[Intl] Throws exception on grandfather and private locale

Bug: v8:9613
Change-Id: Ie91a5bd39c82b6baf33fd84dee8420d2c4a5f504
Reviewed-on: https://chromium-review.googlesource.com/c/v8/v8/+/1803783
Commit-Queue: Frank Tang <ftang@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#63815}

src/objects/intl-objects.cc

@@ -20,6 +20,7 @@
 #include "src/objects/js-collator-inl.h"
 #include "src/objects/js-date-time-format-inl.h"
 #include "src/objects/js-locale-inl.h"
+#include "src/objects/js-locale.h"
 #include "src/objects/js-number-format-inl.h"
 #include "src/objects/objects-inl.h"
 #include "src/objects/property-descriptor.h"
@@ -784,6 +785,11 @@ Maybe<std::string> Intl::CanonicalizeLanguageTag(Isolate* isolate,
   }
   std::string locale(locale_str->ToCString().get());
 
+  if (!IsStructurallyValidLanguageTag(locale)) {
+    THROW_NEW_ERROR_RETURN_VALUE(
+        isolate, NewRangeError(MessageTemplate::kLocaleBadParameters),
+        Nothing<std::string>());
+  }
   return Intl::CanonicalizeLanguageTag(isolate, locale);
 }
 
@@ -2096,5 +2102,9 @@ MaybeHandle<String> Intl::FormattedToString(
   return Intl::ToString(isolate, result);
 }
 
+bool Intl::IsStructurallyValidLanguageTag(const std::string& tag) {
+  return JSLocale::StartsWithUnicodeLanguageId(tag);
+}
+
 }  // namespace internal
 }  // namespace v8

src/objects/intl-objects.h

@@ -337,6 +337,8 @@ class Intl {
   static const std::set<std::string>& GetAvailableLocalesForLocale();
 
   static const std::set<std::string>& GetAvailableLocalesForDateFormat();
+
+  static bool IsStructurallyValidLanguageTag(const std::string& tag);
 };
 
 }  // namespace internal

src/objects/js-locale.cc

@@ -168,10 +168,11 @@ bool IsUnicodeVariantSubtag(const std::string& value) {
 bool IsExtensionSingleton(const std::string& value) {
   return IsAlphanum(value, 1, 1);
 }
+}  // namespace
 
 // TODO(ftang) Replace the following check w/ icu::LocaleBuilder
 // once ICU64 land in March 2019.
-bool StartsWithUnicodeLanguageId(const std::string& value) {
+bool JSLocale::StartsWithUnicodeLanguageId(const std::string& value) {
   // unicode_language_id =
   // unicode_language_subtag (sep unicode_script_subtag)?
   //   (sep unicode_region_subtag)? (sep unicode_variant_subtag)* ;
@@ -207,6 +208,7 @@ bool StartsWithUnicodeLanguageId(const std::string& value) {
   return true;
 }
 
+namespace {
 Maybe<bool> ApplyOptionsToTag(Isolate* isolate, Handle<String> tag,
                               Handle<JSReceiver> options,
                               icu::LocaleBuilder* builder) {
@@ -223,7 +225,7 @@ Maybe<bool> ApplyOptionsToTag(Isolate* isolate, Handle<String> tag,
   CHECK_NOT_NULL(*bcp47_tag);
   // 2. If IsStructurallyValidLanguageTag(tag) is false, throw a RangeError
   // exception.
-  if (!StartsWithUnicodeLanguageId(*bcp47_tag)) {
+  if (!JSLocale::StartsWithUnicodeLanguageId(*bcp47_tag)) {
     return Just(false);
   }
   UErrorCode status = U_ZERO_ERROR;

src/objects/js-locale.h

@@ -49,6 +49,9 @@ class JSLocale : public JSObject {
   static Handle<String> ToString(Isolate* isolate, Handle<JSLocale> locale);
   static std::string ToString(Handle<JSLocale> locale);
 
+  // Help function to validate locale by other Intl objects.
+  static bool StartsWithUnicodeLanguageId(const std::string& value);
+
   DECL_CAST(JSLocale)
 
   DECL_ACCESSORS(icu_locale, Managed<icu::Locale>)

test/intl/general/case-mapping.js

@@ -125,9 +125,6 @@ assertEquals("abci\u0307", "aBcI\u0307".toLowerCase());
 // Anything other than 'tr' and 'az' behave like root for U+0307.
 assertEquals("abci\u0307", "aBcI\u0307".toLocaleLowerCase("fil"));
 assertEquals("abci\u0307", "aBcI\u0307".toLocaleLowerCase("zh-Hant-TW"));
-assertEquals("abci\u0307", "aBcI\u0307".toLocaleLowerCase("i-klingon"));
-assertEquals("abci\u0307", "aBcI\u0307".toLocaleLowerCase("i-enochian"));
-assertEquals("abci\u0307", "aBcI\u0307".toLocaleLowerCase("x-foobar"));
 
