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 (void)currentBundle; // suppress compiler warnings about unused variables
458 UErrorCode status = U_ZERO_ERROR;
459 uint32_t expectedLCID;
460 char lcidStringC[64] = {0};
461 int32_t len;
462
463 expectedLCID = uloc_getLCID(localeName);
464 if (expectedLCID == 0) {
465 log_verbose("INFO: %-5s does not have any LCID mapping\n",
466 localeName);
467 return;
468 }
469
470 status = U_ZERO_ERROR;
471 len = uprv_convertToPosix(expectedLCID, lcidStringC, UPRV_LENGTHOF(lcidStringC) - 1, &status);
472 if (U_FAILURE(status)) {
473 log_err("ERROR: %.4x does not have a POSIX mapping due to %s\n",
474 expectedLCID, u_errorName(status));
475 }
476 lcidStringC[len] = 0;
477
478 if(strcmp(localeName, lcidStringC) != 0) {
479 char langName[1024];
480 char langLCID[1024];
481 uloc_getLanguage(localeName, langName, sizeof(langName), &status);
482 uloc_getLanguage(lcidStringC, langLCID, sizeof(langLCID), &status);
483
484 if (strcmp(langName, langLCID) == 0) {
485 log_verbose("WARNING: %-5s resolves to %s (0x%.4x)\n",
486 localeName, lcidStringC, expectedLCID);
487 }
488 else if (!(strcmp(localeName, "ku") == 0 && log_knownIssue("20181", "ICU-20181 Fix LCID mapping for ckb vs ku"))) {
489 log_err("ERROR: %-5s has 0x%.4x and the number resolves wrongfully to %s\n",
490 localeName, expectedLCID, lcidStringC);
491 }
492 }
493 }
494
495 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
496 static void
TestLocaleStructure(void)497 TestLocaleStructure(void) {
498 // This test checks the locale structure against a key file located
499 // at source/test/testdata/structLocale.txt. When adding new data to
500 // a locale file such as en.txt, the structLocale.txt file must be changed
501 // too to include the the template of the new data. Otherwise this test
502 // will fail!
503
504 UResourceBundle *root, *currentLocale;
505 int32_t locCount = uloc_countAvailable();
506 int32_t locIndex;
507 UErrorCode errorCode = U_ZERO_ERROR;
508 const char *currLoc, *resolvedLoc;
509
510 /* TODO: Compare against parent's data too. This code can't handle fallbacks that some tools do already. */
511 /* char locName[ULOC_FULLNAME_CAPACITY];
512 char *locNamePtr;
513
514 for (locIndex = 0; locIndex < locCount; locIndex++) {
515 errorCode=U_ZERO_ERROR;
516 strcpy(locName, uloc_getAvailable(locIndex));
517 locNamePtr = strrchr(locName, '_');
518 if (locNamePtr) {
519 *locNamePtr = 0;
520 }
521 else {
522 strcpy(locName, "root");
523 }
524
525 root = ures_openDirect(NULL, locName, &errorCode);
526 if(U_FAILURE(errorCode)) {
527 log_err("Can't open %s\n", locName);
528 continue;
529 }
530 */
531 if (locCount <= 1) {
532 log_data_err("At least root needs to be installed\n");
533 }
534
535 root = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode);
536 if(U_FAILURE(errorCode)) {
537 log_data_err("Can't open structLocale\n");
538 return;
539 }
540 for (locIndex = 0; locIndex < locCount; locIndex++) {
541 errorCode=U_ZERO_ERROR;
542 currLoc = uloc_getAvailable(locIndex);
543 currentLocale = ures_open(NULL, currLoc, &errorCode);
544 if(errorCode != U_ZERO_ERROR) {
545 if(U_SUCCESS(errorCode)) {
546 /* It's installed, but there is no data.
547 It's installed for the g18n white paper [grhoten] */
548 log_err("ERROR: Locale %-5s not installed, and it should be, err %s\n",
549 uloc_getAvailable(locIndex), u_errorName(errorCode));
550 } else {
551 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
552 u_errorName(errorCode),
553 uloc_getAvailable(locIndex));
554 }
555 ures_close(currentLocale);
556 continue;
557 }
558 ures_getStringByKey(currentLocale, "Version", NULL, &errorCode);
559 if(errorCode != U_ZERO_ERROR) {
560 log_err("No version information is available for locale %s, and it should be!\n",
561 currLoc);
562 }
563 else if (ures_getStringByKey(currentLocale, "Version", NULL, &errorCode)[0] == (UChar)(0x78)) {
564 log_verbose("WARNING: The locale %s is experimental! It shouldn't be listed as an installed locale.\n",
565 currLoc);
566 }
567 resolvedLoc = ures_getLocaleByType(currentLocale, ULOC_ACTUAL_LOCALE, &errorCode);
568 if (strcmp(resolvedLoc, currLoc) != 0) {
569 /* All locales have at least a Version resource.
