[Intl] Move Intl.getCanonicalLocales to C++

Bug: v8:5751
Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng
Change-Id: I20dc0e7eda58f2d9211d0e569d553dd6fcc14fd3
Reviewed-on: https://chromium-review.googlesource.com/c/1275807
Commit-Queue: Frank Tang <ftang@chromium.org>
Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
Cr-Commit-Position: refs/heads/master@{#56747}

src/bootstrapper.cc

@@ -2913,6 +2913,9 @@ void Genesis::InitializeGlobal(Handle<JSGlobalObject> global_object,
         factory->NewJSObject(isolate_->object_function(), TENURED);
     JSObject::AddProperty(isolate_, global, name, intl, DONT_ENUM);
 
+    SimpleInstallFunction(isolate(), intl, "getCanonicalLocales",
+                          Builtins::kIntlGetCanonicalLocales, 1, false);
+
     {
       Handle<JSFunction> date_time_format_constructor = InstallFunction(
           isolate_, intl, "DateTimeFormat", JS_INTL_DATE_TIME_FORMAT_TYPE,

src/builtins/builtins-definitions.h

@@ -1371,6 +1371,8 @@ namespace internal {
   CPP(DateTimeFormatPrototypeResolvedOptions)                          \
   /* ecma402 #sec-intl.datetimeformat.supportedlocalesof */            \
   CPP(DateTimeFormatSupportedLocalesOf)                                \
+  /* ecma402 #sec-intl.getcanonicallocales */                          \
+  CPP(IntlGetCanonicalLocales)                                         \
   /* ecma402 #sec-intl-listformat-constructor */                       \
   CPP(ListFormatConstructor)                                           \
   /* ecma402 #sec-intl-list-format.prototype.format */                 \

src/builtins/builtins-intl.cc

@@ -485,6 +485,14 @@ BUILTIN(DateTimeFormatInternalFormat) {
                                         isolate, date_format_holder, date));
 }
 
+BUILTIN(IntlGetCanonicalLocales) {
+  HandleScope scope(isolate);
+  Handle<Object> locales = args.atOrUndefined(isolate, 1);
+
+  RETURN_RESULT_OR_FAILURE(isolate,
+                           Intl::GetCanonicalLocales(isolate, locales));
+}
+
 BUILTIN(ListFormatConstructor) {
   HandleScope scope(isolate);
 

src/js/intl.js

@@ -17,85 +17,13 @@
 // -------------------------------------------------------------------
 // Imports
 
-var ArrayPush;
 var GlobalIntl = global.Intl;
 var GlobalIntlDateTimeFormat = GlobalIntl.DateTimeFormat;
 var GlobalIntlNumberFormat = GlobalIntl.NumberFormat;
 var GlobalIntlCollator = GlobalIntl.Collator;
-var GlobalIntlPluralRules = GlobalIntl.PluralRules;
-var GlobalIntlv8BreakIterator = GlobalIntl.v8BreakIterator;
-var InternalArray = utils.InternalArray;
-
-utils.Import(function(from) {
-  ArrayPush = from.ArrayPush;
-});
 
 // -------------------------------------------------------------------
 
-/**
- * Caches available locales for each service.
- */
-var AVAILABLE_LOCALES = {
-   __proto__ : null,
-  'collator': UNDEFINED,
-  'numberformat': UNDEFINED,
-  'dateformat': UNDEFINED,
-  'breakiterator': UNDEFINED,
-  'pluralrules': UNDEFINED,
-  'relativetimeformat': UNDEFINED,
-  'listformat': UNDEFINED,
-  'segmenter': UNDEFINED,
-};
-
-/* Make JS array[] out of InternalArray */
-function makeArray(input) {
-  var array = [];
-  %MoveArrayContents(input, array);
-  return array;
-}
-
-/**
- * Returns an InternalArray where all locales are canonicalized and duplicates
- * removed.
- * Throws on locales that are not well formed BCP47 tags.
- * ECMA 402 8.2.1 steps 1 (ECMA 402 9.2.1) and 2.
- */
-function canonicalizeLocaleList(locales) {
-  var seen = new InternalArray();
-  if (!IS_UNDEFINED(locales)) {
-    // We allow single string localeID.
-    if (typeof locales === 'string') {
-      %_Call(ArrayPush, seen, %CanonicalizeLanguageTag(locales));
-      return seen;
-    }
-
-    var o = TO_OBJECT(locales);
-    var len = TO_LENGTH(o.length);
-
-    for (var k = 0; k < len; k++) {
-      if (k in o) {
-        var value = o[k];
-
-        var tag = %CanonicalizeLanguageTag(value);
-
-        if (%ArrayIndexOf(seen, tag, 0) === -1) {
-          %_Call(ArrayPush, seen, tag);
-        }
-      }
-    }
-  }
-
-  return seen;
-}
-
-// ECMA 402 section 8.2.1
-DEFINE_METHOD(
-  GlobalIntl,
-  getCanonicalLocales(locales) {
-    return makeArray(canonicalizeLocaleList(locales));
-  }
-);
-
 // Save references to Intl objects and methods we use, for added security.
 var savedObjects = {
   __proto__: null,

src/objects/intl-objects.cc

@@ -1304,12 +1304,35 @@ std::vector<std::string> BestFitSupportedLocales(
   return LookupSupportedLocales(available_locales, requested_locales);
 }
 
