1 // © 2016 and later: Unicode, Inc. and others.
2 // License & terms of use: http://www.unicode.org/copyright.html
3 /********************************************************************
5 * Copyright (c) 1997-2016, International Business Machines Corporation and
6 * others. All Rights Reserved.
7 ********************************************************************/
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"
24 returns a new UnicodeSet that is a flattened form of the original
28 createFlattenSet(USet
*origSet
, UErrorCode
*status
) {
32 int32_t origItemCount
= 0;
33 int32_t idx
, graphmeSize
;
36 if (U_FAILURE(*status
)) {
37 log_err("createFlattenSet called with %s\n", u_errorName(*status
));
40 newSet
= uset_open(1, 0);
41 origItemCount
= uset_getItemCount(origSet
);
42 for (idx
= 0; idx
< origItemCount
; idx
++) {
43 graphmeSize
= uset_getItem(origSet
, idx
,
45 graphme
, UPRV_LENGTHOF(graphme
),
47 if (U_FAILURE(*status
)) {
48 log_err("ERROR: uset_getItem returned %s\n", u_errorName(*status
));
49 *status
= U_ZERO_ERROR
;
52 uset_addAllCodePoints(newSet
, graphme
, graphmeSize
);
55 uset_addRange(newSet
, start
, end
);
58 uset_closeOver(newSet
,USET_CASE_INSENSITIVE
);
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){
75 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
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
;
82 ures_resetIterator(root
);
83 ures_resetIterator(currentBundle
);
84 while (ures_hasNext(currentBundle
)) {
85 const char *subBundleKey
= NULL
;
86 const char *currentBundleKey
= NULL
;
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
));
96 subBundleKey
= ures_getKey(subBundle
);
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",
103 ures_getKey(currentBundle
),
106 ures_close(subBundle
);
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",
113 ures_getKey(currentBundle
),
115 ures_getType(subRootBundle
),
116 ures_getType(subBundle
));
117 ures_close(subBundle
);
120 else if (ures_getType(subBundle
) == URES_INT_VECTOR
) {
122 int32_t subBundleSize
;
124 UBool sameArray
= TRUE
;
125 const int32_t *subRootBundleArr
= ures_getIntVector(subRootBundle
, &minSize
, &errorCode
);
126 const int32_t *subBundleArr
= ures_getIntVector(subBundle
, &subBundleSize
, &errorCode
);
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",
132 ures_getKey(currentBundle
),
136 for (idx
= 0; idx
< minSize
&& sameArray
; idx
++) {
137 if (subRootBundleArr
[idx
] != subBundleArr
[idx
]) {
140 if (strcmp(subBundleKey
, "DateTimeElements") == 0
141 && (subBundleArr
[idx
] < 1 || 7 < subBundleArr
[idx
]))
143 log_err("Value out of range with key \"%s\" at index %d in \"%s\" for locale \"%s\"\n",
146 ures_getKey(currentBundle
),
150 /* Special exception es_US and DateTimeElements */
152 && !(strcmp(locale
, "es_US") == 0 && strcmp(subBundleKey
, "DateTimeElements") == 0))
154 log_err("Integer vectors are the same with key \"%s\" in \"%s\" from root for locale \"%s\"\n",
156 ures_getKey(currentBundle
),
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
);
164 if (U_SUCCESS(errorCode
)
165 && (ures_getType(subSubBundle
) == URES_ARRAY
|| ures_getType(subSubRootBundle
) == URES_ARRAY
))
167 /* Here is one of the recursive parts */
168 TestKeyInRootRecursive(subRootBundle
, rootName
, subBundle
, locale
);
171 int32_t minSize
= ures_getSize(subRootBundle
);
173 UBool sameArray
= TRUE
;
175 if (minSize
> ures_getSize(subBundle
)) {
176 minSize
= ures_getSize(subBundle
);
179 if ((subBundleKey
== NULL
180 || (subBundleKey
!= NULL
&& strcmp(subBundleKey
, "LocaleScript") != 0 && !isCurrencyPreEuro(subBundleKey
)))
181 && ures_getSize(subRootBundle
) != ures_getSize(subBundle
))
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",
186 ures_getKey(currentBundle
),
188 ures_getSize(subRootBundle
),
189 ures_getSize(subBundle
));
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",
195 ures_getKey(currentBundle),
197 ures_getSize(subBundle));
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) {
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",
218 ures_getKey(currentBundle
),
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",
227 ures_getKey(currentBundle
),
233 localeStr
= ures_getStringByIndex(arr
,0,&localeStrLen
,&errorCode
);
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",
238 ures_getKey(currentBundle
),
244 log_err("Got a NULL string with key \"%s\" in \"%s\" at index %d for root or locale \"%s\"\n",
246 ures_getKey(currentBundle
),
252 if (localeStr
[0] == (UChar
)0x20) {
253 log_err("key \"%s\" at index %d in \"%s\" starts with a space in locale \"%s\"\n",
256 ures_getKey(currentBundle
),
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",
263 ures_getKey(currentBundle
),
266 else if (subBundleKey
!= NULL
267 && strcmp(subBundleKey
, "DateTimePatterns") == 0)
270 const UChar
*localeStrItr
= localeStr
;
271 while (*localeStrItr
) {
272 if (*localeStrItr
== (UChar
)0x27 /* ' */) {
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 */))
287 log_err("key \"%s\" at index %d has time pattern chars in date for locale \"%s\"\n",
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 */))
303 log_err("key \"%s\" at index %d has date pattern chars in time for locale \"%s\"\n",
312 else if (idx
== 4 && subBundleKey
!= NULL
313 && strcmp(subBundleKey
, "NumberElements") == 0
314 && u_charDigitValue(localeStr
[0]) != 0)
316 log_err("key \"%s\" at index %d has a non-zero based number for locale \"%s\"\n",
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",
326 ures_getKey(currentBundle),
330 ures_close(subSubBundle
);
331 ures_close(subSubRootBundle
);
333 else if (ures_getType(subBundle
) == URES_STRING
) {
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",
339 ures_getKey(currentBundle
),
341 } else if (string
[0] == (UChar
)0x20) {
342 log_err("key \"%s\" in \"%s\" starts with a space in locale \"%s\"\n",
344 ures_getKey(currentBundle
),
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",
350 ures_getKey(currentBundle
),
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)
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
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",
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",
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",
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. */
395 else if (strcmp(locale
, "root") != 0 && errorCode
== U_ZERO_ERROR
) {
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
));
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
),
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
);
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
);
426 log_verbose("Skipping key %s in %s\n", subBundleKey
, locale
);
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",
435 ures_getKey(currentBundle
),
438 /* Testing for MeasurementSystem is done in VerifyTranslation */
441 log_err("Type %d for key \"%s\" in \"%s\" is unknown for locale \"%s\"\n",
442 ures_getType(subBundle
),
444 ures_getKey(currentBundle
),
447 ures_close(subRootBundle
);
448 ures_close(subBundle
);
454 testLCID(UResourceBundle
*currentBundle
,
455 const char *localeName
)
457 (void)currentBundle
; // suppress compiler warnings about unused variables
458 UErrorCode status
