fix: ensure values of en-US-POSIX canonicalize correctly

src/ecma402/locale.cpp

@@ -63,6 +63,7 @@ namespace {
 	int getNumberingSystemsForLocale(char *localeId, const char **values);
 	int getTimeZonesForLocale(char *localeId, const char **values);
 	int languageTagForLocaleId(const char *localeId, char *languageTag, ecma402_errorStatus *status);
+	int languageTagForLocaleIdAlt(const char *localeId, char *languageTag, ecma402_errorStatus *status);
 
 } // namespace
 
@@ -868,54 +869,100 @@ namespace {
 
 	int languageTagForLocaleId(const char *localeId, char *languageTag, ecma402_errorStatus *status)
 	{
-		icu::Locale canonicalLocale;
 		UErrorCode icuStatus = U_ZERO_ERROR;
-		UBool const strict = 1;
-		char *unicodeLocaleId;
+		icu::Locale unicodeLocale;
+		std::string bcp47Locale;
+		char *resolvedLocale;
+		int resolvedLocaleLength;
 
-		canonicalLocale = icu::Locale::createCanonical(localeId);
-		if (canonicalLocale == nullptr) {
-			ecma402_ecmaError(status, CANNOT_CREATE_LOCALE_ID, "Invalid language tag \"%s\"", localeId);
-			return -1;
+		unicodeLocale = icu::Locale::createCanonical(localeId);
+		if (unicodeLocale == nullptr) {
+			// If icu::Locale::createCanonical() fails to create a locale, then
+			// use an alternate form of creation.
+			// See: https://unicode-org.atlassian.net/browse/ICU-22486
+			return languageTagForLocaleIdAlt(localeId, languageTag, status);
 		}
 
-		std::string const locale = canonicalLocale.toLanguageTag<std::string>(icuStatus);
-		if (U_FAILURE(icuStatus) != U_ZERO_ERROR) {
+		bcp47Locale = unicodeLocale.toLanguageTag<std::string>(icuStatus);
+		if (icuStatus > U_ZERO_ERROR) {
 			ecma402_icuError(status, icuStatus, "Invalid language tag \"%s\"", localeId);
 			return -1;
 		}
 
-		// If the input localeId is not "und," but we got "und," then return 0.
-		if (strcasecmp(locale.c_str(), UNDETERMINED_LANGUAGE) == 0 &&
+		// If the input localeId is not "und," but we got "und," then return -1.
+		if (strcasecmp(bcp47Locale.c_str(), UNDETERMINED_LANGUAGE) == 0 &&
 		    strcasecmp(localeId, UNDETERMINED_LANGUAGE) != 0) {
 			ecma402_ecmaError(status, UNDEFINED_LOCALE_ID, "Invalid language tag \"%s\"", localeId);
 			return -1;
 		}
 
 		// This additional conversion step forces tags like "en-latn-us-co-foo" and
 		// "de-de_euro" to result in failures, which is the desired result.
-		unicodeLocaleId = (char *)malloc(sizeof(char) * ULOC_FULLNAME_CAPACITY);
-		int const length =
-			uloc_toLanguageTag(locale.c_str(), unicodeLocaleId, ULOC_FULLNAME_CAPACITY, strict, &icuStatus);
+		resolvedLocale = (char *)malloc(sizeof(char) * ULOC_FULLNAME_CAPACITY);
+		resolvedLocaleLength =
+			uloc_toLanguageTag(bcp47Locale.c_str(), resolvedLocale, ULOC_FULLNAME_CAPACITY, 1, &icuStatus);
 
-		if (U_FAILURE(icuStatus) != U_ZERO_ERROR || strlen(unicodeLocaleId) == 0 || unicodeLocaleId == nullptr) {
-			if (U_FAILURE(icuStatus) != U_ZERO_ERROR) {
+		if (icuStatus > U_ZERO_ERROR || strlen(resolvedLocale) == 0 || resolvedLocale == nullptr) {
+			if (icuStatus > U_ZERO_ERROR) {
 				ecma402_icuError(status, icuStatus, "Invalid language tag \"%s\"", localeId);
 			} else {
 				ecma402_ecmaError(status, INVALID_LOCALE_ID, "Invalid language tag \"%s\"", localeId);
 			}
 