570 If it's absolutely empty, then the previous test will fail too.*/
571 log_err("Locale resolves to different locale. Is %s an alias of %s?\n",
572 currLoc, resolvedLoc);
573 }
574 TestKeyInRootRecursive(root, "root", currentLocale, currLoc);
575
576 testLCID(currentLocale, currLoc);
577
578 ures_close(currentLocale);
579 }
580
581 ures_close(root);
582 }
583 #endif
584
585 static void
compareArrays(const char * keyName,UResourceBundle * fromArray,const char * fromLocale,UResourceBundle * toArray,const char * toLocale,int32_t start,int32_t end)586 compareArrays(const char *keyName,
587 UResourceBundle *fromArray, const char *fromLocale,
588 UResourceBundle *toArray, const char *toLocale,
589 int32_t start, int32_t end)
590 {
591 int32_t fromSize = ures_getSize(fromArray);
592 int32_t toSize = ures_getSize(fromArray);
593 int32_t idx;
594 UErrorCode errorCode = U_ZERO_ERROR;
595
596 if (fromSize > toSize) {
597 fromSize = toSize;
598 log_err("Arrays are different size from \"%s\" to \"%s\"\n",
599 fromLocale,
600 toLocale);
601 }
602
603 for (idx = start; idx <= end; idx++) {
604 const UChar *fromBundleStr = ures_getStringByIndex(fromArray, idx, NULL, &errorCode);
605 const UChar *toBundleStr = ures_getStringByIndex(toArray, idx, NULL, &errorCode);
606 if (fromBundleStr && toBundleStr && u_strcmp(fromBundleStr, toBundleStr) != 0)
607 {
608 log_err("Difference for %s at index %d from %s= \"%s\" to %s= \"%s\"\n",
609 keyName,
610 idx,
611 fromLocale,
612 austrdup(fromBundleStr),
613 toLocale,
614 austrdup(toBundleStr));
615 }
616 }
617 }
618
619 static void
compareConsistentCountryInfo(const char * fromLocale,const char * toLocale)620 compareConsistentCountryInfo(const char *fromLocale, const char *toLocale) {
621 UErrorCode errorCode = U_ZERO_ERROR;
622 UResourceBundle *fromArray, *toArray;
623 UResourceBundle *fromLocaleBund = ures_open(NULL, fromLocale, &errorCode);
624 UResourceBundle *toLocaleBund = ures_open(NULL, toLocale, &errorCode);
625 UResourceBundle *toCalendar, *fromCalendar, *toGregorian, *fromGregorian;
626
627 if(U_FAILURE(errorCode)) {
628 log_err("Can't open resource bundle %s or %s - %s\n", fromLocale, toLocale, u_errorName(errorCode));
629 return;
630 }
631 fromCalendar = ures_getByKey(fromLocaleBund, "calendar", NULL, &errorCode);
632 fromGregorian = ures_getByKeyWithFallback(fromCalendar, "gregorian", NULL, &errorCode);
633
634 toCalendar = ures_getByKey(toLocaleBund, "calendar", NULL, &errorCode);
635 toGregorian = ures_getByKeyWithFallback(toCalendar, "gregorian", NULL, &errorCode);
636
637 fromArray = ures_getByKey(fromLocaleBund, "CurrencyElements", NULL, &errorCode);
638 toArray = ures_getByKey(toLocaleBund, "CurrencyElements", NULL, &errorCode);
639 if (strcmp(fromLocale, "en_CA") != 0)
640 {
641 /* The first one is probably localized. */
642 compareArrays("CurrencyElements", fromArray, fromLocale, toArray, toLocale, 1, 2);
643 }
644 ures_close(fromArray);
645 ures_close(toArray);
646
647 fromArray = ures_getByKey(fromLocaleBund, "NumberPatterns", NULL, &errorCode);
648 toArray = ures_getByKey(toLocaleBund, "NumberPatterns", NULL, &errorCode);
649 if (strcmp(fromLocale, "en_CA") != 0)
650 {
651 compareArrays("NumberPatterns", fromArray, fromLocale, toArray, toLocale, 0, 3);
652 }
653 ures_close(fromArray);
654 ures_close(toArray);
655
656 /* Difficult to test properly */
657 /*
658 fromArray = ures_getByKey(fromLocaleBund, "DateTimePatterns", NULL, &errorCode);
659 toArray = ures_getByKey(toLocaleBund, "DateTimePatterns", NULL, &errorCode);
660 {
661 compareArrays("DateTimePatterns", fromArray, fromLocale, toArray, toLocale);
662 }
663 ures_close(fromArray);
664 ures_close(toArray);*/
665
666 fromArray = ures_getByKey(fromLocaleBund, "NumberElements", NULL, &errorCode);
667 toArray = ures_getByKey(toLocaleBund, "NumberElements", NULL, &errorCode);
668 if (strcmp(fromLocale, "en_CA") != 0)
669 {
670 compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 0, 3);
671 /* Index 4 is a script based 0 */
672 compareArrays("NumberElements", fromArray, fromLocale, toArray, toLocale, 5, 10);
673 }
674 ures_close(fromArray);
675 ures_close(toArray);
676 ures_close(fromCalendar);
677 ures_close(toCalendar);
678 ures_close(fromGregorian);
679 ures_close(toGregorian);
680
681 ures_close(fromLocaleBund);
682 ures_close(toLocaleBund);
683 }
684
685 static void
TestConsistentCountryInfo(void)686 TestConsistentCountryInfo(void) {
687 /* UResourceBundle *fromLocale, *toLocale;*/
688 int32_t locCount = uloc_countAvailable();
689 int32_t fromLocIndex, toLocIndex;
690
691 int32_t fromCountryLen, toCountryLen;
692 char fromCountry[ULOC_FULLNAME_CAPACITY], toCountry[ULOC_FULLNAME_CAPACITY];
693
694 int32_t fromVariantLen, toVariantLen;
695 char fromVariant[ULOC_FULLNAME_CAPACITY], toVariant[ULOC_FULLNAME_CAPACITY];
696
697 UErrorCode errorCode = U_ZERO_ERROR;
698
699 for (fromLocIndex = 0; fromLocIndex < locCount; fromLocIndex++) {
700 const char *fromLocale = uloc_getAvailable(fromLocIndex);
701
702 errorCode=U_ZERO_ERROR;
703 fromCountryLen = uloc_getCountry(fromLocale, fromCountry, ULOC_FULLNAME_CAPACITY, &errorCode);
704 if (fromCountryLen <= 0) {
705 /* Ignore countryless locales */
706 continue;
707 }
708 fromVariantLen = uloc_getVariant(fromLocale, fromVariant, ULOC_FULLNAME_CAPACITY, &errorCode);
709 if (fromVariantLen > 0) {
710 /* Most variants are ignorable like collation variants. */
711 continue;
712 }
713 /* Start comparing only after the current index.
