ZERO_ERROR;
929 ULocaleData * uld = ulocdata_open(currLoc, &exemplarStatus);
930 if (U_SUCCESS(exemplarStatus)) {
931 USet * exemplarSet = ulocdata_getExemplarSet(uld, NULL, USET_ADD_CASE_MAPPINGS, ULOCDATA_ES_STANDARD, &exemplarStatus);
932 if (U_SUCCESS(exemplarStatus)) {
933 mergedExemplarSet = uset_cloneAsThawed(exemplarSet);
934 uset_close(exemplarSet);
935 exemplarSet = ulocdata_getExemplarSet(uld, NULL, USET_ADD_CASE_MAPPINGS, ULOCDATA_ES_AUXILIARY, &exemplarStatus);
936 if (U_SUCCESS(exemplarStatus)) {
937 uset_addAll(mergedExemplarSet, exemplarSet);
938 uset_close(exemplarSet);
939 }
940 exemplarStatus = U_ZERO_ERROR;
941 exemplarSet = ulocdata_getExemplarSet(uld, NULL, 0, ULOCDATA_ES_PUNCTUATION, &exemplarStatus);
942 if (U_SUCCESS(exemplarStatus)) {
943 uset_addAll(mergedExemplarSet, exemplarSet);
944 uset_close(exemplarSet);
945 }
946 } else {
947 log_err("error ulocdata_getExemplarSet (main) for locale %s returned %s\n", currLoc, u_errorName(errorCode));
948 }
949 ulocdata_close(uld);
950 } else {
951 log_err("error ulocdata_open for locale %s returned %s\n", currLoc, u_errorName(errorCode));
952 }
953 }
954 if (mergedExemplarSet == NULL /*|| (getTestOption(QUICK_OPTION) && uset_size() > 2048)*/) {
955 log_verbose("skipping test for %s\n", currLoc);
956 }
957 //else if (uprv_strncmp(currLoc,"bem",3) == 0 || uprv_strncmp(currLoc,"mgo",3) == 0 || uprv_strncmp(currLoc,"nl",2) == 0) {
958 // log_verbose("skipping test for %s, some month and country names known to use aux exemplars\n", currLoc);
959 //}
960 else {
961 UChar langBuffer[128];
962 int32_t langSize;
963 int32_t strIdx;
964 UChar32 badChar;
965 langSize = uloc_getDisplayLanguage(currLoc, currLoc, langBuffer, UPRV_LENGTHOF(langBuffer), &errorCode);
966 if (U_FAILURE(errorCode)) {
967 log_err("error uloc_getDisplayLanguage returned %s\n", u_errorName(errorCode));
968 }
969 else {
970 strIdx = findStringSetMismatch(currLoc, langBuffer, langSize, mergedExemplarSet, FALSE, &badChar);
971 if (strIdx >= 0) {
972 log_err("getDisplayLanguage(%s) at index %d returned characters not in the exemplar characters: %04X.\n",
973 currLoc, strIdx, badChar);
974 }
975 }
976 langSize = uloc_getDisplayCountry(currLoc, currLoc, langBuffer, UPRV_LENGTHOF(langBuffer), &errorCode);
977 if (U_FAILURE(errorCode)) {
978 log_err("error uloc_getDisplayCountry returned %s\n", u_errorName(errorCode));
979 }
980 {
981 UResourceBundle* cal = ures_getByKey(currentLocale, "calendar", NULL, &errorCode);
982 UResourceBundle* greg = ures_getByKeyWithFallback(cal, "gregorian", NULL, &errorCode);
983 UResourceBundle* names = ures_getByKeyWithFallback(greg, "dayNames", NULL, &errorCode);
984 UResourceBundle* format = ures_getByKeyWithFallback(names, "format", NULL, &errorCode);
985 resArray = ures_getByKeyWithFallback(format, "wide", NULL, &errorCode);
986
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 if ((uprv_strncmp(currLoc,"lrc",3) == 0 || uprv_strncmp(currLoc,"mzn",3) == 0) &&
998 log_knownIssue("cldrbug:8899", "lrc and mzn locales don't have translated day names")) {
999 end = 0;
1000 }
1001
1002 for (idx = 0; idx < end; idx++) {
1003 const UChar *fromBundleStr = ures_getStringByIndex(resArray, idx, &langSize, &errorCode);
