LibJS: Start implementing Intl.Segmenter

Author: https://github.com/IdanHo
Commit: https://github.com/SerenityOS/serenity/commit/a3bc06bb239
Pull-request: https://github.com/SerenityOS/serenity/pull/12211
Reviewed-by: https://github.com/linusg ✅
Reviewed-by: https://github.com/trflynn89 ✅

Userland/Libraries/LibJS/CMakeLists.txt

@@ -115,6 +115,9 @@ set(SOURCES
     Runtime/Intl/RelativeTimeFormat.cpp
     Runtime/Intl/RelativeTimeFormatConstructor.cpp
     Runtime/Intl/RelativeTimeFormatPrototype.cpp
+    Runtime/Intl/Segmenter.cpp
+    Runtime/Intl/SegmenterConstructor.cpp
+    Runtime/Intl/SegmenterPrototype.cpp
     Runtime/IteratorOperations.cpp
     Runtime/IteratorPrototype.cpp
     Runtime/JSONObject.cpp

Userland/Libraries/LibJS/Forward.h

@@ -67,15 +67,16 @@
     __JS_ENUMERATE(Float32Array, float32_array, Float32ArrayPrototype, Float32ArrayConstructor, float)                          \
     __JS_ENUMERATE(Float64Array, float64_array, Float64ArrayPrototype, Float64ArrayConstructor, double)
 
-#define JS_ENUMERATE_INTL_OBJECTS                                                                        \
-    __JS_ENUMERATE(Collator, collator, CollatorPrototype, CollatorConstructor)                           \
-    __JS_ENUMERATE(DateTimeFormat, date_time_format, DateTimeFormatPrototype, DateTimeFormatConstructor) \
-    __JS_ENUMERATE(DisplayNames, display_names, DisplayNamesPrototype, DisplayNamesConstructor)          \
-    __JS_ENUMERATE(ListFormat, list_format, ListFormatPrototype, ListFormatConstructor)                  \
-    __JS_ENUMERATE(Locale, locale, LocalePrototype, LocaleConstructor)                                   \
-    __JS_ENUMERATE(NumberFormat, number_format, NumberFormatPrototype, NumberFormatConstructor)          \
-    __JS_ENUMERATE(PluralRules, plural_rules, PluralRulesPrototype, PluralRulesConstructor)              \
-    __JS_ENUMERATE(RelativeTimeFormat, relative_time_format, RelativeTimeFormatPrototype, RelativeTimeFormatConstructor)
+#define JS_ENUMERATE_INTL_OBJECTS                                                                                        \
+    __JS_ENUMERATE(Collator, collator, CollatorPrototype, CollatorConstructor)                                           \
+    __JS_ENUMERATE(DateTimeFormat, date_time_format, DateTimeFormatPrototype, DateTimeFormatConstructor)                 \
+    __JS_ENUMERATE(DisplayNames, display_names, DisplayNamesPrototype, DisplayNamesConstructor)                          \
+    __JS_ENUMERATE(ListFormat, list_format, ListFormatPrototype, ListFormatConstructor)                                  \
+    __JS_ENUMERATE(Locale, locale, LocalePrototype, LocaleConstructor)                                                   \
+    __JS_ENUMERATE(NumberFormat, number_format, NumberFormatPrototype, NumberFormatConstructor)                          \
+    __JS_ENUMERATE(PluralRules, plural_rules, PluralRulesPrototype, PluralRulesConstructor)                              \
+    __JS_ENUMERATE(RelativeTimeFormat, relative_time_format, RelativeTimeFormatPrototype, RelativeTimeFormatConstructor) \
+    __JS_ENUMERATE(Segmenter, segmenter, SegmenterPrototype, SegmenterConstructor)
 
 #define JS_ENUMERATE_TEMPORAL_OBJECTS                                                                    \
     __JS_ENUMERATE(Calendar, calendar, CalendarPrototype, CalendarConstructor)                           \

Userland/Libraries/LibJS/Runtime/CommonPropertyNames.h

@@ -240,6 +240,7 @@ namespace JS {
     P(getYear)                               \
     P(global)                                \
     P(globalThis)                            \
+    P(granularity)                           \
     P(group)                                 \
     P(groupBy)                               \
     P(groupByToMap)                          \

Userland/Libraries/LibJS/Runtime/GlobalObject.cpp

@@ -67,6 +67,8 @@
 #include <LibJS/Runtime/Intl/PluralRulesPrototype.h>
 #include <LibJS/Runtime/Intl/RelativeTimeFormatConstructor.h>
 #include <LibJS/Runtime/Intl/RelativeTimeFormatPrototype.h>
+#include <LibJS/Runtime/Intl/SegmenterConstructor.h>
+#include <LibJS/Runtime/Intl/SegmenterPrototype.h>
 #include <LibJS/Runtime/IteratorPrototype.h>
 #include <LibJS/Runtime/JSONObject.h>
 #include <LibJS/Runtime/MapConstructor.h>

