Implement String.prototype.{ isWellFormed, toWellFormed }

Signed-off-by: Seonghyun Kim <[email protected]>
Samsung · Jul 4, 2024 · d59154a · d59154a
1 parent 18ec8bc
commit d59154a
Show file tree

Hide file tree

Showing 5 changed files with 107 additions and 16 deletions.
diff --git a/src/builtins/BuiltinString.cpp b/src/builtins/BuiltinString.cpp
@@ -1575,6 +1575,36 @@ static Value builtinStringIncludes(ExecutionState& state, Value thisValue, size_
     return Value(true);
 }
 
+// https://tc39.es/ecma262/multipage/text-processing.html#sec-string.prototype.iswellformed
+static Value builtinStringIsWellFormed(ExecutionState& state, Value thisValue, size_t argc, Value* argv, Optional<Object*> newTarget)
+{
+    // Let O be ? RequireObjectCoercible(this value).
+    // Let S be ? ToString(O).
+    RESOLVE_THIS_BINDING_TO_STRING(S, String, isWellFormed);
+    // Return IsStringWellFormedUnicode(S)
+    return Value(S->isWellFormed());
+}
+
+// https://tc39.es/ecma262/multipage/text-processing.html#sec-string.prototype.towellformed
+static Value builtinStringToWellFormed(ExecutionState& state, Value thisValue, size_t argc, Value* argv, Optional<Object*> newTarget)
+{
+    // Let O be ? RequireObjectCoercible(this value).
+    // Let S be ? ToString(O).
+    RESOLVE_THIS_BINDING_TO_STRING(S, String, toWellFormed);
+    // Let strLen be the length of S.
+    // Let k be 0.
+    // Let result be the empty String.
+    // Repeat, while k < strLen,
+    //   Let cp be CodePointAt(S, k).
+    //   If cp.[[IsUnpairedSurrogate]] is true, then
+    //     Set result to the string-concatenation of result and 0xFFFD (REPLACEMENT CHARACTER).
+    //   Else,
+    //     Set result to the string-concatenation of result and UTF16EncodeCodePoint(cp.[[CodePoint]]).
+    //   Set k to k + cp.[[CodeUnitCount]].
+    // Return result.
+    return S->toWellFormed();
+}
+
 static Value builtinStringIteratorNext(ExecutionState& state, Value thisValue, size_t argc, Value* argv, Optional<Object*> newTarget)
 {
     if (!thisValue.isObject() || !thisValue.asObject()->isStringIteratorObject()) {
@@ -1745,6 +1775,14 @@ void GlobalObject::installString(ExecutionState& state)
                                                ObjectPropertyDescriptor(new NativeFunctionObject(state, NativeFunctionInfo(strings->at, builtinStringAt, 1, NativeFunctionInfo::Strict)),
                                                                         (ObjectPropertyDescriptor::PresentAttribute)(ObjectPropertyDescriptor::WritablePresent | ObjectPropertyDescriptor::ConfigurablePresent)));
 
+    m_stringPrototype->directDefineOwnProperty(state, ObjectPropertyName(strings->isWellFormed),
+                                               ObjectPropertyDescriptor(new NativeFunctionObject(state, NativeFunctionInfo(strings->isWellFormed, builtinStringIsWellFormed, 0, NativeFunctionInfo::Strict)),
+                                                                        (ObjectPropertyDescriptor::PresentAttribute)(ObjectPropertyDescriptor::WritablePresent | ObjectPropertyDescriptor::ConfigurablePresent)));
+
+    m_stringPrototype->directDefineOwnProperty(state, ObjectPropertyName(strings->toWellFormed),
+                                               ObjectPropertyDescriptor(new NativeFunctionObject(state, NativeFunctionInfo(strings->toWellFormed, builtinStringToWellFormed, 0, NativeFunctionInfo::Strict)),
+                                                                        (ObjectPropertyDescriptor::PresentAttribute)(ObjectPropertyDescriptor::WritablePresent | ObjectPropertyDescriptor::ConfigurablePresent)));
+
 #define DEFINE_STRING_ADDITIONAL_HTML_FUNCTION(fnName, argLength)                                                                                                                                           \
     m_stringPrototype->directDefineOwnProperty(state, ObjectPropertyName(strings->fnName),                                                                                                                  \
                                                ObjectPropertyDescriptor(new NativeFunctionObject(state, NativeFunctionInfo(strings->fnName, builtinString##fnName, argLength, NativeFunctionInfo::Strict)), \

diff --git a/src/runtime/StaticStrings.h b/src/runtime/StaticStrings.h
@@ -303,6 +303,7 @@ namespace Escargot {
     F(isPrototypeOf)              \
     F(isSafeInteger)              \
     F(isSealed)                   \
+    F(isWellFormed)               \
     F(isView)                     \
     F(italics)                    \
     F(iterator)                   \
@@ -468,6 +469,7 @@ namespace Escargot {
     F(toTimeString)               \
     F(toUTCString)                \
     F(toUpperCase)                \
+    F(toWellFormed)               \
     F(transfer)                   \
     F(trim)                       \
     F(trimEnd)                    \

diff --git a/src/runtime/String.cpp b/src/runtime/String.cpp
@@ -359,6 +359,30 @@ size_t utf32ToUtf16(char32_t i, char16_t* u)
     }
 }
 