= U_ZERO_ERROR
;
459 uint32_t expectedLCID
;
460 char lcidStringC
[64] = {0};
463 expectedLCID
= uloc_getLCID(localeName
);
464 if (expectedLCID
== 0) {
465 log_verbose("INFO: %-5s does not have any LCID mapping\n",
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
));
476 lcidStringC
[len
] = 0;
478 if(strcmp(localeName
, lcidStringC
) != 0) {
481 uloc_getLanguage(localeName
, langName
, sizeof(langName
), &status
);
482 uloc_getLanguage(lcidStringC
, langLCID
, sizeof(langLCID
), &status
);
484 if (strcmp(langName
, langLCID
) == 0) {
485 log_verbose("WARNING: %-5s resolves to %s (0x%.4x)\n",
486 localeName
, lcidStringC
, expectedLCID
);
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
);
495 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
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
504 UResourceBundle
*root
, *currentLocale
;
505 int32_t locCount
= uloc_countAvailable();
507 UErrorCode errorCode
= U_ZERO_ERROR
;
508 const char *currLoc
, *resolvedLoc
;
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];
514 for (locIndex = 0; locIndex < locCount; locIndex++) {
515 errorCode=U_ZERO_ERROR;
516 strcpy(locName, uloc_getAvailable(locIndex));
517 locNamePtr = strrchr(locName, '_');
522 strcpy(locName, "root");
525 root = ures_openDirect(NULL, locName, &errorCode);
526 if(U_FAILURE(errorCode)) {
527 log_err("Can't open %s\n", locName);
532 log_data_err("At least root needs to be installed\n");
535 root
= ures_openDirect(loadTestData(&errorCode
), "structLocale", &errorCode
);
536 if(U_FAILURE(errorCode
)) {
537 log_data_err("Can't open structLocale\n");
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
));
551 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
552 u_errorName(errorCode
),
553 uloc_getAvailable(locIndex
));
555 ures_close(currentLocale
);
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",
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",
567 resolvedLoc
= ures_getLocaleByType(currentLocale
, ULOC_ACTUAL_LOCALE
, &errorCode
);
568 if (strcmp(resolvedLoc
, currLoc
) != 0 && strcmp(currLoc
, "ars") != 0 && strcmp(currLoc
, "wuu") != 0 // /* ars,wuu are aliased locales */
569 && strcmp(currLoc
, "ur_Arab_IN") != 0 && strcmp(currLoc
, "ur_Aran_IN") != 0 /* so are ur_Ara?_IN <rdar://problem/47494884> */
570 && strcmp(currLoc
, "pa_Aran") != 0) { /* and pa_Aran <rdar://problem/51418203> */
571 /* All locales have at least a Version resource.
572 If it's absolutely empty, then the previous test will fail too.*/
573 log_err("Locale resolves to different locale. Is %s an alias of %s?\n",
574 currLoc
, resolvedLoc
);
576 TestKeyInRootRecursive(root
, "root", currentLocale
, currLoc
);
578 testLCID(currentLocale
, currLoc
);
580 ures_close(currentLocale
);
588 compareArrays(const char *keyName
,
589 UResourceBundle
*fromArray
, const char *fromLocale
,
590 UResourceBundle
*toArray
, const char *toLocale
,
591 int32_t start
, int32_t end
)
593 int32_t fromSize
= ures_getSize(fromArray
);
594 int32_t toSize
= ures_getSize(fromArray
);
596 UErrorCode errorCode
= U_ZERO_ERROR
;
598 if (fromSize
> toSize
) {
600 log_err("Arrays are different size from \"%s\" to \"%s\"\n",
605 for (idx
= start
; idx
<= end
; idx
++) {
606 const UChar
*fromBundleStr
= ures_getStringByIndex(fromArray
, idx
, NULL
, &errorCode
);
607 const UChar
*toBundleStr
= ures_getStringByIndex(toArray
, idx
, NULL
, &errorCode
);
608 if (fromBundleStr
&& toBundleStr
&& u_strcmp(fromBundleStr
, toBundleStr
) != 0)
610 log_err("Difference for %s at index %d from %s= \"%s\" to %s= \"%s\"\n",
614 austrdup(fromBundleStr
),
616 austrdup(toBundleStr
));
622 compareConsistentCountryInfo(const char *fromLocale
, const char *toLocale
) {
623 UErrorCode errorCode
= U_ZERO_ERROR
;
624 UResourceBundle
*fromArray
, *toArray
;
625 UResourceBundle
*fromLocaleBund
= ures_open(NULL
, fromLocale
, &errorCode
);
626 UResourceBundle
*toLocaleBund
= ures_open(NULL
, toLocale
, &errorCode
);
627 UResourceBundle
*toCalendar
, *fromCalendar
, *toGregorian
, *fromGregorian
;
629 if(U_FAILURE(errorCode
)) {
630 log_err("Can't open resource bundle %s or %s - %s\n", fromLocale
, toLocale
, u_errorName(errorCode
));
633 fromCalendar
= ures_getByKey(fromLocaleBund
, "calendar", NULL
, &errorCode
);
634 fromGregorian
= ures_getByKeyWithFallback(fromCalendar
, "gregorian", NULL
, &errorCode
);
636 toCalendar
= ures_getByKey(toLocaleBund
, "calendar", NULL
, &errorCode
);
637 toGregorian
= ures_getByKeyWithFallback(toCalendar
, "gregorian", NULL
, &errorCode
);
639 fromArray
= ures_getByKey(fromLocaleBund
, "CurrencyElements", NULL
, &errorCode
);
640 toArray
= ures_getByKey(toLocaleBund
, "CurrencyElements", NULL
, &errorCode
);
641 if (strcmp(fromLocale
, "en_CA") != 0)
643 /* The first one is probably localized. */
644 compareArrays("CurrencyElements", fromArray
, fromLocale
, toArray
, toLocale
, 1, 2);
646 ures_close(fromArray
);
649 fromArray
= ures_getByKey(fromLocaleBund
, "NumberPatterns", NULL
, &errorCode
);
650 toArray
= ures_getByKey(toLocaleBund
, "NumberPatterns", NULL
, &errorCode
);
651 if (strcmp(fromLocale
, "en_CA") != 0)
653 compareArrays("NumberPatterns", fromArray
, fromLocale
, toArray
, toLocale
, 0, 3);
655 ures_close(fromArray
);
658 /* Difficult to test properly */
660 fromArray = ures_getByKey(fromLocaleBund, "DateTimePatterns", NULL, &errorCode);
661 toArray = ures_getByKey(toLocaleBund, "DateTimePatterns", NULL, &errorCode);
663 compareArrays("DateTimePatterns", fromArray, fromLocale, toArray, toLocale);
665 ures_close(fromArray);
666 ures_close(toArray);*/
668 fromArray
= ures_getByKey(fromLocaleBund
, "NumberElements", NULL
, &errorCode
);
669 toArray
= ures_getByKey(toLocaleBund
, "NumberElements", NULL
, &errorCode
);
670 if (strcmp(fromLocale
, "en_CA") != 0)
672 compareArrays("NumberElements", fromArray
, fromLocale
, toArray
, toLocale
, 0, 3);
673 /* Index 4 is a script based 0 */
674 compareArrays("NumberElements", fromArray
, fromLocale
, toArray
, toLocale
, 5, 10);
676 ures_close(fromArray
);
678 ures_close(fromCalendar
);
679 ures_close(toCalendar
);
680 ures_close(fromGregorian
);
681 ures_close(toGregorian
);
683 ures_close(fromLocaleBund
);
684 ures_close(toLocaleBund
);
688 TestConsistentCountryInfo(void) {
689 /* UResourceBundle *fromLocale, *toLocale;*/
690 int32_t locCount
= uloc_countAvailable();
691 int32_t fromLocIndex
, toLocIndex
;
693 int32_t fromCountryLen
, toCountryLen
;
694 char fromCountry
[ULOC_FULLNAME_CAPACITY
], toCountry
[ULOC_FULLNAME_CAPACITY
];
696 int32_t fromVariantLen
, toVariantLen
;
697 char fromVariant
[ULOC_FULLNAME_CAPACITY
], toVariant
[ULOC_FULLNAME_CAPACITY
];
699 UErrorCode errorCode
= U_ZERO_ERROR
;
701 for (fromLocIndex
= 0; fromLocIndex
< locCount
; fromLocIndex
++) {
702 const char *fromLocale
= uloc_getAvailable(fromLocIndex
);
704 errorCode
=U_ZERO_ERROR
;
705 fromCountryLen
= uloc_getCountry(fromLocale
, fromCountry
, ULOC_FULLNAME_CAPACITY
, &errorCode
);
706 if (fromCountryLen
<= 0) {
707 /* Ignore countryless locales */
710 fromVariantLen
= uloc_getVariant(fromLocale
, fromVariant
, ULOC_FULLNAME_CAPACITY
, &errorCode
);
711 if (fromVariantLen
> 0) {
712 /* Most variants are ignorable like collation variants. */
715 /* Start comparing only after the current index.
