diff --git a/BUILD.gn b/BUILD.gn index e35142380f..03c62ce558 100644 --- a/BUILD.gn +++ b/BUILD.gn @@ -2874,6 +2874,9 @@ v8_source_set("v8_base_without_compiler") { "src/objects/js-segmenter-inl.h", "src/objects/js-segmenter.cc", "src/objects/js-segmenter.h", + "src/objects/js-segments-inl.h", + "src/objects/js-segments.cc", + "src/objects/js-segments.h", "src/objects/js-weak-refs-inl.h", "src/objects/js-weak-refs.h", "src/objects/keys.cc", @@ -3724,6 +3727,9 @@ v8_source_set("v8_base_without_compiler") { "src/objects/js-segmenter-inl.h", "src/objects/js-segmenter.cc", "src/objects/js-segmenter.h", + "src/objects/js-segments-inl.h", + "src/objects/js-segments.cc", + "src/objects/js-segments.h", "src/runtime/runtime-intl.cc", "src/strings/char-predicates.cc", ] diff --git a/src/builtins/builtins-definitions.h b/src/builtins/builtins-definitions.h index 0feb5bcf40..e51f057ca7 100644 --- a/src/builtins/builtins-definitions.h +++ b/src/builtins/builtins-definitions.h @@ -1035,16 +1035,12 @@ namespace internal { CPP(SegmenterPrototypeSegment) \ /* ecma402 #sec-Intl.Segmenter.supportedLocalesOf */ \ CPP(SegmenterSupportedLocalesOf) \ - /* ecma402 #sec-segment-iterator-prototype-breakType */ \ - CPP(SegmentIteratorPrototypeBreakType) \ - /* ecma402 #sec-segment-iterator-prototype-following */ \ - CPP(SegmentIteratorPrototypeFollowing) \ - /* ecma402 #sec-segment-iterator-prototype-preceding */ \ - CPP(SegmentIteratorPrototypePreceding) \ - /* ecma402 #sec-segment-iterator-prototype-index */ \ - CPP(SegmentIteratorPrototypeIndex) \ /* ecma402 #sec-segment-iterator-prototype-next */ \ CPP(SegmentIteratorPrototypeNext) \ + /* ecma402 #sec-%segmentsprototype%.containing */ \ + CPP(SegmentsPrototypeContaining) \ + /* ecma402 #sec-%segmentsprototype%-@@iterator */ \ + CPP(SegmentsPrototypeIterator) \ /* ES #sec-string.prototype.normalize */ \ CPP(StringPrototypeNormalizeIntl) \ /* ecma402 #sup-string.prototype.tolocalelowercase */ \ diff --git a/src/builtins/builtins-intl.cc b/src/builtins/builtins-intl.cc index d483c08054..6cccaf0175 100644 --- a/src/builtins/builtins-intl.cc +++ b/src/builtins/builtins-intl.cc @@ -28,10 +28,10 @@ #include "src/objects/js-relative-time-format-inl.h" #include "src/objects/js-segment-iterator-inl.h" #include "src/objects/js-segmenter-inl.h" +#include "src/objects/js-segments-inl.h" #include "src/objects/objects-inl.h" #include "src/objects/property-descriptor.h" #include "src/objects/smi.h" - #include "unicode/brkiter.h" namespace v8 { @@ -968,32 +968,9 @@ BUILTIN(CollatorInternalCompare) { return *Intl::CompareStrings(isolate, *icu_collator, string_x, string_y); } -// ecma402 #sec-segment-iterator-prototype-breakType -BUILTIN(SegmentIteratorPrototypeBreakType) { - const char* const method = "get %SegmentIteratorPrototype%.breakType"; - HandleScope scope(isolate); - - CHECK_RECEIVER(JSSegmentIterator, segment_iterator, method); - return *segment_iterator->BreakType(); -} - -// ecma402 #sec-segment-iterator-prototype-following -BUILTIN(SegmentIteratorPrototypeFollowing) { - const char* const method = "%SegmentIteratorPrototype%.following"; - HandleScope scope(isolate); - CHECK_RECEIVER(JSSegmentIterator, segment_iterator, method); - - Handle from = args.atOrUndefined(isolate, 1); - - Maybe success = - JSSegmentIterator::Following(isolate, segment_iterator, from); - MAYBE_RETURN(success, ReadOnlyRoots(isolate).exception()); - return *isolate->factory()->ToBoolean(success.FromJust()); -} - -// ecma402 #sec-segment-iterator-prototype-next +// ecma402 #sec-%segmentiteratorprototype%.next BUILTIN(SegmentIteratorPrototypeNext) { - const char* const method = "%SegmentIteratorPrototype%.next"; + const char* const method = "%SegmentIterator.prototype%.next"; HandleScope scope(isolate); CHECK_RECEIVER(JSSegmentIterator, segment_iterator, method); @@ -1001,29 +978,7 @@ BUILTIN(SegmentIteratorPrototypeNext) { JSSegmentIterator::Next(isolate, segment_iterator)); } -// ecma402 #sec-segment-iterator-prototype-preceding -BUILTIN(SegmentIteratorPrototypePreceding) { - const char* const method = "%SegmentIteratorPrototype%.preceding"; - HandleScope scope(isolate); - CHECK_RECEIVER(JSSegmentIterator, segment_iterator, method); - - Handle from = args.atOrUndefined(isolate, 1); - - Maybe success = - JSSegmentIterator::Preceding(isolate, segment_iterator, from); - MAYBE_RETURN(success, ReadOnlyRoots(isolate).exception()); - return *isolate->factory()->ToBoolean(success.FromJust()); -} - -// ecma402 #sec-segment-iterator-prototype-index -BUILTIN(SegmentIteratorPrototypeIndex) { - const char* const method = "get %SegmentIteratorPrototype%.index"; - HandleScope scope(isolate); - - CHECK_RECEIVER(JSSegmentIterator, segment_iterator, method); - return *JSSegmentIterator::Index(isolate, segment_iterator); -} - +// ecma402 #sec-intl.segmenter BUILTIN(SegmenterConstructor) { HandleScope scope(isolate); @@ -1032,6 +987,7 @@ BUILTIN(SegmenterConstructor) { "Intl.Segmenter"); } +// ecma402 #sec-intl.segmenter.supportedlocalesof BUILTIN(SegmenterSupportedLocalesOf) { HandleScope scope(isolate); Handle locales = args.atOrUndefined(isolate, 1); @@ -1043,30 +999,52 @@ BUILTIN(SegmenterSupportedLocalesOf) { JSSegmenter::GetAvailableLocales(), locales, options)); } +// ecma402 #sec-intl.segmenter.prototype.resolvedoptions BUILTIN(SegmenterPrototypeResolvedOptions) { HandleScope scope(isolate); - CHECK_RECEIVER(JSSegmenter, segmenter_holder, + CHECK_RECEIVER(JSSegmenter, segmenter, "Intl.Segmenter.prototype.resolvedOptions"); - return *JSSegmenter::ResolvedOptions(isolate, segmenter_holder); + return *JSSegmenter::ResolvedOptions(isolate, segmenter); } -// ecma402 #sec-Intl.Segmenter.prototype.segment +// ecma402 #sec-intl.segmenter.prototype.segment BUILTIN(SegmenterPrototypeSegment) { HandleScope scope(isolate); - CHECK_RECEIVER(JSSegmenter, segmenter_holder, - "Intl.Segmenter.prototype.segment"); + CHECK_RECEIVER(JSSegmenter, segmenter, "Intl.Segmenter.prototype.segment"); Handle input_text = args.atOrUndefined(isolate, 1); // 3. Let string be ? ToString(string). - Handle text; - ASSIGN_RETURN_FAILURE_ON_EXCEPTION(isolate, text, + Handle string; + ASSIGN_RETURN_FAILURE_ON_EXCEPTION(isolate, string, Object::ToString(isolate, input_text)); - // 4. Return ? CreateSegmentIterator(segment, string). + // 4. Return ? CreateSegmentsObject(segmenter, string). + RETURN_RESULT_OR_FAILURE(isolate, + JSSegments::Create(isolate, segmenter, string)); +} + +// ecma402 #sec-%segmentsprototype%.containing +BUILTIN(SegmentsPrototypeContaining) { + const char* const method = "%Segments.prototype%.containing"; + HandleScope scope(isolate); + CHECK_RECEIVER(JSSegments, segments, method); + Handle index = args.atOrUndefined(isolate, 1); + + // 6. Let n be ? ToInteger(index). + ASSIGN_RETURN_FAILURE_ON_EXCEPTION(isolate, index, + Object::ToInteger(isolate, index)); + double const n = index->Number(); + RETURN_RESULT_OR_FAILURE( - isolate, - JSSegmentIterator::Create( - isolate, segmenter_holder->icu_break_iterator().raw()->clone(), - segmenter_holder->granularity(), text)); + isolate, JSSegments::Containing(isolate, segments, static_cast(n))); +} + +// ecma402 #sec-%segmentsprototype%-@@iterator +BUILTIN(SegmentsPrototypeIterator) { + const char* const method = "%SegmentIsPrototype%[@@iterator]"; + HandleScope scope(isolate); + CHECK_RECEIVER(JSSegments, segments, method); + RETURN_RESULT_OR_FAILURE( + isolate, JSSegments::CreateSegmentIterator(isolate, segments)); } BUILTIN(V8BreakIteratorConstructor) { diff --git a/src/compiler/code-assembler.h b/src/compiler/code-assembler.h index 42690c1b30..bf7e2139fe 100644 --- a/src/compiler/code-assembler.h +++ b/src/compiler/code-assembler.h @@ -62,6 +62,7 @@ class JSRegExpStringIterator; class JSRelativeTimeFormat; class JSSegmentIterator; class JSSegmenter; +class JSSegments; class JSV8BreakIterator; class JSWeakCollection; class JSFinalizationRegistry; diff --git a/src/compiler/types.cc b/src/compiler/types.cc index 3c6bcaba92..bda9b8253a 100644 --- a/src/compiler/types.cc +++ b/src/compiler/types.cc @@ -218,6 +218,7 @@ Type::bitset BitsetType::Lub(const MapRefLike& map) { case JS_RELATIVE_TIME_FORMAT_TYPE: case JS_SEGMENT_ITERATOR_TYPE: case JS_SEGMENTER_TYPE: + case JS_SEGMENTS_TYPE: #endif // V8_INTL_SUPPORT case JS_CONTEXT_EXTENSION_OBJECT_TYPE: case JS_GENERATOR_OBJECT_TYPE: diff --git a/src/diagnostics/objects-debug.cc b/src/diagnostics/objects-debug.cc index a212af94c3..7c710e8fd2 100644 --- a/src/diagnostics/objects-debug.cc +++ b/src/diagnostics/objects-debug.cc @@ -54,6 +54,7 @@ #include "src/objects/js-relative-time-format-inl.h" #include "src/objects/js-segment-iterator-inl.h" #include "src/objects/js-segmenter-inl.h" +#include "src/objects/js-segments-inl.h" #endif // V8_INTL_SUPPORT #include "src/objects/js-weak-refs-inl.h" #include "src/objects/literal-objects-inl.h" diff --git a/src/diagnostics/objects-printer.cc b/src/diagnostics/objects-printer.cc index 4e3c3f0cf3..ca9f0dcf32 100644 --- a/src/diagnostics/objects-printer.cc +++ b/src/diagnostics/objects-printer.cc @@ -50,6 +50,7 @@ #include "src/objects/js-relative-time-format-inl.h" #include "src/objects/js-segment-iterator-inl.h" #include "src/objects/js-segmenter-inl.h" +#include "src/objects/js-segments-inl.h" #endif // V8_INTL_SUPPORT #include "src/compiler/node.h" #include "src/objects/js-weak-refs-inl.h" @@ -2107,18 +2108,25 @@ void JSRelativeTimeFormat::JSRelativeTimeFormatPrint( void JSSegmentIterator::JSSegmentIteratorPrint(std::ostream& os) { // NOLINT JSObjectPrintHeader(os, *this, "JSSegmentIterator"); os << "\n - icu break iterator: " << Brief(icu_break_iterator()); - os << "\n - unicode string: " << Brief(unicode_string()); - os << "\n - granularity: " << GranularityAsString(); + os << "\n - granularity: " << GranularityAsString(GetIsolate()); os << "\n"; } void JSSegmenter::JSSegmenterPrint(std::ostream& os) { // NOLINT JSObjectPrintHeader(os, *this, "JSSegmenter"); os << "\n - locale: " << Brief(locale()); - os << "\n - granularity: " << GranularityAsString(); + os << "\n - granularity: " << GranularityAsString(GetIsolate()); os << "\n - icu break iterator: " << Brief(icu_break_iterator()); JSObjectPrintBody(os, *this); } + +void JSSegments::JSSegmentsPrint(std::ostream& os) { // NOLINT + JSObjectPrintHeader(os, *this, "JSSegments"); + os << "\n - icu break iterator: " << Brief(icu_break_iterator()); + os << "\n - unicode string: " << Brief(unicode_string()); + os << "\n - granularity: " << GranularityAsString(GetIsolate()); + JSObjectPrintBody(os, *this); +} #endif // V8_INTL_SUPPORT namespace { diff --git a/src/init/bootstrapper.cc b/src/init/bootstrapper.cc index e1514dfad9..96c764de6b 100644 --- a/src/init/bootstrapper.cc +++ b/src/init/bootstrapper.cc @@ -53,6 +53,7 @@ #include "src/objects/js-relative-time-format.h" #include "src/objects/js-segment-iterator.h" #include "src/objects/js-segmenter.h" +#include "src/objects/js-segments.h" #endif // V8_INTL_SUPPORT #include "src/objects/js-weak-refs.h" #include "src/objects/ordered-hash-table.h" @@ -4315,6 +4316,12 @@ void Genesis::InitializeGlobal_harmony_intl_segmenter() { Handle prototype( JSObject::cast(segmenter_fun->instance_prototype()), isolate()); + // #sec-intl.segmenter.prototype-@@tostringtag + // + // Intl.Segmenter.prototype [ @@toStringTag ] + // + // The initial value of the @@toStringTag property is the String value + // "Intl.Segmenter". InstallToStringTag(isolate(), prototype, "Intl.Segmenter"); SimpleInstallFunction(isolate(), prototype, "resolvedOptions", @@ -4325,6 +4332,32 @@ void Genesis::InitializeGlobal_harmony_intl_segmenter() { Builtins::kSegmenterPrototypeSegment, 1, false); } + { + // Setup %SegmentsPrototype%. + Handle prototype = factory()->NewJSObject( + isolate()->object_function(), AllocationType::kOld); + + Handle name_string = + Name::ToFunctionName(isolate(), isolate()->factory()->Segments_string()) + .ToHandleChecked(); + Handle segments_fun = CreateFunction( + isolate(), name_string, JS_SEGMENTS_TYPE, JSSegments::kHeaderSize, 0, + prototype, Builtins::kIllegal); + segments_fun->shared().set_native(false); + segments_fun->shared().set_length(0); + segments_fun->shared().DontAdaptArguments(); + + SimpleInstallFunction(isolate(), prototype, "containing", + Builtins::kSegmentsPrototypeContaining, 1, false); + + InstallFunctionAtSymbol( + isolate_, prototype, factory()->iterator_symbol(), "[Symbol.iterator]", + Builtins::kSegmentsPrototypeIterator, 0, true, DONT_ENUM); + + Handle segments_map(segments_fun->initial_map(), isolate()); + native_context()->set_intl_segments_map(*segments_map); + } + { // Setup %SegmentIteratorPrototype%. Handle iterator_prototype( @@ -4334,26 +4367,17 @@ void Genesis::InitializeGlobal_harmony_intl_segmenter() { isolate()->object_function(), AllocationType::kOld); JSObject::ForceSetPrototype(prototype, iterator_prototype); - InstallToStringTag(isolate(), prototype, - factory()->SegmentIterator_string()); + // #sec-%segmentiteratorprototype%.