1004 if (U_FAILURE(errorCode)) {
1005 log_err("error ures_getStringByIndex(%d) returned %s\n", idx, u_errorName(errorCode));
1006 continue;
1007 }
1008 strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, mergedExemplarSet, TRUE, &badChar);
1009 if ( strIdx >= 0 ) {
1010 log_err("getDayNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n",
1011 currLoc, idx, strIdx, badChar);
1012 }
1013 }
1014 ures_close(resArray);
1015 ures_close(format);
1016 ures_close(names);
1017
1018 names = ures_getByKeyWithFallback(greg, "monthNames", NULL, &errorCode);
1019 format = ures_getByKeyWithFallback(names,"format", NULL, &errorCode);
1020 resArray = ures_getByKeyWithFallback(format, "wide", NULL, &errorCode);
1021 if (U_FAILURE(errorCode)) {
1022 log_err("error ures_getByKey returned %s\n", u_errorName(errorCode));
1023 }
1024 if (getTestOption(QUICK_OPTION)) {
1025 end = 1;
1026 }
1027 else {
1028 end = ures_getSize(resArray);
1029 }
1030
1031 for (idx = 0; idx < end; idx++) {
1032 const UChar *fromBundleStr = ures_getStringByIndex(resArray, idx, &langSize, &errorCode);
1033 if (U_FAILURE(errorCode)) {
1034 log_err("error ures_getStringByIndex(%d) returned %s\n", idx, u_errorName(errorCode));
1035 continue;
1036 }
1037 strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, mergedExemplarSet, TRUE, &badChar);
1038 if (strIdx >= 0) {
1039 log_err("getMonthNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n",
1040 currLoc, idx, strIdx, badChar);
1041 }
1042 }
1043 ures_close(resArray);
1044 ures_close(format);
1045 ures_close(names);
1046 ures_close(greg);
1047 ures_close(cal);
1048 }
1049 errorCode = U_ZERO_ERROR;
1050 numScripts = uscript_getCode(currLoc, scripts, UPRV_LENGTHOF(scripts), &errorCode);
1051 if (strcmp(currLoc, "yi") == 0 && numScripts > 0 && log_knownIssue("11217", "Fix result of uscript_getCode for yi: USCRIPT_YI -> USCRIPT_HEBREW")) {
1052 scripts[0] = USCRIPT_HEBREW;
1053 }
1054 if (numScripts == 0) {
1055 log_err("uscript_getCode(%s) doesn't work.\n", currLoc);
1056 }else if(scripts[0] == USCRIPT_COMMON){
1057 log_err("uscript_getCode(%s) returned USCRIPT_COMMON.\n", currLoc);
1058 }
1059
1060 /* test that the scripts are a superset of exemplar characters. */
1061 {
1062 ULocaleData *uld = ulocdata_open(currLoc,&errorCode);
1063 USet *exemplarSet = ulocdata_getExemplarSet(uld, NULL, 0, ULOCDATA_ES_STANDARD, &errorCode);
1064 /* test if exemplar characters are part of script code */
1065 findSetMatch(scripts, numScripts, exemplarSet, currLoc);
1066 uset_close(exemplarSet);
1067 ulocdata_close(uld);
1068 }
1069
1070 /* test that the paperSize API works */
1071 {
1072 int32_t height=0, width=0;
1073 ulocdata_getPaperSize(currLoc, &height, &width, &errorCode);
1074 if(U_FAILURE(errorCode)){
1075 log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1076 }
1077 if(strstr(currLoc, "_US")!=NULL && height != 279 && width != 216 ){
1078 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc);
1079 }
1080 }
1081 /* test that the MeasurementSystem API works */
1082 {
1083 char fullLoc[ULOC_FULLNAME_CAPACITY];
1084 UMeasurementSystem measurementSystem;
1085 int32_t height = 0, width = 0;
1086
1087 uloc_addLikelySubtags(currLoc, fullLoc, ULOC_FULLNAME_CAPACITY, &errorCode);