716 Previous loop should have already compared fromLocIndex.
718 for (toLocIndex
= fromLocIndex
+ 1; toLocIndex
< locCount
; toLocIndex
++) {
719 const char *toLocale
= uloc_getAvailable(toLocIndex
);
721 toCountryLen
= uloc_getCountry(toLocale
, toCountry
, ULOC_FULLNAME_CAPACITY
, &errorCode
);
722 if(U_FAILURE(errorCode
)) {
723 log_err("Unknown failure fromLocale=%s toLocale=%s errorCode=%s\n",
724 fromLocale
, toLocale
, u_errorName(errorCode
));
728 if (toCountryLen
<= 0) {
729 /* Ignore countryless locales */
732 toVariantLen
= uloc_getVariant(toLocale
, toVariant
, ULOC_FULLNAME_CAPACITY
, &errorCode
);
733 if (toVariantLen
> 0) {
734 /* Most variants are ignorable like collation variants. */
735 /* They're a variant for a reason. */
738 if (strcmp(fromCountry
, toCountry
) == 0) {
739 log_verbose("comparing fromLocale=%s toLocale=%s\n",
740 fromLocale
, toLocale
);
741 compareConsistentCountryInfo(fromLocale
, toLocale
);
748 findStringSetMismatch(const char *currLoc
, const UChar
*string
, int32_t langSize
,
749 USet
* mergedExemplarSet
,
750 UBool ignoreNumbers
, UChar32
* badCharPtr
) {
751 UErrorCode errorCode
= U_ZERO_ERROR
;
754 if (mergedExemplarSet
== NULL
) {
757 exemplarSet
= createFlattenSet(mergedExemplarSet
, &errorCode
);
758 if (U_FAILURE(errorCode
)) {
759 log_err("%s: error createFlattenSet returned %s\n", currLoc
, u_errorName(errorCode
));
763 for (strIdx
= 0; strIdx
< langSize
;) {
765 U16_NEXT(string
, strIdx
, langSize
, testChar
);
766 if (!uset_contains(exemplarSet
, testChar
)
767 && testChar
!= 0x0020 && testChar
!= 0x00A0 && testChar
!= 0x002e && testChar
!= 0x002c && testChar
!= 0x002d && testChar
!= 0x0027
768 && testChar
!= 0x005B && testChar
!= 0x005D && testChar
!= 0x2019 && testChar
!= 0x0f0b && testChar
!= 0x200C && testChar
!= 0x200D) {
769 if (!ignoreNumbers
|| (ignoreNumbers
&& (testChar
< 0x30 || testChar
> 0x39))) {
770 uset_close(exemplarSet
);
772 *badCharPtr
= testChar
;
778 uset_close(exemplarSet
);
784 /* include non-invariant chars */
786 myUCharsToChars(const UChar
* us
, char* cs
, int32_t len
){
798 findSetMatch( UScriptCode
*scriptCodes
, int32_t scriptsLen
,
801 USet
*scripts
[10]= {0};
802 char pattern
[256] = { '[', ':', 0x000 };
804 UChar uPattern
[256] = {0};
805 UErrorCode status
= U_ZERO_ERROR
;
808 /* create the sets with script codes */
809 for(i
= 0; i
<scriptsLen
; i
++){
810 strcat(pattern
, uscript_getShortName(scriptCodes
[i
]));
811 strcat(pattern
, ":]");
812 patternLen
= (int32_t)strlen(pattern
);
813 u_charsToUChars(pattern
, uPattern
, patternLen
);
814 scripts
[i
] = uset_openPattern(uPattern
, patternLen
, &status
);
815 if(U_FAILURE(status
)){
816 log_err("Could not create set for pattern %s. Error: %s\n", pattern
, u_errorName(status
));
821 if (strcmp(locale
, "uk") == 0 || strcmp(locale
, "uk_UA") == 0) {
822 /* Special addition. Add the modifying apostrophe, which isn't in Cyrillic. */
823 uset_add(scripts
[0], 0x2bc);
825 if(U_SUCCESS(status
)){
826 UBool existsInScript
= FALSE
;
827 /* iterate over the exemplarSet and ascertain if all
828 * UChars in exemplarSet belong to the scripts returned
831 int32_t count
= uset_getItemCount(exemplarSet
);
833 for( i
=0; i
< count
; i
++){
837 int32_t strCapacity
= 0;
839 strCapacity
= uset_getItem(exemplarSet
, i
, &start
, &end
, str
, strCapacity
, &status
);
840 if(U_SUCCESS(status
)){
842 if(strCapacity
== 0){
843 /* ok the item is a range */
844 for( j
= 0; j
< scriptsLen
; j
++){
845 if(uset_containsRange(scripts
[j
], start
, end
) == TRUE
){
846 existsInScript
= TRUE
;
849 if(existsInScript
== FALSE
){
850 for( j
= 0; j
< scriptsLen
; j
++){
851 UChar toPattern
[500]={'\0'};
852 char pat
[500]={'\0'};
853 int32_t len
= uset_toPattern(scripts
[j
], toPattern
, 500, TRUE
, &status
);
854 len
= myUCharsToChars(toPattern
, pat
, len
);
855 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
));
857 log_err("Pattern: %s\n",pat
);
860 log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale
);
863 strCapacity
++; /* increment for NUL termination */
864 /* allocate the str and call the api again */
865 str
= (UChar
*) malloc(U_SIZEOF_UCHAR
* strCapacity
);
866 strCapacity
= uset_getItem(exemplarSet
, i
, &start
, &end
, str
, strCapacity
, &status
);
867 /* iterate over the scripts and figure out if the string contained is actually
870 for( j
= 0; j
< scriptsLen
; j
++){
871 if(uset_containsString(scripts
[j
],str
, strCapacity
) == TRUE
){
872 existsInScript
= TRUE
;
875 if(existsInScript
== FALSE
){
876 log_err("ExemplarCharacters and LocaleScript containment test failed for locale %s. \n", locale
);
885 for(i
= 0; i
<scriptsLen
; i
++){
886 uset_close(scripts
[i
]);
890 static void VerifyTranslation(void) {
891 UResourceBundle
*root
, *currentLocale
;
892 int32_t locCount
= uloc_countAvailable();
894 UErrorCode errorCode
= U_ZERO_ERROR
;
896 UScriptCode scripts
[USCRIPT_CODE_LIMIT
];
900 UResourceBundle
*resArray
;
903 log_data_err("At least root needs to be installed\n");
906 root
= ures_openDirect(NULL
, "root", &errorCode
);
907 if(U_FAILURE(errorCode
)) {
908 log_data_err("Can't open root\n");
911 for (locIndex
= 0; locIndex
< locCount
; locIndex
++) {
912 USet
* mergedExemplarSet
= NULL
;
913 errorCode
=U_ZERO_ERROR
;
914 currLoc
= uloc_getAvailable(locIndex
);
915 currentLocale
= ures_open(NULL
, currLoc
, &errorCode
);
916 if(errorCode
!= U_ZERO_ERROR
) {
917 if(U_SUCCESS(errorCode
)) {
918 /* It's installed, but there is no data.