+bool isWellFormed(const char16_t*& utf16, const char16_t* bufferEnd)
+{
+    if (utf16[0] >= 0xd800 && utf16[0] <= 0xdbff) {
+        if (utf16 + 1 < bufferEnd) {
+            if (utf16[1] >= 0xdc00 && utf16[1] <= 0xdfff) {
+                utf16 += 2;
+                return true;
+            } else {
+                utf16 += 1;
+                return false;
+            }
+        } else {
+            utf16 += 1;
+            return false;
+        }
+    } else if (utf16[0] >= 0xdc00 && utf16[0] <= 0xdfff) {
+        utf16 += 1;
+        return false;
+    } else {
+        utf16 += 1;
+        return true;
+    }
+}
+
 bool StringBufferAccessData::equals16Bit(const char16_t* c1, const char* c2, size_t len)
 {
     while (len > 0) {
@@ -844,6 +868,45 @@ uint32_t String::tryToUseAsIndexProperty() const
     return tryToUseAsIndex32();
 }
 
+bool String::isWellFormed() const
+{
+    auto bad = bufferAccessData();
+    if (bad.has8BitContent) {
+        return true;
+    }
+    auto utf16 = bad.bufferAs16Bit;
+    auto end = bad.bufferAs16Bit + bad.length;
+    while (utf16 < end) {
+        if (!::Escargot::isWellFormed(utf16, end)) {
+            return false;
+        }
+    }
+    return true;
+}
+
+String* String::toWellFormed()
+{
+    if (isWellFormed()) {
+        return this;
+    }
+    auto bad = bufferAccessData();
+    auto utf16 = bad.bufferAs16Bit;
+    auto end = bad.bufferAs16Bit + bad.length;
+    UTF16StringData result;
+    while (utf16 < end) {
+        auto start = utf16;
+        if (::Escargot::isWellFormed(utf16, end)) {
+            while (start < utf16) {
+                result.pushBack(*start);
+                start++;
+            }
+        } else {
+            result.pushBack(0xfffd);
+        }
+    }
+    return new UTF16String(std::move(result));
+}
+
 size_t String::find(String* str, size_t pos) const
 {
     const size_t srcStrLen = str->length();

diff --git a/src/runtime/String.h b/src/runtime/String.h
@@ -62,6 +62,7 @@ ASCIIStringData utf16StringToASCIIString(const char16_t* buf, const size_t len);
 ASCIIStringDataNonGCStd dtoa(double number);
 size_t utf32ToUtf8(char32_t uc, char* UTF8);
 size_t utf32ToUtf16(char32_t i, char16_t* u);
+bool isWellFormed(const char16_t*& utf16, const char16_t* bufferEnd);
 
 // these functions only care ascii range(0~127)
 bool islower(char32_t ch);
@@ -534,6 +535,9 @@ class String : public PointerValue {
         return has8BitContent();
     }
 
+    bool isWellFormed() const;
+    String* toWellFormed();
+
     template <typename Any>
     const Any* characters() const
     {

diff --git a/tools/test/test262/excludelist.orig.xml b/tools/test/test262/excludelist.orig.xml
@@ -1189,24 +1189,8 @@
     <test id="built-ins/ShadowRealm/prototype/importValue/throws-typeerror-import-throws"><reason>TODO</reason></test>
     <test id="built-ins/ShadowRealm/prototype/importValue/validates-realm-object"><reason>TODO</reason></test>
     <test id="built-ins/ShadowRealm/prototype/proto"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/isWellFormed/length"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/isWellFormed/name"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/isWellFormed/not-a-constructor"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/isWellFormed/prop-desc"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/isWellFormed/return-abrupt-from-this"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/isWellFormed/returns-boolean"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/isWellFormed/to-string"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/isWellFormed/to-string-primitive"><reason>TODO</reason></test>
     <test id="built-ins/String/prototype/match/duplicate-named-groups-properties"><reason>TODO</reason></test>
     <test id="built-ins/String/prototype/match/duplicate-named-indices-groups-properties"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/toWellFormed/length"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/toWellFormed/name"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/toWellFormed/not-a-constructor"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/toWellFormed/prop-desc"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/toWellFormed/return-abrupt-from-this"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/toWellFormed/returns-well-formed-string"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/toWellFormed/to-string"><reason>TODO</reason></test>
-    <test id="built-ins/String/prototype/toWellFormed/to-string-primitive"><reason>TODO</reason></test>
     <test id="built-ins/Temporal/Calendar/argument-wrong-type"><reason>TODO</reason></test>
     <test id="built-ins/Temporal/Calendar/from/calendar-case-insensitive"><reason>TODO</reason></test>
     <test id="built-ins/Temporal/Calendar/from/calendar-number"><reason>TODO</reason></test>