Issue #2259 - Improve ResolveLocale performance when initializing the default Intl objects

Based-on: m-c 1365650

10 files changed, 189 insertions, 91 deletions

diff --git a/js/src/builtin/intl/Collator.cpp b/js/src/builtin/intl/Collator.cpp
index ea04b4ee3a..ad95a5712c 100644
--- a/js/src/builtin/intl/Collator.cpp
+++ b/js/src/builtin/intl/Collator.cpp
@@ -233,7 +233,6 @@ js::intl_availableCollations(JSContext* cx, unsigned argc, Value* vp)
     if (!DefineElement(cx, collations, index++, NullHandleValue))

         return false;

 

-    RootedString jscollation(cx);

     RootedValue element(cx);

 

     for (uint32_t i = 0; i < count; i++) {

@@ -251,8 +250,7 @@ js::intl_availableCollations(JSContext* cx, unsigned argc, Value* vp)
             continue;

 

         // ICU returns old-style keyword values; map them to BCP 47 equivalents.

-        jscollation = JS_NewStringCopyZ(cx, uloc_toUnicodeLocaleType("co", collation));

-

+        JSString* jscollation = JS_NewStringCopyZ(cx, uloc_toUnicodeLocaleType("co", collation));

         if (!jscollation)

             return false;

         element = StringValue(jscollation);

diff --git a/js/src/builtin/intl/Collator.js b/js/src/builtin/intl/Collator.js
index eba09d3c83..ee6ea9a9b8 100644
--- a/js/src/builtin/intl/Collator.js
+++ b/js/src/builtin/intl/Collator.js
@@ -82,15 +82,11 @@ function resolveCollatorInternals(lazyCollatorData)
     // Steps 21-22.

     var s = lazyCollatorData.rawSensitivity;

     if (s === undefined) {

-        if (collatorIsSorting) {

-            // Step 21.a.

-            s = "variant";

-        } else {

-            // Step 21.b.

-            var dataLocale = r.dataLocale;

-            var dataLocaleData = localeData(dataLocale);

-            s = dataLocaleData.sensitivity;

-        }

+        // In theory the default sensitivity for the "search" collator is

+        // locale dependent; in reality the CLDR/ICU default strength is

+        // always tertiary. Therefore use "variant" as the default value for

+        // both collation modes.

+        s = "variant";

     }

     internalProps.sensitivity = s;

 

@@ -263,49 +259,96 @@ var collatorInternalProperties = {
 

 

 /**

- * Returns the default caseFirst values for the given locale and usage. The

- * first element in the returned array denotes the default value per ES2017

- * Intl, 9.1 Internal slots of Service Constructors.

+ * Returns the actual locale used when a collator for |locale| is constructed.

  */

-function collatorCaseFirst(locale, usage) {

+function collatorActualLocale(locale) {

     assert(typeof locale === "string", "locale should be string");

-    assert(usage === "sort" || usage === "search", "invalid usage option");

-

-    if (usage === "sort") {

-        // If |locale| is the default locale (e.g. da-DK), but only supported

-        // through a fallback (da), we need to get the actual locale before we

-        // can call intl_isUpperCaseFirst. Also see BestAvailableLocaleHelper.

-        var availableLocales = callFunction(collatorInternalProperties.availableLocales,

-                                            collatorInternalProperties);

-        var actualLocale = BestAvailableLocaleIgnoringDefault(availableLocales, locale);

-

-        if (intl_isUpperCaseFirst(actualLocale))

-            return ["upper", "false", "lower"];

-    }

+

+    // If |locale| is the default locale (e.g. da-DK), but only supported

+    // through a fallback (da), we need to get the actual locale before we

+    // can call intl_isUpperCaseFirst. Also see BestAvailableLocaleHelper.

+    var availableLocales = callFunction(collatorInternalProperties.availableLocales,

+                                        collatorInternalProperties);

+    return BestAvailableLocaleIgnoringDefault(availableLocales, locale);

+}

+

+

+/**

+ * Returns the default caseFirst values for the given locale. The first

+ * element in the returned array denotes the default value per ES2017 Intl,

+ * 9.1 Internal slots of Service Constructors.

