lib,src: replace toUSVString with toWellFormed()

Author: anonrig

benchmark/url/usvstring.js

@@ -1949,6 +1949,10 @@ The encoding supported by the `TextEncoder` instance. Always set to `'utf-8'`.
 added:
   - v16.8.0
   - v14.18.0
+changes:
+  - version: REPLACEME
+    pr-url: https://github.com/nodejs/node/pull/47342
+    description: Implementation uses String.prototype.toWellFormed.
 -->
 
 * `string` {string}

doc/api/util.md

@@ -15,7 +15,6 @@ const {
   customInspectSymbol: kInspect,
   kEnumerableProperty,
   kEmptyObject,
-  toUSVString,
 } = require('internal/util');
 
 const {
@@ -55,7 +54,7 @@ class File extends Blob {
       lastModified = DateNow();
     }
 
-    this.#name = toUSVString(fileName);
+    this.#name = `${fileName}`.toWellFormed();
     this.#lastModified = lastModified;
   }
 

lib/internal/file.js

@@ -43,7 +43,6 @@ const {
 const {
   getConstructorOf,
   removeColors,
-  toUSVString,
   kEnumerableProperty,
   SideEffectFreeRegExpPrototypeSymbolReplace,
 } = require('internal/util');
@@ -181,7 +180,7 @@ class URLSearchParams {
           }
           const convertedPair = [];
           for (const element of pair)
-            ArrayPrototypePush(convertedPair, toUSVString(element));
+            ArrayPrototypePush(convertedPair, `${element}`.toWellFormed());
           ArrayPrototypePush(pairs, convertedPair);
         }
 
@@ -202,10 +201,10 @@ class URLSearchParams {
           const key = keys[i];
           const desc = ReflectGetOwnPropertyDescriptor(init, key);
           if (desc !== undefined && desc.enumerable) {
-            const typedKey = toUSVString(key);
-            const typedValue = toUSVString(init[key]);
+            const typedKey = `${key}`.toWellFormed();
+            const typedValue = `${init[key]}`.toWellFormed();
 
-            // Two different key may result same after `toUSVString()`, we only
+            // Two different key may result same after `String.prototype.toWellFormed`, we only
             // leave the later one. Refers to WPT.
             if (visited[typedKey] !== undefined) {
               this[searchParams][visited[typedKey]] = typedValue;
@@ -219,7 +218,7 @@ class URLSearchParams {
       }
     } else {
       // USVString
-      init = toUSVString(init);
+      init = `${init}`.toWellFormed();
       this[searchParams] = init ? parseParams(init) : [];
     }
 
@@ -277,8 +276,8 @@ class URLSearchParams {
       throw new ERR_MISSING_ARGS('name', 'value');
     }
 
-    name = toUSVString(name);
-    value = toUSVString(value);
+    name = `${name}`.toWellFormed();
+    value = `${value}`.toWellFormed();
     ArrayPrototypePush(this[searchParams], name, value);
     if (this[context]) {
       this[context].search = this.toString();
@@ -294,7 +293,7 @@ class URLSearchParams {
     }
 
     const list = this[searchParams];
-    name = toUSVString(name);
+    name = `${name}`.toWellFormed();
     for (let i = 0; i < list.length;) {
       const cur = list[i];
       if (cur === name) {
@@ -317,7 +316,7 @@ class URLSearchParams {
     }
 
     const list = this[searchParams];
-    name = toUSVString(name);
+    name = `${name}`.toWellFormed();
     for (let i = 0; i < list.length; i += 2) {
       if (list[i] === name) {
         return list[i + 1];
@@ -336,7 +335,7 @@ class URLSearchParams {
 
     const list = this[searchParams];
     const values = [];
-    name = toUSVString(name);
+    name = `${name}`.toWellFormed();
     for (let i = 0; i < list.length; i += 2) {
       if (list[i] === name) {
         values.push(list[i + 1]);
@@ -354,7 +353,7 @@ class URLSearchParams {
     }
 
     const list = this[searchParams];
-    name = toUSVString(name);
+    name = `${name}`.toWellFormed();
     for (let i = 0; i < list.length; i += 2) {
       if (list[i] === name) {
         return true;
@@ -372,8 +371,8 @@ class URLSearchParams {
     }
 
     const list = this[searchParams];
-    name = toUSVString(name);
-    value = toUSVString(value);
+    name = `${name}`.toWellFormed();
+    value = `${value}`.toWellFormed();
 
     // If there are any name-value pairs whose name is `name`, in `list`, set
     // the value of the first such name-value pair to `value` and remove the
@@ -549,7 +548,7 @@ class URL {
   #searchParams;
 
   constructor(input, base = undefined) {
-    // toUSVString is not needed.
+    // String.prototype.toWellFormed is not needed.
     input = `${input}`;
 
     if (base !== undefined) {
@@ -690,7 +689,7 @@ class URL {
   }
 
   set search(value) {
-    updateUrl(this.#context.href, updateActions.kSearch, toUSVString(value), this.#onParseComplete);
+    updateUrl(this.#context.href, updateActions.kSearch, `${value}`.toWellFormed(), this.#onParseComplete);
   }
 