714 Previous loop should have already compared fromLocIndex.
715 */
716 for (toLocIndex = fromLocIndex + 1; toLocIndex < locCount; toLocIndex++) {
717 const char *toLocale = uloc_getAvailable(toLocIndex);
718
719 toCountryLen = uloc_getCountry(toLocale, toCountry, ULOC_FULLNAME_CAPACITY, &errorCode);
720 if(U_FAILURE(errorCode)) {
721 log_err("Unknown failure fromLocale=%s toLocale=%s errorCode=%s\n",
722 fromLocale, toLocale, u_errorName(errorCode));
723 continue;
724 }
725
726 if (toCountryLen <= 0) {
727 /* Ignore countryless locales */
728 continue;
729 }
730 toVariantLen = uloc_getVariant(toLocale, toVariant, ULOC_FULLNAME_CAPACITY, &errorCode);
731 if (toVariantLen > 0) {
732 /* Most variants are ignorable like collation variants. */
733 /* They're a variant for a reason. */
734 continue;
735 }
736 if (strcmp(fromCountry, toCountry) == 0) {
737 log_verbose("comparing fromLocale=%s toLocale=%s\n",
738 fromLocale, toLocale);
739 compareConsistentCountryInfo(fromLocale, toLocale);
740 }
741 }
742 }
743 }
744
745 static int32_t
findStringSetMismatch(const char * currLoc,const UChar * string,int32_t langSize,USet * mergedExemplarSet,UBool ignoreNumbers,UChar32 * badCharPtr)746 findStringSetMismatch(const char *currLoc, const UChar *string, int32_t langSize,
747 USet * mergedExemplarSet,
748 UBool ignoreNumbers, UChar32* badCharPtr) {
749 UErrorCode errorCode = U_ZERO_ERROR;
750 USet *exemplarSet;
751 int32_t strIdx;
752 if (mergedExemplarSet == NULL) {
753 return -1;
754 }
755 exemplarSet = createFlattenSet(mergedExemplarSet, &errorCode);
756 if (U_FAILURE(errorCode)) {
757 log_err("%s: error createFlattenSet returned %s\n", currLoc, u_errorName(errorCode));
758 return -1;
759 }
760
761 for (strIdx = 0; strIdx < langSize;) {
762 UChar32 testChar;
763 U16_NEXT(string, strIdx, langSize, testChar);
764 if (!uset_contains(exemplarSet, testChar)
765 && testChar != 0x0020 && testChar != 0x00A0 && testChar != 0x002e && testChar != 0x002c && testChar != 0x002d && testChar != 0x0027
766 && testChar != 0x005B && testChar != 0x005D && testChar != 0x2019 && testChar != 0x0f0b && testChar != 0x200C && testChar != 0x200D) {
767 if (!ignoreNumbers || (ignoreNumbers && (testChar < 0x30 || testChar > 0x39))) {
768 uset_close(exemplarSet);
769 if (badCharPtr) {
770 *badCharPtr = testChar;
771 }
772 return strIdx;
773 }
774 }
775 }
776 uset_close(exemplarSet);
777 if (badCharPtr) {
778 *badCharPtr = 0;
779 }
780 return -1;
781 }
782 /* include non-invariant chars */
783 static int32_t
myUCharsToChars(const UChar * us,char * cs,int32_t len)784 myUCharsToChars(const UChar* us, char* cs, int32_t len){
785 int32_t i=0;
786 for(; i< len; i++){
787 if(us[i] < 0x7f){
788 cs[i] = (char)us[i];
789 }else{
790 return -1;
791 }
792 }
793 return i;
794 }
795 static void
findSetMatch(UScriptCode * scriptCodes,int32_t scriptsLen,USet * exemplarSet,const char * locale)796 findSetMatch( UScriptCode *scriptCodes, int32_t scriptsLen,
797 USet *exemplarSet,
798 const char *locale){
799 USet *scripts[10]= {0};
800 char pattern[256] = { '[', ':', 0x000 };
801 int32_t patternLen;
802 UChar uPattern[256] = {0};
803 UErrorCode status = U_ZERO_ERROR;
804 int32_t i;
805
806 /* create the sets with script codes */
807 for(i = 0; i<scriptsLen; i++){
808 strcat(pattern, uscript_getShortName(scriptCodes[i]));
809 strcat(pattern, ":]");
810 patternLen = (int32_t)strlen(pattern);
811 u_charsToUChars(pattern, uPattern, patternLen);
812 scripts[i] = uset_openPattern(uPattern, patternLen, &status);
813 if(U_FAILURE(status)){
814 log_err("Could not create set for pattern %s. Error: %s\n", pattern, u_errorName(status));
815 return;
816 }
817 pattern[2] = 0;
818 }
819 if (strcmp(locale, "uk") == 0 || strcmp(locale, "uk_UA") == 0) {
820 /* Special addition. Add the modifying apostrophe, which isn't in Cyrillic. */
821 uset_add(scripts[0], 0x2bc);
822 }
823 if(U_SUCCESS(status)){
824 UBool existsInScript = FALSE;
825 /* iterate over the exemplarSet and ascertain if all
826 * UChars in exemplarSet belong to the scripts returned
827 * by getScript
828 */
829 int32_t count = uset_getItemCount(exemplarSet);
830
831 for( i=0; i < count; i++){
832 UChar32 start = 0;
833 UChar32 end = 0;
834 UChar *str = NULL;
835 int32_t strCapacity = 0;
836
837 strCapacity = uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status);
838 if(U_SUCCESS(status)){
839 int32_t j;
840 if(strCapacity == 0){
841 /* ok the item is a range */
842 for( j = 0; j < scriptsLen; j++){
843 if(uset_containsRange(scripts[j], start, end) == TRUE){
844 existsInScript = TRUE;
845 }
846 }
847 if(existsInScript == FALSE){
848 for( j = 0; j < scriptsLen; j++){
849 UChar toPattern[500]={'\0'};
850 char pat[500]={'\0'};
851 int32_t len = uset_toPattern(scripts[j], toPattern, 500, TRUE, &status);
852 len = myUCharsToChars(toPattern, pat, len);
853 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));
854 if(len!=-1){
855 log_err("Pattern: %s\n",pat);
856 }
857 }
858 log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale);
859 }
860 }else{
861 strCapacity++; /* increment for NUL termination */
862 /* allocate the str and call the api again */
863 str = (UChar*) malloc(U_SIZEOF_UCHAR * strCapacity);
864 strCapacity = uset_getItem(exemplarSet, i, &start, &end, str, strCapacity, &status);
865 /* iterate over the scripts and figure out if the string contained is actually
866 * in the script set
867 */
868 for( j = 0; j < scriptsLen; j++){
869 if(uset_containsString(scripts[j],str, strCapacity) == TRUE){
870 existsInScript = TRUE;
871 }
872 }
873 if(existsInScript == FALSE){
874 log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale);
875 }
876 }
877 }
878 }
879
880 }
881
882 /* close the sets */
883 for(i = 0; i<scriptsLen; i++){
884 uset_close(scripts[i]);
885 }
886 }
887
VerifyTranslation(void)888 static void VerifyTranslation(void) {
889 UResourceBundle *root, *currentLocale;
890 int32_t locCount = uloc_countAvailable();
891 int32_t locIndex;
892 UErrorCode errorCode = U_ZERO_ERROR;
893 const char *currLoc;
894 UScriptCode scripts[USCRIPT_CODE_LIMIT];
895 int32_t numScripts;
896 int32_t idx;
897 int32_t end;
898 UResourceBundle *resArray;
899
900 if (locCount <= 1) {
901 log_data_err("At least root needs to be installed\n");
902 }
903
904 root = ures_openDirect(NULL, "root", &errorCode);
905 if(U_FAILURE(errorCode)) {
906 log_data_err("Can't open root\n");
907 return;
908 }
909 for (locIndex = 0; locIndex < locCount; locIndex++) {
910 USet * mergedExemplarSet = NULL;
911 errorCode=U_ZERO_ERROR;
912 currLoc = uloc_getAvailable(locIndex);
913 currentLocale = ures_open(NULL, currLoc, &errorCode);
914 if(errorCode != U_ZERO_ERROR) {
915 if(U_SUCCESS(errorCode)) {
916 /* It's installed, but there is no data.