+ */

+function collatorSortCaseFirst(locale) {

+    var actualLocale = collatorActualLocale(locale);

+    if (intl_isUpperCaseFirst(actualLocale))

+        return ["upper", "false", "lower"];

 

     // Default caseFirst values for all other languages.

     return ["false", "lower", "upper"];

 }

 

 

-function collatorSortLocaleData(locale) {

+/**

+ * Returns the default caseFirst value for the given locale.

+ */

+function collatorSortCaseFirstDefault(locale) {

+    var actualLocale = collatorActualLocale(locale);

+    if (intl_isUpperCaseFirst(actualLocale))

+        return "upper";

+

+    // Default caseFirst value for all other languages.

+    return "false";

+}

+

+function collatorSortLocaleData() {

+    /* eslint-disable object-shorthand */

     return {

-        co: intl_availableCollations(locale),

-        kn: ["false", "true"],

-        kf: collatorCaseFirst(locale, "sort"),

+        co: intl_availableCollations,

+        kn: function() {

+            return ["false", "true"];

+        },

+        kf: collatorSortCaseFirst,

+        default: {

+            co: function() {

+                // The first element of the collations array must be |null|

+                // per ES2017 Intl, 10.2.3 Internal Slots.

+                return null;

+            },

+            kn: function() {

+                return "false";

+            },

+            kf: collatorSortCaseFirstDefault,

+        }

     };

+    /* eslint-enable object-shorthand */

 }

 

 

-function collatorSearchLocaleData(locale) {

+function collatorSearchLocaleData() {

+    /* eslint-disable object-shorthand */

     return {

-        co: [null],

-        kn: ["false", "true"],

-        kf: collatorCaseFirst(locale, "search"),

-        // In theory the default sensitivity is locale dependent;

-        // in reality the CLDR/ICU default strength is always tertiary.

-        sensitivity: "variant"

+        co: function() {

+            return [null];

+        },

+        kn: function() {

+            return ["false", "true"];

+        },

+        kf: function() {

+            return ["false", "lower", "upper"];

+        },

+        default: {

+            co: function() {

+                return null;

+            },

+            kn: function() {

+                return "false";

+            },

+            kf: function() {

+                return "false";

+            },

+        }

     };

+    /* eslint-enable object-shorthand */

 }

 

 

@@ -354,6 +397,7 @@ function Intl_Collator_compare_get() {
     // Step 2.

     return internals.boundCompare;

 }

+_SetCanonicalName(Intl_Collator_compare_get, "get compare");

 

 

 /**

diff --git a/js/src/builtin/intl/CommonFunctions.cpp b/js/src/builtin/intl/CommonFunctions.cpp
index be07ae4de7..e8723f3cfe 100644
--- a/js/src/builtin/intl/CommonFunctions.cpp
+++ b/js/src/builtin/intl/CommonFunctions.cpp
@@ -92,7 +92,6 @@ js::intl::GetAvailableLocales(JSContext* cx, CountAvailable countAvailable,
         return false;

 

     uint32_t count = countAvailable();

-    RootedValue t(cx, BooleanValue(true));

     for (uint32_t i = 0; i < count; i++) {

         const char* locale = getAvailable(i);

         auto lang = DuplicateString(cx, locale);

@@ -104,7 +103,7 @@ js::intl::GetAvailableLocales(JSContext* cx, CountAvailable countAvailable,
         RootedAtom a(cx, Atomize(cx, lang.get(), strlen(lang.get())));

         if (!a)

             return false;

-        if (!DefineProperty(cx, locales, a->asPropertyName(), t, nullptr, nullptr,

+        if (!DefineProperty(cx, locales, a->asPropertyName(), TrueHandleValue, nullptr, nullptr,

                             JSPROP_ENUMERATE))

         {

             return false;

diff --git a/js/src/builtin/intl/CommonFunctions.js b/js/src/builtin/intl/CommonFunctions.js
index 59fbeaea3b..cf5a615721 100644
--- a/js/src/builtin/intl/CommonFunctions.js
+++ b/js/src/builtin/intl/CommonFunctions.js
@@ -767,8 +767,6 @@ function BestAvailableLocaleIgnoringDefault(availableLocales, locale) {
     return BestAvailableLocaleHelper(availableLocales, locale, false);
 }
 
-var noRelevantExtensionKeys = [];
-
 /**
  * Compares a BCP 47 language priority list against the set of locales in
  * availableLocales and determines the best available language to meet the
@@ -925,21 +923,17 @@ function ResolveLocale(availableLocales, requestedLocales, options, relevantExte
     // Step 8.
     var supportedExtension = "-u";
 