@@tostringtag + // + // %SegmentIteratorPrototype% [ @@toStringTag ] + // + // The initial value of the @@toStringTag property is the String value + // "Segmenter String Iterator". + InstallToStringTag(isolate(), prototype, "Segmenter String Iterator"); SimpleInstallFunction(isolate(), prototype, "next", Builtins::kSegmentIteratorPrototypeNext, 0, false); - SimpleInstallFunction(isolate(), prototype, "following", - Builtins::kSegmentIteratorPrototypeFollowing, 0, - false); - - SimpleInstallFunction(isolate(), prototype, "preceding", - Builtins::kSegmentIteratorPrototypePreceding, 0, - false); - - SimpleInstallGetter(isolate(), prototype, factory()->index_string(), - Builtins::kSegmentIteratorPrototypeIndex, false); - - SimpleInstallGetter(isolate(), prototype, factory()->breakType_string(), - Builtins::kSegmentIteratorPrototypeBreakType, false); - // Setup SegmentIterator constructor. Handle name_string = Name::ToFunctionName(isolate(), diff --git a/src/init/heap-symbols.h b/src/init/heap-symbols.h index d09392cf3c..76f59ae5ba 100644 --- a/src/init/heap-symbols.h +++ b/src/init/heap-symbols.h @@ -54,6 +54,7 @@ V(_, ignorePunctuation_string, "ignorePunctuation") \ V(_, Invalid_Date_string, "Invalid Date") \ V(_, integer_string, "integer") \ + V(_, isWordLike_string, "isWordLike") \ V(_, kana_string, "kana") \ V(_, language_string, "language") \ V(_, letter_string, "letter") \ @@ -88,6 +89,7 @@ V(_, second_string, "second") \ V(_, segment_string, "segment") \ V(_, SegmentIterator_string, "Segment Iterator") \ + V(_, Segments_string, "Segments") \ V(_, sensitivity_string, "sensitivity") \ V(_, sep_string, "sep") \ V(_, shared_string, "shared") \ diff --git a/src/objects/class-definitions-tq-deps-inl.h b/src/objects/class-definitions-tq-deps-inl.h index dafba941ea..ccfd8c3aa1 100644 --- a/src/objects/class-definitions-tq-deps-inl.h +++ b/src/objects/class-definitions-tq-deps-inl.h @@ -38,6 +38,7 @@ #include "src/objects/js-relative-time-format-inl.h" #include "src/objects/js-segment-iterator-inl.h" #include "src/objects/js-segmenter-inl.h" +#include "src/objects/js-segments-inl.h" #endif #endif // V8_OBJECTS_CLASS_DEFINITIONS_TQ_DEPS_INL_H_ diff --git a/src/objects/contexts.h b/src/objects/contexts.h index 484257af90..4821dea1db 100644 --- a/src/objects/contexts.h +++ b/src/objects/contexts.h @@ -153,6 +153,7 @@ enum ContextLookupFlags { V(INTL_RELATIVE_TIME_FORMAT_FUNCTION_INDEX, JSFunction, \ intl_relative_time_format_function) \ V(INTL_SEGMENTER_FUNCTION_INDEX, JSFunction, intl_segmenter_function) \ + V(INTL_SEGMENTS_MAP_INDEX, Map, intl_segments_map) \ V(INTL_SEGMENT_ITERATOR_MAP_INDEX, Map, intl_segment_iterator_map) \ V(ITERATOR_RESULT_MAP_INDEX, Map, iterator_result_map) \ V(JS_ARRAY_PACKED_SMI_ELEMENTS_MAP_INDEX, Map, \ diff --git a/src/objects/intl-objects.tq b/src/objects/intl-objects.tq index 41db48dae3..cbfcc360bb 100644 --- a/src/objects/intl-objects.tq +++ b/src/objects/intl-objects.tq @@ -14,6 +14,7 @@ #include 'src/objects/js-relative-time-format.h' #include 'src/objects/js-segment-iterator.h' #include 'src/objects/js-segmenter.h' +#include 'src/objects/js-segments.h' type DateTimeStyle extends int32 constexpr 'JSDateTimeFormat::DateTimeStyle'; type HourCycle extends int32 constexpr 'JSDateTimeFormat::HourCycle'; @@ -115,15 +116,24 @@ extern class JSSegmenter extends JSObject { flags: SmiTagged; } +bitfield struct JSSegmentsFlags extends uint31 { + granularity: JSSegmenterGranularity: 2 bit; +} + +@generateCppClass +extern class JSSegments extends JSObject { + icu_break_iterator: Foreign; // Managed + unicode_string: Foreign; // Managed + flags: SmiTagged; +} + bitfield struct JSSegmentIteratorFlags extends uint31 { granularity: JSSegmenterGranularity: 2 bit; - break_type_set: bool: 1 bit; } @generateCppClass extern class JSSegmentIterator extends JSObject { icu_break_iterator: Foreign; // Managed - unicode_string: Foreign; // Managed flags: SmiTagged; } diff --git a/src/objects/js-function.cc b/src/objects/js-function.cc index 49d5b8a87e..9897fcd332 100644 --- a/src/objects/js-function.cc +++ b/src/objects/js-function.cc @@ -668,6 +668,7 @@ bool CanSubclassHaveInobjectProperties(InstanceType instance_type) { case JS_RELATIVE_TIME_FORMAT_TYPE: case JS_SEGMENT_ITERATOR_TYPE: case JS_SEGMENTER_TYPE: + case JS_SEGMENTS_TYPE: case JS_V8_BREAK_ITERATOR_TYPE: #endif case JS_ASYNC_FUNCTION_OBJECT_TYPE: diff --git a/src/objects/js-objects.cc b/src/objects/js-objects.cc index dd1e5a9b91..7f5378e6c1 100644 --- a/src/objects/js-objects.cc +++ b/src/objects/js-objects.cc @@ -53,6 +53,7 @@ #include "src/objects/js-relative-time-format.h" #include "src/objects/js-segment-iterator.h" #include "src/objects/js-segmenter.h" +#include "src/objects/js-segments.h" #endif // V8_INTL_SUPPORT #include "src/objects/js-weak-refs.h" #include "src/objects/map-inl.h" @@ -2173,6 +2174,8 @@ int JSObject::GetHeaderSize(InstanceType type, return JSSegmentIterator::kHeaderSize; case JS_SEGMENTER_TYPE: return JSSegmenter::kHeaderSize; + case JS_SEGMENTS_TYPE: + return JSSegments::kHeaderSize; #endif // V8_INTL_SUPPORT case WASM_GLOBAL_OBJECT_TYPE: return WasmGlobalObject::kHeaderSize; diff --git a/src/objects/js-segment-iterator-inl.h b/src/objects/js-segment-iterator-inl.h index 2e32cc5d87..5a010a186f 100644 --- a/src/objects/js-segment-iterator-inl.h +++ b/src/objects/js-segment-iterator-inl.h @@ -1,14 +1,13 @@ -// Copyright 2018 the V8 project authors. All rights reserved. +// Copyright 2020 the V8 project authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. +#ifndef V8_OBJECTS_JS_SEGMENT_ITERATOR_INL_H_ +#define V8_OBJECTS_JS_SEGMENT_ITERATOR_INL_H_ #ifndef V8_INTL_SUPPORT #error Internationalization is expected to be enabled. #endif // V8_INTL_SUPPORT -#ifndef V8_OBJECTS_JS_SEGMENT_ITERATOR_INL_H_ -#define V8_OBJECTS_JS_SEGMENT_ITERATOR_INL_H_ - #include "src/objects/js-segment-iterator.h" #include "src/objects/objects-inl.h" @@ -23,11 +22,6 @@ TQ_OBJECT_CONSTRUCTORS_IMPL(JSSegmentIterator) // Base segment iterator accessors. ACCESSORS(JSSegmentIterator, icu_break_iterator, Managed, kIcuBreakIteratorOffset) -ACCESSORS(JSSegmentIterator, unicode_string, Managed, - kUnicodeStringOffset) - -BIT_FIELD_ACCESSORS(JSSegmentIterator, flags, is_break_type_set, - JSSegmentIterator::BreakTypeSetBit) inline void JSSegmentIterator::set_granularity( JSSegmenter::Granularity granularity) { diff --git a/src/objects/js-segment-iterator.cc b/src/objects/js-segment-iterator.cc index d893f3705f..29a7c52225 100644 --- a/src/objects/js-segment-iterator.cc +++ b/src/objects/js-segment-iterator.cc @@ -16,6 +16,7 @@ #include "src/heap/factory.h" #include "src/objects/intl-objects.h" #include "src/objects/js-segment-iterator-inl.h" +#include "src/objects/js-segments.h" #include "src/objects/managed.h" #include "src/objects/objects-inl.h" #include "unicode/brkiter.h" @@ -23,36 +24,22 @@ namespace v8 { namespace internal { -MaybeHandle JSSegmentIterator::GetSegment(Isolate* isolate, - int32_t start, - int32_t end) const { - return Intl::ToString(isolate, *(unicode_string().raw()), start, end); -} - -Handle JSSegmentIterator::GranularityAsString() const { - switch (granularity()) { - case JSSegmenter::Granularity::GRAPHEME: - return GetReadOnlyRoots().grapheme_string_handle(); - case JSSegmenter::Granularity::WORD: - return GetReadOnlyRoots().word_string_handle(); - case JSSegmenter::Granularity::SENTENCE: - return GetReadOnlyRoots().sentence_string_handle(); - } - UNREACHABLE(); +Handle JSSegmentIterator::GranularityAsString(Isolate* isolate) const { + return JSSegmenter::GetGranularityString(isolate, granularity()); } +// ecma402 #sec-createsegmentiterator MaybeHandle JSSegmentIterator::Create( Isolate* isolate, icu::BreakIterator* break_iterator, - JSSegmenter::Granularity granularity, Handle text) { - CHECK_NOT_NULL(break_iterator); - // 1. Let iterator be ObjectCreate(%SegmentIteratorPrototype%). + JSSegmenter::Granularity granularity) { + DCHECK_NOT_NULL(break_iterator); Handle map = Handle( isolate->native_context()->intl_segment_iterator_map(), isolate); + // 5. Set iterator.[[IteratedStringNextSegmentCodeUnitIndex]] to 0. + break_iterator->first(); Handle> managed_break_iterator = Managed::FromRawPtr(isolate, 0, break_iterator); - Handle> unicode_string = - Intl::SetTextToBreakIterator(isolate, text, break_iterator); // Now all properties are ready, so we can allocate the result object. Handle result = isolate->factory()->NewJSObjectFromMap(map); @@ -62,230 +49,47 @@ MaybeHandle JSSegmentIterator::Create( segment_iterator->set_flags(0); segment_iterator->set_granularity(granularity); - // 2. Let iterator.[[SegmentIteratorSegmenter]] be segmenter. segment_iterator->set_icu_break_iterator(*managed_break_iterator); - // 3. Let iterator.[[SegmentIteratorString]] be string. - segment_iterator->set_unicode_string(*unicode_string); - - // 4. Let iterator.[[SegmentIteratorIndex]] be 0. - // step 4 is stored inside break_iterator. - - // 5. Let iterator.