   // readonly
@@ -1106,15 +1105,15 @@ function domainToASCII(domain) {
   if (arguments.length < 1)
     throw new ERR_MISSING_ARGS('domain');
 
-  // toUSVString is not needed.
+  // String.prototype.toWellFormed is not needed.
   return _domainToASCII(`${domain}`);
 }
 
 function domainToUnicode(domain) {
   if (arguments.length < 1)
     throw new ERR_MISSING_ARGS('domain');
 
-  // toUSVString is not needed.
+  // String.prototype.toWellFormed is not needed.
   return _domainToUnicode(`${domain}`);
 }
 
@@ -1287,7 +1286,6 @@ function toPathIfFileURL(fileURLOrPath) {
 }
 
 module.exports = {
-  toUSVString,
   fileURLToPath,
   pathToFileURL,
   toPathIfFileURL,

lib/internal/url.js

@@ -59,7 +59,6 @@ const {
     decorated_private_symbol,
   },
   sleep: _sleep,
-  toUSVString: _toUSVString,
 } = internalBinding('util');
 const { isNativeError } = internalBinding('types');
 
@@ -69,10 +68,6 @@ const experimentalWarnings = new SafeSet();
 
 const colorRegExp = /\u001b\[\d\d?m/g; // eslint-disable-line no-control-regex
 
-function toUSVString(val) {
-  return _toUSVString(`${val}`);
-}
-
 let uvBinding;
 
 function lazyUv() {
@@ -786,7 +781,6 @@ module.exports = {
   sleep,
   spliceOne,
   setupCoverageHooks,
-  toUSVString,
   removeColors,
 
   // Symbol used to customize promisify conversion

lib/internal/util.js

@@ -74,7 +74,6 @@ const {
   getSystemErrorMap,
   getSystemErrorName: internalErrorName,
   promisify,
-  toUSVString,
   defineLazyProperties,
 } = require('internal/util');
 
@@ -346,6 +345,14 @@ function getSystemErrorName(err) {
   return internalErrorName(err);
 }
 
+/**
+ * @param {string} input
+ * @returns {string}
+ */
+function toUSVString(input) {
+  return `${input}`.toWellFormed();
+}
+
 // Keep the `exports =` so that various functions can still be monkeypatched
 module.exports = {
   _errnoException: errnoException,

lib/util.js

@@ -36,9 +36,6 @@ using v8::String;
 using v8::Uint32;
 using v8::Value;
 
-// Used in ToUSVString().
-constexpr char16_t kUnicodeReplacementCharacter = 0xFFFD;
-
 // If a UTF-16 character is a low/trailing surrogate.
 CHAR_TEST(16, IsUnicodeTrail, (ch & 0xFC00) == 0xDC00)
 
@@ -320,36 +317,6 @@ static void GuessHandleType(const FunctionCallbackInfo<Value>& args) {
   args.GetReturnValue().Set(OneByteString(env->isolate(), type));
 }
 
-static void ToUSVString(const FunctionCallbackInfo<Value>& args) {
-  Environment* env = Environment::GetCurrent(args);
-  CHECK_GE(args.Length(), 1);
-  CHECK(args[0]->IsString());
-
-  TwoByteValue value(env->isolate(), args[0]);
-
-  for (size_t i = 0; i < value.length(); i++) {
-    char16_t c = value[i];
-    if (!IsUnicodeSurrogate(c)) {
-      continue;
-    } else if (IsUnicodeSurrogateTrail(c) || i == value.length() - 1) {
-      value[i] = kUnicodeReplacementCharacter;
-    } else {
-      char16_t d = value[i + 1];
-      if (IsUnicodeTrail(d)) {
-        i++;
-      } else {
-        value[i] = kUnicodeReplacementCharacter;
-      }
-    }
-  }
-
-  args.GetReturnValue().Set(
-      String::NewFromTwoByte(env->isolate(),
-                             *value,
-                             v8::NewStringType::kNormal,
-                             value.length()).ToLocalChecked());
-}
-
 void RegisterExternalReferences(ExternalReferenceRegistry* registry) {
   registry->Register(GetPromiseDetails);
   registry->Register(GetProxyDetails);
@@ -366,7 +333,6 @@ void RegisterExternalReferences(ExternalReferenceRegistry* registry) {
   registry->Register(WeakReference::IncRef);
   registry->Register(WeakReference::DecRef);
   registry->Register(GuessHandleType);
-  registry->Register(ToUSVString);
 }
 
 void Initialize(Local<Object> target,
@@ -472,8 +438,6 @@ void Initialize(Local<Object> target,
   SetConstructorFunction(context, target, "WeakReference", weak_ref);
 
   SetMethod(context, target, "guessHandleType", GuessHandleType);
-
-  SetMethodNoSideEffect(context, target, "toUSVString", ToUSVString);
 }
 
 }  // namespace util