1088
1089 errorCode = U_ZERO_ERROR;
1090 measurementSystem = ulocdata_getMeasurementSystem(currLoc, &errorCode);
1091 if (U_FAILURE(errorCode)) {
1092 log_err("ulocdata_getMeasurementSystem failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1093 } else {
1094 if ( strstr(fullLoc, "_US")!=NULL || strstr(fullLoc, "_MM")!=NULL || strstr(fullLoc, "_LR")!=NULL ) {
1095 if(measurementSystem != UMS_US){
1096 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1097 }
1098 } else if ( strstr(fullLoc, "_GB")!=NULL ) {
1099 if(measurementSystem != UMS_UK){
1100 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1101 }
1102 } else if (measurementSystem != UMS_SI) {
1103 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1104 }
1105 }
1106
1107 errorCode = U_ZERO_ERROR;
1108 ulocdata_getPaperSize(currLoc, &height, &width, &errorCode);
1109 if (U_FAILURE(errorCode)) {
1110 log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1111 } else {
1112 if ( strstr(fullLoc, "_US")!=NULL || strstr(fullLoc, "_BZ")!=NULL || strstr(fullLoc, "_CA")!=NULL || strstr(fullLoc, "_CL")!=NULL ||
1113 strstr(fullLoc, "_CO")!=NULL || strstr(fullLoc, "_CR")!=NULL || strstr(fullLoc, "_GT")!=NULL || strstr(fullLoc, "_MX")!=NULL ||
1114 strstr(fullLoc, "_NI")!=NULL || strstr(fullLoc, "_PA")!=NULL || strstr(fullLoc, "_PH")!=NULL || strstr(fullLoc, "_PR")!=NULL ||
1115 strstr(fullLoc, "_SV")!=NULL || strstr(fullLoc, "_VE")!=NULL ) {
1116 if (height != 279 || width != 216) {
1117 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc);
1118 }
1119 } else if (height != 297 || width != 210) {
1120 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc);
1121 }
1122 }
1123 }
1124 }
1125 if (mergedExemplarSet != NULL) {
1126 uset_close(mergedExemplarSet);
1127 }
1128 ures_close(currentLocale);
1129 }
1130
1131 ures_close(root);
1132 }
1133
1134 /* adjust this limit as appropriate */
1135 #define MAX_SCRIPTS_PER_LOCALE 8
1136
TestExemplarSet(void)1137 static void TestExemplarSet(void){
1138 int32_t i, j, k, m, n;
1139 int32_t equalCount = 0;
1140 UErrorCode ec = U_ZERO_ERROR;
1141 UEnumeration* avail;
1142 USet* exemplarSets[2];
1143 USet* unassignedSet;
1144 UScriptCode code[MAX_SCRIPTS_PER_LOCALE];
1145 USet* codeSets[MAX_SCRIPTS_PER_LOCALE];
1146 int32_t codeLen;
1147 char cbuf[32]; /* 9 should be enough */
1148 UChar ubuf[64]; /* adjust as needed */
1149 UBool existsInScript;
1150 int32_t itemCount;
1151 int32_t strLen;
1152 UChar32 start, end;
1153
1154 unassignedSet = NULL;
1155 exemplarSets[0] = NULL;
1156 exemplarSets[1] = NULL;
1157 for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) {
1158 codeSets[i] = NULL;
1159 }
1160
1161 avail = ures_openAvailableLocales(NULL, &ec);
1162 if (!assertSuccess("ures_openAvailableLocales", &ec)) goto END;
1163 n = uenum_count(avail, &ec);
1164 if (!assertSuccess("uenum_count", &ec)) goto END;
1165
1166 u_uastrcpy(ubuf, "[:unassigned:]");
1167 unassignedSet = uset_openPattern(ubuf, -1, &ec);
1168 if (!assertSuccess("uset_openPattern", &ec)) goto END;
1169
1170 for(i=0; i<n; i++){
1171 const char* locale = uenum_next(avail, NULL, &ec);
1172 if (!assertSuccess("uenum_next", &ec)) goto END;
1173 log_verbose("%s\n", locale);
1174 for (k=0; k<2; ++k) {