+    // In this implementation, localeData is a function, not an object.
+    var localeDataProvider = localeData();
+
     // Steps 9-12.
-    var i = 0;
-    var len = relevantExtensionKeys.length;
-    while (i < len) {
+    for (var i = 0; i < relevantExtensionKeys.length; i++) {
         // Steps 12.a-c.
         var key = relevantExtensionKeys[i];
 
-        // In this implementation, localeData is a function, not an object.
-        var foundLocaleData = localeData(foundLocale);
-        var keyLocaleData = foundLocaleData[key];
-
-        // Locale data provides default value.
-        // Step 12.d.
-        var value = keyLocaleData[0];
-        assert(typeof value === "string" || value === null, "unexpected locale data value");
+        // Steps 12.b-d (The locale data is only computed when needed).
+        var keyLocaleData = undefined;
+        var value = undefined;
 
         // Locale tag may override.
 
@@ -958,6 +952,9 @@ function ResolveLocale(availableLocales, requestedLocales, options, relevantExte
 
             // Step 12.f.ii.
             if (requestedValue !== undefined) {
+                // Steps 12.b-c.
+                keyLocaleData = callFunction(localeDataProvider[key], null, foundLocale);
+
                 // Step 12.f.ii.1.
                 if (requestedValue !== "") {
                     // Step 12.f.ii.1.a.
@@ -983,18 +980,29 @@ function ResolveLocale(availableLocales, requestedLocales, options, relevantExte
         var optionsValue = options[key];
 
         // Step 12.g, 12.gg.ii.
-        if (optionsValue !== undefined &&
-            optionsValue !== value &&
-            callFunction(ArrayIndexOf, keyLocaleData, optionsValue) !== -1)
-        {
-            value = optionsValue;
-            supportedExtensionAddition = "";
+        if (optionsValue !== undefined && optionsValue !== value) {
+            // Steps 12.b-c.
+            if (keyLocaleData === undefined)
+                keyLocaleData = callFunction(localeDataProvider[key], null, foundLocale);
+
+            if (callFunction(ArrayIndexOf, keyLocaleData, optionsValue) !== -1) {
+                value = optionsValue;
+                supportedExtensionAddition = "";
+            }
+        }
+
+        // Locale data provides default value.
+        if (value === undefined) {
+            // Steps 12.b-d.
+            value = keyLocaleData === undefined
+                    ? callFunction(localeDataProvider.default[key], null, foundLocale)
+                    : keyLocaleData[0];
         }
 
         // Steps 12.h-j.
+        assert(typeof value === "string" || value === null, "unexpected locale data value");
         result[key] = value;
         supportedExtension += supportedExtensionAddition;
-        i++;
     }
 
     // Step 13.
diff --git a/js/src/builtin/intl/DateTimeFormat.cpp b/js/src/builtin/intl/DateTimeFormat.cpp
index c649c3c416..9b0a10f1c3 100644
--- a/js/src/builtin/intl/DateTimeFormat.cpp
+++ b/js/src/builtin/intl/DateTimeFormat.cpp
@@ -241,6 +241,30 @@ js::intl_DateTimeFormat_availableLocales(JSContext* cx, unsigned argc, Value* vp
     return true;

 }

 

+static bool

+DefaultCalendar(JSContext* cx, const JSAutoByteString& locale, MutableHandleValue rval)

+{

+    UErrorCode status = U_ZERO_ERROR;

+    UCalendar* cal = ucal_open(nullptr, 0, locale.ptr(), UCAL_DEFAULT, &status);

+

+    // This correctly handles nullptr |cal| when opening failed.

+    ScopedICUObject<UCalendar, ucal_close> closeCalendar(cal);

+

+    const char* calendar = ucal_getType(cal, &status);

+    if (U_FAILURE(status)) {

+        JS_ReportErrorNumberASCII(cx, GetErrorMessage, nullptr, JSMSG_INTERNAL_INTL_ERROR);

+        return false;

+    }

+

+    // ICU returns old-style keyword values; map them to BCP 47 equivalents

+    JSString* str = JS_NewStringCopyZ(cx, uloc_toUnicodeLocaleType("ca", calendar));

+    if (!str)

+        return false;

+

+    rval.setString(str);

+    return true;

+}

+

 struct CalendarAlias

 {

     const char* const calendar;

@@ -269,31 +293,15 @@ js::intl_availableCalendars(JSContext* cx, unsigned argc, Value* vp)
     uint32_t index = 0;

 

     // We need the default calendar for the locale as the first result.

