ICU-12615 LocaleValidityChecker to handle the special case '-u-va-posix'

X-SVN-Rev: 39806

diff --git a/icu4j/main/classes/core/src/com/ibm/icu/impl/locale/LocaleValidityChecker.java b/icu4j/main/classes/core/src/com/ibm/icu/impl/locale/LocaleValidityChecker.java
index fbe826149faed8ef7ef02a752511ba7f984c3bb1..a08b0684ff9a941b108ff6b8360bc8dbdde9223c 100644 (file)
--- a/icu4j/main/classes/core/src/com/ibm/icu/impl/locale/LocaleValidityChecker.java
+++ b/icu4j/main/classes/core/src/com/ibm/icu/impl/locale/LocaleValidityChecker.java
@@ -75,7 +75,7 @@ public class LocaleValidityChecker {
         final Set<Character> extensionKeys = locale.getExtensionKeys();
         //        if (language.isEmpty()) {
         //            // the only case where this is valid is if there is only an 'x' extension string
-        //            if (!script.isEmpty() || !region.isEmpty() || variantString.isEmpty() 
+        //            if (!script.isEmpty() || !region.isEmpty() || variantString.isEmpty()
         //                    || extensionKeys.size() != 1 || !extensionKeys.contains('x')) {
         //                return where.set(Datatype.x, "Null language only with x-...");
         //            }
@@ -135,8 +135,8 @@ public class LocaleValidityChecker {
     }
 
     /**
-     * @param locale 
-     * @param datatype 
+     * @param locale
+     * @param datatype
      * @param extension
      * @param where
      * @return
@@ -154,7 +154,7 @@ public class LocaleValidityChecker {
         // TODO: is empty -u- valid?
 
         for (String subtag : SEPARATOR.split(extensionString)) {
-            if (subtag.length() == 2 
+            if (subtag.length() == 2
                     && (tBuffer == null || subtag.charAt(1) <= '9')) {
                 // if we have accumulated a t buffer, check that first
                 if (tBuffer != null) {
@@ -182,7 +182,7 @@ public class LocaleValidityChecker {
             } else {
                 ++typeCount;
                 switch (valueType) {
-                case single: 
+                case single:
                     if (typeCount > 1) {
                         return where.set(datatype, key+"-"+subtag);
                     }
@@ -203,9 +203,9 @@ public class LocaleValidityChecker {
                     break;
                 }
                 switch (specialCase) {
-                case anything: 
+                case anything:
                     continue;
-                case codepoints: 
+                case codepoints:
                     try {
                         if (Integer.parseInt(subtag,16) > 0x10FFFF) {
                             return where.set(datatype, key+"-"+subtag);
@@ -324,14 +324,30 @@ public class LocaleValidityChecker {
     }
 
     /**
-     * @param language
-     * @param language2
+     * @param datatype
+     * @param code
+     * @param where
      * @return
      */
     private boolean isValid(Datatype datatype, String code, Where where) {
-        return code.isEmpty() ? true :
-            ValidIdentifiers.isValid(datatype, datasubtypes, code) != null ? true : 
-                where == null ? false 
-                        : where.set(datatype, code);
+        if (code.isEmpty()) {
+            return true;
+        }
+
+        // Note:
+        // BCP 47 -u- locale extension '-u-va-posix' is mapped to variant 'posix' automatically.
+        // For example, ULocale.forLanguageTag("en-u-va-posix").getVariant() returns "posix".
+        // This is only the exceptional case when -u- locale extension is mapped to a subtag type
+        // other than keyword.
+        //
+        // The locale validity data is based on IANA language subtag registry data and "posix"
+        // is not a valid variant. So we need to handle this specific case here. There are no
+        // othe exceptions.
+        if (datatype == Datatype.variant && "posix".equalsIgnoreCase(code)) {
+            return true;
+        }
+
+        return ValidIdentifiers.isValid(datatype, datasubtypes, code) != null ?
+                true : (where == null ? false : where.set(datatype, code));
     }
 }

diff --git a/icu4j/main/tests/core/src/com/ibm/icu/dev/test/util/TestLocaleValidity.java b/icu4j/main/tests/core/src/com/ibm/icu/dev/test/util/TestLocaleValidity.java
index 4b0e821cac41a882e11bc5fbe9278b03b8b411ec..929ef67a58961a81403ddd6040cac860ee780fe9 100644 (file)
--- a/icu4j/main/tests/core/src/com/ibm/icu/dev/test/util/TestLocaleValidity.java
+++ b/icu4j/main/tests/core/src/com/ibm/icu/dev/test/util/TestLocaleValidity.java
@@ -73,7 +73,7 @@ public class TestLocaleValidity extends TestFmwk {
                 {"OK", "en-u-sd-usny"},
                 {"OK", "en-u-tz-adalv"},
                 {"OK", "en-u-va-posix"},
-                
+
                 {"OK", "en-t-d0-accents"},
                 {"OK", "en-u-em-default"},
                 {"OK", "en-t-i0-handwrit"},
@@ -100,9 +100,9 @@ public class TestLocaleValidity extends TestFmwk {
                 {"OK", "en-u-sd-uszzzz"},
 
                 // really long case
-                
+
                 {"OK", "en-u-ca-buddhist-ca-islamic-umalqura-cf-account-co-big5han-cu-adp-fw-fri-hc-h11-ka-noignore-kb-false-kc-false-kf-false-kk-false-kn-false-kr-latn-digit-symbol-ks-identic-kv-currency-nu-ahom-sd-usny-tz-adalv-va-posix"},
-                
+
                 // bad case (for language tag)
                 {"{language, root}", "root"},
 