1175 uint32_t option = (k==0) ? 0 : USET_CASE_INSENSITIVE;
1176 ULocaleData *uld = ulocdata_open(locale,&ec);
1177 USet* exemplarSet = ulocdata_getExemplarSet(uld,NULL, option, ULOCDATA_ES_STANDARD, &ec);
1178 uset_close(exemplarSets[k]);
1179 ulocdata_close(uld);
1180 exemplarSets[k] = exemplarSet;
1181 if (!assertSuccess("ulocaledata_getExemplarSet", &ec)) goto END;
1182
1183 if (uset_containsSome(exemplarSet, unassignedSet)) {
1184 log_err("ExemplarSet contains unassigned characters for locale : %s\n", locale);
1185 }
1186 codeLen = uscript_getCode(locale, code, 8, &ec);
1187 if (strcmp(locale, "yi") == 0 && codeLen > 0 && log_knownIssue("11217", "Fix result of uscript_getCode for yi: USCRIPT_YI -> USCRIPT_HEBREW")) {
1188 code[0] = USCRIPT_HEBREW;
1189 }
1190 if (!assertSuccess("uscript_getCode", &ec)) goto END;
1191
1192 for (j=0; j<MAX_SCRIPTS_PER_LOCALE; ++j) {
1193 uset_close(codeSets[j]);
1194 codeSets[j] = NULL;
1195 }
1196 for (j=0; j<codeLen; ++j) {
1197 uprv_strcpy(cbuf, "[:");
1198 if(code[j]==-1){
1199 log_err("USCRIPT_INVALID_CODE returned for locale: %s\n", locale);
1200 continue;
1201 }
1202 uprv_strcat(cbuf, uscript_getShortName(code[j]));
1203 uprv_strcat(cbuf, ":]");
1204 u_uastrcpy(ubuf, cbuf);
1205 codeSets[j] = uset_openPattern(ubuf, -1, &ec);
1206 }
1207 if (!assertSuccess("uset_openPattern", &ec)) goto END;
1208
1209 existsInScript = FALSE;
1210 itemCount = uset_getItemCount(exemplarSet);
1211 for (m=0; m<itemCount && !existsInScript; ++m) {
1212 strLen = uset_getItem(exemplarSet, m, &start, &end, ubuf,
1213 UPRV_LENGTHOF(ubuf), &ec);
1214 /* failure here might mean str[] needs to be larger */
1215 if (!assertSuccess("uset_getItem", &ec)) goto END;
1216 if (strLen == 0) {
1217 for (j=0; j<codeLen; ++j) {
1218 if (codeSets[j]!=NULL && uset_containsRange(codeSets[j], start, end)) {
1219 existsInScript = TRUE;
1220 break;
1221 }
1222 }
1223 } else {
1224 for (j=0; j<codeLen; ++j) {
1225 if (codeSets[j]!=NULL && uset_containsString(codeSets[j], ubuf, strLen)) {
1226 existsInScript = TRUE;
1227 break;
1228 }
1229 }
1230 }
1231 }
1232
1233 if (existsInScript == FALSE){
1234 log_err("ExemplarSet containment failed for locale : %s\n", locale);
1235 }
1236 }
1237 assertTrue("case-folded is a superset",
1238 uset_containsAll(exemplarSets[1], exemplarSets[0]));
1239 if (uset_equals(exemplarSets[1], exemplarSets[0])) {
1240 ++equalCount;
1241 }
1242 }
1243 /* Note: The case-folded set should sometimes be a strict superset
1244 and sometimes be equal. */
1245 assertTrue("case-folded is sometimes a strict superset, and sometimes equal",
1246 equalCount > 0 && equalCount < n);
1247
1248 END:
1249 uenum_close(avail);
1250 uset_close(exemplarSets[0]);
1251 uset_close(exemplarSets[1]);
1252 uset_close(unassignedSet);
1253 for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) {
1254 uset_close(codeSets[i]);
1255 }
1256 }
1257
1258 enum { kUBufMax = 32 };
TestLocaleDisplayPattern(void)1259 static void TestLocaleDisplayPattern(void){
1260 UErrorCode status;
1261 UChar pattern[kUBufMax] = {0,};
1262 UChar separator[kUBufMax] = {0,};
1263 ULocaleData *uld;
1264 static const UChar enExpectPat[] = { 0x007B,0x0030,0x007D,0x0020,0x0028,0x007B,0x0031,0x007D,0x0029,0 }; /* "{0} ({1})" */