[[SegmentIteratorBreakType]] be undefined. - segment_iterator->set_is_break_type_set(false); - return segment_iterator; } -// ecma402 #sec-segment-iterator-prototype-breakType -Handle JSSegmentIterator::BreakType() const { - if (!is_break_type_set()) { - return GetReadOnlyRoots().undefined_value_handle(); - } - icu::BreakIterator* break_iterator = icu_break_iterator().raw(); - int32_t rule_status = break_iterator->getRuleStatus(); - switch (granularity()) { - case JSSegmenter::Granularity::GRAPHEME: - return GetReadOnlyRoots().undefined_value_handle(); - case JSSegmenter::Granularity::WORD: - if (rule_status >= UBRK_WORD_NONE && rule_status < UBRK_WORD_NONE_LIMIT) { - // "words" that do not fit into any of other categories. Includes spaces - // and most punctuation. - return GetReadOnlyRoots().none_string_handle(); - } - if ((rule_status >= UBRK_WORD_NUMBER && - rule_status < UBRK_WORD_NUMBER_LIMIT) || - (rule_status >= UBRK_WORD_LETTER && - rule_status < UBRK_WORD_LETTER_LIMIT) || - (rule_status >= UBRK_WORD_KANA && - rule_status < UBRK_WORD_KANA_LIMIT) || - (rule_status >= UBRK_WORD_IDEO && - rule_status < UBRK_WORD_IDEO_LIMIT)) { - // words that appear to be numbers, letters, kana characters, - // ideographic characters, etc - return GetReadOnlyRoots().word_string_handle(); - } - return GetReadOnlyRoots().undefined_value_handle(); - case JSSegmenter::Granularity::SENTENCE: - if (rule_status >= UBRK_SENTENCE_TERM && - rule_status < UBRK_SENTENCE_TERM_LIMIT) { - // sentences ending with a sentence terminator ('.', '?', '!', etc.) - // character, possibly followed by a hard separator (CR, LF, PS, etc.) - return GetReadOnlyRoots().term_string_handle(); - } - if ((rule_status >= UBRK_SENTENCE_SEP && - rule_status < UBRK_SENTENCE_SEP_LIMIT)) { - // sentences that do not contain an ending sentence terminator ('.', - // '?', '!', etc.) character, but are ended only by a hard separator - // (CR, LF, PS, etc.) hard, or mandatory line breaks - return GetReadOnlyRoots().sep_string_handle(); - } - return GetReadOnlyRoots().undefined_value_handle(); - } - UNREACHABLE(); -} - -// ecma402 #sec-segment-iterator-prototype-index -Handle JSSegmentIterator::Index( - Isolate* isolate, Handle segment_iterator) { - icu::BreakIterator* icu_break_iterator = - segment_iterator->icu_break_iterator().raw(); - CHECK_NOT_NULL(icu_break_iterator); - return isolate->factory()->NewNumberFromInt(icu_break_iterator->current()); -} - -// ecma402 #sec-segment-iterator-prototype-next +// ecma402 #sec-%segmentiteratorprototype%.next MaybeHandle JSSegmentIterator::Next( Isolate* isolate, Handle segment_iterator) { Factory* factory = isolate->factory(); icu::BreakIterator* icu_break_iterator = segment_iterator->icu_break_iterator().raw(); - // 3. Let _previousIndex be iterator.[[SegmentIteratorIndex]]. - int32_t prev = icu_break_iterator->current(); - // 4. Let done be AdvanceSegmentIterator(iterator, forwards). - int32_t index = icu_break_iterator->next(); - segment_iterator->set_is_break_type_set(true); - if (index == icu::BreakIterator::DONE) { - // 5. If done is true, return CreateIterResultObject(undefined, true). + // 5. Let startIndex be iterator.[[IteratedStringNextSegmentCodeUnitIndex]]. + int32_t start_index = icu_break_iterator->current(); + // 6. Let endIndex be ! FindBoundary(segmenter, string, startIndex, after). + int32_t end_index = icu_break_iterator->next(); + + // 7. If endIndex is not finite, then + if (end_index == icu::BreakIterator::DONE) { + // a. Return ! CreateIterResultObject(undefined, true). return factory->NewJSIteratorResult(isolate->factory()->undefined_value(), true); } - // 6. Let newIndex be iterator.[[SegmentIteratorIndex]]. - Handle new_index = factory->NewNumberFromInt(index); - // 8. Let segment be the substring of string from previousIndex to - // newIndex, inclusive of previousIndex and exclusive of newIndex. - Handle segment; - ASSIGN_RETURN_ON_EXCEPTION(isolate, segment, - segment_iterator->GetSegment(isolate, prev, index), - JSReceiver); + // 8. Set iterator.[[IteratedStringNextSegmentCodeUnitIndex]] to endIndex. - // 9. Let breakType be iterator.[[SegmentIteratorBreakType]]. - Handle break_type = segment_iterator->BreakType(); + // 9. Let segmentData be ! CreateSegmentDataObject(segmenter, string, + // startIndex, endIndex). - // 10. Let result be ! ObjectCreate(%ObjectPrototype%). - Handle result = factory->NewJSObject(isolate->object_function()); + icu::UnicodeString string; + icu_break_iterator->getText().getText(string); - // 11. Perform ! CreateDataProperty(result "segment", segment). - CHECK(JSReceiver::CreateDataProperty(isolate, result, - factory->segment_string(), segment, - Just(kDontThrow)) - .FromJust()); + Handle segment_data; + ASSIGN_RETURN_ON_EXCEPTION( + isolate, segment_data, + JSSegments::CreateSegmentDataObject( + isolate, segment_iterator->granularity(), icu_break_iterator, string, + start_index, end_index), + JSReceiver); - // 12. Perform ! CreateDataProperty(result, "breakType", breakType). - CHECK(JSReceiver::CreateDataProperty(isolate, result, - factory->breakType_string(), break_type, - Just(kDontThrow)) - .FromJust()); - - // 13. Perform ! CreateDataProperty(result, "index", newIndex). - CHECK(JSReceiver::CreateDataProperty(isolate, result, factory->index_string(), - new_index, Just(kDontThrow)) - .FromJust()); - - // 14. Return CreateIterResultObject(result, false). - return factory->NewJSIteratorResult(result, false); -} - -// ecma402 #sec-segment-iterator-prototype-following -Maybe JSSegmentIterator::Following( - Isolate* isolate, Handle segment_iterator, - Handle from_obj) { - Factory* factory = isolate->factory(); - icu::BreakIterator* icu_break_iterator = - segment_iterator->icu_break_iterator().raw(); - // 3. If from is not undefined, - if (!from_obj->IsUndefined()) { - // a. Let from be ? ToIndex(from). - uint32_t from; - Handle index; - ASSIGN_RETURN_ON_EXCEPTION_VALUE( - isolate, index, - Object::ToIndex(isolate, from_obj, MessageTemplate::kInvalidIndex), - Nothing()); - if (!index->ToArrayIndex(&from)) { - THROW_NEW_ERROR_RETURN_VALUE( - isolate, - NewRangeError(MessageTemplate::kParameterOfFunctionOutOfRange, - factory->from_string(), - factory->NewStringFromStaticChars("following"), index), - Nothing()); - } - // b. Let length be the length of iterator.[[SegmentIteratorString]]. - uint32_t length = - static_cast(icu_break_iterator->getText().getLength()); - - // c. If from ≥ length, throw a RangeError exception. - if (from >= length) { - THROW_NEW_ERROR_RETURN_VALUE( - isolate, - NewRangeError(MessageTemplate::kParameterOfFunctionOutOfRange, - factory->from_string(), - factory->NewStringFromStaticChars("following"), - from_obj), - Nothing()); - } - - // d. Let iterator.[[SegmentIteratorPosition]] be from. - segment_iterator->set_is_break_type_set(true); - icu_break_iterator->following(from); - return Just(false); - } - // 4. return AdvanceSegmentIterator(iterator, forward). - // 4. .... or if direction is backwards and position is 0, return true. - // 4. If direction is forwards and position is the length of string ... return - // true. - segment_iterator->set_is_break_type_set(true); - return Just(icu_break_iterator->next() == icu::BreakIterator::DONE); -} - -// ecma402 #sec-segment-iterator-prototype-preceding -Maybe JSSegmentIterator::Preceding( - Isolate* isolate, Handle segment_iterator, - Handle from_obj) { - Factory* factory = isolate->factory(); - icu::BreakIterator* icu_break_iterator = - segment_iterator->icu_break_iterator().raw(); - // 3. If from is not undefined, - if (!from_obj->IsUndefined()) { - // a. Let from be ? ToIndex(from). - uint32_t from; - Handle index; - ASSIGN_RETURN_ON_EXCEPTION_VALUE( - isolate, index, - Object::ToIndex(isolate, from_obj, MessageTemplate::kInvalidIndex), - Nothing()); - - if (!index->ToArrayIndex(&from)) { - THROW_NEW_ERROR_RETURN_VALUE( - isolate, - NewRangeError(MessageTemplate::kParameterOfFunctionOutOfRange, - factory->from_string(), - factory->NewStringFromStaticChars("preceding"), index), - Nothing()); - } - // b. Let length be the length of iterator.[[SegmentIteratorString]]. - uint32_t length = - static_cast(icu_break_iterator->getText().getLength()); - // c. If from > length or from = 0, throw a RangeError exception. - if (from > length || from == 0) { - THROW_NEW_ERROR_RETURN_VALUE( - isolate, - NewRangeError(MessageTemplate::kParameterOfFunctionOutOfRange, - factory->from_string(), - factory->NewStringFromStaticChars("preceding"), - from_obj), - Nothing()); - } - // d. Let iterator.[[SegmentIteratorIndex]] be from. - segment_iterator->set_is_break_type_set(true); - icu_break_iterator->preceding(from); - return Just(false); - } - // 4. return AdvanceSegmentIterator(iterator, backwards). - // 4. .... or if direction is backwards and position is 0, return true. - segment_iterator->set_is_break_type_set(true); - return Just(icu_break_iterator->previous() == icu::BreakIterator::DONE); + // 10. Return ! CreateIterResultObject(segmentData, false). + return factory->NewJSIteratorResult(segment_data, false); } } // namespace internal diff --git a/src/objects/js-segment-iterator.h b/src/objects/js-segment-iterator.h index f1310233e1..1e70298797 100644 --- a/src/objects/js-segment-iterator.h +++ b/src/objects/js-segment-iterator.h @@ -1,14 +1,13 @@ -// Copyright 2018 the V8 project authors. All rights reserved. +// Copyright 2020 the V8 project authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. +#ifndef V8_OBJECTS_JS_SEGMENT_ITERATOR_H_ +#define V8_OBJECTS_JS_SEGMENT_ITERATOR_H_ #ifndef V8_INTL_SUPPORT #error Internationalization is expected to be enabled. #endif // V8_INTL_SUPPORT -#ifndef V8_OBJECTS_JS_SEGMENT_ITERATOR_H_ -#define V8_OBJECTS_JS_SEGMENT_ITERATOR_H_ - #include "src/base/bit-field.h" #include "src/execution/isolate.h" #include "src/heap/factory.h" @@ -34,40 +33,16 @@ class JSSegmentIterator // ecma402 #sec-CreateSegmentIterator V8_WARN_UNUSED_RESULT static MaybeHandle Create( Isolate* isolate, icu::BreakIterator* icu_break_iterator, - JSSegmenter::Granularity granularity, Handle string); + JSSegmenter::Granularity granularity); // ecma402 #sec-segment-iterator-prototype-next V8_WARN_UNUSED_RESULT static MaybeHandle Next( Isolate* isolate, Handle segment_iterator_holder); - // ecma402 #sec-segment-iterator-prototype-following - static Maybe Following( - Isolate* isolate, Handle segment_iterator_holder, - Handle from); - - // ecma402 #sec-segment-iterator-prototype-preceding - static Maybe Preceding( - Isolate* isolate, Handle segment_iterator_holder, - Handle from); - - // ecma402 #sec-segment-iterator-prototype-index - static Handle Index( - Isolate* isolate, Handle segment_iterator_holder); - - Handle GranularityAsString() const; - - DECL_BOOLEAN_ACCESSORS(is_break_type_set) - - // ecma402 #sec-segment-iterator-prototype-breakType - Handle BreakType() const; - - V8_WARN_UNUSED_RESULT MaybeHandle GetSegment(Isolate* isolate, - int32_t start, - int32_t end) const; + Handle GranularityAsString(Isolate* isolate) const; // SegmentIterator accessors. DECL_ACCESSORS(icu_break_iterator, Managed) - DECL_ACCESSORS(unicode_string, Managed) DECL_PRINTER(JSSegmentIterator) diff --git a/src/objects/js-segmenter-inl.h b/src/objects/js-segmenter-inl.h index ebf4002e70..98bc2e863b 100644 --- a/src/objects/js-segmenter-inl.h +++ b/src/objects/js-segmenter-inl.h @@ -1,14 +1,13 @@ -// Copyright 2018 the V8 project authors. All rights reserved. +// Copyright 2020 the V8 project authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. +#ifndef V8_OBJECTS_JS_SEGMENTER_INL_H_ +#define V8_OBJECTS_JS_SEGMENTER_INL_H_ #ifndef V8_INTL_SUPPORT #error Internationalization is expected to be enabled. #endif // V8_INTL_SUPPORT -#ifndef V8_OBJECTS_JS_SEGMENTER_INL_H_ -#define V8_OBJECTS_JS_SEGMENTER_INL_H_ - #include "src/objects/js-segmenter.h" #include "src/objects/objects-inl.h" diff --git a/src/objects/js-segmenter.cc b/src/objects/js-segmenter.cc index f12bac2031..aec804102f 100644 --- a/src/objects/js-segmenter.cc +++ b/src/objects/js-segmenter.cc @@ -26,36 +26,37 @@ namespace internal { MaybeHandle JSSegmenter::New(Isolate* isolate, Handle map, Handle locales, Handle input_options) { - // 3. Let requestedLocales be ? CanonicalizeLocaleList(locales). + // 4. Let requestedLocales be ? CanonicalizeLocaleList(locales). Maybe> maybe_requested_locales = Intl::CanonicalizeLocaleList(isolate, locales); MAYBE_RETURN(maybe_requested_locales, Handle()); std::vector requested_locales = maybe_requested_locales.FromJust(); - // 11. If options is undefined, then + // 5. If options is undefined, then Handle options; if (input_options->IsUndefined(isolate)) { - // 11. a. Let options be ObjectCreate(null). + // a. Let options be ObjectCreate(null). options = isolate->factory()->NewJSObjectWithNullProto(); - // 12. Else - } else { - // 23. a. Let options be ? ToObject(options). + } else { // 6. Else + // a. Let options be ? ToObject(options). ASSIGN_RETURN_ON_EXCEPTION(isolate, options, Object::ToObject(isolate, input_options), JSSegmenter); } - // 4. Let opt be a new Record. - // 5. Let matcher be ? GetOption(options, "localeMatcher", "string", + // 7. Let opt be a new Record. + // 8. Let matcher be ? GetOption(options, "localeMatcher", "string", // « "lookup", "best fit" », "best fit"). - // 6. Set opt.