919 It's installed for the g18n white paper [grhoten] */
920 log_err("ERROR: Locale %-5s not installed, and it should be!\n",
921 uloc_getAvailable(locIndex
));
923 log_err("%%%%%%% Unexpected error %d in %s %%%%%%%",
924 u_errorName(errorCode
),
925 uloc_getAvailable(locIndex
));
927 ures_close(currentLocale
);
931 UErrorCode exemplarStatus
= U_ZERO_ERROR
;
932 ULocaleData
* uld
= ulocdata_open(currLoc
, &exemplarStatus
);
933 if (U_SUCCESS(exemplarStatus
)) {
934 USet
* exemplarSet
= ulocdata_getExemplarSet(uld
, NULL
, USET_ADD_CASE_MAPPINGS
, ULOCDATA_ES_STANDARD
, &exemplarStatus
);
935 if (U_SUCCESS(exemplarStatus
)) {
936 mergedExemplarSet
= uset_cloneAsThawed(exemplarSet
);
937 uset_close(exemplarSet
);
938 exemplarSet
= ulocdata_getExemplarSet(uld
, NULL
, USET_ADD_CASE_MAPPINGS
, ULOCDATA_ES_AUXILIARY
, &exemplarStatus
);
939 if (U_SUCCESS(exemplarStatus
)) {
940 uset_addAll(mergedExemplarSet
, exemplarSet
);
941 uset_close(exemplarSet
);
943 exemplarStatus
= U_ZERO_ERROR
;
944 exemplarSet
= ulocdata_getExemplarSet(uld
, NULL
, 0, ULOCDATA_ES_PUNCTUATION
, &exemplarStatus
);
945 if (U_SUCCESS(exemplarStatus
)) {
946 uset_addAll(mergedExemplarSet
, exemplarSet
);
947 uset_close(exemplarSet
);
950 log_err("error ulocdata_getExemplarSet (main) for locale %s returned %s\n", currLoc
, u_errorName(errorCode
));
954 log_err("error ulocdata_open for locale %s returned %s\n", currLoc
, u_errorName(errorCode
));
957 if (mergedExemplarSet
== NULL
/*|| (getTestOption(QUICK_OPTION) && uset_size() > 2048)*/) {
958 log_verbose("skipping test for %s\n", currLoc
);
960 //else if (uprv_strncmp(currLoc,"bem",3) == 0 || uprv_strncmp(currLoc,"mgo",3) == 0 || uprv_strncmp(currLoc,"nl",2) == 0) {
961 // log_verbose("skipping test for %s, some month and country names known to use aux exemplars\n", currLoc);
964 UChar langBuffer
[128];
968 langSize
= uloc_getDisplayLanguage(currLoc
, currLoc
, langBuffer
, UPRV_LENGTHOF(langBuffer
), &errorCode
);
969 if (U_FAILURE(errorCode
)) {
970 log_err("error uloc_getDisplayLanguage returned %s\n", u_errorName(errorCode
));
971 } else if (uprv_strncmp(currLoc
,"gez",3) == 0 || uprv_strcmp(currLoc
,"sa") == 0 || uprv_strncmp(currLoc
,"sa_",3) == 0) { // Apple xtra locales
972 log_verbose("skipping DisplayLanguage test for %s, name or exemplars may need adjustment\n", currLoc
);
974 strIdx
= findStringSetMismatch(currLoc
, langBuffer
, langSize
, mergedExemplarSet
, FALSE
, &badChar
);
977 log_err("getDisplayLanguage(%s) at index %d returned characters not in the exemplar characters: %04X in \"%s\"\n",
978 currLoc
, strIdx
, badChar
, u_austrncpy(bbuf
,langBuffer
,langSize
));
981 langSize
= uloc_getDisplayCountry(currLoc
, currLoc
, langBuffer
, UPRV_LENGTHOF(langBuffer
), &errorCode
);
982 if (U_FAILURE(errorCode
)) {
983 log_err("error uloc_getDisplayCountry returned %s\n", u_errorName(errorCode
));
985 if ( uprv_strcmp(currLoc
,"ba") == 0 || uprv_strncmp(currLoc
,"ba_",3) == 0 ||
986 uprv_strcmp(currLoc
,"cv") == 0 || uprv_strncmp(currLoc
,"cv_",3) == 0 ||
987 uprv_strcmp(currLoc
,"dv") == 0 || uprv_strncmp(currLoc
,"dv_",3) == 0 ||
988 uprv_strcmp(currLoc
,"sa") == 0 || uprv_strncmp(currLoc
,"sa_",3) == 0 ||
989 uprv_strncmp(currLoc
,"kaj",3) == 0 || uprv_strncmp(currLoc
,"kpe",3) == 0 ||
990 uprv_strncmp(currLoc
,"nqo",3) == 0 || uprv_strncmp(currLoc
,"sat",3) == 0 ||
991 uprv_strncmp(currLoc
,"syr",3) == 0 ||
992 uprv_strncmp(currLoc
,"ks_Deva",7) == 0 || uprv_strncmp(currLoc
,"mni_Mtei",8) == 0 ||
993 uprv_strncmp(currLoc
,"sd_Deva",7) == 0 ) {
994 log_verbose("skipping day/month tests for %s, missing some translated names\n", currLoc
);
996 UResourceBundle
* cal
= ures_getByKey(currentLocale
, "calendar", NULL
, &errorCode
);
997 UResourceBundle
* greg
= ures_getByKeyWithFallback(cal
, "gregorian", NULL
, &errorCode
);
998 UResourceBundle
* names
= ures_getByKeyWithFallback(greg
, "dayNames", NULL
, &errorCode
);
999 UResourceBundle
* format
= ures_getByKeyWithFallback(names
, "format", NULL
, &errorCode
);
1000 resArray
= ures_getByKeyWithFallback(format
, "wide", NULL
, &errorCode
);
1002 if (U_FAILURE(errorCode
)) {
1003 log_err("error ures_getByKey returned %s\n", u_errorName(errorCode
));
1005 if (getTestOption(QUICK_OPTION
)) {
1009 end
= ures_getSize(resArray
);
1012 if ((uprv_strncmp(currLoc
,"lrc",3) == 0 || uprv_strncmp(currLoc
,"mzn",3) == 0) &&
1013 log_knownIssue("cldrbug:8899", "lrc and mzn locales don't have translated day names")) {
1017 for (idx
= 0; idx
< end
; idx
++) {
1018 const UChar
*fromBundleStr
= ures_getStringByIndex(resArray
, idx
, &langSize
, &errorCode
);
1019 if (U_FAILURE(errorCode
)) {
1020 log_err("error ures_getStringByIndex(%d) returned %s\n", idx
, u_errorName(errorCode
));
1023 strIdx
= findStringSetMismatch(currLoc
, fromBundleStr
, langSize
, mergedExemplarSet
, TRUE
, &badChar
);
1024 if ( strIdx
>= 0 ) {
1025 log_err("getDayNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n",
1026 currLoc
, idx
, strIdx
, badChar
);
1029 ures_close(resArray
);
1033 names
= ures_getByKeyWithFallback(greg
, "monthNames", NULL
, &errorCode
);
1034 format
= ures_getByKeyWithFallback(names
,"format", NULL
, &errorCode
);
1035 resArray
= ures_getByKeyWithFallback(format
, "wide", NULL
, &errorCode
);
1036 if (U_FAILURE(errorCode
)) {
1037 log_err("error ures_getByKey returned %s\n", u_errorName(errorCode
));
1039 if (getTestOption(QUICK_OPTION
)) {
1043 end
= ures_getSize(resArray
);
1046 for (idx
= 0; idx
< end
; idx
++) {
1047 const UChar
*fromBundleStr
= ures_getStringByIndex(resArray