-			if (unicodeLocaleId != nullptr) {
-				free(unicodeLocaleId);
+			if (resolvedLocale != nullptr) {
+				free(resolvedLocale);
 			}
 
 			return -1;
 		}
 
-		memcpy(languageTag, unicodeLocaleId, length + 1);
-		free(unicodeLocaleId);
+		memcpy(languageTag, resolvedLocale, resolvedLocaleLength + 1);
+		free(resolvedLocale);
+
+		return resolvedLocaleLength;
+	}
+
+	int languageTagForLocaleIdAlt(const char *localeId, char *languageTag, ecma402_errorStatus *status)
+	{
+		UErrorCode icuStatus = U_ZERO_ERROR;
+		char *unicodeLocale, *bcp47Locale;
+		int bcp47LocaleLength;
+
+		unicodeLocale = (char *)malloc(sizeof(char) * ULOC_FULLNAME_CAPACITY);
+		uloc_canonicalize(localeId, unicodeLocale, ULOC_FULLNAME_CAPACITY, &icuStatus);
+
+		if (icuStatus > U_ZERO_ERROR) {
+			ecma402_icuError(status, icuStatus, "Invalid language tag \"%s\"", localeId);
+			free(unicodeLocale);
+			return -1;
+		}
+
+		bcp47Locale = (char *)malloc(sizeof(char) * ULOC_FULLNAME_CAPACITY);
+		bcp47LocaleLength = uloc_toLanguageTag(unicodeLocale, bcp47Locale, ULOC_FULLNAME_CAPACITY, 1, &icuStatus);
+		free(unicodeLocale);
+
+		if (icuStatus > U_ZERO_ERROR) {
+			ecma402_icuError(status, icuStatus, "Invalid language tag \"%s\"", localeId);
+			free(bcp47Locale);
+			return -1;
+		}
+
+		// If the input localeId is not "und," but we got "und," then return -1.
+		if (strcasecmp(bcp47Locale, UNDETERMINED_LANGUAGE) == 0 && strcasecmp(localeId, UNDETERMINED_LANGUAGE) != 0) {
+			ecma402_ecmaError(status, UNDEFINED_LOCALE_ID, "Invalid language tag \"%s\"", localeId);
+			return -1;
+		}
+
+		if (bcp47Locale == nullptr || strlen(bcp47Locale) == 0) {
+			ecma402_ecmaError(status, INVALID_LOCALE_ID, "Invalid language tag \"%s\"", localeId);
+			free(bcp47Locale);
+			return -1;
+		}
+
+		memcpy(languageTag, bcp47Locale, bcp47LocaleLength + 1);
+		free(bcp47Locale);
 
-		return length;
+		return bcp47LocaleLength;
 	}
 
 } // namespace

tests/criterion/ecma402/locale_canonicalization_test.c

@@ -77,7 +77,7 @@ Test(TEST_SUITE, canonicalizeLocaleIdHasErrorForStructurallyInvalidLocaleId)
 
 ParameterizedTestParameters(TEST_SUITE, canonicalizeLocaleIdCanonicalizes)
 {
-	START_STRING_TEST_PARAMS(24)
+	START_STRING_TEST_PARAMS(29)
 	STRING_TEST("de", "de")
 	STRING_TEST("DE-de", "de-DE")
 	STRING_TEST("de-DE", "de-DE")
@@ -102,6 +102,11 @@ ParameterizedTestParameters(TEST_SUITE, canonicalizeLocaleIdCanonicalizes)
 	STRING_TEST("zh-xiang-u-nu-thai-x-0", "hsn-u-nu-thai-x-0")
 	STRING_TEST("de-DE-u-cu-FOO", "de-DE-u-cu-foo")
 	STRING_TEST("de-DE-u-cu-FOOBAR", "de-DE-u-cu-foobar")
+	STRING_TEST("de-POSIX", "de-u-va-posix")
+	STRING_TEST("de-DE-POSIX", "de-DE-u-va-posix")
+	STRING_TEST("en-POSIX", "en-u-va-posix")
+	STRING_TEST("en-GB-POSIX", "en-GB-u-va-posix")
+	STRING_TEST("en-US-POSIX", "en-US-u-va-posix")
 	END_STRING_TEST_PARAMS;
 }
 

tests/phpt/Intl/Locale-001.phpt

@@ -20,6 +20,7 @@ $tests = [
     'en-u-kn-false',
     'en-u-nu-arab',
     'en-latn-us-u-ca-gregory-kf-upper-co-emoji-hc-h23-nu-latn-kn-true-cu-USD',
+    'en-US-POSIX',
 ];
 
 foreach ($tests as $test) {
@@ -38,3 +39,4 @@ en-u-hc-h12
 en-u-kn-false
 en-u-nu-arab
 en-Latn-US-u-ca-gregory-co-emoji-cu-usd-hc-h23-kf-upper-kn-nu-latn
+en-US-u-va-posix

tests/phpt/Intl/Locale-008.phpt

@@ -34,6 +34,7 @@ $tests = [
     )],
     ['en-latn-us-u-ca-gregory-kf-upper-co-emoji-hc-h23-nu-latn-kn-true', new Options()],
     ['en-US-u-cu-usd', new Options(currency: 'EUR')],
+    ['en-US-POSIX', new Options(region: 'gb')],
 ];
 
