lib,src: replace toUSVString with toWellFormed()

PR-URL: https://github.com/nodejs/node/pull/47342
Reviewed-By: Michaël Zasso <targos@protonmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: Chengzhong Wu <legendecas@gmail.com>

benchmark/url/usvstring.js

@@ -1,27 +0,0 @@
-'use strict';
-const common = require('../common.js');
-
-const inputs = {
-  valid: 'adsfadsfadsf',
-  validsurr: '\uda23\ude23\uda1f\udfaa\ud800\udfff\uda23\ude23\uda1f\udfaa' +
-             '\ud800\udfff',
-  someinvalid: 'asasfdfasd\uda23',
-  allinvalid: '\udc45\uda23 \udf00\udc00 \udfaa\uda12 \udc00\udfaa',
-  nonstring: { toString() { return 'asdf'; } },
-};
-const bench = common.createBenchmark(main, {
-  input: Object.keys(inputs),
-  n: [5e7],
-}, {
-  flags: ['--expose-internals'],
-});
-
-function main({ input, n }) {
-  const { toUSVString } = require('internal/url');
-  const str = inputs[input];
-
-  bench.start();
-  for (let i = 0; i < n; i++)
-    toUSVString(str);
-  bench.end(n);
-}

benchmark/util/to-usv-string.js

@@ -1,21 +0,0 @@
-'use strict';
-
-const common = require('../common');
-
-const BASE = 'string\ud801';
-
-const bench = common.createBenchmark(main, {
-  n: [1e5],
-  size: [10, 100, 500],
-});
-
-function main({ n, size }) {
-  const { toUSVString } = require('util');
-  const str = BASE.repeat(size);
-
-  bench.start();
-  for (let i = 0; i < n; i++) {
-    toUSVString(str);
-  }
-  bench.end(n);
-}

lib/internal/file.js

@@ -4,6 +4,7 @@ const {
   DateNow,
   NumberIsNaN,
   ObjectDefineProperties,
+  StringPrototypeToWellFormed,
   SymbolToStringTag,
 } = primordials;
 
@@ -15,7 +16,6 @@ const {
   customInspectSymbol: kInspect,
   kEnumerableProperty,
   kEmptyObject,
-  toUSVString,
 } = require('internal/util');
 
 const {
@@ -55,7 +55,7 @@ class File extends Blob {
       lastModified = DateNow();
     }
 
-    this.#name = toUSVString(fileName);
+    this.#name = StringPrototypeToWellFormed(`${fileName}`);
     this.#lastModified = lastModified;
   }
 

lib/internal/url.js

@@ -26,6 +26,7 @@ const {
   StringPrototypeIndexOf,
   StringPrototypeSlice,
   StringPrototypeStartsWith,
+  StringPrototypeToWellFormed,
   Symbol,
   SymbolIterator,
   SymbolToStringTag,
@@ -42,7 +43,6 @@ const {
 const {
   getConstructorOf,
   removeColors,
-  toUSVString,
   kEnumerableProperty,
   SideEffectFreeRegExpPrototypeSymbolReplace,
 } = require('internal/util');
@@ -366,7 +366,11 @@ class URLSearchParams {
               throw new ERR_INVALID_TUPLE('Each query pair', '[name, value]');
             }
             // Append (innerSequence[0], innerSequence[1]) to querys list.
-            ArrayPrototypePush(this.#searchParams, toUSVString(pair[0]), toUSVString(pair[1]));
+            ArrayPrototypePush(
+              this.#searchParams,
+              StringPrototypeToWellFormed(`${pair[0]}`),
+              StringPrototypeToWellFormed(`${pair[1]}`),
+            );
           } else {
             if (((typeof pair !== 'object' && typeof pair !== 'function') ||
                 typeof pair[SymbolIterator] !== 'function')) {
@@ -377,7 +381,7 @@ class URLSearchParams {
 
             for (const element of pair) {
               length++;
-              ArrayPrototypePush(this.#searchParams, toUSVString(element));
+              ArrayPrototypePush(this.#searchParams, StringPrototypeToWellFormed(`${element}`));
             }
 