, idx
, &langSize
, &errorCode
);
1048 if (U_FAILURE(errorCode
)) {
1049 log_err("error ures_getStringByIndex(%d) returned %s\n", idx
, u_errorName(errorCode
));
1052 strIdx
= findStringSetMismatch(currLoc
, fromBundleStr
, langSize
, mergedExemplarSet
, TRUE
, &badChar
);
1054 log_err("getMonthNames(%s, %d) at index %d returned characters not in the exemplar characters: %04X.\n",
1055 currLoc
, idx
, strIdx
, badChar
);
1058 ures_close(resArray
);
1064 errorCode
= U_ZERO_ERROR
;
1065 numScripts
= uscript_getCode(currLoc
, scripts
, UPRV_LENGTHOF(scripts
), &errorCode
);
1066 if (strcmp(currLoc
, "yi") == 0 && numScripts
> 0 && log_knownIssue("11217", "Fix result of uscript_getCode for yi: USCRIPT_YI -> USCRIPT_HEBREW")) {
1067 scripts
[0] = USCRIPT_HEBREW
;
1069 if (numScripts
== 0) {
1070 log_err("uscript_getCode(%s) doesn't work.\n", currLoc
);
1071 }else if(scripts
[0] == USCRIPT_COMMON
){
1072 log_err("uscript_getCode(%s) returned USCRIPT_COMMON.\n", currLoc
);
1075 /* test that the scripts are a superset of exemplar characters. */
1077 ULocaleData
*uld
= ulocdata_open(currLoc
,&errorCode
);
1078 USet
*exemplarSet
= ulocdata_getExemplarSet(uld
, NULL
, 0, ULOCDATA_ES_STANDARD
, &errorCode
);
1079 /* test if exemplar characters are part of script code */
1080 findSetMatch(scripts
, numScripts
, exemplarSet
, currLoc
);
1081 uset_close(exemplarSet
);
1082 ulocdata_close(uld
);
1085 /* test that the paperSize API works */
1087 int32_t height
=0, width
=0;
1088 ulocdata_getPaperSize(currLoc
, &height
, &width
, &errorCode
);
1089 if(U_FAILURE(errorCode
)){
1090 log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc
, u_errorName(errorCode
));
1092 if(strstr(currLoc
, "_US")!=NULL
&& height
!= 279 && width
!= 216 ){
1093 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc
);
1096 /* test that the MeasurementSystem API works */
1098 char fullLoc
[ULOC_FULLNAME_CAPACITY
];
1099 UMeasurementSystem measurementSystem
;
1100 int32_t height
= 0, width
= 0;
1102 uloc_addLikelySubtags(currLoc
, fullLoc
, ULOC_FULLNAME_CAPACITY
, &errorCode
);
1104 errorCode
= U_ZERO_ERROR
;
1105 measurementSystem
= ulocdata_getMeasurementSystem(currLoc
, &errorCode
);
1106 if (U_FAILURE(errorCode
)) {
1107 log_err("ulocdata_getMeasurementSystem failed for locale %s with error: %s \n", currLoc
, u_errorName(errorCode
));
1109 if ( strstr(fullLoc
, "_US")!=NULL
|| strstr(fullLoc
, "_MM")!=NULL
|| strstr(fullLoc
, "_LR")!=NULL
) {
1110 if(measurementSystem
!= UMS_US
){
1111 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc
);
1113 } else if ( strstr(fullLoc
, "_GB")!=NULL
) {
1114 if(measurementSystem
!= UMS_UK
){
1115 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc
);
1117 } else if (measurementSystem
!= UMS_SI
) {
1118 log_err("ulocdata_getMeasurementSystem did not return expected data for locale %s \n", currLoc
);
1122 errorCode
= U_ZERO_ERROR
;
1123 ulocdata_getPaperSize(currLoc
, &height
, &width
, &errorCode
);
1124 if (U_FAILURE(errorCode
)) {
1125 log_err("ulocdata_getPaperSize failed for locale %s with error: %s \n", currLoc
, u_errorName(errorCode
));
1127 if ( strstr(fullLoc
, "_US")!=NULL
|| strstr(fullLoc
, "_BZ")!=NULL
|| strstr(fullLoc
, "_CA")!=NULL
|| strstr(fullLoc
, "_CL")!=NULL
||
1128 strstr(fullLoc
, "_CO")!=NULL
|| strstr(fullLoc
, "_CR")!=NULL
|| strstr(fullLoc
, "_GT")!=NULL
|| strstr(fullLoc
, "_MX")!=NULL
||
1129 strstr(fullLoc
, "_NI")!=NULL
|| strstr(fullLoc
, "_PA")!=NULL
|| strstr(fullLoc
, "_PH")!=NULL
|| strstr(fullLoc
, "_PR")!=NULL
||
1130 strstr(fullLoc
, "_SV")!=NULL
|| strstr(fullLoc
, "_VE")!=NULL
) {
1131 if (height
!= 279 || width
!= 216) {
1132 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc
);
1134 } else if (height
!= 297 || width
!= 210) {
1135 log_err("ulocdata_getPaperSize did not return expected data for locale %s \n", currLoc
);
1140 if (mergedExemplarSet
!= NULL
) {
1141 uset_close(mergedExemplarSet
);
1143 ures_close(currentLocale
);
1149 /* adjust this limit as appropriate */
1150 #define MAX_SCRIPTS_PER_LOCALE 8
1152 static void TestExemplarSet(void){
1153 int32_t i
, j
, k
, m
, n
;
1154 int32_t equalCount
= 0;
1155 UErrorCode ec
= U_ZERO_ERROR
;
1156 UEnumeration
* avail
;
1157 USet
* exemplarSets
[2];
1158 USet
* unassignedSet
;
1159 UScriptCode code
[MAX_SCRIPTS_PER_LOCALE
];
1160 USet
* codeSets
[MAX_SCRIPTS_PER_LOCALE
];
1162 char cbuf
[32]; /* 9 should be enough */
1163 UChar ubuf
[64]; /* adjust as needed */
1164 UBool existsInScript
;
1169 unassignedSet
= NULL
;
1170 exemplarSets
[0] = NULL
;
1171 exemplarSets
[1] = NULL
;
1172 for (i
=0; i
<MAX_SCRIPTS_PER_LOCALE
; ++i
) {
1176 avail
= ures_openAvailableLocales(NULL
, &ec
);
1177 if (!assertSuccess("ures_openAvailableLocales", &ec
)) goto END
;
1178 n
= uenum_count(avail
, &ec
);
1179 if (!assertSuccess("uenum_count", &ec
)) goto END
;
1181 u_uastrcpy(ubuf
, "[:unassigned:]");
1182 unassignedSet
= uset_openPattern(ubuf
, -1, &ec
);
1183 if (!assertSuccess("uset_openPattern", &ec
)) goto END
;
1186 const char* locale
= uenum_next(avail
, NULL
, &ec
);
1187 if (!assertSuccess("uenum_next", &ec
)) goto END
;
1188 log_verbose("%s\n", locale
);
1189 for (k
=0; k
<2; ++k
) {
1190 uint32_t option
= (k
==0) ? 0 : USET_CASE_INSENSITIVE
;
1191 ULocaleData
*uld
= ulocdata_open(locale
,&ec
);
1192 USet
* exemplarSet
= ulocdata_getExemplarSet(uld
,NULL
, option
, ULOCDATA_ES_STANDARD
, &ec
);
1193 uset_close(exemplarSets
[k
]);
1194 ulocdata_close(uld
);
1195 exemplarSets
[k
] = exemplarSet
;
1196 if (!assertSuccess("ulocaledata_getExemplarSet", &ec