 foreach ($tests as $params) {
@@ -54,3 +55,4 @@ en-u-nu-latn
 es-Some-MX-u-ca-buddhist-co-phonebk-cu-mxn-hc-h11-kf-false-kn-false-nu-arab
 en-Latn-US-u-ca-gregory-co-emoji-hc-h23-kf-upper-kn-nu-latn
 en-US-u-cu-eur
+en-GB-u-va-posix

tests/phpt/Intl/Locale-009.phpt

@@ -41,6 +41,7 @@ $tests = [
 
     'hy-arevela',
     'hy-arevmda',
+    'en-US-POSIX',
 ];
 
 foreach ($tests as $test) {
@@ -109,3 +110,7 @@ minimized: hy
 canonical: hyw
 maximized: hyw
 minimized: hyw
+
+canonical: en-US-u-va-posix
+maximized: en-Latn-US-u-va-posix
+minimized: en-u-va-posix

tests/phpt/Intl/Locale-010.phpt

@@ -41,6 +41,7 @@ $tests = [
 
     'hy-arevela',
     'hy-arevmda',
+    'en-US-POSIX',
 ];
 
 foreach ($tests as $test) {
@@ -109,3 +110,7 @@ minimized: hy
 canonical: hyw
 maximized: hyw-Armn-AM
 minimized: hyw
+
+canonical: en-US-u-va-posix
+maximized: en-Latn-US-u-va-posix
+minimized: en-u-va-posix

tests/phpt/Intl_getCanonicalLocales-006.phpt

@@ -44,6 +44,7 @@ $generate = function () use ($makeStringable, $makeImplicitStringable): iterable
         'cmn-hans-cn-u-ca-t-ca-x-t-u',
         $makeImplicitStringable('de-gregory-u-ca-gregory'),
         'sgn-GR',
+        'en-US-POSIX',
         'ji',
         'de-DD',
         $makeStringable('in'),
@@ -73,11 +74,12 @@ array (
   7 => 'zh-Hans-CN-t-ca-u-ca-x-t-u',
   8 => 'de-gregory-u-ca-gregory',
   9 => 'gss',
-  10 => 'yi',
-  11 => 'id',
-  12 => 'sr-Cyrl-ekavsk',
-  13 => 'en-CA-newfound',
-  14 => 'sl-1994-biske-rozaj',
-  15 => 'da-u-attr',
-  16 => 'da-u-attr-co-search',
+  10 => 'en-US-u-va-posix',
+  11 => 'yi',
+  12 => 'id',
+  13 => 'sr-Cyrl-ekavsk',
+  14 => 'en-CA-newfound',
+  15 => 'sl-1994-biske-rozaj',
+  16 => 'da-u-attr',
+  17 => 'da-u-attr-co-search',
 )

tests/phpt/Intl_getCanonicalLocales-007.phpt

@@ -31,6 +31,7 @@ $tags = [
     $makeStringable('cmn-hans-cn-u-ca-t-ca-x-t-u'),
     $makeStringable('de-gregory-u-ca-gregory'),
     'sgn-GR',
+    $makeStringable('en-US-POSIX'),
     $makeStringable('ji'),
     'de-DD',
     'in',
@@ -66,11 +67,12 @@ array (
   7 => 'zh-Hans-CN-t-ca-u-ca-x-t-u',
   8 => 'de-gregory-u-ca-gregory',
   9 => 'gss',
-  10 => 'yi',
-  11 => 'id',
-  12 => 'sr-Cyrl-ekavsk',
-  13 => 'en-CA-newfound',
-  14 => 'sl-1994-biske-rozaj',
-  15 => 'da-u-attr',
-  16 => 'da-u-attr-co-search',
+  10 => 'en-US-u-va-posix',
+  11 => 'yi',
+  12 => 'id',
+  13 => 'sr-Cyrl-ekavsk',
+  14 => 'en-CA-newfound',
+  15 => 'sl-1994-biske-rozaj',
+  16 => 'da-u-attr',
+  17 => 'da-u-attr-co-search',
 )

tests/phpt/Intl_getCanonicalLocales-008.phpt

@@ -32,6 +32,7 @@ $generate = function () use ($makeStringable): iterable {
         'cmn-hans-cn-u-ca-t-ca-x-t-u',
         'de-gregory-u-ca-gregory',
         'sgn-GR',
+        'en-US-POSIX',
         'ji',
         'de-DD',
         'in',

tests/phpt/Intl_getCanonicalLocales-009.phpt

@@ -31,6 +31,7 @@ $tags = [
     $makeStringable('cmn-hans-cn-u-ca-t-ca-x-t-u'),
     $makeStringable('de-gregory-u-ca-gregory'),
     $makeStringable('sgn-GR'),
+    'en-US-POSIX',
     'ji',
     'de-DD',
     new stdClass(),                      // Here is the invalid value.