1265 static const UChar enExpectSep[] = { 0x002C,0x0020,0 }; /* ", " */
1266 static const UChar zhExpectPat[] = { 0x007B,0x0030,0x007D,0xFF08,0x007B,0x0031,0x007D,0xFF09,0 };
1267 static const UChar zhExpectSep[] = { 0xFF0C,0 };
1268
1269 status = U_ZERO_ERROR;
1270 uld = ulocdata_open("en", &status);
1271 if(U_FAILURE(status)){
1272 log_data_err("ulocdata_open en error %s", u_errorName(status));
1273 } else {
1274 ulocdata_getLocaleDisplayPattern(uld, pattern, kUBufMax, &status);
1275 if (U_FAILURE(status)){
1276 log_err("ulocdata_getLocaleDisplayPattern en error %s", u_errorName(status));
1277 } else if (u_strcmp(pattern, enExpectPat) != 0) {
1278 log_err("ulocdata_getLocaleDisplayPattern en returns unexpected pattern");
1279 }
1280 status = U_ZERO_ERROR;
1281 ulocdata_getLocaleSeparator(uld, separator, kUBufMax, &status);
1282 if (U_FAILURE(status)){
1283 log_err("ulocdata_getLocaleSeparator en error %s", u_errorName(status));
1284 } else if (u_strcmp(separator, enExpectSep) != 0) {
1285 log_err("ulocdata_getLocaleSeparator en returns unexpected string ");
1286 }
1287 ulocdata_close(uld);
1288 }
1289
1290 status = U_ZERO_ERROR;
1291 uld = ulocdata_open("zh", &status);
1292 if(U_FAILURE(status)){
1293 log_data_err("ulocdata_open zh error %s", u_errorName(status));
1294 } else {
1295 ulocdata_getLocaleDisplayPattern(uld, pattern, kUBufMax, &status);
1296 if (U_FAILURE(status)){
1297 log_err("ulocdata_getLocaleDisplayPattern zh error %s", u_errorName(status));
1298 } else if (u_strcmp(pattern, zhExpectPat) != 0) {
1299 log_err("ulocdata_getLocaleDisplayPattern zh returns unexpected pattern");
1300 }
1301 status = U_ZERO_ERROR;
1302 ulocdata_getLocaleSeparator(uld, separator, kUBufMax, &status);
1303 if (U_FAILURE(status)){
1304 log_err("ulocdata_getLocaleSeparator zh error %s", u_errorName(status));
1305 } else if (u_strcmp(separator, zhExpectSep) != 0) {
1306 log_err("ulocdata_getLocaleSeparator zh returns unexpected string ");
1307 }
1308 ulocdata_close(uld);
1309 }
1310 }
1311
TestCoverage(void)1312 static void TestCoverage(void){
1313 ULocaleDataDelimiterType types[] = {
1314 ULOCDATA_QUOTATION_START, /* Quotation start */
1315 ULOCDATA_QUOTATION_END, /* Quotation end */
1316 ULOCDATA_ALT_QUOTATION_START, /* Alternate quotation start */
1317 ULOCDATA_ALT_QUOTATION_END, /* Alternate quotation end */
1318 ULOCDATA_DELIMITER_COUNT
1319 };
1320 int i;
1321 UBool sub;
1322 UErrorCode status = U_ZERO_ERROR;
1323 ULocaleData *uld = ulocdata_open(uloc_getDefault(), &status);
1324
1325 if(U_FAILURE(status)){
1326 log_data_err("ulocdata_open error");
1327 return;
1328 }
1329
1330
1331 for(i = 0; i < ULOCDATA_DELIMITER_COUNT; i++){
1332 UChar result[32] = {0,};
1333 status = U_ZERO_ERROR;
1334 ulocdata_getDelimiter(uld, types[i], result, 32, &status);
1335 if (U_FAILURE(status)){
1336 log_err("ulocdata_getgetDelimiter error with type %d", types[i]);
1337 }
1338 }
1339
1340 sub = ulocdata_getNoSubstitute(uld);
1341 ulocdata_setNoSubstitute(uld,sub);
1342 ulocdata_close(uld);
1343 }
1344
TestIndexChars(void)1345 static void TestIndexChars(void) {
1346 /* Very basic test of ULOCDATA_ES_INDEX.
1347 * No comprehensive test of data, just basic check that the code path is alive.