)) goto END
;
1198 if (uset_containsSome(exemplarSet
, unassignedSet
)) {
1199 log_err("ExemplarSet contains unassigned characters for locale : %s\n", locale
);
1201 codeLen
= uscript_getCode(locale
, code
, 8, &ec
);
1202 if (strcmp(locale
, "yi") == 0 && codeLen
> 0 && log_knownIssue("11217", "Fix result of uscript_getCode for yi: USCRIPT_YI -> USCRIPT_HEBREW")) {
1203 code
[0] = USCRIPT_HEBREW
;
1205 if (!assertSuccess("uscript_getCode", &ec
)) goto END
;
1207 for (j
=0; j
<MAX_SCRIPTS_PER_LOCALE
; ++j
) {
1208 uset_close(codeSets
[j
]);
1211 for (j
=0; j
<codeLen
; ++j
) {
1212 uprv_strcpy(cbuf
, "[:");
1214 log_err("USCRIPT_INVALID_CODE returned for locale: %s\n", locale
);
1217 uprv_strcat(cbuf
, uscript_getShortName(code
[j
]));
1218 uprv_strcat(cbuf
, ":]");
1219 u_uastrcpy(ubuf
, cbuf
);
1220 codeSets
[j
] = uset_openPattern(ubuf
, -1, &ec
);
1222 if (!assertSuccess("uset_openPattern", &ec
)) goto END
;
1224 existsInScript
= FALSE
;
1225 itemCount
= uset_getItemCount(exemplarSet
);
1226 for (m
=0; m
<itemCount
&& !existsInScript
; ++m
) {
1227 strLen
= uset_getItem(exemplarSet
, m
, &start
, &end
, ubuf
,
1228 UPRV_LENGTHOF(ubuf
), &ec
);
1229 /* failure here might mean str[] needs to be larger */
1230 if (!assertSuccess("uset_getItem", &ec
)) goto END
;
1232 for (j
=0; j
<codeLen
; ++j
) {
1233 if (codeSets
[j
]!=NULL
&& uset_containsRange(codeSets
[j
], start
, end
)) {
1234 existsInScript
= TRUE
;
1239 for (j
=0; j
<codeLen
; ++j
) {
1240 if (codeSets
[j
]!=NULL
&& uset_containsString(codeSets
[j
], ubuf
, strLen
)) {
1241 existsInScript
= TRUE
;
1248 if (existsInScript
== FALSE
){
1249 log_err("ExemplarSet containment failed for locale : %s\n", locale
);
1252 assertTrue("case-folded is a superset",
1253 uset_containsAll(exemplarSets
[1], exemplarSets
[0]));
1254 if (uset_equals(exemplarSets
[1], exemplarSets
[0])) {
1258 /* Note: The case-folded set should sometimes be a strict superset
1259 and sometimes be equal. */
1260 assertTrue("case-folded is sometimes a strict superset, and sometimes equal",
1261 equalCount
> 0 && equalCount
< n
);
1265 uset_close(exemplarSets
[0]);
1266 uset_close(exemplarSets
[1]);
1267 uset_close(unassignedSet
);
1268 for (i
=0; i
<MAX_SCRIPTS_PER_LOCALE
; ++i
) {
1269 uset_close(codeSets
[i
]);
1273 enum { kUBufMax
= 32 };
1274 static void TestLocaleDisplayPattern(void){
1276 UChar pattern
[kUBufMax
] = {0,};
1277 UChar separator
[kUBufMax
] = {0,};
1279 static const UChar enExpectPat
[] = { 0x007B,0x0030,0x007D,0x0020,0x0028,0x007B,0x0031,0x007D,0x0029,0 }; /* "{0} ({1})" */
1280 static const UChar enExpectSep
[] = { 0x002C,0x0020,0 }; /* ", " */
1281 static const UChar zhExpectPat
[] = { 0x007B,0x0030,0x007D,0xFF08,0x007B,0x0031,0x007D,0xFF09,0 };
1282 static const UChar zhExpectSep
[] = { 0xFF0C,0 };
1284 status
= U_ZERO_ERROR
;
1285 uld
= ulocdata_open("en", &status
);
1286 if(U_FAILURE(status
)){
1287 log_data_err("ulocdata_open en error %s", u_errorName(status
));
1289 ulocdata_getLocaleDisplayPattern(uld
, pattern
, kUBufMax
, &status
);
1290 if (U_FAILURE(status
)){
1291 log_err("ulocdata_getLocaleDisplayPattern en error %s", u_errorName(status
));
1292 } else if (u_strcmp(pattern
, enExpectPat
) != 0) {
1293 log_err("ulocdata_getLocaleDisplayPattern en returns unexpected pattern");
1295 status
= U_ZERO_ERROR
;
1296 ulocdata_getLocaleSeparator(uld
, separator
, kUBufMax
, &status
);
1297 if (U_FAILURE(status
)){
1298 log_err("ulocdata_getLocaleSeparator en error %s", u_errorName(status
));
1299 } else if (u_strcmp(separator
, enExpectSep
) != 0) {
1300 log_err("ulocdata_getLocaleSeparator en returns unexpected string ");
1302 ulocdata_close(uld
);
1305 status
= U_ZERO_ERROR
;
1306 uld
= ulocdata_open("zh", &status
);
1307 if(U_FAILURE(status
)){
1308 log_data_err("ulocdata_open zh error %s", u_errorName(status
));
1310 ulocdata_getLocaleDisplayPattern(uld
, pattern
, kUBufMax
, &status
);
1311 if (U_FAILURE(status
)){
1312 log_err("ulocdata_getLocaleDisplayPattern zh error %s", u_errorName(status
));
1313 } else if (u_strcmp(pattern
, zhExpectPat
) != 0) {
1314 log_err("ulocdata_getLocaleDisplayPattern zh returns unexpected pattern");
1316 status
= U_ZERO_ERROR
;
1317 ulocdata_getLocaleSeparator(uld
, separator
, kUBufMax
, &status
);
1318 if (U_FAILURE(status
)){
1319 log_err("ulocdata_getLocaleSeparator zh error %s", u_errorName(status
));
1320 } else if (u_strcmp(separator
, zhExpectSep
) != 0) {
1321 log_err("ulocdata_getLocaleSeparator zh returns unexpected string ");
1323 ulocdata_close(uld
);
1327 static void TestCoverage(void){
1328 ULocaleDataDelimiterType types
[] = {
1329 ULOCDATA_QUOTATION_START
, /* Quotation start */
1330 ULOCDATA_QUOTATION_END
, /* Quotation end */
1331 ULOCDATA_ALT_QUOTATION_START
, /* Alternate quotation start */
1332 ULOCDATA_ALT_QUOTATION_END
, /* Alternate quotation end */
1333 ULOCDATA_DELIMITER_COUNT
1337 UErrorCode status
= U_ZERO_ERROR
;
1338 ULocaleData
*uld
= ulocdata_open(uloc_getDefault(), &status
);
1340 if(U_FAILURE(status
)){
1341 log_data_err("ulocdata_open error");
1346 for(i
= 0; i
< ULOCDATA_DELIMITER_COUNT
; i
++){
1347 UChar result
[32] = {0,};
1348 status
= U_ZERO_ERROR
;
1349 ulocdata_getDelimiter(uld
, types
[i
], result
, 32, &status
);
1350 if (U_FAILURE(status
)){
1351 log_err("ulocdata_getgetDelimiter error with type %d", types
[i
]);
1355 sub
= ulocdata_getNoSubstitute(uld
);
1356 ulocdata_setNoSubstitute(uld
,sub
);
1357 ulocdata_close(uld
);
1360 static void TestIndexChars(void) {
1361 /* Very basic test of ULOCDATA_ES_INDEX.