 // Up to 8 chars are allowed for the primary language tag in BCP 47.
 assertEquals("abci\u0307", "aBcI\u0307".toLocaleLowerCase("longlang"));

test/intl/general/grandfathered_tags_without_preferred_value.js

@@ -8,18 +8,8 @@
   // v8 works around that ICU issue.
   // See https://www.iana.org/assignments/language-subtag-registry/language-subtag-registry .
   ["cel-gaulish", "cel-gaulish"],
-  ["i-default", "i-default"],
-  ["i-mingo", "i-mingo"],
-  ["i-enochian", "i-enochian"],
-  ["zh-min", "zh-min"],
 
   // Matching should be case-insensitive.
-  ["I-default", "i-default"],
-  ["i-DEFAULT", "i-default"],
-  ["I-DEFAULT", "i-default"],
-  ["i-DEfauLT", "i-default"],
-  ["zh-Min", "zh-min"],
-  ["Zh-min", "zh-min"],
 ].forEach(([inputLocale, expectedLocale]) => {
   const canonicalLocales = Intl.getCanonicalLocales(inputLocale);
   assertEquals(canonicalLocales.length, 1);

test/intl/general/language_tags_with_preferred_values.js

@@ -9,8 +9,6 @@
 
   // Matching should be case-insensitive.
   ["sgn-De", "gsg"],
-  ["sgn-BE-FR", "sfb"],
-  ["Sgn-bE-Fr", "sfb"],
 
   // deprecated region tag
   ["und-Latn-dd", "und-Latn-DE"],

test/intl/general/supported-locales-of.js

@@ -83,16 +83,15 @@ for (const service of services) {
   privateuseLocale = service.supportedLocalesOf("en-US-x-twain");
   assertEquals("en-US-x-twain", privateuseLocale[0]);
 
-  privateuseLocale2 = service.supportedLocalesOf("x-twain");
-  assertEquals(undefined, privateuseLocale2[0]);
+  assertThrows(() => service.supportedLocalesOf("x-twain"), RangeError);
+
 
   if (service != Intl.PluralRules) {
     grandfatheredLocale = service.supportedLocalesOf("art-lojban");
     assertEquals(undefined, grandfatheredLocale[0]);
   }
 
-  grandfatheredLocale2 = service.supportedLocalesOf("i-pwn");
-  assertEquals(undefined, grandfatheredLocale2[0]);
+  assertThrows(() => service.supportedLocalesOf("x-pwn"), RangeError);
 
   unicodeInPrivateuseLocale = service.supportedLocalesOf(
     "en-US-x-u-co-phonebk"

test/intl/list-format/resolved-options.js

@@ -144,7 +144,3 @@ assertEquals(
 assertEquals(
     'ar',
     (new Intl.ListFormat(['xyz', 'ar'])).resolvedOptions().locale);
-
-assertEquals(
-    'ar',
-    (new Intl.ListFormat(['i-default', 'ar'])).resolvedOptions().locale);

test/intl/regress-8725514.js

@@ -6,5 +6,5 @@ Object.prototype.__defineGetter__('x', function () {
     return -2147483648;
   });
 
-var f = ["x-u-foo"];
+var f = ["en-US"];
 Intl.NumberFormat(f);

test/intl/relative-time-format/resolved-options.js

@@ -156,7 +156,3 @@ assertEquals(
   assertThrows(() =>
      Intl.RelativeTimeFormat.prototype.resolvedOptions.call(receiver), TypeError);
 }
-
-assertEquals(
-    'ar',
-    (new Intl.RelativeTimeFormat(['i-default', 'ar'])).resolvedOptions().locale);

test/test262/test262.status

@@ -530,20 +530,15 @@
   # https://bugs.chromium.org/p/v8/issues/detail?id=9613
   'intl402/Intl/getCanonicalLocales/canonicalized-tags': [FAIL],
   'intl402/Intl/getCanonicalLocales/grandfathered': [FAIL],
-  'intl402/Intl/getCanonicalLocales/invalid-tags': [FAIL],
   'intl402/Intl/getCanonicalLocales/non-iana-canon': [FAIL],
   'intl402/Intl/getCanonicalLocales/preferred-grandfathered': [FAIL],
   'intl402/Intl/getCanonicalLocales/preferred-variant': [FAIL],
-  'intl402/language-tags-invalid': [FAIL],
-  'intl402/ListFormat/constructor/constructor/locales-valid': [FAIL],
   'intl402/Locale/constructor-non-iana-canon': [FAIL],
   'intl402/Locale/constructor-options-region-valid': [FAIL],
   'intl402/Locale/constructor-tag': [FAIL],
   'intl402/Locale/getters': [FAIL],
   'intl402/Locale/likely-subtags-grandfathered': [FAIL],
   'intl402/PluralRules/prototype/resolvedOptions/order': [FAIL],
-  'intl402/RelativeTimeFormat/constructor/constructor/locales-valid': [FAIL],
-  'intl402/Segmenter/constructor/constructor/locales-valid': [FAIL],
 
   ######################## NEEDS INVESTIGATION ###########################