917 It's installed for the g18n white paper [grhoten] */
918 log_err("ERROR: Locale %-5s not installed, and it should be!\n",
919 uloc_getAvailable(locIndex));
920 } else {
921 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
922 u_errorName(errorCode),
923 uloc_getAvailable(locIndex));
924 }
925 ures_close(currentLocale);
926 continue;
927 }
928 {
929 UErrorCode exemplarStatus = U_ZERO_ERROR;
930 ULocaleData * uld = ulocdata_open(currLoc, &exemplarStatus);
931 if (U_SUCCESS(exemplarStatus)) {
932 USet * exemplarSet = ulocdata_getExemplarSet(uld, NULL, USET_ADD_CASE_MAPPINGS, ULOCDATA_ES_STANDARD, &exemplarStatus);
933 if (U_SUCCESS(exemplarStatus)) {
934 mergedExemplarSet = uset_cloneAsThawed(exemplarSet);
935 uset_close(exemplarSet);
936 exemplarSet = ulocdata_getExemplarSet(uld, NULL, USET_ADD_CASE_MAPPINGS, ULOCDATA_ES_AUXILIARY, &exemplarStatus);
937 if (U_SUCCESS(exemplarStatus)) {
938 uset_addAll(mergedExemplarSet, exemplarSet);
939 uset_close(exemplarSet);
940 }
941 exemplarStatus = U_ZERO_ERROR;
942 exemplarSet = ulocdata_getExemplarSet(uld, NULL, 0, ULOCDATA_ES_PUNCTUATION, &exemplarStatus);
943 if (U_SUCCESS(exemplarStatus)) {
944 uset_addAll(mergedExemplarSet, exemplarSet);
945 uset_close(exemplarSet);
946 }
947 } else {
948 log_err("error ulocdata_getExemplarSet (main) for locale %s returned %s\n", currLoc, u_errorName(errorCode));
949 }
950 ulocdata_close(uld);
951 } else {
952 log_err("error ulocdata_open for locale %s returned %s\n", currLoc, u_errorName(errorCode));
953 }
954 }
955 if (mergedExemplarSet == NULL /*|| (getTestOption(QUICK_OPTION) && uset_size() > 2048)*/) {
956 log_verbose("skipping test for %s\n", currLoc);
957 }
958 //else if (uprv_strncmp(currLoc,"bem",3) == 0 || uprv_strncmp(currLoc,"mgo",3) == 0 || uprv_strncmp(currLoc,"nl",2) == 0) {
959 // log_verbose("skipping test for %s, some month and country names known to use aux exemplars\n", currLoc);
960 //}
961 else {
962 UChar langBuffer[128];
963 int32_t langSize;
964 int32_t strIdx;
965 UChar32 badChar;
966 langSize = uloc_getDisplayLanguage(currLoc, currLoc, langBuffer, UPRV_LENGTHOF(langBuffer), &errorCode);
967 if (U_FAILURE(errorCode)) {
968 log_err("error uloc_getDisplayLanguage returned %s\n", u_errorName(errorCode));
969 }
970 else {
971 strIdx = findStringSetMismatch(currLoc, langBuffer, langSize, mergedExemplarSet, FALSE, &badChar);
972 if (strIdx >= 0) {
973 log_err("getDisplayLanguage(%s) at index %d returned characters not in the exemplar characters: %04X.\n",
974 currLoc, strIdx, badChar);
975 }
976 }
977 langSize = uloc_getDisplayCountry(currLoc, currLoc, langBuffer, UPRV_LENGTHOF(langBuffer), &errorCode);
978 if (U_FAILURE(errorCode)) {
979 log_err("error uloc_getDisplayCountry returned %s\n", u_errorName(errorCode));
980 }
981 {
982 UResourceBundle* cal = ures_getByKey(currentLocale, "calendar", NULL, &errorCode);
983 UResourceBundle* greg = ures_getByKeyWithFallback(cal, "gregorian", NULL, &errorCode);
984 UResourceBundle* names = ures_getByKeyWithFallback(greg, "dayNames", NULL, &errorCode);
985 UResourceBundle* format = ures_getByKeyWithFallback(names, "format", NULL, &errorCode);
986 resArray = ures_getByKeyWithFallback(format, "wide", NULL, &errorCode);
987
988 if (U_FAILURE(errorCode)) {
989 log_err("error ures_getByKey returned %s\n", u_errorName(errorCode));
990 }
991 if (getTestOption(QUICK_OPTION)) {
992 end = 1;
993 }
994 else {
995 end = ures_getSize(resArray);
996 }
997
998 if ((uprv_strncmp(currLoc,"lrc",3) == 0 || uprv_strncmp(currLoc,"mzn",3) == 0) &&
999 log_knownIssue("cldrbug:8899", "lrc and mzn locales don't have translated day names")) {
1000 end = 0;
1001 }
1002
1003 for (idx = 0; idx < end; idx++) {
1004 const UChar *fromBundleStr = ures_getStringByIndex(resArray, idx, &langSize, &errorCode);
1005 if (U_FAILURE(errorCode)) {
1006 log_err("error ures_getStringByIndex(%d) returned %s\n", idx, u_errorName(errorCode));
1007 continue;
1008 }
1009 strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, mergedExemplarSet, TRUE, &badChar);
1010 if ( strIdx >= 0 ) {
1011 log_err("getDayNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n",
1012 currLoc, idx, strIdx, badChar);
1013 }
1014 }
1015 ures_close(resArray);
1016 ures_close(format);
1017 ures_close(names);
1018
1019 names = ures_getByKeyWithFallback(greg, "monthNames", NULL, &errorCode);
1020 format = ures_getByKeyWithFallback(names,"format", NULL, &errorCode);
1021 resArray = ures_getByKeyWithFallback(format, "wide", NULL, &errorCode);
1022 if (U_FAILURE(errorCode)) {
1023 log_err("error ures_getByKey returned %s\n", u_errorName(errorCode));
1024 }
1025 if (getTestOption(QUICK_OPTION)) {
1026 end = 1;
1027 }
1028 else {
1029 end = ures_getSize(resArray);
1030 }
1031
1032 for (idx = 0; idx < end; idx++) {
1033 const UChar *fromBundleStr = ures_getStringByIndex(resArray, idx, &langSize, &errorCode);
1034 if (U_FAILURE(errorCode)) {