Userland/Libraries/LibJS/Runtime/Intl/Intl.cpp

@@ -16,6 +16,7 @@
 #include <LibJS/Runtime/Intl/NumberFormatConstructor.h>
 #include <LibJS/Runtime/Intl/PluralRulesConstructor.h>
 #include <LibJS/Runtime/Intl/RelativeTimeFormatConstructor.h>
+#include <LibJS/Runtime/Intl/SegmenterConstructor.h>
 
 namespace JS::Intl {
 
@@ -43,6 +44,7 @@ void Intl::initialize(GlobalObject& global_object)
     define_direct_property(vm.names.NumberFormat, global_object.intl_number_format_constructor(), attr);
     define_direct_property(vm.names.PluralRules, global_object.intl_plural_rules_constructor(), attr);
     define_direct_property(vm.names.RelativeTimeFormat, global_object.intl_relative_time_format_constructor(), attr);
+    define_direct_property(vm.names.Segmenter, global_object.intl_segmenter_constructor(), attr);
 
     define_native_function(vm.names.getCanonicalLocales, get_canonical_locales, 1, attr);
 }

Userland/Libraries/LibJS/Runtime/Intl/Segmenter.cpp

@@ -0,0 +1,43 @@
+/*
+ * Copyright (c) 2022, Idan Horowitz <idan.horowitz@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include <LibJS/Runtime/Intl/Segmenter.h>
+
+namespace JS::Intl {
+
+// 18 Segmenter Objects, https://tc39.es/ecma402/#segmenter-objects
+Segmenter::Segmenter(Object& prototype)
+    : Object(prototype)
+{
+}
+
+void Segmenter::set_segmenter_granularity(StringView segmenter_granularity)
+{
+    if (segmenter_granularity == "grapheme"sv)
+        m_segmenter_granularity = SegmenterGranularity::Grapheme;
+    else if (segmenter_granularity == "word"sv)
+        m_segmenter_granularity = SegmenterGranularity::Word;
+    else if (segmenter_granularity == "sentence"sv)
+        m_segmenter_granularity = SegmenterGranularity::Sentence;
+    else
+        VERIFY_NOT_REACHED();
+}
+
+StringView Segmenter::segmenter_granularity_string() const
+{
+    switch (m_segmenter_granularity) {
+    case SegmenterGranularity::Grapheme:
+        return "grapheme"sv;
+    case SegmenterGranularity::Word:
+        return "word"sv;
+    case SegmenterGranularity::Sentence:
+        return "sentence"sv;
+    default:
+        VERIFY_NOT_REACHED();
+    }
+}
+
+}

Userland/Libraries/LibJS/Runtime/Intl/Segmenter.h

@@ -0,0 +1,39 @@
+/*
+ * Copyright (c) 2022, Idan Horowitz <idan.horowitz@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <AK/String.h>
+#include <LibJS/Runtime/Object.h>
+
+namespace JS::Intl {
+
+class Segmenter final : public Object {
+    JS_OBJECT(Segmenter, Object);
+
+public:
+    enum class SegmenterGranularity {
+        Grapheme,
+        Word,
+        Sentence,
+    };
+
+    explicit Segmenter(Object& prototype);
+    virtual ~Segmenter() override = default;
+
+    String const& locale() const { return m_locale; }
+    void set_locale(String locale) { m_locale = move(locale); }
+
+    SegmenterGranularity segmenter_granularity() const { return m_segmenter_granularity; }
+    void set_segmenter_granularity(StringView);
+    StringView segmenter_granularity_string() const;
+
+private:
+    String m_locale;                                                                 // [[Locale]]
+    SegmenterGranularity m_segmenter_granularity { SegmenterGranularity::Grapheme }; // [[SegmenterGranularity]]
+};
+
+}

