AK+LibJS+LibWeb: Use simdutf to create well-formed strings

Author: https://github.com/trflynn89
Commit: 1375e6bf39
Pull-request: https://github.com/LadybirdBrowser/ladybird/pull/5608
Reviewed-by: https://github.com/gmta ✅

AK/Utf16String.cpp

@@ -114,6 +114,20 @@ Utf16String Utf16String::repeated(u32 code_point, size_t count)
     return builder.to_utf16_string();
 }
 
+Utf16String Utf16String::to_well_formed() const
+{
+    if (utf16_view().validate(AllowLonelySurrogates::No))
+        return *this;
+    return Utf16String { Detail::Utf16StringData::to_well_formed(*this) };
+}
+
+String Utf16String::to_well_formed_utf8() const
+{
+    if (utf16_view().validate(AllowLonelySurrogates::No))
+        return to_utf8(AllowLonelySurrogates::No);
+    return to_well_formed().to_utf8(AllowLonelySurrogates::No);
+}
+
 ErrorOr<void> Formatter<Utf16String>::format(FormatBuilder& builder, Utf16String const& utf16_string)
 {
     if (utf16_string.has_long_utf16_storage())

AK/Utf16String.h

@@ -138,6 +138,9 @@ public:
         return from_string_builder_without_validation(builder);
     }
 
+    Utf16String to_well_formed() const;
+    String to_well_formed_utf8() const;
+
     // These methods require linking LibUnicode.
     Utf16String to_lowercase(Optional<StringView> const& locale = {}) const;
     Utf16String to_uppercase(Optional<StringView> const& locale = {}) const;

AK/Utf16StringData.cpp

@@ -158,6 +158,16 @@ NonnullRefPtr<Utf16StringData> Utf16StringData::from_string_builder(StringBuilde
     return adopt_ref(*new (buffer->buffer.data()) Utf16StringData { storage_type, code_unit_length });
 }
 
+NonnullRefPtr<Utf16StringData> Utf16StringData::to_well_formed(Utf16View const& utf16_string)
+{
+    VERIFY(!utf16_string.has_ascii_storage());
+
+    auto string = create_uninitialized(StorageType::UTF16, utf16_string.length_in_code_units());
+    simdutf::to_well_formed_utf16(utf16_string.utf16_span().data(), utf16_string.length_in_code_units(), string->m_utf16_data);
+
+    return string;
+}
+
 size_t Utf16StringData::calculate_code_point_length() const
 {
     ASSERT(!has_ascii_storage());

AK/Utf16StringData.h

@@ -35,6 +35,8 @@ public:
     static NonnullRefPtr<Utf16StringData> from_utf32(Utf32View const&);
     static NonnullRefPtr<Utf16StringData> from_string_builder(StringBuilder&);
 
+    static NonnullRefPtr<Utf16StringData> to_well_formed(Utf16View const&);
+
     ~Utf16StringData()
     {
         if (is_fly_string())

Libraries/LibJS/Runtime/StringPrototype.cpp

@@ -1362,44 +1362,17 @@ JS_DEFINE_NATIVE_FUNCTION(StringPrototype::to_well_formed)
     // 2. Let S be ? ToString(O).
     auto string = TRY(primitive_string_from(vm));
 