-    UErrorCode status = U_ZERO_ERROR;

-    RootedString jscalendar(cx);

-    {

-        UCalendar* cal = ucal_open(nullptr, 0, locale.ptr(), UCAL_DEFAULT, &status);

-

-        // This correctly handles nullptr |cal| when opening failed.

-        ScopedICUObject<UCalendar, ucal_close> closeCalendar(cal);

-

-        const char* calendar = ucal_getType(cal, &status);

-        if (U_FAILURE(status)) {

-            intl::ReportInternalError(cx);

-            return false;

-        }

-

-        // ICU returns old-style keyword values; map them to BCP 47 equivalents

-        jscalendar = JS_NewStringCopyZ(cx, uloc_toUnicodeLocaleType("ca", calendar));

-        if (!jscalendar)

-            return false;

-    }

+    RootedValue element(cx);

+    if (!DefaultCalendar(cx, locale, &element))

+        return false;

 

-    RootedValue element(cx, StringValue(jscalendar));

     if (!DefineElement(cx, calendars, index++, element))

         return false;

 

     // Now get the calendars that "would make a difference", i.e., not the default.

+    UErrorCode status = U_ZERO_ERROR;

     UEnumeration* values = ucal_getKeywordValuesForLocale("ca", locale.ptr(), false, &status);

     if (U_FAILURE(status)) {

         intl::ReportInternalError(cx);

@@ -317,7 +325,7 @@ js::intl_availableCalendars(JSContext* cx, unsigned argc, Value* vp)
         // ICU returns old-style keyword values; map them to BCP 47 equivalents

         calendar = uloc_toUnicodeLocaleType("ca", calendar);

 

-        jscalendar = JS_NewStringCopyZ(cx, calendar);

+        JSString* jscalendar = JS_NewStringCopyZ(cx, calendar);

         if (!jscalendar)

             return false;

         element = StringValue(jscalendar);

@@ -327,7 +335,7 @@ js::intl_availableCalendars(JSContext* cx, unsigned argc, Value* vp)
         // ICU doesn't return calendar aliases, append them here.

         for (const auto& calendarAlias : calendarAliases) {

             if (StringsAreEqual(calendar, calendarAlias.calendar)) {

-                jscalendar = JS_NewStringCopyZ(cx, calendarAlias.alias);

+                JSString* jscalendar = JS_NewStringCopyZ(cx, calendarAlias.alias);

                 if (!jscalendar)

                     return false;

                 element = StringValue(jscalendar);

@@ -342,6 +350,20 @@ js::intl_availableCalendars(JSContext* cx, unsigned argc, Value* vp)
 }

 

 bool

+js::intl_defaultCalendar(JSContext* cx, unsigned argc, Value* vp)

+{

+    CallArgs args = CallArgsFromVp(argc, vp);

+    MOZ_ASSERT(args.length() == 1);

+    MOZ_ASSERT(args[0].isString());

+

+    JSAutoByteString locale(cx, args[0].toString());

+    if (!locale)

+        return false;

+

+    return DefaultCalendar(cx, locale, args.rval());

+}

+

+bool

 js::intl_IsValidTimeZoneName(JSContext* cx, unsigned argc, Value* vp)

 {

     CallArgs args = CallArgsFromVp(argc, vp);

diff --git a/js/src/builtin/intl/DateTimeFormat.h b/js/src/builtin/intl/DateTimeFormat.h
index db468269dc..b4a6a218de 100644
--- a/js/src/builtin/intl/DateTimeFormat.h
+++ b/js/src/builtin/intl/DateTimeFormat.h
@@ -77,6 +77,16 @@ extern MOZ_MUST_USE bool
 intl_availableCalendars(JSContext* cx, unsigned argc, Value* vp);

 

 /**

+ * Returns the calendar type identifier per Unicode Technical Standard 35,

+ * Unicode Locale Data Markup Language, for the default calendar for the given

+ * locale.