             // If innerSequence's size is not 2, then throw a TypeError.
@@ -395,8 +399,8 @@ 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 = StringPrototypeToWellFormed(key);
+            const typedValue = StringPrototypeToWellFormed(`${init[key]}`);
 
             // Two different keys may become the same USVString after normalization.
             // In that case, we retain the later one. Refer to WPT.
@@ -413,7 +417,7 @@ class URLSearchParams {
       }
     } else {
       // https://url.spec.whatwg.org/#dom-urlsearchparams-urlsearchparams
-      init = toUSVString(init);
+      init = StringPrototypeToWellFormed(`${init}`);
       this.#searchParams = init ? parseParams(init) : [];
     }
   }
@@ -468,8 +472,8 @@ class URLSearchParams {
       throw new ERR_MISSING_ARGS('name', 'value');
     }
 
-    name = toUSVString(name);
-    value = toUSVString(value);
+    name = StringPrototypeToWellFormed(`${name}`);
+    value = StringPrototypeToWellFormed(`${value}`);
     ArrayPrototypePush(this.#searchParams, name, value);
     if (this.#context) {
       this.#context.search = this.toString();
@@ -485,10 +489,10 @@ class URLSearchParams {
     }
 
     const list = this.#searchParams;
-    name = toUSVString(name);
+    name = StringPrototypeToWellFormed(`${name}`);
 
     if (value !== undefined) {
-      value = toUSVString(value);
+      value = StringPrototypeToWellFormed(`${value}`);
       for (let i = 0; i < list.length;) {
         if (list[i] === name && list[i + 1] === value) {
           list.splice(i, 2);
@@ -519,7 +523,7 @@ class URLSearchParams {
     }
 
     const list = this.#searchParams;
-    name = toUSVString(name);
+    name = StringPrototypeToWellFormed(`${name}`);
     for (let i = 0; i < list.length; i += 2) {
       if (list[i] === name) {
         return list[i + 1];
@@ -538,7 +542,7 @@ class URLSearchParams {
 
     const list = this.#searchParams;
     const values = [];
-    name = toUSVString(name);
+    name = StringPrototypeToWellFormed(`${name}`);
     for (let i = 0; i < list.length; i += 2) {
       if (list[i] === name) {
         values.push(list[i + 1]);
@@ -556,10 +560,10 @@ class URLSearchParams {
     }
 
     const list = this.#searchParams;
-    name = toUSVString(name);
+    name = StringPrototypeToWellFormed(`${name}`);
 
     if (value !== undefined) {
-      value = toUSVString(value);
+      value = StringPrototypeToWellFormed(`${value}`);
     }
 
     for (let i = 0; i < list.length; i += 2) {
@@ -582,8 +586,8 @@ class URLSearchParams {
     }
 
     const list = this.#searchParams;
-    name = toUSVString(name);
-    value = toUSVString(value);
+    name = StringPrototypeToWellFormed(`${name}`);
+    value = StringPrototypeToWellFormed(`${value}`);
 
     // 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
@@ -773,7 +777,7 @@ class URL {
       throw new ERR_MISSING_ARGS('url');
     }
 
-    // toUSVString is not needed.
+    // StringPrototypeToWellFormed is not needed.
     input = `${input}`;
 
     if (base !== undefined) {
@@ -1006,7 +1010,7 @@ class URL {
   }
 
   set search(value) {
-    const href = bindingUrl.update(this.#context.href, updateActions.kSearch, toUSVString(value));
+    const href = bindingUrl.update(this.#context.href, updateActions.kSearch, StringPrototypeToWellFormed(`${value}`));
     if (href) {
       this.#updateContext(href);
     }
@@ -1297,7 +1301,7 @@ function domainToASCII(domain) {
   if (arguments.length < 1)
     throw new ERR_MISSING_ARGS('domain');
 
