[Intl] Improve Intl constructor performance

Lazy initialize some data once.

$ python -u tools/run_perf.py --binary-override-path   out/x64.release/d8 --filter "JSTests/Intl"   test/js-perf-test/JSTests.json
INFO      >>> Running suite: JSTests/Intl
INFO      >>> Stdout (#1):
NewIntlCollator-Intl(Score): 2047
NewIntlDateTimeFormat-Intl(Score): 26.5
NewIntlNumberFormat-Intl(Score): 449
NewIntlPluralRules-Intl(Score): 670
NewIntlListFormat-Intl(Score): 1762
NewIntlRelativeTimeFormat-Intl(Score): 918


Compare to performance baseline scores in
https://chromium-review.googlesource.com/c/v8/v8/+/1455717

Bug: chromium:928098
Change-Id: I7f213f50eb261474fb38d9592725a101939020d8
Reviewed-on: https://chromium-review.googlesource.com/c/1455727
Commit-Queue: Frank Tang <ftang@chromium.org>
Reviewed-by: Jakob Kummerow <jkummerow@chromium.org>
Cr-Commit-Position: refs/heads/master@{#59425}

src/objects/intl-objects.h

@@ -250,6 +250,30 @@ class Intl {
       const std::vector<std::string>& requested_locales, MatcherOption options,
       const std::set<std::string>& relevant_extension_keys);
 
+  // A helper template to implement the GetAvailableLocales
+  // Usage in src/objects/js-XXX.cc
+  //
+  // const std::set<std::string>& JSXxx::GetAvailableLocales() {
+  //   static base::LazyInstance<Intl::AvailableLocales<icu::YYY>>::type
+  //       available_locales = LAZY_INSTANCE_INITIALIZER;
+  //   return available_locales.Pointer()->Get();
+  // }
+  template <typename T>
+  class AvailableLocales {
+   public:
+    AvailableLocales() {
+      int32_t num_locales = 0;
+      const icu::Locale* icu_available_locales =
+          T::getAvailableLocales(num_locales);
+      set = Intl::BuildLocaleSet(icu_available_locales, num_locales);
+    }
+    virtual ~AvailableLocales() {}
+    const std::set<std::string>& Get() const { return set; }
+
+   private:
+    std::set<std::string> set;
+  };
+
   // Utility function to set text to BreakIterator.
   static Managed<icu::UnicodeString> SetTextToBreakIterator(
       Isolate* isolate, Handle<String> text,

src/objects/js-break-iterator.cc

@@ -186,11 +186,10 @@ String JSV8BreakIterator::BreakType(Isolate* isolate,
   return ReadOnlyRoots(isolate).unknown_string();
 }
 
-std::set<std::string> JSV8BreakIterator::GetAvailableLocales() {
-  int32_t num_locales = 0;
-  const icu::Locale* icu_available_locales =
-      icu::BreakIterator::getAvailableLocales(num_locales);
-  return Intl::BuildLocaleSet(icu_available_locales, num_locales);
+const std::set<std::string>& JSV8BreakIterator::GetAvailableLocales() {
+  static base::LazyInstance<Intl::AvailableLocales<icu::BreakIterator>>::type
+      available_locales = LAZY_INSTANCE_INITIALIZER;
+  return available_locales.Pointer()->Get();
 }
 
 }  // namespace internal

src/objects/js-break-iterator.h

@@ -35,7 +35,7 @@ class JSV8BreakIterator : public JSObject {
   static Handle<JSObject> ResolvedOptions(
       Isolate* isolate, Handle<JSV8BreakIterator> break_iterator);
 
-  static std::set<std::string> GetAvailableLocales();
+  static const std::set<std::string>& GetAvailableLocales();
 
   static void AdoptText(Isolate* isolate,
                         Handle<JSV8BreakIterator> break_iterator,

src/objects/js-collator.cc

@@ -484,11 +484,10 @@ MaybeHandle<JSCollator> JSCollator::Initialize(Isolate* isolate,
   return collator;
 }
 
-std::set<std::string> JSCollator::GetAvailableLocales() {
-  int32_t num_locales = 0;
-  const icu::Locale* icu_available_locales =
-      icu::Collator::getAvailableLocales(num_locales);
-  return Intl::BuildLocaleSet(icu_available_locales, num_locales);
+const std::set<std::string>& JSCollator::GetAvailableLocales() {
+  static base::LazyInstance<Intl::AvailableLocales<icu::Collator>>::type
+      available_locales = LAZY_INSTANCE_INITIALIZER;
+  return available_locales.Pointer()->Get();
 }
 