-    // NOTE: Rest of steps in to_well_formed below
-    return PrimitiveString::create(vm, to_well_formed_string(string->utf16_string()));
-}
-
-// https://tc39.es/ecma262/#sec-string.prototype.towellformed
-String to_well_formed_string(Utf16String const& string)
-{
     // 3. Let strLen be the length of S.
-    auto length = string.length_in_code_units();
-
     // 4. Let k be 0.
-    size_t k = 0;
-
     // 5. Let result be the empty String.
-    StringBuilder result;
-
     // 6. Repeat, while k < strLen,
-    while (k < length) {
-        // a. Let cp be CodePointAt(S, k).
-        auto code_point = JS::code_point_at(string, k);
-
-        // b. If cp.[[IsUnpairedSurrogate]] is true, then
-        if (code_point.is_unpaired_surrogate) {
-            // i. Set result to the string-concatenation of result and 0xFFFD (REPLACEMENT CHARACTER).
-            result.append_code_point(0xfffd);
-        }
-        // c. Else,
-        else {
-            // i. Set result to the string-concatenation of result and UTF16EncodeCodePoint(cp.[[CodePoint]]).
-            result.append_code_point(code_point.code_point);
-        }
-
-        // d. Set k to k + cp.[[CodeUnitCount]].
-        k += code_point.code_unit_count;
-    }
-
-    // 7. Return result.
-    return MUST(result.to_string());
+    //     a. Let cp be CodePointAt(S, k).
+    //     b. If cp.[[IsUnpairedSurrogate]] is true, then
+    //         i. Set result to the string-concatenation of result and 0xFFFD (REPLACEMENT CHARACTER).
+    //     c. Else,
+    //         i. Set result to the string-concatenation of result and UTF16EncodeCodePoint(cp.[[CodePoint]]).
+    //     d. Set k to k + cp.[[CodeUnitCount]].
+    return PrimitiveString::create(vm, string->utf16_string().to_well_formed());
 }
 
 // 22.1.3.32.1 TrimString ( string, where ), https://tc39.es/ecma262/#sec-trimstring

Libraries/LibJS/Runtime/StringPrototype.h

@@ -19,7 +19,6 @@ struct CodePoint {
 
 Optional<size_t> string_index_of(Utf16View const& string, Utf16View const& search_value, size_t from_index);
 CodePoint code_point_at(Utf16View const& string, size_t position);
-String to_well_formed_string(Utf16String const&);
 static constexpr Utf8View whitespace_characters = Utf8View("\x09\x0A\x0B\x0C\x0D\x20\xC2\xA0\xE1\x9A\x80\xE2\x80\x80\xE2\x80\x81\xE2\x80\x82\xE2\x80\x83\xE2\x80\x84\xE2\x80\x85\xE2\x80\x86\xE2\x80\x87\xE2\x80\x88\xE2\x80\x89\xE2\x80\x8A\xE2\x80\xAF\xE2\x81\x9F\xE3\x80\x80\xE2\x80\xA8\xE2\x80\xA9\xEF\xBB\xBF"sv);
 ThrowCompletionOr<String> trim_string(VM&, Value string, TrimMode where);
 

Libraries/LibJS/Runtime/Value.cpp

@@ -458,11 +458,6 @@ ThrowCompletionOr<Utf16String> Value::to_utf16_string(VM& vm) const
     return Utf16String::from_utf8(utf8_string);
 }
 
-ThrowCompletionOr<String> Value::to_well_formed_string(VM& vm) const
-{
-    return ::JS::to_well_formed_string(TRY(to_utf16_string(vm)));
-}
-
 // 7.1.2 ToBoolean ( argument ), https://tc39.es/ecma262/#sec-toboolean
 bool Value::to_boolean_slow_case() const
 {

Libraries/LibJS/Runtime/Value.h

@@ -352,7 +352,6 @@ public:
     ThrowCompletionOr<String> to_string(VM&) const;
     ThrowCompletionOr<ByteString> to_byte_string(VM&) const;
     ThrowCompletionOr<Utf16String> to_utf16_string(VM&) const;
-    ThrowCompletionOr<String> to_well_formed_string(VM&) const;
     ThrowCompletionOr<GC::Ref<PrimitiveString>> to_primitive_string(VM&);
     ThrowCompletionOr<Value> to_primitive(VM&, PreferredType preferred_type = PreferredType::Default) const;
     ThrowCompletionOr<GC::Ref<Object>> to_object(VM&) const;

Libraries/LibWeb/WebIDL/AbstractOperations.cpp

@@ -241,7 +241,7 @@ JS::ThrowCompletionOr<Utf16String> to_utf16_string(JS::VM& vm, JS::Value value)
 
 JS::ThrowCompletionOr<String> to_usv_string(JS::VM& vm, JS::Value value)
 {
-    return value.to_well_formed_string(vm);
+    return TRY(value.to_utf16_string(vm)).to_well_formed_utf8();
 }
 
 // https://webidl.spec.whatwg.org/#invoke-a-callback-function