1362 * No comprehensive test of data, just basic check that the code path is alive.
1364 UErrorCode status
= U_ZERO_ERROR
;
1366 USet
*exemplarChars
;
1369 uld
= ulocdata_open("en", &status
);
1370 exemplarChars
= uset_openEmpty();
1371 indexChars
= uset_openEmpty();
1372 ulocdata_getExemplarSet(uld
, exemplarChars
, 0, ULOCDATA_ES_STANDARD
, &status
);
1373 ulocdata_getExemplarSet(uld
, indexChars
, 0, ULOCDATA_ES_INDEX
, &status
);
1374 if (U_FAILURE(status
)) {
1375 log_data_err("File %s, line %d, Failure opening exemplar chars: %s", __FILE__
, __LINE__
, u_errorName(status
));
1378 /* en data, standard exemplars are [a-z], lower case. */
1379 /* en data, index characters are [A-Z], upper case. */
1380 if ((uset_contains(exemplarChars
, (UChar32
)0x41) || uset_contains(indexChars
, (UChar32
)0x61))) {
1381 log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__
, __LINE__
);
1384 if (!(uset_contains(exemplarChars
, (UChar32
)0x61) && uset_contains(indexChars
, (UChar32
)0x41) )) {
1385 log_err("File %s, line %d, Exemplar characters incorrect.", __FILE__
, __LINE__
);
1390 uset_close(exemplarChars
);
1391 uset_close(indexChars
);
1392 ulocdata_close(uld
);
1397 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
1398 static void TestCurrencyList(void){
1399 #if !UCONFIG_NO_FORMATTING
1400 UErrorCode errorCode
= U_ZERO_ERROR
;
1401 int32_t structLocaleCount
, currencyCount
;
1402 UEnumeration
*en
= ucurr_openISOCurrencies(UCURR_ALL
, &errorCode
);
1403 const char *isoCode
, *structISOCode
;
1404 UResourceBundle
*subBundle
;
1405 UResourceBundle
*currencies
= ures_openDirect(loadTestData(&errorCode
), "structLocale", &errorCode
);
1406 if(U_FAILURE(errorCode
)) {
1407 log_data_err("Can't open structLocale\n");
1410 currencies
= ures_getByKey(currencies
, "Currencies", currencies
, &errorCode
);
1411 currencyCount
= uenum_count(en
, &errorCode
);
1412 structLocaleCount
= ures_getSize(currencies
);
1413 if (currencyCount
!= structLocaleCount
) {
1414 log_err("structLocale(%d) and ISO4217(%d) currency list are out of sync.\n", structLocaleCount
, currencyCount
);
1415 #if U_CHARSET_FAMILY == U_ASCII_FAMILY
1416 ures_resetIterator(currencies
);
1417 while ((isoCode
= uenum_next(en
, NULL
, &errorCode
)) != NULL
&& ures_hasNext(currencies
)) {
1418 subBundle
= ures_getNextResource(currencies
, NULL
, &errorCode
);
1419 structISOCode
= ures_getKey(subBundle
);
1420 ures_close(subBundle
);
1421 if (strcmp(structISOCode
, isoCode
) != 0) {
1422 log_err("First difference found at structLocale(%s) and ISO4217(%s).\n", structISOCode
, isoCode
);
1428 ures_close(currencies
);
1434 static void TestAvailableIsoCodes(void){
1435 #if !UCONFIG_NO_FORMATTING
1436 UErrorCode errorCode
= U_ZERO_ERROR
;
1437 const char* eurCode
= "EUR";
1438 const char* usdCode
= "USD";
1439 const char* lastCode
= "RHD";
1440 const char* zzzCode
= "ZZZ";
1441 UDate date1950
= (UDate
)-630720000000.0;/* year 1950 */
1442 UDate date1970
= (UDate
)0.0; /* year 1970 */
1443 UDate date1975
= (UDate
)173448000000.0; /* year 1975 */
1444 UDate date1978
= (UDate
)260172000000.0; /* year 1978 */
1445 UDate date1981
= (UDate
)346896000000.0; /* year 1981 */
1446 UDate date1992
= (UDate
)693792000000.0; /* year 1992 */
1447 UChar
* isoCode
= (UChar
*)malloc(sizeof(UChar
) * (uprv_strlen(usdCode
) + 1));
1449 /* testing available codes with no time ranges */
1450 u_charsToUChars(eurCode
, isoCode
, (int32_t)uprv_strlen(usdCode
) + 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", eurCode
);
1455 u_charsToUChars(usdCode
, isoCode
, (int32_t)uprv_strlen(zzzCode
) + 1);
1456 if (ucurr_isAvailable(isoCode
, U_DATE_MIN
, U_DATE_MAX
, &errorCode
) == FALSE
) {
1457 log_data_err("FAIL: ISO code (%s) is not found.\n", usdCode
);
1460 u_charsToUChars(zzzCode
, isoCode
, (int32_t)uprv_strlen(zzzCode
) + 1);
1461 if (ucurr_isAvailable(isoCode
, U_DATE_MIN
, U_DATE_MAX
, &errorCode
) == TRUE
) {
1462 log_err("FAIL: ISO code (%s) is reported as available, but it doesn't exist.\n", zzzCode
);
1465 u_charsToUChars(lastCode
, isoCode
, (int32_t)uprv_strlen(zzzCode
) + 1);
1466 if (ucurr_isAvailable(isoCode
, U_DATE_MIN
, U_DATE_MAX
, &errorCode
) == FALSE
) {
1467 log_data_err("FAIL: ISO code (%s) is not found.\n", lastCode
);
1470 /* RHD was used from 1970-02-17 to 1980-04-18*/
1473 if (ucurr_isAvailable(isoCode
, date1970
, U_DATE_MAX
, &errorCode
) == FALSE
) {
1474 log_data_err("FAIL: ISO code (%s) was available in time range >1970-01-01.\n", lastCode
);
1477 if (ucurr_isAvailable(isoCode
, date1975
, U_DATE_MAX
, &errorCode
) == FALSE
) {
1478 log_data_err("FAIL: ISO code (%s) was available in time range >1975.\n", lastCode
);
1481 if (ucurr_isAvailable(isoCode
, date1981
, U_DATE_MAX
, &errorCode