[[localeMatcher]] to matcher. + // 9. Set opt.[[localeMatcher]] to matcher. Maybe maybe_locale_matcher = Intl::GetLocaleMatcher(isolate, options, "Intl.Segmenter"); MAYBE_RETURN(maybe_locale_matcher, MaybeHandle()); Intl::MatcherOption matcher = maybe_locale_matcher.FromJust(); - // 9. Let r be ResolveLocale(%Segmenter%.[[AvailableLocales]], + // 10. Let localeData be %Segmenter%.[[LocaleData]]. + + // 11. Let r be ResolveLocale(%Segmenter%.[[AvailableLocales]], // requestedLocales, opt, %Segmenter%.[[RelevantExtensionKeys]]). Maybe maybe_resolve_locale = Intl::ResolveLocale(isolate, JSSegmenter::GetAvailableLocales(), @@ -66,6 +67,7 @@ MaybeHandle JSSegmenter::New(Isolate* isolate, Handle map, } Intl::ResolvedLocale r = maybe_resolve_locale.FromJust(); + // 12. Set segmenter.[[Locale]] to the value of r.[[locale]]. Handle locale_str = isolate->factory()->NewStringFromAsciiChecked(r.locale.c_str()); @@ -100,32 +102,34 @@ MaybeHandle JSSegmenter::New(Isolate* isolate, Handle map, break; } - CHECK(U_SUCCESS(status)); - CHECK_NOT_NULL(icu_break_iterator.get()); + DCHECK(U_SUCCESS(status)); + DCHECK_NOT_NULL(icu_break_iterator.get()); Handle> managed_break_iterator = Managed::FromUniquePtr(isolate, 0, std::move(icu_break_iterator)); // Now all properties are ready, so we can allocate the result object. - Handle segmenter_holder = Handle::cast( + Handle segmenter = Handle::cast( isolate->factory()->NewFastOrSlowJSObjectFromMap(map)); DisallowHeapAllocation no_gc; - segmenter_holder->set_flags(0); + segmenter->set_flags(0); - // 10. Set segmenter.[[Locale]] to the value of r.[[Locale]]. - segmenter_holder->set_locale(*locale_str); + // 12. Set segmenter.[[Locale]] to the value of r.[[Locale]]. + segmenter->set_locale(*locale_str); // 14. Set segmenter.[[SegmenterGranularity]] to granularity. - segmenter_holder->set_granularity(granularity_enum); + segmenter->set_granularity(granularity_enum); - segmenter_holder->set_icu_break_iterator(*managed_break_iterator); - return segmenter_holder; + segmenter->set_icu_break_iterator(*managed_break_iterator); + + // 15. Return segmenter. + return segmenter; } // ecma402 #sec-Intl.Segmenter.prototype.resolvedOptions -Handle JSSegmenter::ResolvedOptions( - Isolate* isolate, Handle segmenter_holder) { +Handle JSSegmenter::ResolvedOptions(Isolate* isolate, + Handle segmenter) { Factory* factory = isolate->factory(); // 3. Let options be ! ObjectCreate(%ObjectPrototype%). Handle result = factory->NewJSObject(isolate->object_function()); @@ -141,23 +145,29 @@ Handle JSSegmenter::ResolvedOptions( // [[Locale]] "locale" // [[SegmenterGranularity]] "granularity" - Handle locale(segmenter_holder->locale(), isolate); + Handle locale(segmenter->locale(), isolate); JSObject::AddProperty(isolate, result, factory->locale_string(), locale, NONE); JSObject::AddProperty(isolate, result, factory->granularity_string(), - segmenter_holder->GranularityAsString(), NONE); + segmenter->GranularityAsString(isolate), NONE); // 5. Return options. return result; } -Handle JSSegmenter::GranularityAsString() const { - switch (granularity()) { +Handle JSSegmenter::GranularityAsString(Isolate* isolate) const { + return GetGranularityString(isolate, granularity()); +} + +Handle JSSegmenter::GetGranularityString(Isolate* isolate, + Granularity granularity) { + Factory* factory = isolate->factory(); + switch (granularity) { case Granularity::GRAPHEME: - return GetReadOnlyRoots().grapheme_string_handle(); + return factory->grapheme_string(); case Granularity::WORD: - return GetReadOnlyRoots().word_string_handle(); + return factory->word_string(); case Granularity::SENTENCE: - return GetReadOnlyRoots().sentence_string_handle(); + return factory->sentence_string(); } UNREACHABLE(); } diff --git a/src/objects/js-segmenter.h b/src/objects/js-segmenter.h index 39b32480ba..23890d1770 100644 --- a/src/objects/js-segmenter.h +++ b/src/objects/js-segmenter.h @@ -1,14 +1,13 @@ -// Copyright 2018 the V8 project authors. All rights reserved. +// Copyright 2020 the V8 project authors. All rights reserved. // Use of this source code is governed by a BSD-style license that can be // found in the LICENSE file. +#ifndef V8_OBJECTS_JS_SEGMENTER_H_ +#define V8_OBJECTS_JS_SEGMENTER_H_ #ifndef V8_INTL_SUPPORT #error Internationalization is expected to be enabled. #endif // V8_INTL_SUPPORT -#ifndef V8_OBJECTS_JS_SEGMENTER_H_ -#define V8_OBJECTS_JS_SEGMENTER_H_ - #include #include @@ -42,7 +41,7 @@ class JSSegmenter : public TorqueGeneratedJSSegmenter { V8_EXPORT_PRIVATE static const std::set& GetAvailableLocales(); - Handle GranularityAsString() const; + Handle GranularityAsString(Isolate* isolate) const; // Segmenter accessors. DECL_ACCESSORS(icu_break_iterator, Managed) @@ -58,6 +57,9 @@ class JSSegmenter : public TorqueGeneratedJSSegmenter { inline void set_granularity(Granularity granularity); inline Granularity granularity() const; + Handle static GetGranularityString(Isolate* isolate, + Granularity granularity); + // Bit positions in |flags|. DEFINE_TORQUE_GENERATED_JS_SEGMENTER_FLAGS() diff --git a/src/objects/js-segments-inl.h b/src/objects/js-segments-inl.h new file mode 100644 index 0000000000..ceabd6741d --- /dev/null +++ b/src/objects/js-segments-inl.h @@ -0,0 +1,44 @@ +// Copyright 2020 the V8 project authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +#ifndef V8_OBJECTS_JS_SEGMENTS_INL_H_ +#define V8_OBJECTS_JS_SEGMENTS_INL_H_ + +#ifndef V8_INTL_SUPPORT +#error Internationalization is expected to be enabled. +#endif // V8_INTL_SUPPORT + +#include "src/objects/js-segments.h" +#include "src/objects/objects-inl.h" + +// Has to be the last include (doesn't have include guards): +#include "src/objects/object-macros.h" + +namespace v8 { +namespace internal { + +TQ_OBJECT_CONSTRUCTORS_IMPL(JSSegments) + +// Base segments accessors. +ACCESSORS(JSSegments, icu_break_iterator, Managed, + kIcuBreakIteratorOffset) +ACCESSORS(JSSegments, unicode_string, Managed, + kUnicodeStringOffset) + +inline void JSSegments::set_granularity(JSSegmenter::Granularity granularity) { + DCHECK_GE(GranularityBits::kMax, granularity); + int hints = flags(); + hints = GranularityBits::update(hints, granularity); + set_flags(hints); +} + +inline JSSegmenter::Granularity JSSegments::granularity() const { + return GranularityBits::decode(flags()); +} + +} // namespace internal +} // namespace v8 + +#include "src/objects/object-macros-undef.h" + +#endif // V8_OBJECTS_JS_SEGMENTS_INL_H_ diff --git a/src/objects/js-segments.cc b/src/objects/js-segments.cc new file mode 100644 index 0000000000..96dbcc349d --- /dev/null +++ b/src/objects/js-segments.cc @@ -0,0 +1,183 @@ +// Copyright 2020 the V8 project authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. + +#ifndef V8_INTL_SUPPORT +#error Internationalization is expected to be enabled. +#endif // V8_INTL_SUPPORT + +#include "src/objects/js-segments.h" + +#include +#include +#include + +#include "src/execution/isolate.h" +#include "src/heap/factory.h" +#include "src/objects/intl-objects.h" +#include "src/objects/js-segment-iterator.h" +#include "src/objects/js-segmenter-inl.h" +#include "src/objects/js-segments-inl.h" +#include "src/objects/managed.h" +#include "src/objects/objects-inl.h" +#include "unicode/brkiter.h" + +namespace v8 { +namespace internal { + +// ecma402 #sec-createsegmentsobject +MaybeHandle JSSegments::Create(Isolate* isolate, + Handle segmenter, + Handle string) { + icu::BreakIterator* break_iterator = + segmenter->icu_break_iterator().raw()->clone(); + DCHECK_NOT_NULL(break_iterator); + + Handle> unicode_string = + Intl::SetTextToBreakIterator(isolate, string, break_iterator); + Handle> managed_break_iterator = + Managed::FromRawPtr(isolate, 0, break_iterator); + + // 1. Let internalSlotsList be « [[SegmentsSegmenter]], [[SegmentsString]] ». + // 2. Let segments be ! ObjectCreate(%Segments.prototype%, internalSlotsList). + Handle map(isolate->native_context()->intl_segments_map(), isolate); + Handle result = isolate->factory()->NewJSObjectFromMap(map); + + Handle segments = Handle::cast(result); + segments->set_flags(0); + + // 3. Set segments.[[SegmentsSegmenter]] to segmenter. + segments->set_icu_break_iterator(*managed_break_iterator); + segments->set_granularity(segmenter->granularity()); + + // 4. Set segments.[[SegmentsString]] to string. + segments->set_unicode_string(*unicode_string); + + // 5. Return segments. + return segments; +} + +// ecma402 #sec-createsegmentiterator +MaybeHandle JSSegments::CreateSegmentIterator( + Isolate* isolate, Handle segments) { + return JSSegmentIterator::Create( + isolate, segments->icu_break_iterator().raw()->clone(), + segments->granularity()); +} + +// ecma402 #sec-%segmentsprototype%.containing +MaybeHandle JSSegments::Containing(Isolate* isolate, + Handle segments, + int32_t n) { + // 5. Let len be the length of string. + int32_t len = segments->unicode_string().raw()->length(); + + // 7. If n < 0 or n ≥ len, return undefined. + if (n < 0 || n >= len) { + return isolate->factory()->undefined_value(); + } + + icu::BreakIterator* break_iterator = segments->icu_break_iterator().raw(); + // 8. Let startIndex be ! FindBoundary(segmenter, string, n, before). + int32_t start_index = + break_iterator->isBoundary(n) ? n : break_iterator->preceding(n); + + // 9. Let endIndex be ! FindBoundary(segmenter, string, n, after). + int32_t end_index = break_iterator->following(n); + + // 10. Return ! CreateSegmentDataObject(segmenter, string, startIndex, + // endIndex). + return CreateSegmentDataObject( + isolate, segments->granularity(), break_iterator, + *(segments->unicode_string().raw()), start_index, end_index); +} + +namespace { + +bool CurrentSegmentIsWordLike(icu::BreakIterator* break_iterator) { + int32_t rule_status = break_iterator->getRuleStatus(); + return (rule_status >= UBRK_WORD_NUMBER && + rule_status < UBRK_WORD_NUMBER_LIMIT) || + (rule_status >= UBRK_WORD_LETTER && + rule_status < UBRK_WORD_LETTER_LIMIT) || + (rule_status >= UBRK_WORD_KANA && + rule_status < UBRK_WORD_KANA_LIMIT) || + (rule_status >= UBRK_WORD_IDEO && rule_status < UBRK_WORD_IDEO_LIMIT); +} + +} // namespace + +// ecma402 #sec-createsegmentdataobject +MaybeHandle