 }  // namespace internal

src/objects/js-collator.h

@@ -40,7 +40,7 @@ class JSCollator : public JSObject {
   static Handle<JSObject> ResolvedOptions(Isolate* isolate,
                                           Handle<JSCollator> collator);
 
-  static std::set<std::string> GetAvailableLocales();
+  static const std::set<std::string>& GetAvailableLocales();
 
   DECL_CAST(JSCollator)
   DECL_PRINTER(JSCollator)

src/objects/js-date-time-format.cc

@@ -54,7 +54,7 @@ class PatternItem {
   std::vector<const char*> allowed_values;
 };
 
-const std::vector<PatternItem> GetPatternItems() {
+static const std::vector<PatternItem> BuildPatternItems() {
   const std::vector<const char*> kLongShort = {"long", "short"};
   const std::vector<const char*> kNarrowLongShort = {"narrow", "long", "short"};
   const std::vector<const char*> k2DigitNumeric = {"2-digit", "numeric"};
@@ -107,6 +107,22 @@ const std::vector<PatternItem> GetPatternItems() {
   return kPatternItems;
 }
 
+class PatternItems {
+ public:
+  PatternItems() : data(BuildPatternItems()) {}
+  virtual ~PatternItems() {}
+  const std::vector<PatternItem>& Get() const { return data; }
+
+ private:
+  const std::vector<PatternItem> data;
+};
+
+static const std::vector<PatternItem>& GetPatternItems() {
+  static base::LazyInstance<PatternItems>::type items =
+      LAZY_INSTANCE_INITIALIZER;
+  return items.Pointer()->Get();
+}
+
 class PatternData {
  public:
   PatternData(const std::string property, std::vector<PatternMap> pairs,
@@ -154,23 +170,57 @@ const std::vector<PatternData> CreateData(const char* digit2,
 //                                  kk   24
 //   K      hour in am/pm (0~11)    K    0
 //                                  KK   00
-const std::vector<PatternData> GetPatternData(Intl::HourCycle hour_cycle) {
-  const std::vector<PatternData> data = CreateData("jj", "j");
-  const std::vector<PatternData> data_h11 = CreateData("KK", "K");
-  const std::vector<PatternData> data_h12 = CreateData("hh", "h");
-  const std::vector<PatternData> data_h23 = CreateData("HH", "H");
-  const std::vector<PatternData> data_h24 = CreateData("kk", "k");
+
+class Pattern {
+ public:
+  Pattern(const char* d1, const char* d2) : data(CreateData(d1, d2)) {}
+  virtual ~Pattern() {}
+  virtual const std::vector<PatternData>& Get() const { return data; }
+
+ private:
+  std::vector<PatternData> data;
+};
+
+#define DEFFINE_TRAIT(name, d1, d2)              \
+  struct name {                                  \
+    static void Construct(void* allocated_ptr) { \
+      new (allocated_ptr) Pattern(d1, d2);       \
+    }                                            \
+  };
+DEFFINE_TRAIT(H11Trait, "KK", "K")
+DEFFINE_TRAIT(H12Trait, "hh", "h")
+DEFFINE_TRAIT(H23Trait, "HH", "H")
+DEFFINE_TRAIT(H24Trait, "kk", "k")
+DEFFINE_TRAIT(HDefaultTrait, "jj", "j")
+#undef DEFFINE_TRAIT
+
+const std::vector<PatternData>& GetPatternData(Intl::HourCycle hour_cycle) {
   switch (hour_cycle) {
-    case Intl::HourCycle::kH11:
-      return data_h11;
-    case Intl::HourCycle::kH12:
-      return data_h12;
-    case Intl::HourCycle::kH23:
-      return data_h23;
-    case Intl::HourCycle::kH24:
-      return data_h24;
-    case Intl::HourCycle::kUndefined:
-      return data;
+    case Intl::HourCycle::kH11: {
+      static base::LazyInstance<Pattern, H11Trait>::type h11 =
+          LAZY_INSTANCE_INITIALIZER;
+      return h11.Pointer()->Get();
+    }
+    case Intl::HourCycle::kH12: {
+      static base::LazyInstance<Pattern, H12Trait>::type h12 =
+          LAZY_INSTANCE_INITIALIZER;
+      return h12.Pointer()->Get();
+    }
+    case Intl::HourCycle::kH23: {
+      static base::LazyInstance<Pattern, H23Trait>::type h23 =
+          LAZY_INSTANCE_INITIALIZER;
+      return h23.Pointer()->Get();
+    }
+    case Intl::HourCycle::kH24: {
+      static base::LazyInstance<Pattern, H24Trait>::type h24 =
+          LAZY_INSTANCE_INITIALIZER;
+      return h24.Pointer()->Get();
+    }
+    case Intl::HourCycle::kUndefined: {
+      static base::LazyInstance<Pattern, HDefaultTrait>::type hDefault =
+          LAZY_INSTANCE_INITIALIZER;
+      return hDefault.Pointer()->Get();
+    }
     default:
       UNREACHABLE();
   }
@@ -1362,11 +1412,10 @@ MaybeHandle<Object> JSDateTimeFormat::FormatToParts(
   return result;
 }
 
