LibJS: Implement String.prototype.isWellFormed

Author: https://github.com/trflynn89
Commit: https://github.com/SerenityOS/serenity/commit/0bb46235a7
Pull-request: https://github.com/SerenityOS/serenity/pull/16278
Reviewed-by: https://github.com/linusg

Userland/Libraries/LibJS/Runtime/CommonPropertyNames.h

@@ -308,6 +308,7 @@ namespace JS {
     P(isoNanosecond)                         \
     P(isoSecond)                             \
     P(isoYear)                               \
+    P(isWellFormed)                          \
     P(isWordLike)                            \
     P(italics)                               \
     P(join)                                  \

Userland/Libraries/LibJS/Runtime/StringPrototype.cpp

@@ -81,6 +81,32 @@ static Optional<size_t> string_index_of(Utf16View const& string, Utf16View const
     return {};
 }
 
+// 7.2.9 Static Semantics: IsStringWellFormedUnicode ( string )
+static bool is_string_well_formed_unicode(Utf16View string)
+{
+    // 1. Let strLen be the length of string.
+    auto length = string.length_in_code_units();
+
+    // 2. Let k be 0.
+    size_t k = 0;
+
+    // 3. Repeat, while k ≠ strLen,
+    while (k != length) {
+        // a. Let cp be CodePointAt(string, k).
+        auto code_point = code_point_at(string, k);
+
+        // b. If cp.[[IsUnpairedSurrogate]] is true, return false.
+        if (code_point.is_unpaired_surrogate)
+            return false;
+
+        // c. Set k to k + cp.[[CodeUnitCount]].
+        k += code_point.code_unit_count;
+    }
+
+    // 4. Return true.
+    return true;
+}
+
 // 11.1.4 CodePointAt ( string, position ), https://tc39.es/ecma262/#sec-codepointat
 CodePoint code_point_at(Utf16View const& string, size_t position)
 {
@@ -124,6 +150,7 @@ void StringPrototype::initialize(Realm& realm)
     define_native_function(realm, vm.names.endsWith, ends_with, 1, attr);
     define_native_function(realm, vm.names.includes, includes, 1, attr);
     define_native_function(realm, vm.names.indexOf, index_of, 1, attr);
+    define_native_function(realm, vm.names.isWellFormed, is_well_formed, 0, attr);
     define_native_function(realm, vm.names.lastIndexOf, last_index_of, 1, attr);
     define_native_function(realm, vm.names.localeCompare, locale_compare, 1, attr);
     define_native_function(realm, vm.names.match, match, 1, attr);
@@ -345,6 +372,17 @@ JS_DEFINE_NATIVE_FUNCTION(StringPrototype::index_of)
     return index.has_value() ? Value(*index) : Value(-1);
 }
 
+// 22.1.3.10 String.prototype.isWellFormed ( ), https://tc39.es/proposal-is-usv-string/#sec-string.prototype.iswellformed
+JS_DEFINE_NATIVE_FUNCTION(StringPrototype::is_well_formed)
+{
+    // 1. Let O be ? RequireObjectCoercible(this value).
+    // 2. Let S be ? ToString(O).
+    auto string = TRY(utf16_string_from(vm));
+
+    // 3. Return IsStringWellFormedUnicode(S).
+    return is_string_well_formed_unicode(string.view());
+}
+
 // 22.1.3.10 String.prototype.lastIndexOf ( searchString [ , position ] ), https://tc39.es/ecma262/#sec-string.prototype.lastindexof
 JS_DEFINE_NATIVE_FUNCTION(StringPrototype::last_index_of)
 {

Userland/Libraries/LibJS/Runtime/StringPrototype.h

@@ -38,6 +38,7 @@ private:
     JS_DECLARE_NATIVE_FUNCTION(ends_with);
     JS_DECLARE_NATIVE_FUNCTION(includes);
     JS_DECLARE_NATIVE_FUNCTION(index_of);
+    JS_DECLARE_NATIVE_FUNCTION(is_well_formed);
     JS_DECLARE_NATIVE_FUNCTION(last_index_of);
     JS_DECLARE_NATIVE_FUNCTION(locale_compare);
     JS_DECLARE_NATIVE_FUNCTION(match);

Userland/Libraries/LibJS/Tests/builtins/String/String.prototype.isWellFormed.js

@@ -0,0 +1,47 @@
+describe("errors", () => {
+    test("called with value that cannot be converted to a string", () => {
+        expect(() => {
+            String.prototype.isWellFormed.call(Symbol.hasInstance);
+        }).toThrowWithMessage(TypeError, "Cannot convert symbol to string");
+    });
+});
+
+describe("basic functionality", () => {
+    test("ascii strings", () => {
+        expect("".isWellFormed()).toBeTrue();
+        expect("foo".isWellFormed()).toBeTrue();
+        expect("abcdefghi".isWellFormed()).toBeTrue();
+    });
+
+    test("valid UTF-16 strings", () => {
+        expect("😀".isWellFormed()).toBeTrue();
+        expect("\ud83d\ude00".isWellFormed()).toBeTrue();
+    });
+
+    test("invalid UTF-16 strings", () => {
+        expect("😀".slice(0, 1).isWellFormed()).toBeFalse();
+        expect("😀".slice(1, 2).isWellFormed()).toBeFalse();
+        expect("\ud83d".isWellFormed()).toBeFalse();
+        expect("\ude00".isWellFormed()).toBeFalse();
+        expect("a\ud83d".isWellFormed()).toBeFalse();
+        expect("a\ude00".isWellFormed()).toBeFalse();
+        expect("\ud83da".isWellFormed()).toBeFalse();
+        expect("\ude00a".isWellFormed()).toBeFalse();
+        expect("a\ud83da".isWellFormed()).toBeFalse();
+        expect("a\ude00a".isWellFormed()).toBeFalse();
+    });
+
+    test("object converted to a string", () => {
+        let toStringCalled = false;
+
+        const obj = {
+            toString: function () {
+                toStringCalled = true;
+                return "toString";
+            },
+        };
+
+        expect(String.prototype.isWellFormed.call(obj)).toBeTrue();
+        expect(toStringCalled).toBeTrue();
+    });
+});