JSSegments::CreateSegmentDataObject( + Isolate* isolate, JSSegmenter::Granularity granularity, + icu::BreakIterator* break_iterator, const icu::UnicodeString& string, + int32_t start_index, int32_t end_index) { + Factory* factory = isolate->factory(); + + // 1. Let len be the length of string. + // 2. Assert: startIndex ≥ 0. + DCHECK_GE(start_index, 0); + // 3. Assert: endIndex ≤ len. + DCHECK_LE(end_index, string.length()); + // 4. Assert: startIndex < endIndex. + DCHECK_LT(start_index, end_index); + + // 5. Let result be ! ObjectCreate(%ObjectPrototype%). + Handle result = factory->NewJSObject(isolate->object_function()); + + // 6. Let segment be the String value equal to the substring of string + // consisting of the code units at indices startIndex (inclusive) through + // endIndex (exclusive). + Handle segment; + ASSIGN_RETURN_ON_EXCEPTION( + isolate, segment, Intl::ToString(isolate, string, start_index, end_index), + JSObject); + + // 7. Perform ! CreateDataPropertyOrThrow(result, "segment", segment). + Maybe maybe_create_segment = JSReceiver::CreateDataProperty( + isolate, result, factory->segment_string(), segment, Just(kDontThrow)); + DCHECK(maybe_create_segment.FromJust()); + USE(maybe_create_segment); + + // 8. Perform ! CreateDataPropertyOrThrow(result, "index", startIndex). + Maybe maybe_create_index = JSReceiver::CreateDataProperty( + isolate, result, factory->index_string(), + factory->NewNumberFromInt(start_index), Just(kDontThrow)); + DCHECK(maybe_create_index.FromJust()); + USE(maybe_create_index); + + // 9. Perform ! CreateDataPropertyOrThrow(result, "input", string). + Handle input_string; + ASSIGN_RETURN_ON_EXCEPTION(isolate, input_string, + Intl::ToString(isolate, string), JSObject); + Maybe maybe_create_input = JSReceiver::CreateDataProperty( + isolate, result, factory->input_string(), input_string, Just(kDontThrow)); + DCHECK(maybe_create_input.FromJust()); + USE(maybe_create_input); + + Handle is_word_like; + // 10. Let granularity be segmenter.[[SegmenterGranularity]]. + // 11. If granularity is "word", then + if (granularity == JSSegmenter::Granularity::WORD) { + // a. Let isWordLike be a Boolean value indicating whether the word segment + // segment in string is "word-like" according to locale + // segmenter.[[Locale]]. + is_word_like = CurrentSegmentIsWordLike(break_iterator) + ? factory->true_value() + : factory->false_value(); + // b. Perform ! CreateDataPropertyOrThrow(result, "isWordLike", isWordLike). + Maybe maybe_create_is_word_like = JSReceiver::CreateDataProperty( + isolate, result, factory->isWordLike_string(), is_word_like, + Just(kDontThrow)); + DCHECK(maybe_create_is_word_like.FromJust()); + USE(maybe_create_is_word_like); + } + return result; +} + +Handle JSSegments::GranularityAsString(Isolate* isolate) const { + return JSSegmenter::GetGranularityString(isolate, granularity()); +} + +} // namespace internal +} // namespace v8 diff --git a/src/objects/js-segments.h b/src/objects/js-segments.h new file mode 100644 index 0000000000..484a7b8d24 --- /dev/null +++ b/src/objects/js-segments.h @@ -0,0 +1,80 @@ +// Copyright 2020 the V8 project authors. All rights reserved. +// Use of this source code is governed by a BSD-style license that can be +// found in the LICENSE file. +#ifndef V8_OBJECTS_JS_SEGMENTS_H_ +#define V8_OBJECTS_JS_SEGMENTS_H_ + +#ifndef V8_INTL_SUPPORT +#error Internationalization is expected to be enabled. +#endif // V8_INTL_SUPPORT + +#include "src/base/bit-field.h" +#include "src/execution/isolate.h" +#include "src/heap/factory.h" +#include "src/objects/js-segmenter.h" +#include "src/objects/managed.h" +#include "src/objects/objects.h" +#include "unicode/uversion.h" + +// Has to be the last include (doesn't have include guards): +#include "src/objects/object-macros.h" + +namespace U_ICU_NAMESPACE { +class BreakIterator; +class UnicodeString; +} // namespace U_ICU_NAMESPACE + +namespace v8 { +namespace internal { + +class JSSegments : public TorqueGeneratedJSSegments { + public: + // ecma402 #sec-createsegmentsobject + V8_WARN_UNUSED_RESULT static MaybeHandle Create( + Isolate* isolate, Handle segmenter, Handle string); + + // ecma402 #sec-%segmentsprototype%.containing + V8_WARN_UNUSED_RESULT static MaybeHandle Containing( + Isolate* isolate, Handle segments_holder, int32_t index); + + // ecma402 #sec-%segmentsprototype%-@@iterator + V8_WARN_UNUSED_RESULT static MaybeHandle CreateSegmentIterator( + Isolate* isolate, Handle segments_holder); + + // ecma402 #sec-get-%segmentsprototype%.string + V8_WARN_UNUSED_RESULT static MaybeHandle GetString( + Isolate* isolate, Handle segments_holder); + + // ecma402 #sec-createsegmentdataobject + V8_WARN_UNUSED_RESULT static MaybeHandle CreateSegmentDataObject( + Isolate* isolate, JSSegmenter::Granularity granularity, + icu::BreakIterator* break_iterator, const icu::UnicodeString& string, + int32_t start_index, int32_t end_index); + + Handle GranularityAsString(Isolate* isolate) const; + + // SegmentIterator accessors. + DECL_ACCESSORS(icu_break_iterator, Managed) + DECL_ACCESSORS(unicode_string, Managed) + + DECL_PRINTER(JSSegments) + + inline void set_granularity(JSSegmenter::Granularity granularity); + inline JSSegmenter::Granularity granularity() const; + + // Bit positions in |flags|. + DEFINE_TORQUE_GENERATED_JS_SEGMENT_ITERATOR_FLAGS() + + STATIC_ASSERT(JSSegmenter::Granularity::GRAPHEME <= GranularityBits::kMax); + STATIC_ASSERT(JSSegmenter::Granularity::WORD <= GranularityBits::kMax); + STATIC_ASSERT(JSSegmenter::Granularity::SENTENCE <= GranularityBits::kMax); + + TQ_OBJECT_CONSTRUCTORS(JSSegments) +}; + +} // namespace internal +} // namespace v8 + +#include "src/objects/object-macros-undef.h" + +#endif // V8_OBJECTS_JS_SEGMENTS_H_ diff --git a/src/objects/map.cc b/src/objects/map.cc index 2474928180..cd21eb1d12 100644 --- a/src/objects/map.cc +++ b/src/objects/map.cc @@ -306,6 +306,7 @@ VisitorId Map::GetVisitorId(Map map) { case JS_RELATIVE_TIME_FORMAT_TYPE: case JS_SEGMENT_ITERATOR_TYPE: case JS_SEGMENTER_TYPE: + case JS_SEGMENTS_TYPE: #endif // V8_INTL_SUPPORT case WASM_EXCEPTION_OBJECT_TYPE: case WASM_GLOBAL_OBJECT_TYPE: diff --git a/src/objects/object-list-macros.h b/src/objects/object-list-macros.h index 3dfc3b2212..d7db982eab 100644 --- a/src/objects/object-list-macros.h +++ b/src/objects/object-list-macros.h @@ -252,7 +252,8 @@ class ZoneForwardList; V(JSPluralRules) \ V(JSRelativeTimeFormat) \ V(JSSegmentIterator) \ - V(JSSegmenter) + V(JSSegmenter) \ + V(JSSegments) #else #define HEAP_OBJECT_ORDINARY_TYPE_LIST(V) HEAP_OBJECT_ORDINARY_TYPE_LIST_BASE(V) #endif // V8_INTL_SUPPORT diff --git a/src/objects/objects-body-descriptors-inl.h b/src/objects/objects-body-descriptors-inl.h index 6868e474e4..f45052b05b 100644 --- a/src/objects/objects-body-descriptors-inl.h +++ b/src/objects/objects-body-descriptors-inl.h @@ -1033,6 +1033,7 @@ ReturnType BodyDescriptorApply(InstanceType type, T1 p1, T2 p2, T3 p3, T4 p4) { case JS_RELATIVE_TIME_FORMAT_TYPE: case JS_SEGMENT_ITERATOR_TYPE: case JS_SEGMENTER_TYPE: + case JS_SEGMENTS_TYPE: #endif // V8_INTL_SUPPORT case WASM_EXCEPTION_OBJECT_TYPE: case WASM_GLOBAL_OBJECT_TYPE: diff --git a/src/objects/objects.cc b/src/objects/objects.cc index 99528a63c9..8f25b3b5e8 100644 --- a/src/objects/objects.cc +++ b/src/objects/objects.cc @@ -94,6 +94,7 @@ #include "src/objects/js-relative-time-format.h" #include "src/objects/js-segment-iterator.h" #include "src/objects/js-segmenter.h" +#include "src/objects/js-segments.h" #endif // V8_INTL_SUPPORT #include "src/codegen/source-position-table.h" #include "src/objects/js-weak-refs-inl.h" diff --git a/src/objects/objects.h b/src/objects/objects.h index a52865c23b..8d287f1268 100644 --- a/src/objects/objects.h +++ b/src/objects/objects.h @@ -78,6 +78,7 @@ // - JSPluralRules // If V8_INTL_SUPPORT enabled. // - JSRelativeTimeFormat // If V8_INTL_SUPPORT enabled. // - JSSegmenter // If V8_INTL_SUPPORT enabled. +// - JSSegments // If V8_INTL_SUPPORT enabled. // - JSSegmentIterator // If V8_INTL_SUPPORT enabled. // - JSV8BreakIterator // If V8_INTL_SUPPORT enabled. // - WasmExceptionObject diff --git a/src/torque/implementation-visitor.cc b/src/torque/implementation-visitor.cc index 57538cbfc9..944d8c50b9 100644 --- a/src/torque/implementation-visitor.cc +++ b/src/torque/implementation-visitor.cc @@ -4788,6 +4788,7 @@ void ImplementationVisitor::GenerateExportedMacrosAssembler( cc_contents << "#include \"src/objects/js-relative-time-format.h\"\n"; cc_contents << "#include \"src/objects/js-segment-iterator.h\"\n"; cc_contents << "#include \"src/objects/js-segmenter.h\"\n"; + cc_contents << "#include \"src/objects/js-segments.h\"\n"; } cc_contents << "#include \"torque-generated/" << file_name << ".h\"\n"; diff --git a/test/intl/segmenter/segment-grapheme-following.js b/test/intl/segmenter/segment-grapheme-following.js deleted file mode 100644 index 4cfd8f4aef..0000000000 --- a/test/intl/segmenter/segment-grapheme-following.js +++ /dev/null @@ -1,38 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "grapheme"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - const iter = seg.segment(text); - let prev = 0; - let segments = []; - while (!iter.following()) { - assertEquals(undefined, iter.breakType); - assertTrue(iter.index >= 0); - assertTrue(iter.index <= text.length); - assertTrue(iter.index > prev); - segments.push(text.substring(prev, iter.index)); - prev = iter.index; - } - assertEquals(text, segments.join("")); -} diff --git a/test/intl/segmenter/segment-grapheme-iterable.js b/test/intl/segmenter/segment-grapheme-iterable.js deleted file mode 100644 index cd18590ee1..0000000000 --- a/test/intl/segmenter/segment-grapheme-iterable.js +++ /dev/null @@ -1,45 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "grapheme"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - let segments = []; - // Create another %SegmentIterator% to compare with result from the one that - // created in the for of loop. - let iter = seg.segment(text); - let prev = 0; - for (const v of seg.segment(text)) { - assertEquals(undefined, v.breakType); - assertEquals("string", typeof v.segment); - assertTrue(v.segment.length > 0); - segments.push(v.segment); - - // manually advance the iter. - assertFalse(iter.following()); - assertEquals(iter.breakType, v.breakType); - assertEquals(text.substring(prev, iter.index), v.segment); - prev = iter.index; - } - assertTrue(iter.following()); - assertEquals(text, segments.join('')); -} diff --git a/test/intl/segmenter/segment-grapheme-next.js b/test/intl/segmenter/segment-grapheme-next.js deleted file mode 100644 index df7f82acff..0000000000 --- a/test/intl/segmenter/segment-grapheme-next.js +++ /dev/null @@ -1,40 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "grapheme"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - const iter = seg.segment(text); - let segments = []; - let oldPos = -1; - for (let result = iter.next(); !result.done; result = iter.next()) { - const v = result.value; - assertEquals(undefined, v.breakType); - assertEquals("string", typeof v.segment); - assertTrue(v.segment.length > 0); - segments.push(v.segment); - assertEquals("number", typeof v.index); - assertTrue(oldPos < v.index); - oldPos = v.index; - } - assertEquals(text, segments.join('')); -} diff --git a/test/intl/segmenter/segment-grapheme-preceding.js b/test/intl/segmenter/segment-grapheme-preceding.js deleted file