1035 log_err("error ures_getStringByIndex(%d) returned %s\n", idx, u_errorName(errorCode));
1036 continue;
1037 }
1038 strIdx = findStringSetMismatch(currLoc, fromBundleStr, langSize, mergedExemplarSet, TRUE, &badChar);
1039 if (strIdx >= 0) {
1040 log_err("getMonthNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n",
1041 currLoc, idx, strIdx, badChar);
1042 }
1043 }
1044 ures_close(resArray);
1045 ures_close(format);
1046 ures_close(names);
1047 ures_close(greg);
1048 ures_close(cal);
1049 }
1050 errorCode = U_ZERO_ERROR;
1051 numScripts = uscript_getCode(currLoc, scripts, UPRV_LENGTHOF(scripts), &errorCode);
1052 if (strcmp(currLoc, "yi") == 0 && numScripts > 0 && log_knownIssue("11217", "Fix result of uscript_getCode for yi: USCRIPT_YI -> USCRIPT_HEBREW")) {
1053 scripts[0] = USCRIPT_HEBREW;
1054 }
1055 if (numScripts == 0) {
1056 log_err("uscript_getCode(%s) doesn't work.\n", currLoc);
1057 }else if(scripts[0] == USCRIPT_COMMON){
1058 log_err("uscript_getCode(%s) returned USCRIPT_COMMON.\n", currLoc);
1059 }
1060
1061 /* test that the scripts are a superset of exemplar characters. */
1062 {
1063 ULocaleData *uld = ulocdata_open(currLoc,&errorCode);
1064 USet *exemplarSet = ulocdata_getExemplarSet(uld, NULL, 0, ULOCDATA_ES_STANDARD, &errorCode);
1065 /* test if exemplar characters are part of script code */
1066 findSetMatch(scripts, numScripts, exemplarSet, currLoc);
1067 uset_close(exemplarSet);
1068 ulocdata_close(uld);
1069 }
1070
1071 /* test that the paperSize API works */
1072 {
1073 int32_t height=0, width=0;
1074 ulocdata_getPaperSize(currLoc, &height, &width, &errorCode);
1075 if(U_FAILURE(errorCode)){
1076 log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1077 }
1078 if(strstr(currLoc, "_US")!=NULL && height != 279 && width != 216 ){
1079 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc);
1080 }
1081 }
1082 /* test that the MeasurementSystem API works */
1083 {
1084 char fullLoc[ULOC_FULLNAME_CAPACITY];
1085 UMeasurementSystem measurementSystem;
1086 int32_t height = 0, width = 0;
1087
1088 uloc_addLikelySubtags(currLoc, fullLoc, ULOC_FULLNAME_CAPACITY, &errorCode);
1089
1090 errorCode = U_ZERO_ERROR;
1091 measurementSystem = ulocdata_getMeasurementSystem(currLoc, &errorCode);
1092 if (U_FAILURE(errorCode)) {
1093 log_err("ulocdata_getMeasurementSystem failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1094 } else {
1095 if ( strstr(fullLoc, "_US")!=NULL || strstr(fullLoc, "_MM")!=NULL || strstr(fullLoc, "_LR")!=NULL ) {
1096 if(measurementSystem != UMS_US){
1097 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1098 }
1099 } else if ( strstr(fullLoc, "_GB")!=NULL ) {
1100 if(measurementSystem != UMS_UK){
1101 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1102 }
1103 } else if (measurementSystem != UMS_SI) {
1104 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc);
1105 }
1106 }
1107
1108 errorCode = U_ZERO_ERROR;
1109 ulocdata_getPaperSize(currLoc, &height, &width, &errorCode);
1110 if (U_FAILURE(errorCode)) {
1111 log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc, u_errorName(errorCode));
1112 } else {
1113 if ( strstr(fullLoc, "_US")!=NULL || strstr(fullLoc, "_BZ")!=NULL || strstr(fullLoc, "_CA")!=NULL || strstr(fullLoc, "_CL")!=NULL ||
1114 strstr(fullLoc, "_CO")!=NULL || strstr(fullLoc, "_CR")!=NULL || strstr(fullLoc, "_GT")!=NULL || strstr(fullLoc, "_MX")!=NULL ||
1115 strstr(fullLoc, "_NI")!=NULL || strstr(fullLoc, "_PA")!=NULL || strstr(fullLoc, "_PH")!=NULL || strstr(fullLoc, "_PR")!=NULL ||
1116 strstr(fullLoc, "_SV")!=NULL || strstr(fullLoc, "_VE")!=NULL ) {
1117 if (height != 279 || width != 216) {
1118 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc);
1119 }
1120 } else if (height != 297 || width != 210) {
1121 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc);
1122 }
1123 }
1124 }
1125 }
1126 if (mergedExemplarSet != NULL) {
1127 uset_close(mergedExemplarSet);
1128 }
1129 ures_close(currentLocale);
1130 }
1131
1132 ures_close(root);
1133 }
1134
1135 /* adjust this limit as appropriate */
1136 #define MAX_SCRIPTS_PER_LOCALE 8
1137
TestExemplarSet(void)1138 static void TestExemplarSet(void){
1139 int32_t i, j, k, m, n;
1140 int32_t equalCount = 0;
1141 UErrorCode ec = U_ZERO_ERROR;
1142 UEnumeration* avail;
1143 USet* exemplarSets[2];
1144 USet* unassignedSet;
1145 UScriptCode code[MAX_SCRIPTS_PER_LOCALE];
1146 USet* codeSets[MAX_SCRIPTS_PER_LOCALE];
1147 int32_t codeLen;
1148 char cbuf[32]; /* 9 should be enough */
1149 UChar ubuf[64]; /* adjust as needed */
1150 UBool existsInScript;
1151 int32_t itemCount;
1152 int32_t strLen;
1153 UChar32 start, end;
1154
1155 unassignedSet = NULL;
1156 exemplarSets[0] = NULL;
1157 exemplarSets[1] = NULL;
1158 for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) {
1159 codeSets[i] = NULL;
1160 }
1161
1162 avail = ures_openAvailableLocales(NULL, &ec);
1163 if (!assertSuccess("ures_openAvailableLocales", &ec)) goto END;