1348 */
1349 UErrorCode status = U_ZERO_ERROR;
1350 ULocaleData *uld;
1351 USet *exemplarChars;
1352 USet *indexChars;
1353
1354 uld = ulocdata_open("en", &status);
1355 exemplarChars = uset_openEmpty();
1356 indexChars = uset_openEmpty();
1357 ulocdata_getExemplarSet(uld, exemplarChars, 0, ULOCDATA_ES_STANDARD, &status);
1358 ulocdata_getExemplarSet(uld, indexChars, 0, ULOCDATA_ES_INDEX, &status);
1359 if (U_FAILURE(status)) {
1360 log_data_err("File %s, line %d, Failure opening exemplar chars: %s", __FILE__, __LINE__, u_errorName(status));
1361 goto close_sets;
1362 }
1363 /* en data, standard exemplars are [a-z], lower case. */
1364 /* en data, index characters are [A-Z], upper case. */
1365 if ((uset_contains(exemplarChars, (UChar32)0x41) || uset_contains(indexChars, (UChar32)0x61))) {
1366 log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__, __LINE__ );
1367 goto close_sets;
1368 }
1369 if (!(uset_contains(exemplarChars, (UChar32)0x61) && uset_contains(indexChars, (UChar32)0x41) )) {
1370 log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__, __LINE__ );
1371 goto close_sets;
1372 }
1373
1374 close_sets:
1375 uset_close(exemplarChars);
1376 uset_close(indexChars);
1377 ulocdata_close(uld);
1378 }
1379
1380
1381
1382 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
TestCurrencyList(void)1383 static void TestCurrencyList(void){
1384 #if !UCONFIG_NO_FORMATTING
1385 UErrorCode errorCode = U_ZERO_ERROR;
1386 int32_t structLocaleCount, currencyCount;
1387 UEnumeration *en = ucurr_openISOCurrencies(UCURR_ALL, &errorCode);
1388 const char *isoCode, *structISOCode;
1389 UResourceBundle *subBundle;
1390 UResourceBundle *currencies = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode);
1391 if(U_FAILURE(errorCode)) {
1392 log_data_err("Can't open structLocale\n");
1393 return;
1394 }
1395 currencies = ures_getByKey(currencies, "Currencies", currencies, &errorCode);
1396 currencyCount = uenum_count(en, &errorCode);
1397 structLocaleCount = ures_getSize(currencies);
1398 if (currencyCount != structLocaleCount) {
1399 log_err("structLocale(%d) and ISO4217(%d) currency list are out of sync.\n", structLocaleCount, currencyCount);
1400 #if U_CHARSET_FAMILY == U_ASCII_FAMILY
1401 ures_resetIterator(currencies);
1402 while ((isoCode = uenum_next(en, NULL, &errorCode)) != NULL && ures_hasNext(currencies)) {
1403 subBundle = ures_getNextResource(currencies, NULL, &errorCode);
1404 structISOCode = ures_getKey(subBundle);
1405 ures_close(subBundle);
1406 if (strcmp(structISOCode, isoCode) != 0) {
1407 log_err("First difference found at structLocale(%s) and ISO4217(%s).\n", structISOCode, isoCode);
1408 break;
1409 }
1410 }
1411 #endif
1412 }
1413 ures_close(currencies);
1414 uenum_close(en);
1415 #endif
1416 }
1417 #endif
1418
TestAvailableIsoCodes(void)1419 static void TestAvailableIsoCodes(void){
1420 #if !UCONFIG_NO_FORMATTING
1421 UErrorCode errorCode = U_ZERO_ERROR;
1422 const char* eurCode = "EUR";
1423 const char* usdCode = "USD";
1424 const char* lastCode = "RHD";
1425 const char* zzzCode = "ZZZ";
1426 UDate date1950 = (UDate)-630720000000.0;/* year 1950 */
1427 UDate date1970 = (UDate)0.0; /* year 1970 */
1428 UDate date1975 = (UDate)173448000000.0; /* year 1975 */
1429 UDate date1978 = (UDate)260172000000.0; /* year 1978 */
1430 UDate date1981 = (UDate)346896000000.0; /* year 1981 */
1431 UDate date1992 = (UDate)693792000000.0; /* year 1992 */
1432 UChar* isoCode = (UChar*)malloc(sizeof(UChar) * (uprv_strlen(usdCode) + 1));
1433
1434 /* testing available codes with no time ranges */
1435 u_charsToUChars(eurCode, isoCode, uprv_strlen(usdCode) + 1);
1436 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) {
1437 log_data_err("FAIL: ISO code (%s) is not found.\n", eurCode);
1438 }
1439
1440 u_charsToUChars(usdCode, 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", usdCode);