mode 100644 index 9ee7b7c811..0000000000 --- a/test/intl/segmenter/segment-grapheme-preceding.js +++ /dev/null @@ -1,44 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "grapheme"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - const iter = seg.segment(text); - let prev = text.length; - let segments = []; - iter.preceding(prev) - assertEquals(undefined, iter.breakType) - assertTrue(iter.index >= 0); - assertTrue(iter.index < prev); - segments.push(text.substring(iter.index, prev)); - prev = iter.index; - while (!iter.preceding()) { - assertEquals(undefined, iter.breakType); - assertTrue(iter.index >= 0); - assertTrue(iter.index <= text.length); - assertTrue(iter.index < prev); - segments.push(text.substring(iter.index, prev)); - prev = iter.index; - } - assertEquals(text, segments.reverse().join("")); -} diff --git a/test/intl/segmenter/segment-grapheme.js b/test/intl/segmenter/segment-grapheme.js index 1515f0997d..3e28142e64 100644 --- a/test/intl/segmenter/segment-grapheme.js +++ b/test/intl/segmenter/segment-grapheme.js @@ -23,7 +23,21 @@ for (const text of [ "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean ]) { - const iter = seg.segment(text); - assertEquals(undefined, iter.breakType); - assertEquals(0, iter.index); + const segments = seg.segment(text); + let results = []; + var pos = -1; + for (let s of segments) { + assertEquals(["segment", "index", "input"], Object.keys(s)); + assertEquals(typeof s.index, "number"); + assertEquals(typeof s.segment, "string"); + assertEquals(typeof s.input, "string"); + assertEquals(text, s.input); + assertEquals(text.substring(s.index, s.index + s.segment.length), + s.segment); + assertTrue(pos < s.index); + pos = s.index; + results.push(s.segment); + } + assertTrue(pos < text.length); + assertEquals(text, results.join("")); } diff --git a/test/intl/segmenter/segment-iterator-breakType.js b/test/intl/segmenter/segment-iterator-breakType.js deleted file mode 100644 index ea5f523519..0000000000 --- a/test/intl/segmenter/segment-iterator-breakType.js +++ /dev/null @@ -1,11 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const segmenter = new Intl.Segmenter(); -const text = "Hello World, Test 123! Foo Bar. How are you?"; -const iter = segmenter.segment(text); - -assertEquals(undefined, iter.breakType); diff --git a/test/intl/segmenter/segment-iterator-following.js b/test/intl/segmenter/segment-iterator-following.js deleted file mode 100644 index 14f6dd16c4..0000000000 --- a/test/intl/segmenter/segment-iterator-following.js +++ /dev/null @@ -1,26 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const segmenter = new Intl.Segmenter(); -const text = "Hello World, Test 123! Foo Bar. How are you?"; -const iter = segmenter.segment(text); - -assertEquals("function", typeof iter.following); - -// ToNumber("ABC") return NaN, ToInteger("ABC") return +0, ToIndex("ABC") return 0 -assertDoesNotThrow(() => iter.following("ABC")); -// ToNumber(null) return +0, ToInteger(null) return +0, ToIndex(null) return 0 -assertDoesNotThrow(() => iter.following(null)); -// ToNumber(1.4) return 1.4, ToInteger(1.4) return 1, ToIndex(1.4) return 1 -assertDoesNotThrow(() => iter.following(1.4)); - -assertThrows(() => iter.following(-3), RangeError); - -// 1.5.3.2 %SegmentIteratorPrototype%.following( [ from ] ) -// 3.b If from >= iterator.[[SegmentIteratorString]], throw a RangeError exception. -assertDoesNotThrow(() => iter.following(text.length - 1)); -assertThrows(() => iter.following(text.length), RangeError); -assertThrows(() => iter.following(text.length + 1), RangeError); diff --git a/test/intl/segmenter/segment-iterator-next.js b/test/intl/segmenter/segment-iterator-next.js deleted file mode 100644 index 9aa40494ca..0000000000 --- a/test/intl/segmenter/segment-iterator-next.js +++ /dev/null @@ -1,11 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const segmenter = new Intl.Segmenter(); -const text = "Hello World, Test 123! Foo Bar. How are you?"; -const iter = segmenter.segment(text); - -assertEquals("function", typeof iter.next); diff --git a/test/intl/segmenter/segment-iterator-ownPropertyDescriptor.js b/test/intl/segmenter/segment-iterator-ownPropertyDescriptor.js index 3021c81c63..2aba164c05 100644 --- a/test/intl/segmenter/segment-iterator-ownPropertyDescriptor.js +++ b/test/intl/segmenter/segment-iterator-ownPropertyDescriptor.js @@ -11,6 +11,9 @@ assertTrue(descriptor.writable); assertFalse(descriptor.enumerable); assertTrue(descriptor.configurable); +let segmenterPrototype = Object.getPrototypeOf(seg); +assertEquals("Intl.Segmenter", segmenterPrototype[Symbol.toStringTag]); + // ecma402 #sec-Intl.Segmenter.prototype // Intl.Segmenter.prototype // The value of Intl.Segmenter.prototype is %SegmenterPrototype%. @@ -29,10 +32,11 @@ for (let func of ["segment", "resolvedOptions"]) { assertTrue(descriptor.configurable); } -let segmentIterator = seg.segment('text'); -let prototype = Object.getPrototypeOf(segmentIterator); -for (let func of ["next", "following", "preceding"]) { - let descriptor = Object.getOwnPropertyDescriptor(prototype, func); + +let segments = seg.segment('text'); +let segmentsPrototype = Object.getPrototypeOf(segments); +for (let func of ["containing", Symbol.iterator]) { + let descriptor = Object.getOwnPropertyDescriptor(segmentsPrototype, func); assertTrue(descriptor.writable); assertFalse(descriptor.enumerable); assertTrue(descriptor.configurable); @@ -47,34 +51,44 @@ function checkGetterProperty(prototype, property) { assertTrue(desc.configurable); } -// Test the descriptor is correct for properties. -checkGetterProperty(prototype, 'index'); -checkGetterProperty(prototype, 'breakType'); - -// Test the SegmentIteratorPrototype methods are called with same +// Test the SegmentsPrototype methods are called with same // receiver and won't throw. -assertDoesNotThrow(() => prototype.next.call(segmentIterator)); -assertDoesNotThrow(() => prototype.following.call(segmentIterator)); -assertDoesNotThrow(() => prototype.preceding.call(segmentIterator)); +assertDoesNotThrow(() => segmentsPrototype.containing.call(segments)); +assertDoesNotThrow(() => segmentsPrototype[Symbol.iterator].call(segments)); // Test the SegmentIteratorPrototype methods are called with a different // receiver and correctly throw. var otherReceivers = [ 1, 123.45, undefined, null, "string", true, false, Intl, Intl.Segmenter, Intl.Segmenter.prototype, - prototype, + segmentsPrototype, new Intl.Segmenter(), new Intl.Collator(), new Intl.DateTimeFormat(), new Intl.NumberFormat(), ]; for (let rec of otherReceivers) { - assertThrows(() => prototype.next.call(rec), TypeError); - assertThrows(() => prototype.following.call(rec), TypeError); - assertThrows(() => prototype.preceding.call(rec), TypeError); + assertThrows(() => segmentsPrototype.containing.call(rec), TypeError); } -// Check the property of the return object of next() +let segmentIterator = segments[Symbol.iterator](); +let segmentIteratorPrototype = Object.getPrototypeOf(segmentIterator); +for (let func of ["next"]) { + let descriptor = Object.getOwnPropertyDescriptor(segmentIteratorPrototype, + func); + assertTrue(descriptor.writable); + assertFalse(descriptor.enumerable); + assertTrue(descriptor.configurable); +} + +assertEquals("Segmenter String Iterator", + segmentIteratorPrototype[Symbol.toStringTag]); +let desc = Object.getOwnPropertyDescriptor( + segmentIteratorPrototype, Symbol.toStringTag); +assertFalse(desc.writable); +assertFalse(desc.enumerable); +assertTrue(desc.configurable); + let nextReturn = segmentIterator.next(); function checkProperty(obj, property) { @@ -83,9 +97,3 @@ function checkProperty(obj, property) { assertTrue(desc.enumerable); assertTrue(desc.configurable); } - -checkProperty(nextReturn, 'done'); -checkProperty(nextReturn, 'value'); -checkProperty(nextReturn.value, 'segment'); -checkProperty(nextReturn.value, 'breakType'); -checkProperty(nextReturn.value, 'index'); diff --git a/test/intl/segmenter/segment-iterator-position.js b/test/intl/segmenter/segment-iterator-position.js deleted file mode 100644 index 649076b454..0000000000 --- a/test/intl/segmenter/segment-iterator-position.js +++ /dev/null @@ -1,11 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const segmenter = new Intl.Segmenter(); -const text = "Hello World, Test 123! Foo Bar. How are you?"; -const iter = segmenter.segment(text); - -assertEquals(0, iter.index); diff --git a/test/intl/segmenter/segment-iterator-preceding.js b/test/intl/segmenter/segment-iterator-preceding.js deleted file mode 100644 index 09ba2847cc..0000000000 --- a/test/intl/segmenter/segment-iterator-preceding.js +++ /dev/null @@ -1,30 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const segmenter = new Intl.Segmenter(); -const text = "Hello World, Test 123! Foo Bar. How are you?"; -const iter = segmenter.segment(text); - -assertEquals("function", typeof iter.preceding); - -// ToNumber("ABC") return NaN, ToInteger("ABC") return +0, ToIndex("ABC") return 0 -assertThrows(() => iter.preceding("ABC"), RangeError); -// ToNumber(null) return +0, ToInteger(null) return +0, ToIndex(null) return 0 -assertThrows(() => iter.preceding(null), RangeError); -assertThrows(() => iter.preceding(-3), RangeError); - -// ToNumber(1.4) return 1.4, ToInteger(1.4) return 1, ToIndex(1.4) return 1 -assertDoesNotThrow(() => iter.preceding(1.4)); - -// 1.5.3.3 %SegmentIteratorPrototype%.preceding( [ from ] ) -// 3.b If ... from = 0, throw a RangeError exception. -assertThrows(() => iter.preceding(0), RangeError); - -// 1.5.3.3 %SegmentIteratorPrototype%.preceding( [ from ] ) -// 3.b If from > iterator.[[SegmentIteratorString]] ... , throw a RangeError exception. -assertDoesNotThrow(() => iter.preceding(text.length - 1)); -assertDoesNotThrow(() => iter.preceding(text.length)); -assertThrows(() => iter.preceding(text.length + 1), RangeError); diff --git a/test/intl/segmenter/segment-iterator.js b/test/intl/segmenter/segment-iterator.js deleted file mode 100644 index 696ffab554..0000000000 --- a/test/intl/segmenter/segment-iterator.js +++ /dev/null @@ -1,15 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const text = "Hello World, Test 123! Foo Bar. How are you?"; -for (const granularity of ["grapheme", "word", "sentence"]) { - const segmenter = new Intl.Segmenter("en", { granularity }); - const iter = segmenter.segment(text); - - assertEquals("number", typeof iter.index); - assertEquals(0, iter.index); - assertEquals(undefined, iter.breakType); -} diff --git a/test/intl/segmenter/segment-sentence-following.js b/test/intl/segmenter/segment-sentence-following.js deleted file mode 100644 index c6b4aa25d6..0000000000 --- a/test/intl/segmenter/segment-sentence-following.js +++ /dev/null @@ -1,38 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "sentence"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - const iter = seg.segment(text); - let prev = 0; - let segments = []; - while (!iter.following()) { - assertTrue(["sep", "term"].includes(iter.breakType), iter.breakType); - assertTrue(iter.index >= 0); - assertTrue(iter.index <= text.length); - assertTrue(iter.index > prev); - segments.push(text.substring(prev, iter.index)); - prev = iter.index; - } - assertEquals(text, segments.join("")); -} diff --git a/test/intl/segmenter/segment-sentence-iterable.js b/test/intl/segmenter/segment-sentence-iterable.js deleted file mode 100644 