+ *

+ * Usage: calendar = intl_defaultCalendar(locale)

+ */

+extern MOZ_MUST_USE bool

+intl_defaultCalendar(JSContext* cx, unsigned argc, Value* vp);

+

+/**

  * 6.4.1 IsValidTimeZoneName ( timeZone )

  *

  * Verifies that the given string is a valid time zone name. If it is a valid

diff --git a/js/src/builtin/intl/DateTimeFormat.js b/js/src/builtin/intl/DateTimeFormat.js
index 6057251ca8..79bfbd4f0e 100644
--- a/js/src/builtin/intl/DateTimeFormat.js
+++ b/js/src/builtin/intl/DateTimeFormat.js
@@ -793,10 +793,14 @@ var dateTimeFormatInternalProperties = {
 };

 

 

-function dateTimeFormatLocaleData(locale) {

+function dateTimeFormatLocaleData() {

     return {

-        ca: intl_availableCalendars(locale),

-        nu: getNumberingSystems(locale)

+        ca: intl_availableCalendars,

+        nu: getNumberingSystems,

+        default: {

+            ca: intl_defaultCalendar,

+            nu: intl_numberingSystem,

+        }

     };

 }

 

diff --git a/js/src/builtin/intl/NumberFormat.js b/js/src/builtin/intl/NumberFormat.js
index fca1bdb8ae..bba78d7a0d 100644
--- a/js/src/builtin/intl/NumberFormat.js
+++ b/js/src/builtin/intl/NumberFormat.js
@@ -446,9 +446,12 @@ function getNumberingSystems(locale) {
 }

 

 

-function numberFormatLocaleData(locale) {

+function numberFormatLocaleData() {

     return {

-        nu: getNumberingSystems(locale)

+        nu: getNumberingSystems,

+        default: {

+            nu: intl_numberingSystem,

+        }

     };

 }

 

diff --git a/js/src/builtin/intl/PluralRules.js b/js/src/builtin/intl/PluralRules.js
index 997fa4db7d..1e138a8830 100644
--- a/js/src/builtin/intl/PluralRules.js
+++ b/js/src/builtin/intl/PluralRules.js
@@ -10,6 +10,7 @@
  * Spec: ECMAScript 402 API, PluralRules, 1.3.3.

  */

 var pluralRulesInternalProperties = {

+    localeData: pluralRulesLocaleData,

     _availableLocales: null,

     availableLocales: function()

     {

@@ -20,9 +21,17 @@ var pluralRulesInternalProperties = {
         locales = intl_PluralRules_availableLocales();

         addSpecialMissingLanguageTags(locales);

         return (this._availableLocales = locales);

-    }

+    },

+    relevantExtensionKeys: [],

 };

 

+

+function pluralRulesLocaleData() {

+    // PluralRules don't support any extension keys.

+    return {};

+}

+

+

 /**

  * Compute an internal properties object from |lazyPluralRulesData|.

  */

@@ -39,7 +48,7 @@ function resolvePluralRulesInternals(lazyPluralRulesData) {
     const r = ResolveLocale(callFunction(PluralRules.availableLocales, PluralRules),

                           lazyPluralRulesData.requestedLocales,

                           lazyPluralRulesData.opt,

-                          noRelevantExtensionKeys, undefined);

+                          PluralRules.relevantExtensionKeys, PluralRules.localeData);

 

     // Step 14.

     internalProps.locale = r.locale;

diff --git a/js/src/vm/SelfHosting.cpp b/js/src/vm/SelfHosting.cpp
index 0717bfd490..ebb95c83e9 100644
--- a/js/src/vm/SelfHosting.cpp
+++ b/js/src/vm/SelfHosting.cpp
@@ -2467,6 +2467,7 @@ static const JSFunctionSpec intrinsic_functions[] = {
     JS_FN("intl_CompareStrings", intl_CompareStrings, 3,0),
     JS_FN("intl_DateTimeFormat", intl_DateTimeFormat, 2,0),
     JS_FN("intl_DateTimeFormat_availableLocales", intl_DateTimeFormat_availableLocales, 0,0),
+    JS_FN("intl_defaultCalendar", intl_defaultCalendar, 1,0),
     JS_FN("intl_defaultTimeZone", intl_defaultTimeZone, 0,0),
     JS_FN("intl_defaultTimeZoneOffset", intl_defaultTimeZoneOffset, 0,0),
     JS_FN("intl_FormatDateTime", intl_FormatDateTime, 2,0),