1443 }
1444
1445 u_charsToUChars(zzzCode, isoCode, uprv_strlen(zzzCode) + 1);
1446 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == TRUE) {
1447 log_err("FAIL: ISO code (%s) is reported as available, but it doesn't exist.\n", zzzCode);
1448 }
1449
1450 u_charsToUChars(lastCode, isoCode, uprv_strlen(zzzCode) + 1);
1451 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) {
1452 log_data_err("FAIL: ISO code (%s) is not found.\n", lastCode);
1453 }
1454
1455 /* RHD was used from 1970-02-17 to 1980-04-18*/
1456
1457 /* to = null */
1458 if (ucurr_isAvailable(isoCode, date1970, U_DATE_MAX, &errorCode) == FALSE) {
1459 log_data_err("FAIL: ISO code (%s) was available in time range >1970-01-01.\n", lastCode);
1460 }
1461
1462 if (ucurr_isAvailable(isoCode, date1975, U_DATE_MAX, &errorCode) == FALSE) {
1463 log_data_err("FAIL: ISO code (%s) was available in time range >1975.\n", lastCode);
1464 }
1465
1466 if (ucurr_isAvailable(isoCode, date1981, U_DATE_MAX, &errorCode) == TRUE) {
1467 log_err("FAIL: ISO code (%s) was not available in time range >1981.\n", lastCode);
1468 }
1469
1470 /* from = null */
1471 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1970, &errorCode) == TRUE) {
1472 log_err("FAIL: ISO code (%s) was not available in time range <1970.\n", lastCode);
1473 }
1474
1475 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1975, &errorCode) == FALSE) {
1476 log_data_err("FAIL: ISO code (%s) was available in time range <1975.\n", lastCode);
1477 }
1478
1479 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1981, &errorCode) == FALSE) {
1480 log_data_err("FAIL: ISO code (%s) was available in time range <1981.\n", lastCode);
1481 }
1482
1483 /* full ranges */
1484 if (ucurr_isAvailable(isoCode, date1975, date1978, &errorCode) == FALSE) {
1485 log_data_err("FAIL: ISO code (%s) was available in time range 1975-1978.\n", lastCode);
1486 }
1487
1488 if (ucurr_isAvailable(isoCode, date1970, date1975, &errorCode) == FALSE) {
1489 log_data_err("FAIL: ISO code (%s) was available in time range 1970-1975.\n", lastCode);
1490 }
1491
1492 if (ucurr_isAvailable(isoCode, date1975, date1981, &errorCode) == FALSE) {
1493 log_data_err("FAIL: ISO code (%s) was available in time range 1975-1981.\n", lastCode);
1494 }
1495
1496 if (ucurr_isAvailable(isoCode, date1970, date1981, &errorCode) == FALSE) {
1497 log_data_err("FAIL: ISO code (%s) was available in time range 1970-1981.\n", lastCode);
1498 }
1499
1500 if (ucurr_isAvailable(isoCode, date1981, date1992, &errorCode) == TRUE) {
1501 log_err("FAIL: ISO code (%s) was not available in time range 1981-1992.\n", lastCode);
1502 }
1503
1504 if (ucurr_isAvailable(isoCode, date1950, date1970, &errorCode) == TRUE) {
1505 log_err("FAIL: ISO code (%s) was not available in time range 1950-1970.\n", lastCode);
1506 }
1507
1508 /* wrong range - from > to*/
1509 if (ucurr_isAvailable(isoCode, date1975, date1970, &errorCode) == TRUE) {
1510 log_err("FAIL: Wrong range 1975-1970 for ISO code (%s) was not reported.\n", lastCode);
1511 } else if (errorCode != U_ILLEGAL_ARGUMENT_ERROR) {
1512 log_data_err("FAIL: Error code not reported for wrong range 1975-1970 for ISO code (%s).\n", lastCode);
1513 }
1514
1515 free(isoCode);
1516 #endif
1517 }
1518
1519 #define TESTCASE(name) addTest(root, &name, "tsutil/cldrtest/" #name)
1520
1521 void addCLDRTest(TestNode** root);
1522
addCLDRTest(TestNode ** root)1523 void addCLDRTest(TestNode** root)
1524 {
1525 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
1526 TESTCASE(TestLocaleStructure);
1527 TESTCASE(TestCurrencyList);
1528 #endif
1529 TESTCASE(TestConsistentCountryInfo);
1530 TESTCASE(VerifyTranslation);
1531 TESTCASE(TestExemplarSet);
1532 TESTCASE(TestLocaleDisplayPattern);
1533 TESTCASE(TestCoverage);
1534 TESTCASE(TestIndexChars);
1535 TESTCASE(TestAvailableIsoCodes);
1536 }
1537
1538