+enum MatcherOption { kBestFit, kLookup };
+
+// ecma262 #sec-createarrayfromlist
+Handle<JSArray> CreateArrayFromList(Isolate* isolate,
+                                    std::vector<std::string> elements,
+                                    PropertyAttributes attr) {
+  Factory* factory = isolate->factory();
+  // Let array be ! ArrayCreate(0).
+  Handle<JSArray> array = factory->NewJSArray(0);
+
+  uint32_t length = static_cast<uint32_t>(elements.size());
+  // 3. Let n be 0.
+  // 4. For each element e of elements, do
+  for (uint32_t i = 0; i < length; i++) {
+    // a. Let status be CreateDataProperty(array, ! ToString(n), e).
+    const std::string& part = elements[i];
+    Handle<String> value =
+        factory->NewStringFromUtf8(CStrVector(part.c_str())).ToHandleChecked();
+    JSObject::AddDataElement(array, i, value, attr);
+  }
+  // 5. Return array.
+  return array;
+}
+
 // TODO(bstell): should this be moved somewhere where it is reusable?
 // Implement steps 5, 6, 7 for ECMA 402 9.2.9 SupportedLocales
 // https://tc39.github.io/ecma402/#sec-supportedlocales
 MaybeHandle<JSObject> CreateReadOnlyArray(Isolate* isolate,
                                           std::vector<std::string> elements) {
-  Factory* factory = isolate->factory();
   if (elements.size() >= kMaxUInt32) {
     THROW_NEW_ERROR(
         isolate, NewRangeError(MessageTemplate::kInvalidArrayLength), JSObject);
@@ -1319,17 +1342,9 @@ MaybeHandle<JSObject> CreateReadOnlyArray(Isolate* isolate,
       static_cast<PropertyAttributes>(READ_ONLY | DONT_DELETE);
 
   // 5. Let subset be CreateArrayFromList(elements).
-  // 6. Let keys be subset.[[OwnPropertyKeys]]().
-  Handle<JSArray> subset = factory->NewJSArray(0);
+  Handle<JSArray> subset = CreateArrayFromList(isolate, elements, attr);
 
-  // 7. For each element P of keys in List order, do
-  uint32_t length = static_cast<uint32_t>(elements.size());
-  for (uint32_t i = 0; i < length; i++) {
-    const std::string& part = elements[i];
-    Handle<String> value =
-        factory->NewStringFromUtf8(CStrVector(part.c_str())).ToHandleChecked();
-    JSObject::AddDataElement(subset, i, value, attr);
-  }
+  // 6. Let keys be subset.[[OwnPropertyKeys]]().
 
   // 7.a. Let desc be PropertyDescriptor { [[Configurable]]: false,
   //          [[Writable]]: false }.
@@ -1409,6 +1424,19 @@ MaybeHandle<JSObject> SupportedLocales(
 }
 }  // namespace
 
+// ecma-402 #sec-intl.getcanonicallocales
+MaybeHandle<JSArray> Intl::GetCanonicalLocales(Isolate* isolate,
+                                               Handle<Object> locales) {
+  // 1. Let ll be ? CanonicalizeLocaleList(locales).
+  Maybe<std::vector<std::string>> maybe_ll =
+      CanonicalizeLocaleList(isolate, locales, false);
+  MAYBE_RETURN(maybe_ll, MaybeHandle<JSArray>());
+
+  // 2. Return CreateArrayFromList(ll).
+  PropertyAttributes attr = static_cast<PropertyAttributes>(NONE);
+  return CreateArrayFromList(isolate, maybe_ll.FromJust(), attr);
+}
+
 // ECMA 402 Intl.*.supportedLocalesOf
 MaybeHandle<JSObject> Intl::SupportedLocalesOf(Isolate* isolate,
                                                ICUService service,

src/objects/intl-objects.h

@@ -135,6 +135,10 @@ class Intl {
       Isolate* isolate, Handle<String> locale, Handle<JSObject> options,
       Handle<JSObject> resolved);
 
+  // ecma-402 #sec-intl.getcanonicallocales
+  V8_WARN_UNUSED_RESULT static MaybeHandle<JSArray> GetCanonicalLocales(
+      Isolate* isolate, Handle<Object> locales);
+
   // For locale sensitive functions
   V8_WARN_UNUSED_RESULT static MaybeHandle<String> StringLocaleConvertCase(
       Isolate* isolate, Handle<String> s, bool is_upper,