1164 n = uenum_count(avail, &ec);
1165 if (!assertSuccess("uenum_count", &ec)) goto END;
1166
1167 u_uastrcpy(ubuf, "[:unassigned:]");
1168 unassignedSet = uset_openPattern(ubuf, -1, &ec);
1169 if (!assertSuccess("uset_openPattern", &ec)) goto END;
1170
1171 for(i=0; i<n; i++){
1172 const char* locale = uenum_next(avail, NULL, &ec);
1173 if (!assertSuccess("uenum_next", &ec)) goto END;
1174 log_verbose("%s\n", locale);
1175 for (k=0; k<2; ++k) {
1176 uint32_t option = (k==0) ? 0 : USET_CASE_INSENSITIVE;
1177 ULocaleData *uld = ulocdata_open(locale,&ec);
1178 USet* exemplarSet = ulocdata_getExemplarSet(uld,NULL, option, ULOCDATA_ES_STANDARD, &ec);
1179 uset_close(exemplarSets[k]);
1180 ulocdata_close(uld);
1181 exemplarSets[k] = exemplarSet;
1182 if (!assertSuccess("ulocaledata_getExemplarSet", &ec)) goto END;
1183
1184 if (uset_containsSome(exemplarSet, unassignedSet)) {
1185 log_err("ExemplarSet contains unassigned characters for locale : %s\n", locale);
1186 }
1187 codeLen = uscript_getCode(locale, code, 8, &ec);
1188 if (strcmp(locale, "yi") == 0 && codeLen > 0 && log_knownIssue("11217", "Fix result of uscript_getCode for yi: USCRIPT_YI -> USCRIPT_HEBREW")) {
1189 code[0] = USCRIPT_HEBREW;
1190 }
1191 if (!assertSuccess("uscript_getCode", &ec)) goto END;
1192
1193 for (j=0; j<MAX_SCRIPTS_PER_LOCALE; ++j) {
1194 uset_close(codeSets[j]);
1195 codeSets[j] = NULL;
1196 }
1197 for (j=0; j<codeLen; ++j) {
1198 uprv_strcpy(cbuf, "[:");
1199 if(code[j]==-1){
1200 log_err("USCRIPT_INVALID_CODE returned for locale: %s\n", locale);
1201 continue;
1202 }
1203 uprv_strcat(cbuf, uscript_getShortName(code[j]));
1204 uprv_strcat(cbuf, ":]");
1205 u_uastrcpy(ubuf, cbuf);
1206 codeSets[j] = uset_openPattern(ubuf, -1, &ec);
1207 }
1208 if (!assertSuccess("uset_openPattern", &ec)) goto END;
1209
1210 existsInScript = FALSE;
1211 itemCount = uset_getItemCount(exemplarSet);
1212 for (m=0; m<itemCount && !existsInScript; ++m) {
1213 strLen = uset_getItem(exemplarSet, m, &start, &end, ubuf,
1214 UPRV_LENGTHOF(ubuf), &ec);
1215 /* failure here might mean str[] needs to be larger */
1216 if (!assertSuccess("uset_getItem", &ec)) goto END;
1217 if (strLen == 0) {
1218 for (j=0; j<codeLen; ++j) {
1219 if (codeSets[j]!=NULL && uset_containsRange(codeSets[j], start, end)) {
1220 existsInScript = TRUE;
1221 break;
1222 }
1223 }
1224 } else {
1225 for (j=0; j<codeLen; ++j) {
1226 if (codeSets[j]!=NULL && uset_containsString(codeSets[j], ubuf, strLen)) {
1227 existsInScript = TRUE;
1228 break;
1229 }
1230 }
1231 }
1232 }
1233
1234 if (existsInScript == FALSE){
1235 log_err("ExemplarSet containment failed for locale : %s\n", locale);
1236 }
1237 }
1238 assertTrue("case-folded is a superset",
1239 uset_containsAll(exemplarSets[1], exemplarSets[0]));
1240 if (uset_equals(exemplarSets[1], exemplarSets[0])) {
1241 ++equalCount;
1242 }
1243 }
1244 /* Note: The case-folded set should sometimes be a strict superset
1245 and sometimes be equal. */
1246 assertTrue("case-folded is sometimes a strict superset, and sometimes equal",
1247 equalCount > 0 && equalCount < n);
1248
1249 END:
1250 uenum_close(avail);
1251 uset_close(exemplarSets[0]);
1252 uset_close(exemplarSets[1]);
1253 uset_close(unassignedSet);
1254 for (i=0; i<MAX_SCRIPTS_PER_LOCALE; ++i) {
1255 uset_close(codeSets[i]);
1256 }
1257 }
1258
1259 enum { kUBufMax = 32 };
TestLocaleDisplayPattern(void)1260 static void TestLocaleDisplayPattern(void){
1261 UErrorCode status;
1262 UChar pattern[kUBufMax] = {0,};
1263 UChar separator[kUBufMax] = {0,};
1264 ULocaleData *uld;
1265 static const UChar enExpectPat[] = { 0x007B,0x0030,0x007D,0x0020,0x0028,0x007B,0x0031,0x007D,0x0029,0 }; /* "{0} ({1})" */
1266 static const UChar enExpectSep[] = { 0x002C,0x0020,0 }; /* ", " */
1267 static const UChar zhExpectPat[] = { 0x007B,0x0030,0x007D,0xFF08,0x007B,0x0031,0x007D,0xFF09,0 };
1268 static const UChar zhExpectSep[] = { 0xFF0C,0 };
1269
1270 status = U_ZERO_ERROR;
1271 uld = ulocdata_open("en", &status);
1272 if(U_FAILURE(status)){
1273 log_data_err("ulocdata_open en error %s", u_errorName(status));
1274 } else {
1275 ulocdata_getLocaleDisplayPattern(uld, pattern, kUBufMax, &status);
1276 if (U_FAILURE(status)){
1277 log_err("ulocdata_getLocaleDisplayPattern en error %s", u_errorName(status));
1278 } else if (u_strcmp(pattern, enExpectPat) != 0) {
1279 log_err("ulocdata_getLocaleDisplayPattern en returns unexpected pattern");
1280 }
1281 status = U_ZERO_ERROR;
1282 ulocdata_getLocaleSeparator(uld, separator, kUBufMax, &status);
1283 if (U_FAILURE(status)){
1284 log_err("ulocdata_getLocaleSeparator en error %s", u_errorName(status));
1285 } else if (u_strcmp(separator, enExpectSep) != 0) {
1286 log_err("ulocdata_getLocaleSeparator en returns unexpected string ");
1287 }
1288 ulocdata_close(uld);
1289 }
1290
1291 status = U_ZERO_ERROR;
1292 uld = ulocdata_open("zh", &status);
1293 if(U_FAILURE(status)){