index a84807bfb0..0000000000 --- a/test/intl/segmenter/segment-sentence-iterable.js +++ /dev/null @@ -1,45 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "sentence"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - let segments = []; - // Create another %SegmentIterator% to compare with result from the one that - // created in the for of loop. - let iter = seg.segment(text); - let prev = 0; - for (const v of seg.segment(text)) { - assertTrue(["sep", "term"].includes(v.breakType), v.breakType); - assertEquals("string", typeof v.segment); - assertTrue(v.segment.length > 0); - segments.push(v.segment); - - // manually advance the iter. - assertFalse(iter.following()); - assertEquals(iter.breakType, v.breakType); - assertEquals(text.substring(prev, iter.index), v.segment); - prev = iter.index; - } - assertTrue(iter.following()); - assertEquals(text, segments.join('')); -} diff --git a/test/intl/segmenter/segment-sentence-next.js b/test/intl/segmenter/segment-sentence-next.js deleted file mode 100644 index 466eac54d3..0000000000 --- a/test/intl/segmenter/segment-sentence-next.js +++ /dev/null @@ -1,40 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "sentence"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - const iter = seg.segment(text); - let segments = []; - let oldPos = -1; - for (let result = iter.next(); !result.done; result = iter.next()) { - const v = result.value; - assertTrue(["sep", "term"].includes(iter.breakType), iter.breakType); - assertEquals("string", typeof v.segment); - assertTrue(v.segment.length > 0); - segments.push(v.segment); - assertEquals("number", typeof v.index); - assertTrue(oldPos < v.index); - oldPos = v.index; - } - assertEquals(text, segments.join('')); -} diff --git a/test/intl/segmenter/segment-sentence-preceding.js b/test/intl/segmenter/segment-sentence-preceding.js deleted file mode 100644 index bbc17eecce..0000000000 --- a/test/intl/segmenter/segment-sentence-preceding.js +++ /dev/null @@ -1,44 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "sentence"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - const iter = seg.segment(text); - let prev = text.length; - let segments = []; - iter.preceding(prev); - assertTrue(["sep", "term"].includes(iter.breakType), iter.breakType); - assertTrue(iter.index >= 0); - assertTrue(iter.index < prev); - segments.push(text.substring(iter.index, prev)); - prev = iter.index; - while (!iter.preceding()) { - assertTrue(["sep", "term"].includes(iter.breakType), iter.breakType); - assertTrue(iter.index >= 0); - assertTrue(iter.index <= text.length); - assertTrue(iter.index < prev); - segments.push(text.substring(iter.index, prev)); - prev = iter.index; - } - assertEquals(text, segments.reverse().join("")); -} diff --git a/test/intl/segmenter/segment-sentence.js b/test/intl/segmenter/segment-sentence.js index a802362e0e..6bfcebd2b4 100644 --- a/test/intl/segmenter/segment-sentence.js +++ b/test/intl/segmenter/segment-sentence.js @@ -23,7 +23,21 @@ for (const text of [ "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean ]) { - const iter = seg.segment(text); - assertEquals(undefined, iter.breakType); - assertEquals(0, iter.index); + const segments = seg.segment(text); + let results = []; + var pos = -1; + for (let s of segments) { + assertEquals(["segment", "index", "input"], Object.keys(s)); + assertEquals(typeof s.index, "number"); + assertEquals(typeof s.segment, "string"); + assertEquals(typeof s.input, "string"); + assertEquals(text, s.input); + assertEquals(text.substring(s.index, s.index + s.segment.length), + s.segment); + assertTrue(pos < s.index); + pos = s.index; + results.push(s.segment); + } + assertTrue(pos < text.length); + assertEquals(text, results.join("")); } diff --git a/test/intl/segmenter/segment-word-following.js b/test/intl/segmenter/segment-word-following.js deleted file mode 100644 index cd0950eff5..0000000000 --- a/test/intl/segmenter/segment-word-following.js +++ /dev/null @@ -1,38 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "word"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - const iter = seg.segment(text); - let prev = 0; - let segments = []; - while (!iter.following()) { - assertTrue(["word", "none"].includes(iter.breakType), iter.breakType); - assertTrue(iter.index >= 0); - assertTrue(iter.index <= text.length); - assertTrue(iter.index > prev); - segments.push(text.substring(prev, iter.index)); - prev = iter.index; - } - assertEquals(text, segments.join("")); -} diff --git a/test/intl/segmenter/segment-word-iterable.js b/test/intl/segmenter/segment-word-iterable.js deleted file mode 100644 index 3dab2103c7..0000000000 --- a/test/intl/segmenter/segment-word-iterable.js +++ /dev/null @@ -1,45 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "word"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - let segments = []; - // Create another %SegmentIterator% to compare with result from the one that - // created in the for of loop. - let iter = seg.segment(text); - let prev = 0; - for (const v of seg.segment(text)) { - assertTrue(["word", "none"].includes(v.breakType), v.breakType); - assertEquals("string", typeof v.segment); - assertTrue(v.segment.length > 0); - segments.push(v.segment); - - // manually advance the iter. - assertFalse(iter.following()); - assertEquals(iter.breakType, v.breakType); - assertEquals(text.substring(prev, iter.index), v.segment); - prev = iter.index; - } - assertTrue(iter.following()); - assertEquals(text, segments.join('')); -} diff --git a/test/intl/segmenter/segment-word-next.js b/test/intl/segmenter/segment-word-next.js deleted file mode 100644 index 54fb40a251..0000000000 --- a/test/intl/segmenter/segment-word-next.js +++ /dev/null @@ -1,40 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "word"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - const iter = seg.segment(text); - let segments = []; - let oldPos = -1; - for (let result = iter.next(); !result.done; result = iter.next()) { - const v = result.value; - assertTrue(["word", "none"].includes(iter.breakType), iter.breakType); - assertEquals("string", typeof v.segment); - assertTrue(v.segment.length > 0); - segments.push(v.segment); - assertEquals("number", typeof v.index); - assertTrue(oldPos < v.index); - oldPos = v.index; - } - assertEquals(text, segments.join('')); -} diff --git a/test/intl/segmenter/segment-word-preceding.js b/test/intl/segmenter/segment-word-preceding.js deleted file mode 100644 index 7ca5132dfd..0000000000 --- a/test/intl/segmenter/segment-word-preceding.js +++ /dev/null @@ -1,44 +0,0 @@ -// Copyright 2018 the V8 project authors. All rights reserved. -// Use of this source code is governed by a BSD-style license that can be -// found in the LICENSE file. - -// Flags: --harmony-intl-segmenter - -const seg = new Intl.Segmenter([], {granularity: "word"}) -for (const text of [ - "Hello world!", // English - " Hello world! ", // English with space before/after - " Hello world? Foo bar!", // English - "Jedovatou mambu objevila žena v zahrádkářské kolonii.", // Czech - "Việt Nam: Nhất thể hóa sẽ khác Trung Quốc?", // Vietnamese - "Σοβαρές ενστάσεις Κομισιόν για τον προϋπολογισμό της Ιταλίας", // Greek - "Решение Индии о покупке российских С-400 расценили как вызов США", // Russian - "הרופא שהציל נשים והנערה ששועבדה ע", // Hebrew, - "ترامب للملك سلمان: أنا جاد للغاية.. عليك دفع المزيد", // Arabic - "भारत की एस 400 मिसाइल के मुकाबले पाक की थाड, जानें कौन कितना ताकतवर", // Hindi - "ரெட் அலர்ட் எச்சரிக்கை; புதுச்சேரியில் நாளை அரசு விடுமுறை!", // Tamil - "'ఉత్తర్వులు అందే వరకు ఓటర్ల తుది జాబితాను వెబ్‌సైట్లో పెట్టవద్దు'", // Telugu - "台北》抹黑柯P失敗？朱學恒酸：姚文智氣pupu嗆大老闆", // Chinese - "วัดไทรตีระฆังเบาลงช่วงเข้าพรรษา เจ้าอาวาสเผยคนร้องเรียนรับผลกรรมแล้ว", // Thai - "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese - "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean - ]) { - const iter = seg.segment(text); - let prev = text.length; - let segments = []; - iter.preceding(prev); - assertTrue(["word", "none"].includes(iter.breakType), iter.breakType); - assertTrue(iter.index >= 0); - assertTrue(iter.index < prev); - segments.push(text.substring(iter.index, prev)); - prev = iter.index; - while (!iter.preceding()) { - assertTrue(["word", "none"].includes(iter.breakType), iter.breakType); - assertTrue(iter.index >= 0); - assertTrue(iter.index <= text.length); - assertTrue(iter.index < prev); - segments.push(text.substring(iter.index, prev)); - prev = iter.index; - } - assertEquals(text, segments.reverse().join("")); -} diff --git a/test/intl/segmenter/segment-word.js b/test/intl/segmenter/segment-word.js index b191a67cb9..c2919e5911 100644 --- a/test/intl/segmenter/segment-word.js +++ b/test/intl/segmenter/segment-word.js @@ -23,7 +23,22 @@ for (const text of [ "九州北部の一部が暴風域に入りました(日直予報士 2018年10月06日) - 日本気象協会 tenki.jp", // Japanese "법원 “다스 지분 처분권·수익권 모두 MB가 보유”", // Korean ]) { - const iter = seg.segment(text); - assertEquals(undefined, iter.breakType); - assertEquals(0, iter.index); + const segments = seg.segment(text); + let results = []; + var pos = -1; + for (let s of segments) { + assertEquals(["segment", "index", "input", "isWordLike"], Object.keys(s)); + assertEquals(typeof s.isWordLike, "boolean"); + assertEquals(typeof s.index, "number"); + assertEquals(typeof s.segment, "string"); + assertEquals(typeof s.input, "string"); + assertEquals(text, s.input); + assertEquals(text.substring(s.index, s.index + s.segment.length), + s.segment); + assertTrue(pos < s.index); + pos = s.index; + results.push(s.segment); + } + assertTrue(pos < text.length); + assertEquals(text, results.join("")); } diff --git a/test/intl/segmenter/segment.js b/test/intl/segmenter/segment.js index 4c76b96e54..a4c44f3452 100644 --- a/test/intl/segmenter/segment.js +++ b/test/intl/segmenter/segment.js @@ -8,32 +8,34 @@ assertEquals("function", typeof Intl.Segmenter.prototype.segment); assertEquals(1, Intl.Segmenter.prototype.segment.length); let seg = new Intl.Segmenter("en", {granularity: "word"}) -let res; +let segments; // test with 0 args -assertDoesNotThrow(() => res = seg.segment()) +assertDoesNotThrow(() => segments = seg.segment()) // test with 1 arg -assertDoesNotThrow(() => res = seg.segment("hello")) -assertEquals("hello", res.next().value.segment); +assertDoesNotThrow(() => segments = seg.segment("hello")) +assertEquals("hello", segments.containing(0).input); // test with 2 args -assertDoesNotThrow(() => res = seg.segment("hello world")) -assertEquals("hello", res.next().value.segment); +assertDoesNotThrow(() => segments = seg.segment("hello world")) +assertEquals("hello world", segments.containing(0).input); // test with other types -assertDoesNotThrow(() => res = seg.segment(undefined)) -assertEquals("undefined", res.next().value.segment); -assertDoesNotThrow(() => res = seg.segment(null)) -assertEquals("null", res.next().value.segment); -assertDoesNotThrow(() => res = seg.segment(true)) -assertEquals("true", res.next().value.segment); -assertDoesNotThrow(() => res = seg.segment(false)) -assertEquals("false", res.next().value.segment); -assertDoesNotThrow(() => res = seg.segment(1234)) -assertEquals("1234", res.next().value.segment); -assertDoesNotThrow(() => res = seg.segment(3.1415926)) -assertEquals("3.1415926", res.next().value.segment); -assertDoesNotThrow(() => res = seg.segment(["hello","world"])) -assertEquals("hello", res.next().value.segment); -assertDoesNotThrow(() => res = seg.segment({k: 'v'})) -assertEquals("[", res.next().value.segment); -assertThrows(() => res = seg.segment(Symbol()), TypeError) +assertDoesNotThrow(() => segments = seg.segment(undefined)) +assertEquals("undefined", segments.containing(0).input); +assertDoesNotThrow(() => segments = seg.segment(null)) +assertEquals("null", segments.containing(0).input); +assertDoesNotThrow(() => segments = seg.segment(true)) +assertEquals("true", segments.containing(0).input); +assertDoesNotThrow(() => segments = seg.segment(false)) +assertEquals("false", segments.containing(0).input); +assertDoesNotThrow(() => segments = seg.segment(1234)) +assertEquals("1234", segments.containing(0).input); +assertDoesNotThrow(() => segments = seg.segment(3.1415926)) +assertEquals("3.1415926", segments.containing(0).input); +assertDoesNotThrow(() => segments = seg.segment(98765432109876543210987654321n)) +assertEquals("98765432109876543210987654321", segments.containing(0).input); +assertDoesNotThrow(() => segments = seg.segment(["hello","world"])) +assertEquals("hello,world", segments.containing(0).input); +assertDoesNotThrow(() => segments = seg.segment({k: 'v'})) +assertEquals("[object Object]", segments.containing(0).input); +assertThrows(() => segments = seg.segment(Symbol()), TypeError) diff --git a/test/test262/test262.status b/test/test262/test262.status index 72b6d42b47..abe1043484 100644 --- a/test/test262/test262.status +++ b/test/test262/test262.status @@ -516,8 +516,10 @@ 'intl402/Locale/constructor-options-region-valid': [FAIL], # http://crbug/v8/6891 + 'intl402/Segmenter/prototype/segment/segment-grapheme': [FAIL], + 'intl402/Segmenter/prototype/segment/segment-sentence': [FAIL], 'intl402/Segmenter/prototype/segment/segment-tostring': [FAIL], - 'intl402/Segmenter/prototype/segment/segment-word-iterable': [FAIL], + 'intl402/Segmenter/prototype/segment/segment-word': [FAIL], # https://bugs.chromium.org/p/v8/issues/detail?id=9818 'built-ins/AsyncFunction/proto-from-ctor-realm': [FAIL], diff --git a/tools/heap-stats/categories.js b/tools/heap-stats/categories.js index 9837acce3d..2bd08fad02 100644 --- a/tools/heap-stats/categories.js +++ b/tools/heap-stats/categories.js @@ -60,6 +60,7 @@ export const CATEGORIES = new Map([ 'JS_RELATIVE_TIME_FORMAT_TYPE', 'JS_SEGMENT_ITERATOR_TYPE', 'JS_SEGMENTER_TYPE', + 'JS_SEGMENTS_TYPE', 'JS_V8_BREAK_ITERATOR_TYPE', 'JS_MAP_KEY_ITERATOR_TYPE', 'JS_MAP_KEY_VALUE_ITERATOR_TYPE', diff --git a/tools/v8heapconst.py b/tools/v8heapconst.py index 44d62cfc79..1aff3930d8 100644 --- a/tools/v8heapconst.py +++ b/tools/v8heapconst.py @@ -190,17 +190,18 @@ INSTANCE_TYPES = { 1078: "JS_RELATIVE_TIME_FORMAT_TYPE", 1079: "JS_SEGMENT_ITERATOR_TYPE", 1080: "JS_SEGMENTER_TYPE", - 1081: "JS_STRING_ITERATOR_TYPE", - 1082: "JS_V8_BREAK_ITERATOR_TYPE", - 1083: "JS_WEAK_REF_TYPE", - 1084: "WASM_EXCEPTION_OBJECT_TYPE", - 1085: "WASM_GLOBAL_OBJECT_TYPE", - 1086: "WASM_INSTANCE_OBJECT_TYPE", - 1087: "WASM_MEMORY_OBJECT_TYPE", - 1088: "WASM_MODULE_OBJECT_TYPE", - 1089: "WASM_TABLE_OBJECT_TYPE", - 1090: "JS_BOUND_FUNCTION_TYPE", - 1091: "JS_FUNCTION_TYPE", + 1081: "JS_SEGMENTS_TYPE", + 1082: "JS_STRING_ITERATOR_TYPE", + 1083: "JS_V8_BREAK_ITERATOR_TYPE", + 1084: "JS_WEAK_REF_TYPE", + 1085: "WASM_EXCEPTION_OBJECT_TYPE", + 1086: "WASM_GLOBAL_OBJECT_TYPE", + 1087: "WASM_INSTANCE_OBJECT_TYPE", + 1088: "WASM_MEMORY_OBJECT_TYPE", + 1089: "WASM_MODULE_OBJECT_TYPE", + 1090: "WASM_TABLE_OBJECT_TYPE", + 1091: "JS_BOUND_FUNCTION_TYPE", + 1092: "JS_FUNCTION_TYPE", } # List of known V8 maps. @@ -297,65 +298,65 @@ KNOWN_MAPS = { ("read_only_space", 0x0319d): (96, "EnumCacheMap"), ("read_only_space", 0x031ed): (87, "ArrayBoilerplateDescriptionMap"), ("read_only_space", 0x032d9): (99, "InterceptorInfoMap"), - ("read_only_space", 0x0538d): (72, "PromiseFulfillReactionJobTaskMap"), - ("read_only_space", 0x053b5): (73, "PromiseRejectReactionJobTaskMap"), - ("read_only_space", 0x053dd): (74, "CallableTaskMap"), - ("read_only_space", 0x05405): (75, "CallbackTaskMap"), - ("read_only_space", 0x0542d): (76, "PromiseResolveThenableJobTaskMap"), - ("read_only_space", 0x05455): (79, "FunctionTemplateInfoMap"), - ("read_only_space", 0x0547d): (80, "ObjectTemplateInfoMap"), - ("read_only_space", 0x054a5): (81, "AccessCheckInfoMap"), - ("read_only_space", 0x054cd): (82, "AccessorInfoMap"), - ("read_only_space", 0x054f5): (83, "AccessorPairMap"), - ("read_only_space", 0x0551d): (84, "AliasedArgumentsEntryMap"), - ("read_only_space", 0x05545): (85, "AllocationMementoMap"), - ("read_only_space", 0x0556d): (88, "AsmWasmDataMap"), - ("read_only_space", 0x05595): (89, "AsyncGeneratorRequestMap"), - ("read_only_space", 0x055bd): (90, "BreakPointMap"), - ("read_only_space", 0x055e5): (91, "BreakPointInfoMap"), - ("read_only_space", 0x0560d): (92, "CachedTemplateObjectMap"), - ("read_only_space", 0x05635): (94, "ClassPositionsMap"), - ("read_only_space", 0x0565d): (95, "DebugInfoMap"), - ("read_only_space", 0x05685): (98, "FunctionTemplateRareDataMap"), - ("read_only_space", 0x056ad): (100, "InterpreterDataMap"), - ("read_only_space", 0x056d5): (101, "PromiseCapabilityMap"), - ("read_only_space", 0x056fd): (102, "PromiseReactionMap"), - ("read_only_space", 0x05725): (103, "PropertyDescriptorObjectMap"), - ("read_only_space", 0x0574d): (104, "PrototypeInfoMap"), - ("read_only_space", 0x05775): (105, "ScriptMap"), - ("read_only_space", 0x0579d): (106, "SourceTextModuleInfoEntryMap"), - ("read_only_space", 0x057c5): (107, "StackFrameInfoMap"), - ("read_only_space", 0x057ed): (108, "StackTraceFrameMap"), - ("read_only_space", 0x05815): (109, "TemplateObjectDescriptionMap"), - ("read_only_space", 0x0583d): (110, "Tuple2Map"), - ("read_only_space", 0x05865): (111, "WasmCapiFunctionDataMap"), - ("read_only_space", 0x0588d): (112, "WasmExceptionTagMap"), - ("read_only_space", 0x058b5): (113, "WasmExportedFunctionDataMap"), - ("read_only_space", 0x058dd): (114, "WasmIndirectFunctionTableMap"), - ("read_only_space", 0x05905): (115, "WasmJSFunctionDataMap"), - ("read_only_space", 0x0592d): (116, "WasmValueMap"), - ("read_only_space", 0x05955): (136, "SloppyArgumentsElementsMap"), - ("read_only_space", 0x0597d): (172, "OnHeapBasicBlockProfilerDataMap"), - ("read_only_space", 0x059a5): (169, "InternalClassMap"), - ("read_only_space", 0x059cd): (178, "SmiPairMap"), - ("read_only_space", 0x059f5): (177, "SmiBoxMap"), - ("read_only_space", 0x05a1d): (147, "ExportedSubClassBaseMap"), - ("read_only_space", 0x05a45): (148, "ExportedSubClassMap"), - ("read_only_space", 0x05a6d): (68, "AbstractInternalClassSubclass1Map"), - ("read_only_space", 0x05a95): (69, "AbstractInternalClassSubclass2Map"), - ("read_only_space", 0x05abd): (135, "InternalClassWithSmiElementsMap"), - ("read_only_space", 0x05ae5): (170, "InternalClassWithStructElementsMap"), - ("read_only_space", 0x05b0d): (149, "ExportedSubClass2Map"), - ("read_only_space", 0x05b35): (179, "SortStateMap"), - ("read_only_space", 0x05b5d): (86, "AllocationSiteWithWeakNextMap"), - ("read_only_space", 0x05b85): (86, "AllocationSiteWithoutWeakNextMap"), - ("read_only_space", 0x05bad): (77, "LoadHandler1Map"), - ("read_only_space", 0x05bd5): (77, "LoadHandler2Map"), - ("read_only_space", 0x05bfd): (77, "LoadHandler3Map"), - ("read_only_space", 0x05c25): (78, "StoreHandler0Map"), - ("read_only_space", 0x05c4d): (78, "StoreHandler1Map"), - ("read_only_space", 0x05c75): (78, "StoreHandler2Map"), - ("read_only_space", 0x05c9d): (78, "StoreHandler3Map"), + ("read_only_space", 0x053b9): (72, "PromiseFulfillReactionJobTaskMap"), + ("read_only_space", 0x053e1): (73, "PromiseRejectReactionJobTaskMap"), + ("read_only_space", 0x05409): (74, "CallableTaskMap"), + ("read_only_space", 0x05431): (75, "CallbackTaskMap"), + ("read_only_space", 0x05459): (76, "PromiseResolveThenableJobTaskMap"), + ("read_only_space", 0x05481): (79, "FunctionTemplateInfoMap"), + ("read_only_space", 0x054a9): (80, "ObjectTemplateInfoMap"), + ("read_only_space", 0x054d1): (81, "AccessCheckInfoMap"), + ("read_only_space", 0x054f9): (82, "AccessorInfoMap"), + ("read_only_space", 0x05521): (83, "AccessorPairMap"), + ("read_only_space", 0x05549): (84, "AliasedArgumentsEntryMap"), + ("read_only_space", 0x05571): (85, "AllocationMementoMap"), + ("read_only_space", 0x05599): (88, "AsmWasmDataMap"), + ("read_only_space", 0x055c1): (89, "AsyncGeneratorRequestMap"), + ("read_only_space", 0x055e9): (90, "BreakPointMap"), + ("read_only_space", 0x05611): (91, "BreakPointInfoMap"), + ("read_only_space", 0x05639): (92, "CachedTemplateObjectMap"), + ("read_only_space", 0x05661): (94, "ClassPositionsMap"), + ("read_only_space", 0x05689): (95, "DebugInfoMap"), + ("read_only_space", 0x056b1): (98, "FunctionTemplateRareDataMap"), + ("read_only_space", 0x056d9): (100, "InterpreterDataMap"), + ("read_only_space", 0x05701): (101, "PromiseCapabilityMap"), + ("read_only_space", 0x05729): (102, "PromiseReactionMap"), + ("read_only_space", 0x05751): (103, "PropertyDescriptorObjectMap"), + ("read_only_space", 0x05779): (104, "PrototypeInfoMap"), + ("read_only_space", 0x057a1): (105, "ScriptMap"), + ("read_only_space", 0x057c9): (106, "SourceTextModuleInfoEntryMap"), + ("read_only_space", 0x057f1): (107, "StackFrameInfoMap"), + ("read_only_space", 0x05819): (108, "StackTraceFrameMap"), + ("read_only_space", 0x05841): (109, "TemplateObjectDescriptionMap"), + ("read_only_space", 0x05869): (110, "Tuple2Map"), + ("read_only_space", 0x05891): (111, "WasmCapiFunctionDataMap"), + ("read_only_space", 0x058b9): (112, "WasmExceptionTagMap"), + ("read_only_space", 0x058e1): (113, "WasmExportedFunctionDataMap"), + ("read_only_space", 0x05909): (114, "WasmIndirectFunctionTableMap"), + ("read_only_space", 0x05931): (115, "WasmJSFunctionDataMap"), + ("read_only_space", 0x05959): (116, "WasmValueMap"), + ("read_only_space", 0x05981): (136, "SloppyArgumentsElementsMap"), + ("read_only_space", 0x059a9): (172, "OnHeapBasicBlockProfilerDataMap"), + ("read_only_space", 0x059d1): (169, "InternalClassMap"), + ("read_only_space", 0x059f9): (178, "SmiPairMap"), + ("read_only_space", 0x05a21): (177, "SmiBoxMap"), + ("read_only_space", 0x05a49): (147, "ExportedSubClassBaseMap"), + ("read_only_space", 0x05a71): (148, "ExportedSubClassMap"), + ("read_only_space", 0x05a99): (68, "AbstractInternalClassSubclass1Map"), + ("read_only_space", 0x05ac1): (69, "AbstractInternalClassSubclass2Map"), + ("read_only_space", 0x05ae9): (135, "InternalClassWithSmiElementsMap"), + ("read_only_space", 0x05b11): (170, "InternalClassWithStructElementsMap"), + ("read_only_space", 0x05b39): (149, "ExportedSubClass2Map"), + ("read_only_space", 0x05b61): (179, "SortStateMap"), + ("read_only_space", 0x05b89): (86, "AllocationSiteWithWeakNextMap"), + ("read_only_space", 0x05bb1): (86, "AllocationSiteWithoutWeakNextMap"), + ("read_only_space", 0x05bd9): (77, "LoadHandler1Map"), + ("read_only_space", 0x05c01): (77, "LoadHandler2Map"), + ("read_only_space", 0x05c29): (77, "LoadHandler3Map"), + ("read_only_space", 0x05c51): (78, "StoreHandler0Map"), + ("read_only_space", 0x05c79): (78, "StoreHandler1Map"), + ("read_only_space", 0x05ca1): (78, "StoreHandler2Map"), + ("read_only_space", 0x05cc9): (78, "StoreHandler3Map"), ("map_space", 0x0211d): (1057, "ExternalMap"), ("map_space", 0x02145): (1072, "JSMessageObjectMap"), ("map_space", 0x0216d): (181, "WasmRttEqrefMap"),