Userland/Libraries/LibJS/Runtime/Intl/SegmenterConstructor.cpp

@@ -0,0 +1,86 @@
+/*
+ * Copyright (c) 2022, Idan Horowitz <idan.horowitz@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include <LibJS/Runtime/AbstractOperations.h>
+#include <LibJS/Runtime/GlobalObject.h>
+#include <LibJS/Runtime/Intl/AbstractOperations.h>
+#include <LibJS/Runtime/Intl/Segmenter.h>
+#include <LibJS/Runtime/Intl/SegmenterConstructor.h>
+#include <LibJS/Runtime/Temporal/AbstractOperations.h>
+
+namespace JS::Intl {
+
+// 18.1 The Intl.Segmenter Constructor, https://tc39.es/ecma402/#sec-intl-segmenter-constructor
+SegmenterConstructor::SegmenterConstructor(GlobalObject& global_object)
+    : NativeFunction(vm().names.Segmenter.as_string(), *global_object.function_prototype())
+{
+}
+
+void SegmenterConstructor::initialize(GlobalObject& global_object)
+{
+    NativeFunction::initialize(global_object);
+
+    auto& vm = this->vm();
+
+    // 18.2.1 Intl.Segmenter.prototype, https://tc39.es/ecma402/#sec-intl.segmenter.prototype
+    define_direct_property(vm.names.prototype, global_object.intl_segmenter_prototype(), 0);
+    define_direct_property(vm.names.length, Value(0), Attribute::Configurable);
+}
+
+// 18.1.1 Intl.Segmenter ( [ locales [ , options ] ] ), https://tc39.es/ecma402/#sec-intl.segmenter
+ThrowCompletionOr<Value> SegmenterConstructor::call()
+{
+    // 1. If NewTarget is undefined, throw a TypeError exception.
+    return vm().throw_completion<TypeError>(global_object(), ErrorType::ConstructorWithoutNew, "Intl.Segmenter");
+}
+
+// 18.1.1 Intl.Segmenter ( [ locales [ , options ] ] ), https://tc39.es/ecma402/#sec-intl.segmenter
+ThrowCompletionOr<Object*> SegmenterConstructor::construct(FunctionObject& new_target)
+{
+    auto& vm = this->vm();
+    auto& global_object = this->global_object();
+
+    auto locales = vm.argument(0);
+    auto options_value = vm.argument(1);
+
+    // 2. Let internalSlotsList be « [[InitializedSegmenter]], [[Locale]], [[SegmenterGranularity]] ».
+    // 3. Let segmenter be ? OrdinaryCreateFromConstructor(NewTarget, "%Segmenter.prototype%", internalSlotsList).
+    auto* segmenter = TRY(ordinary_create_from_constructor<Segmenter>(global_object, new_target, &GlobalObject::intl_segmenter_prototype));
+
+    // 4. Let requestedLocales be ? CanonicalizeLocaleList(locales).
+    auto requested_locales = TRY(canonicalize_locale_list(global_object, locales));
+
+    // 5. Set options to ? GetOptionsObject(options).
+    auto* options = TRY(Temporal::get_options_object(global_object, options_value));
+
+    // 6. Let opt be a new Record.
+    LocaleOptions opt {};
+
+    // 7. Let matcher be ? GetOption(options, "localeMatcher", "string", « "lookup", "best fit" », "best fit").
+    auto matcher = TRY(get_option(global_object, *options, vm.names.localeMatcher, Value::Type::String, { "lookup"sv, "best fit"sv }, "best fit"sv));
+
+    // 8. Set opt.[[localeMatcher]] to matcher.
+    opt.locale_matcher = matcher;
+
+    // 9. Let localeData be %Segmenter%.[[LocaleData]].
+
+    // 10. Let r be ResolveLocale(%Segmenter%.[[AvailableLocales]], requestedLocales, opt, %Segmenter%.[[RelevantExtensionKeys]], localeData).
+    auto result = resolve_locale(requested_locales, opt, {});
+
+    // 11. Set segmenter.[[Locale]] to r.[[locale]].
+    segmenter->set_locale(move(result.locale));
+
+    // 12. Let granularity be ? GetOption(options, "granularity", "string", « "grapheme", "word", "sentence" », "grapheme").
+    auto granularity = TRY(get_option(global_object, *options, vm.names.granularity, Value::Type::String, { "grapheme"sv, "word"sv, "sentence"sv }, "grapheme"sv));
+
+    // 13. Set segmenter.[[SegmenterGranularity]] to granularity.
+    segmenter->set_segmenter_granularity(granularity.as_string().string());
+
+    // 14. Return segmenter.
+    return segmenter;
+}
+
+}

Userland/Libraries/LibJS/Runtime/Intl/SegmenterConstructor.h

@@ -0,0 +1,28 @@
+/*
+ * Copyright (c) 2022, Idan Horowitz <idan.horowitz@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <LibJS/Runtime/NativeFunction.h>
+
+namespace JS::Intl {
+
+class SegmenterConstructor final : public NativeFunction {
+    JS_OBJECT(SegmenterConstructor, NativeFunction);
+
+public:
+    explicit SegmenterConstructor(GlobalObject&);
+    virtual void initialize(GlobalObject&) override;
+    virtual ~SegmenterConstructor() override = default;
+
+    virtual ThrowCompletionOr<Value> call() override;
+    virtual ThrowCompletionOr<Object*> construct(FunctionObject& new_target) override;
+
+private:
+    virtual bool has_constructor() const override { return true; }
+};
+
+}