1294 log_data_err("ulocdata_open zh error %s", u_errorName(status));
1295 } else {
1296 ulocdata_getLocaleDisplayPattern(uld, pattern, kUBufMax, &status);
1297 if (U_FAILURE(status)){
1298 log_err("ulocdata_getLocaleDisplayPattern zh error %s", u_errorName(status));
1299 } else if (u_strcmp(pattern, zhExpectPat) != 0) {
1300 log_err("ulocdata_getLocaleDisplayPattern zh returns unexpected pattern");
1301 }
1302 status = U_ZERO_ERROR;
1303 ulocdata_getLocaleSeparator(uld, separator, kUBufMax, &status);
1304 if (U_FAILURE(status)){
1305 log_err("ulocdata_getLocaleSeparator zh error %s", u_errorName(status));
1306 } else if (u_strcmp(separator, zhExpectSep) != 0) {
1307 log_err("ulocdata_getLocaleSeparator zh returns unexpected string ");
1308 }
1309 ulocdata_close(uld);
1310 }
1311 }
1312
TestCoverage(void)1313 static void TestCoverage(void){
1314 ULocaleDataDelimiterType types[] = {
1315 ULOCDATA_QUOTATION_START, /* Quotation start */
1316 ULOCDATA_QUOTATION_END, /* Quotation end */
1317 ULOCDATA_ALT_QUOTATION_START, /* Alternate quotation start */
1318 ULOCDATA_ALT_QUOTATION_END, /* Alternate quotation end */
1319 ULOCDATA_DELIMITER_COUNT
1320 };
1321 int i;
1322 UBool sub;
1323 UErrorCode status = U_ZERO_ERROR;
1324 ULocaleData *uld = ulocdata_open(uloc_getDefault(), &status);
1325
1326 if(U_FAILURE(status)){
1327 log_data_err("ulocdata_open error");
1328 return;
1329 }
1330
1331
1332 for(i = 0; i < ULOCDATA_DELIMITER_COUNT; i++){
1333 UChar result[32] = {0,};
1334 status = U_ZERO_ERROR;
1335 ulocdata_getDelimiter(uld, types[i], result, 32, &status);
1336 if (U_FAILURE(status)){
1337 log_err("ulocdata_getgetDelimiter error with type %d", types[i]);
1338 }
1339 }
1340
1341 sub = ulocdata_getNoSubstitute(uld);
1342 ulocdata_setNoSubstitute(uld,sub);
1343 ulocdata_close(uld);
1344 }
1345
TestIndexChars(void)1346 static void TestIndexChars(void) {
1347 /* Very basic test of ULOCDATA_ES_INDEX.
1348 * No comprehensive test of data, just basic check that the code path is alive.
1349 */
1350 UErrorCode status = U_ZERO_ERROR;
1351 ULocaleData *uld;
1352 USet *exemplarChars;
1353 USet *indexChars;
1354
1355 uld = ulocdata_open("en", &status);
1356 exemplarChars = uset_openEmpty();
1357 indexChars = uset_openEmpty();
1358 ulocdata_getExemplarSet(uld, exemplarChars, 0, ULOCDATA_ES_STANDARD, &status);
1359 ulocdata_getExemplarSet(uld, indexChars, 0, ULOCDATA_ES_INDEX, &status);
1360 if (U_FAILURE(status)) {
1361 log_data_err("File %s, line %d, Failure opening exemplar chars: %s", __FILE__, __LINE__, u_errorName(status));
1362 goto close_sets;
1363 }
1364 /* en data, standard exemplars are [a-z], lower case. */
1365 /* en data, index characters are [A-Z], upper case. */
1366 if ((uset_contains(exemplarChars, (UChar32)0x41) || uset_contains(indexChars, (UChar32)0x61))) {
1367 log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__, __LINE__ );
1368 goto close_sets;
1369 }
1370 if (!(uset_contains(exemplarChars, (UChar32)0x61) && uset_contains(indexChars, (UChar32)0x41) )) {
1371 log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__, __LINE__ );
1372 goto close_sets;
1373 }
1374
1375 close_sets:
1376 uset_close(exemplarChars);
1377 uset_close(indexChars);
1378 ulocdata_close(uld);
1379 }
1380
1381
1382
1383 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
TestCurrencyList(void)1384 static void TestCurrencyList(void){
1385 #if !UCONFIG_NO_FORMATTING
1386 UErrorCode errorCode = U_ZERO_ERROR;
1387 int32_t structLocaleCount, currencyCount;
1388 UEnumeration *en = ucurr_openISOCurrencies(UCURR_ALL, &errorCode);
1389 const char *isoCode, *structISOCode;
1390 UResourceBundle *subBundle;
1391 UResourceBundle *currencies = ures_openDirect(loadTestData(&errorCode), "structLocale", &errorCode);
1392 if(U_FAILURE(errorCode)) {
1393 log_data_err("Can't open structLocale\n");
1394 return;
1395 }
1396 currencies = ures_getByKey(currencies, "Currencies", currencies, &errorCode);
1397 currencyCount = uenum_count(en, &errorCode);
1398 structLocaleCount = ures_getSize(currencies);
1399 if (currencyCount != structLocaleCount) {
1400 log_err("structLocale(%d) and ISO4217(%d) currency list are out of sync.\n", structLocaleCount, currencyCount);
1401 #if U_CHARSET_FAMILY == U_ASCII_FAMILY
1402 ures_resetIterator(currencies);
1403 while ((isoCode = uenum_next(en, NULL, &errorCode)) != NULL && ures_hasNext(currencies)) {
1404 subBundle = ures_getNextResource(currencies, NULL, &errorCode);
1405 structISOCode = ures_getKey(subBundle);
1406 ures_close(subBundle);
1407 if (strcmp(structISOCode, isoCode) != 0) {
1408 log_err("First difference found at structLocale(%s) and ISO4217(%s).\n", structISOCode, isoCode);
1409 break;
1410 }
1411 }
1412 #endif
1413 }
1414 ures_close(currencies);
1415 uenum_close(en);
1416 #endif
1417 }
1418 #endif
1419
TestAvailableIsoCodes(void)1420 static void TestAvailableIsoCodes(void){
1421 #if !UCONFIG_NO_FORMATTING
1422 UErrorCode errorCode = U_ZERO_ERROR;