-std::set<std::string> JSDateTimeFormat::GetAvailableLocales() {
-  int32_t num_locales = 0;
-  const icu::Locale* icu_available_locales =
-      icu::DateFormat::getAvailableLocales(num_locales);
-  return Intl::BuildLocaleSet(icu_available_locales, num_locales);
+const std::set<std::string>& JSDateTimeFormat::GetAvailableLocales() {
+  static base::LazyInstance<Intl::AvailableLocales<icu::DateFormat>>::type
+      available_locales = LAZY_INSTANCE_INITIALIZER;
+  return available_locales.Pointer()->Get();
 }
 
 Handle<String> JSDateTimeFormat::HourCycleAsString() const {

src/objects/js-date-time-format.h

@@ -71,7 +71,7 @@ class JSDateTimeFormat : public JSObject {
       Isolate* isolate, Handle<Object> date, Handle<Object> locales,
       Handle<Object> options, RequiredOption required, DefaultsOption defaults);
 
-  static std::set<std::string> GetAvailableLocales();
+  static const std::set<std::string>& GetAvailableLocales();
 
   Handle<String> HourCycleAsString() const;
   DECL_CAST(JSDateTimeFormat)

src/objects/js-list-format.cc

@@ -411,15 +411,14 @@ MaybeHandle<String> JSListFormat::FormatList(Isolate* isolate,
   return Intl::ToString(isolate, formatted);
 }
 
-std::set<std::string> JSListFormat::GetAvailableLocales() {
-  int32_t num_locales = 0;
+const std::set<std::string>& JSListFormat::GetAvailableLocales() {
   // TODO(ftang): for now just use
   // icu::Locale::getAvailableLocales(count) until we migrate to
   // Intl::GetAvailableLocales().
   // ICU FR at https://unicode-org.atlassian.net/browse/ICU-20015
-  const icu::Locale* icu_available_locales =
-      icu::Locale::getAvailableLocales(num_locales);
-  return Intl::BuildLocaleSet(icu_available_locales, num_locales);
+  static base::LazyInstance<Intl::AvailableLocales<icu::Locale>>::type
+      available_locales = LAZY_INSTANCE_INITIALIZER;
+  return available_locales.Pointer()->Get();
 }
 
 // ecma42 #sec-formatlisttoparts

src/objects/js-list-format.h

@@ -49,7 +49,7 @@ class JSListFormat : public JSObject {
       Isolate* isolate, Handle<JSListFormat> format_holder,
       Handle<JSArray> list);
 
-  static std::set<std::string> GetAvailableLocales();
+  static const std::set<std::string>& GetAvailableLocales();
 
   Handle<String> StyleAsString() const;
   Handle<String> TypeAsString() const;

src/objects/js-number-format.cc

@@ -747,11 +747,10 @@ MaybeHandle<JSArray> JSNumberFormat::FormatToParts(
   return result;
 }
 
-std::set<std::string> JSNumberFormat::GetAvailableLocales() {
-  int32_t num_locales = 0;
-  const icu::Locale* icu_available_locales =
-      icu::NumberFormat::getAvailableLocales(num_locales);
-  return Intl::BuildLocaleSet(icu_available_locales, num_locales);
+const std::set<std::string>& JSNumberFormat::GetAvailableLocales() {
+  static base::LazyInstance<Intl::AvailableLocales<icu::NumberFormat>>::type
+      available_locales = LAZY_INSTANCE_INITIALIZER;
+  return available_locales.Pointer()->Get();
 }
 