Userland/Libraries/LibJS/Runtime/Intl/SegmenterPrototype.cpp

@@ -0,0 +1,29 @@
+/*
+ * Copyright (c) 2022, Idan Horowitz <idan.horowitz@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#include <LibJS/Runtime/GlobalObject.h>
+#include <LibJS/Runtime/Intl/Segmenter.h>
+#include <LibJS/Runtime/Intl/SegmenterPrototype.h>
+
+namespace JS::Intl {
+
+// 18.3 Properties of the Intl.Segmenter Prototype Object, https://tc39.es/ecma402/#sec-properties-of-intl-segmenter-prototype-object
+SegmenterPrototype::SegmenterPrototype(GlobalObject& global_object)
+    : PrototypeObject(*global_object.object_prototype())
+{
+}
+
+void SegmenterPrototype::initialize(GlobalObject& global_object)
+{
+    Object::initialize(global_object);
+
+    auto& vm = this->vm();
+
+    // 18.3.2 Intl.Segmenter.prototype [ @@toStringTag ], https://tc39.es/ecma402/#sec-intl.segmenter.prototype-@@tostringtag
+    define_direct_property(*vm.well_known_symbol_to_string_tag(), js_string(vm, "Intl.Segmenter"), Attribute::Configurable);
+}
+
+}

Userland/Libraries/LibJS/Runtime/Intl/SegmenterPrototype.h

@@ -0,0 +1,23 @@
+/*
+ * Copyright (c) 2022, Idan Horowitz <idan.horowitz@serenityos.org>
+ *
+ * SPDX-License-Identifier: BSD-2-Clause
+ */
+
+#pragma once
+
+#include <LibJS/Runtime/Intl/Segmenter.h>
+#include <LibJS/Runtime/PrototypeObject.h>
+
+namespace JS::Intl {
+
+class SegmenterPrototype final : public PrototypeObject<SegmenterPrototype, Segmenter> {
+    JS_PROTOTYPE_OBJECT(SegmenterPrototype, Segmenter, Segmenter);
+
+public:
+    explicit SegmenterPrototype(GlobalObject&);
+    virtual void initialize(GlobalObject&) override;
+    virtual ~SegmenterPrototype() override = default;
+};
+
+}

Userland/Libraries/LibJS/Tests/builtins/Intl/Segmenter/Segmenter.@@toStringTag.js

@@ -0,0 +1,3 @@
+test("basic functionality", () => {
+    expect(Intl.Segmenter.prototype[Symbol.toStringTag]).toBe("Intl.Segmenter");
+});

Userland/Libraries/LibJS/Tests/builtins/Intl/Segmenter/Segmenter.js

@@ -0,0 +1,65 @@
+describe("errors", () => {
+    test("called without new", () => {
+        expect(() => {
+            Intl.Segmenter();
+        }).toThrowWithMessage(TypeError, "Intl.Segmenter constructor must be called with 'new'");
+    });
+
+    test("structurally invalid tag", () => {
+        expect(() => {
+            new Intl.Segmenter("root");
+        }).toThrowWithMessage(RangeError, "root is not a structurally valid language tag");
+
+        expect(() => {
+            new Intl.Segmenter("en-");
+        }).toThrowWithMessage(RangeError, "en- is not a structurally valid language tag");
+
+        expect(() => {
+            new Intl.Segmenter("Latn");
+        }).toThrowWithMessage(RangeError, "Latn is not a structurally valid language tag");
+
+        expect(() => {
+            new Intl.Segmenter("en-u-aa-U-aa");
+        }).toThrowWithMessage(RangeError, "en-u-aa-U-aa is not a structurally valid language tag");
+    });
+
+    test("options is an invalid type", () => {
+        expect(() => {
+            new Intl.Segmenter("en", null);
+        }).toThrowWithMessage(TypeError, "Options is not an object");
+    });
+
+    test("localeMatcher option is invalid", () => {
+        expect(() => {
+            new Intl.Segmenter("en", { localeMatcher: "hello!" });
+        }).toThrowWithMessage(RangeError, "hello! is not a valid value for option localeMatcher");
+    });
+
+    test("granularity option is invalid", () => {
+        expect(() => {
+            new Intl.Segmenter("en", { granularity: "hello!" });
+        }).toThrowWithMessage(RangeError, "hello! is not a valid value for option granularity");
+    });
+});
+
+describe("normal behavior", () => {
+    test("length is 0", () => {
+        expect(Intl.Segmenter).toHaveLength(0);
+    });
+
+    test("all valid localeMatcher options", () => {
+        ["lookup", "best fit"].forEach(localeMatcher => {
+            expect(() => {
+                new Intl.Segmenter("en", { localeMatcher: localeMatcher });
+            }).not.toThrow();
+        });
+    });
+
+    test("all valid granularity options", () => {
+        ["grapheme", "word", "sentence"].forEach(granularity => {
+            expect(() => {
+                new Intl.Segmenter("en", { granularity: granularity });
+            }).not.toThrow();
+        });
+    });
+});