1423 const char* eurCode = "EUR";
1424 const char* usdCode = "USD";
1425 const char* lastCode = "RHD";
1426 const char* zzzCode = "ZZZ";
1427 UDate date1950 = (UDate)-630720000000.0;/* year 1950 */
1428 UDate date1970 = (UDate)0.0; /* year 1970 */
1429 UDate date1975 = (UDate)173448000000.0; /* year 1975 */
1430 UDate date1978 = (UDate)260172000000.0; /* year 1978 */
1431 UDate date1981 = (UDate)346896000000.0; /* year 1981 */
1432 UDate date1992 = (UDate)693792000000.0; /* year 1992 */
1433 UChar* isoCode = (UChar*)malloc(sizeof(UChar) * (uprv_strlen(usdCode) + 1));
1434
1435 /* testing available codes with no time ranges */
1436 u_charsToUChars(eurCode, isoCode, (int32_t)uprv_strlen(usdCode) + 1);
1437 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) {
1438 log_data_err("FAIL: ISO code (%s) is not found.\n", eurCode);
1439 }
1440
1441 u_charsToUChars(usdCode, isoCode, (int32_t)uprv_strlen(zzzCode) + 1);
1442 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) {
1443 log_data_err("FAIL: ISO code (%s) is not found.\n", usdCode);
1444 }
1445
1446 u_charsToUChars(zzzCode, isoCode, (int32_t)uprv_strlen(zzzCode) + 1);
1447 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == TRUE) {
1448 log_err("FAIL: ISO code (%s) is reported as available, but it doesn't exist.\n", zzzCode);
1449 }
1450
1451 u_charsToUChars(lastCode, isoCode, (int32_t)uprv_strlen(zzzCode) + 1);
1452 if (ucurr_isAvailable(isoCode, U_DATE_MIN, U_DATE_MAX, &errorCode) == FALSE) {
1453 log_data_err("FAIL: ISO code (%s) is not found.\n", lastCode);
1454 }
1455
1456 /* RHD was used from 1970-02-17 to 1980-04-18*/
1457
1458 /* to = null */
1459 if (ucurr_isAvailable(isoCode, date1970, U_DATE_MAX, &errorCode) == FALSE) {
1460 log_data_err("FAIL: ISO code (%s) was available in time range >1970-01-01.\n", lastCode);
1461 }
1462
1463 if (ucurr_isAvailable(isoCode, date1975, U_DATE_MAX, &errorCode) == FALSE) {
1464 log_data_err("FAIL: ISO code (%s) was available in time range >1975.\n", lastCode);
1465 }
1466
1467 if (ucurr_isAvailable(isoCode, date1981, U_DATE_MAX, &errorCode) == TRUE) {
1468 log_err("FAIL: ISO code (%s) was not available in time range >1981.\n", lastCode);
1469 }
1470
1471 /* from = null */
1472 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1970, &errorCode) == TRUE) {
1473 log_err("FAIL: ISO code (%s) was not available in time range <1970.\n", lastCode);
1474 }
1475
1476 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1975, &errorCode) == FALSE) {
1477 log_data_err("FAIL: ISO code (%s) was available in time range <1975.\n", lastCode);
1478 }
1479
1480 if (ucurr_isAvailable(isoCode, U_DATE_MIN, date1981, &errorCode) == FALSE) {
1481 log_data_err("FAIL: ISO code (%s) was available in time range <1981.\n", lastCode);
1482 }
1483
1484 /* full ranges */
1485 if (ucurr_isAvailable(isoCode, date1975, date1978, &errorCode) == FALSE) {
1486 log_data_err("FAIL: ISO code (%s) was available in time range 1975-1978.\n", lastCode);
1487 }
1488
1489 if (ucurr_isAvailable(isoCode, date1970, date1975, &errorCode) == FALSE) {
1490 log_data_err("FAIL: ISO code (%s) was available in time range 1970-1975.\n", lastCode);
1491 }
1492
1493 if (ucurr_isAvailable(isoCode, date1975, date1981, &errorCode) == FALSE) {
1494 log_data_err("FAIL: ISO code (%s) was available in time range 1975-1981.\n", lastCode);
1495 }
1496
1497 if (ucurr_isAvailable(isoCode, date1970, date1981, &errorCode) == FALSE) {
1498 log_data_err("FAIL: ISO code (%s) was available in time range 1970-1981.\n", lastCode);
1499 }
1500
1501 if (ucurr_isAvailable(isoCode, date1981, date1992, &errorCode) == TRUE) {
1502 log_err("FAIL: ISO code (%s) was not available in time range 1981-1992.\n", lastCode);
1503 }
1504
1505 if (ucurr_isAvailable(isoCode, date1950, date1970, &errorCode) == TRUE) {
1506 log_err("FAIL: ISO code (%s) was not available in time range 1950-1970.\n", lastCode);
1507 }
1508
1509 /* wrong range - from > to*/
1510 if (ucurr_isAvailable(isoCode, date1975, date1970, &errorCode) == TRUE) {
1511 log_err("FAIL: Wrong range 1975-1970 for ISO code (%s) was not reported.\n", lastCode);
1512 } else if (errorCode != U_ILLEGAL_ARGUMENT_ERROR) {
1513 log_data_err("FAIL: Error code not reported for wrong range 1975-1970 for ISO code (%s).\n", lastCode);
1514 }
1515
1516 free(isoCode);
1517 #endif
1518 }
1519
1520 #define TESTCASE(name) addTest(root, &name, "tsutil/cldrtest/" #name)
1521
1522 void addCLDRTest(TestNode** root);
1523
addCLDRTest(TestNode ** root)1524 void addCLDRTest(TestNode** root)
1525 {
1526 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
1527 TESTCASE(TestLocaleStructure);
1528 TESTCASE(TestCurrencyList);
1529 #endif
1530 TESTCASE(TestConsistentCountryInfo);
1531 TESTCASE(VerifyTranslation);
1532 TESTCASE(TestExemplarSet);
1533 TESTCASE(TestLocaleDisplayPattern);
1534 TESTCASE(TestCoverage);
1535 TESTCASE(TestIndexChars);
1536 TESTCASE(TestAvailableIsoCodes);
1537 }
1538
1539