@@ -120,16 +120,16 @@ public class TestLocaleValidity extends TestFmwk {
                 {"Incomplete privateuse [at index 0]", "x-abc$defg"},
                 {"Invalid subtag: $ [at index 3]", "EN-$"},
                 {"Invalid subtag: $ [at index 0]", "$"},
-                
+
                 // bad extension
-                
+
                 {"{illegal, q}", "en-q-abcdefg"},
-                
+
                 {"Incomplete privateuse [at index 3]", "en-x-123456789"},
                 {"Empty subtag [at index 14]", "en-x-12345678--a"},
 
                 // bad subtags
-                
+
                 {"{variant, FOOBAR}", "zh-Hant-1606nict-1694acad-foobar"},
                 {"{region, AB}", "zh-Hant-AB"},
                 {"{language, ex}", "ex"},
@@ -137,11 +137,11 @@ public class TestLocaleValidity extends TestFmwk {
                 {"{language, qaa}", "qaa"},
 
                 // bad types for keys
-                
+
                 {"{u, ca-chinesx}", "en-u-ca-chinesx"},
                 {"{script, Latx}", "und-Cyrl-t-und-latx"},
                 {"{u, sd-usca}", "en-AQ-u-sd-usca"},
-                
+
                 {"{u, ca-buddhisx}", "en-u-ca-buddhisx"},
                 {"{u, ca-islamic-umalqurx}", "en-u-ca-islamic-umalqurx"}, // additive
                 {"{u, cf-accounx}", "en-u-cf-accounx"},
@@ -159,16 +159,16 @@ public class TestLocaleValidity extends TestFmwk {
                 {"{u, kr-latn}", "en-u-kr-latn-digit-latn"}, // reorder codes, duplicat
                 {"{u, kr-zzzz}", "en-u-kr-latn-others-digit-Zzzz"}, // reorder codes, duplicat
                 {"{u, kr-zsym}", "en-u-kr-Zsym"}, // reorder codes, duplicat
-                {"{u, kr-qaai}", "en-u-kr-Qaai"}, // reorder codes, duplicat                
+                {"{u, kr-qaai}", "en-u-kr-Qaai"}, // reorder codes, duplicat
                 {"{u, ks-identix}", "en-u-ks-identix"},
                 {"{u, kv-currencx}", "en-u-kv-currencx"},
                 {"{u, nu-ahox}", "en-u-nu-ahox"},
                 {"{u, sd-usnx}", "en-u-sd-usnx"},
                 {"{u, tz-adalx}", "en-u-tz-adalx"},
                 {"{u, va-posit}", "en-u-va-posit"},
-                
+
                 // too many items
-                
+
                 {"{u, cu-usd}", "en-u-cu-adp-usd"},
 
                 // use deprecated subtags. testDeprecated checks if they work when Datasubtype.deprecated is added
@@ -180,7 +180,7 @@ public class TestLocaleValidity extends TestFmwk {
         };
         final LinkedHashSet<String> foundKeys = new LinkedHashSet<String>();
         check(tests, foundKeys, Datasubtype.regular, Datasubtype.unknown);
-        
+
         LinkedHashSet<String> missing = new LinkedHashSet(KeyTypeData.getBcp47Keys());
         missing.removeAll(foundKeys);
         if (!assertEquals("Missing keys", Collections.EMPTY_SET, missing)) {
@@ -243,8 +243,6 @@ public class TestLocaleValidity extends TestFmwk {
         int count = 0;
         LocaleValidityChecker localeValidityChecker = new LocaleValidityChecker(datasubtypes);
         for (String[] test : tests) {
-            if (test[1].endsWith("-va-posix") && logKnownIssue("12615","Validity check wrong for -va-posix?"))
-                continue;
             check(++count, localeValidityChecker, test[0], test[1], keys);
         }
     }
@@ -279,8 +277,8 @@ public class TestLocaleValidity extends TestFmwk {
     private void addKeys(ULocale ulocale, Set<String> keys) {
         for (char cp : ulocale.getExtensionKeys()) {
             switch (cp) {
-            case 't': 
-            case 'u': 
+            case 't':
+            case 'u':
                 String extensionString = ulocale.getExtension(cp);
                 String[] parts = extensionString.split("-");
                 for (String part : parts) {
@@ -321,11 +319,11 @@ public class TestLocaleValidity extends TestFmwk {
     }
 
     private void showValid(Datasubtype expected, Datatype datatype, Set<Datasubtype> datasubtypes, String code) {
-        Datasubtype value = ValidIdentifiers.isValid(datatype, datasubtypes, code);   
+        Datasubtype value = ValidIdentifiers.isValid(datatype, datasubtypes, code);
         assertEquals(datatype + ", " + datasubtypes + ", " + code, expected, value);
     }
     private void showValid(Datasubtype expected, Datatype datatype, Set<Datasubtype> datasubtypes, String code, String code2) {
-        Datasubtype value = ValidIdentifiers.isValid(datatype, datasubtypes, code, code2);   
+        Datasubtype value = ValidIdentifiers.isValid(datatype, datasubtypes, code, code2);
         assertEquals(datatype + ", " + datasubtypes + ", " + code + ", " + code2, expected, value);
     }
 }