) == TRUE
) {
1482 log_err("FAIL: ISO code (%s) was not available in time range >1981.\n", lastCode
);
1486 if (ucurr_isAvailable(isoCode
, U_DATE_MIN
, date1970
, &errorCode
) == TRUE
) {
1487 log_err("FAIL: ISO code (%s) was not available in time range <1970.\n", lastCode
);
1490 if (ucurr_isAvailable(isoCode
, U_DATE_MIN
, date1975
, &errorCode
) == FALSE
) {
1491 log_data_err("FAIL: ISO code (%s) was available in time range <1975.\n", lastCode
);
1494 if (ucurr_isAvailable(isoCode
, U_DATE_MIN
, date1981
, &errorCode
) == FALSE
) {
1495 log_data_err("FAIL: ISO code (%s) was available in time range <1981.\n", lastCode
);
1499 if (ucurr_isAvailable(isoCode
, date1975
, date1978
, &errorCode
) == FALSE
) {
1500 log_data_err("FAIL: ISO code (%s) was available in time range 1975-1978.\n", lastCode
);
1503 if (ucurr_isAvailable(isoCode
, date1970
, date1975
, &errorCode
) == FALSE
) {
1504 log_data_err("FAIL: ISO code (%s) was available in time range 1970-1975.\n", lastCode
);
1507 if (ucurr_isAvailable(isoCode
, date1975
, date1981
, &errorCode
) == FALSE
) {
1508 log_data_err("FAIL: ISO code (%s) was available in time range 1975-1981.\n", lastCode
);
1511 if (ucurr_isAvailable(isoCode
, date1970
, date1981
, &errorCode
) == FALSE
) {
1512 log_data_err("FAIL: ISO code (%s) was available in time range 1970-1981.\n", lastCode
);
1515 if (ucurr_isAvailable(isoCode
, date1981
, date1992
, &errorCode
) == TRUE
) {
1516 log_err("FAIL: ISO code (%s) was not available in time range 1981-1992.\n", lastCode
);
1519 if (ucurr_isAvailable(isoCode
, date1950
, date1970
, &errorCode
) == TRUE
) {
1520 log_err("FAIL: ISO code (%s) was not available in time range 1950-1970.\n", lastCode
);
1523 /* wrong range - from > to*/
1524 if (ucurr_isAvailable(isoCode
, date1975
, date1970
, &errorCode
) == TRUE
) {
1525 log_err("FAIL: Wrong range 1975-1970 for ISO code (%s) was not reported.\n", lastCode
);
1526 } else if (errorCode
!= U_ILLEGAL_ARGUMENT_ERROR
) {
1527 log_data_err("FAIL: Error code not reported for wrong range 1975-1970 for ISO code (%s).\n", lastCode
);
1537 const UChar
* quoteStart
;
1538 const UChar
* quoteEnd
;
1539 const UChar
* altQuoteStart
;
1540 const UChar
* altQuoteEnd
;
1542 static const DelimItem delimItems
[] = {
1543 { "en", u
"“", u
"”", u
"‘", u
"’"},
1544 { "es", u
"«", u
"»", u
"“", u
"”"},
1545 { "es_MX", u
"“", u
"”", u
"‘", u
"’"},
1546 { "ja", u
"「", u
"」", u
"『", u
"』"},
1547 { NULL
,NULL
,NULL
,NULL
,NULL
}
1549 enum {kMaxDelimLen
= 8};
1551 static void TestDelimiters(void){
1552 const DelimItem
* itemPtr
;
1553 for (itemPtr
= delimItems
; itemPtr
->locale
!= NULL
; itemPtr
++) {
1554 UErrorCode status
= U_ZERO_ERROR
;
1555 ULocaleData
* uldat
= ulocdata_open(itemPtr
->locale
, &status
);
1556 if (U_FAILURE(status
)) {
1557 log_data_err("FAIL: ulocdata_open fails for locale %s: %s\n", itemPtr
->locale
, u_errorName(status
));
1559 UChar quoteStart
[kMaxDelimLen
+1];
1560 UChar quoteEnd
[kMaxDelimLen
+1];
1561 UChar altQuoteStart
[kMaxDelimLen
+1];
1562 UChar altQuoteEnd
[kMaxDelimLen
+1];
1564 ulocdata_getDelimiter(uldat
, ULOCDATA_QUOTATION_START
, quoteStart
, kMaxDelimLen
, &status
);
1565 quoteStart
[kMaxDelimLen
] = 0;
1566 ulocdata_getDelimiter(uldat
, ULOCDATA_QUOTATION_END
, quoteEnd
, kMaxDelimLen
, &status
);
1567 quoteEnd
[kMaxDelimLen
] = 0;
1568 ulocdata_getDelimiter(uldat
, ULOCDATA_ALT_QUOTATION_START
, altQuoteStart
, kMaxDelimLen
, &status
);
1569 altQuoteStart
[kMaxDelimLen
] = 0;
1570 ulocdata_getDelimiter(uldat
, ULOCDATA_ALT_QUOTATION_END
, altQuoteEnd
, kMaxDelimLen
, &status
);
1571 altQuoteEnd
[kMaxDelimLen
] = 0;
1572 if (U_FAILURE(status
)) {
1573 log_err("FAIL: ulocdata_getDelimiter fails for locale %s: %s\n", itemPtr
->locale
, u_errorName(status
));
1574 } else if (u_strcmp(quoteStart
, itemPtr
->quoteStart
) != 0 || u_strcmp(quoteEnd
, itemPtr
->quoteEnd
) != 0 ||
1575 u_strcmp(altQuoteStart
, itemPtr
->altQuoteStart
) != 0 || u_strcmp(altQuoteEnd
, itemPtr
->altQuoteEnd
) != 0) {
1576 log_err("FAIL: ulocdata_getDelimiter error for locale %s, one or more delimiters not as expected\n", itemPtr
->locale
);
1578 ulocdata_close(uldat
);
1584 #define TESTCASE(name) addTest(root, &name, "tsutil/cldrtest/" #name)
1586 void addCLDRTest(TestNode
** root
);
1588 void addCLDRTest(TestNode
** root
)
1590 #if !UCONFIG_NO_FILE_IO && !UCONFIG_NO_LEGACY_CONVERSION
1591 TESTCASE(TestLocaleStructure
);
1592 TESTCASE(TestCurrencyList
);
1594 TESTCASE(TestConsistentCountryInfo
);
1595 TESTCASE(VerifyTranslation
);
1596 TESTCASE(TestExemplarSet
);
1597 TESTCASE(TestLocaleDisplayPattern
);
1598 TESTCASE(TestCoverage
);
1599 TESTCASE(TestIndexChars
);
1600 TESTCASE(TestAvailableIsoCodes
);
1601 TESTCASE(TestDelimiters
);