 }  // namespace internal

src/objects/js-number-format.h

@@ -64,7 +64,7 @@ class JSNumberFormat : public JSObject {
       Isolate* isolate, const icu::NumberFormat& number_format,
       Handle<Object> numeric_obj);
 
-  static std::set<std::string> GetAvailableLocales();
+  static const std::set<std::string>& GetAvailableLocales();
 
   Handle<String> StyleAsString() const;
   Handle<String> CurrencyDisplayAsString() const;

src/objects/js-plural-rules.cc

@@ -314,16 +314,15 @@ Handle<JSObject> JSPluralRules::ResolvedOptions(
   return options;
 }
 
-std::set<std::string> JSPluralRules::GetAvailableLocales() {
-  int32_t num_locales = 0;
+const std::set<std::string>& JSPluralRules::GetAvailableLocales() {
   // TODO(ftang): For PluralRules, filter out locales that
   // don't support PluralRules.
   // PluralRules is missing an appropriate getAvailableLocales method,
   // so we should filter from all locales, but it's not clear how; see
   // https://ssl.icu-project.org/trac/ticket/12756
-  const icu::Locale* icu_available_locales =
-      icu::Locale::getAvailableLocales(num_locales);
-  return Intl::BuildLocaleSet(icu_available_locales, num_locales);
+  static base::LazyInstance<Intl::AvailableLocales<icu::Locale>>::type
+      available_locales = LAZY_INSTANCE_INITIALIZER;
+  return available_locales.Pointer()->Get();
 }
 
 }  // namespace internal

src/objects/js-plural-rules.h

@@ -40,7 +40,7 @@ class JSPluralRules : public JSObject {
   V8_WARN_UNUSED_RESULT static MaybeHandle<String> ResolvePlural(
       Isolate* isolate, Handle<JSPluralRules> plural_rules, double number);
 
-  static std::set<std::string> GetAvailableLocales();
+  static const std::set<std::string>& GetAvailableLocales();
 
   // [[Type]] is one of the values "cardinal" or "ordinal",
   // identifying the plural rules used.

src/objects/js-relative-time-format.cc

@@ -409,11 +409,10 @@ MaybeHandle<Object> JSRelativeTimeFormat::Format(
       formatted.length()));
 }
 
-std::set<std::string> JSRelativeTimeFormat::GetAvailableLocales() {
-  int32_t num_locales = 0;
-  const icu::Locale* icu_available_locales =
-      icu::DateFormat::getAvailableLocales(num_locales);
-  return Intl::BuildLocaleSet(icu_available_locales, num_locales);
+const std::set<std::string>& JSRelativeTimeFormat::GetAvailableLocales() {
+  static base::LazyInstance<Intl::AvailableLocales<icu::DateFormat>>::type
+      available_locales = LAZY_INSTANCE_INITIALIZER;
+  return available_locales.Pointer()->Get();
 }
 
 }  // namespace internal

src/objects/js-relative-time-format.h

@@ -50,7 +50,7 @@ class JSRelativeTimeFormat : public JSObject {
       Handle<JSRelativeTimeFormat> format_holder, const char* func_name,
       bool to_parts);
 
-  static std::set<std::string> GetAvailableLocales();
+  static const std::set<std::string>& GetAvailableLocales();
 
   DECL_CAST(JSRelativeTimeFormat)
 

src/objects/js-segmenter.cc

@@ -162,11 +162,10 @@ Handle<String> JSSegmenter::GranularityAsString() const {
   }
 }
 
-std::set<std::string> JSSegmenter::GetAvailableLocales() {
-  int32_t num_locales = 0;
-  const icu::Locale* icu_available_locales =
-      icu::BreakIterator::getAvailableLocales(num_locales);
-  return Intl::BuildLocaleSet(icu_available_locales, num_locales);
+const std::set<std::string>& JSSegmenter::GetAvailableLocales() {
+  static base::LazyInstance<Intl::AvailableLocales<icu::BreakIterator>>::type
+      available_locales = LAZY_INSTANCE_INITIALIZER;
+  return available_locales.Pointer()->Get();
 }
 
 }  // namespace internal

src/objects/js-segmenter.h

@@ -39,7 +39,7 @@ class JSSegmenter : public JSObject {
   V8_WARN_UNUSED_RESULT static Handle<JSObject> ResolvedOptions(
       Isolate* isolate, Handle<JSSegmenter> segmenter_holder);
 
-  static std::set<std::string> GetAvailableLocales();
+  static const std::set<std::string>& GetAvailableLocales();
 
   Handle<String> GranularityAsString() const;