-  // toUSVString is not needed.
+  // StringPrototypeToWellFormed is not needed.
   return bindingUrl.domainToASCII(`${domain}`);
 }
 
@@ -1305,7 +1309,7 @@ function domainToUnicode(domain) {
   if (arguments.length < 1)
     throw new ERR_MISSING_ARGS('domain');
 
-  // toUSVString is not needed.
+  // StringPrototypeToWellFormed is not needed.
   return bindingUrl.domainToUnicode(`${domain}`);
 }
 
@@ -1501,7 +1505,6 @@ function getURLOrigin(url) {
 }
 
 module.exports = {
-  toUSVString,
   fileURLToPath,
   pathToFileURL,
   toPathIfFileURL,

lib/internal/util.js

@@ -62,7 +62,6 @@ const {
     decorated_private_symbol,
   },
   sleep: _sleep,
-  toUSVString: _toUSVString,
 } = internalBinding('util');
 const { isNativeError, isPromise } = internalBinding('types');
 const { getOptionValue } = require('internal/options');
@@ -73,18 +72,6 @@ const experimentalWarnings = new SafeSet();
 
 const colorRegExp = /\u001b\[\d\d?m/g; // eslint-disable-line no-control-regex
 
-const unpairedSurrogateRe =
-  /(?:[^\uD800-\uDBFF]|^)[\uDC00-\uDFFF]|[\uD800-\uDBFF](?![\uDC00-\uDFFF])/;
-function toUSVString(val) {
-  const str = `${val}`;
-  // As of V8 5.5, `str.search()` (and `unpairedSurrogateRe[@@search]()`) are
-  // slower than `unpairedSurrogateRe.exec()`.
-  const match = RegExpPrototypeExec(unpairedSurrogateRe, str);
-  if (!match)
-    return str;
-  return _toUSVString(str, match.index);
-}
-
 let uvBinding;
 
 function lazyUv() {
@@ -913,7 +900,6 @@ module.exports = {
   sleep,
   spliceOne,
   setupCoverageHooks,
-  toUSVString,
   removeColors,
 
   // Symbol used to customize promisify conversion

lib/util.js

@@ -44,6 +44,7 @@ const {
   ObjectValues,
   ReflectApply,
   StringPrototypePadStart,
+  StringPrototypeToWellFormed,
 } = primordials;
 
 const {
@@ -75,7 +76,6 @@ const {
   getSystemErrorMap,
   getSystemErrorName: internalErrorName,
   promisify,
-  toUSVString,
   defineLazyProperties,
 } = require('internal/util');
 
@@ -452,7 +452,9 @@ module.exports = {
                  'DEP0059'),
   promisify,
   stripVTControlCharacters,
-  toUSVString,
+  toUSVString(input) {
+    return StringPrototypeToWellFormed(`${input}`);
+  },
   get transferableAbortSignal() {
     return lazyAbortController().transferableAbortSignal;
   },

src/node_util.cc

@@ -37,9 +37,6 @@ 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)
 
@@ -240,40 +237,6 @@ static uint32_t FastGuessHandleType(Local<Value> receiver, const uint32_t fd) {
 
 CFunction fast_guess_handle_type_(CFunction::Make(FastGuessHandleType));
 
-static void ToUSVString(const FunctionCallbackInfo<Value>& args) {
-  Environment* env = Environment::GetCurrent(args);
-  CHECK_GE(args.Length(), 2);
-  CHECK(args[0]->IsString());
-  CHECK(args[1]->IsNumber());
-
-  TwoByteValue value(env->isolate(), args[0]);
-
-  int64_t start = args[1]->IntegerValue(env->context()).FromJust();
-  CHECK_GE(start, 0);
-
-  for (size_t i = start; 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);
@@ -288,7 +251,6 @@ void RegisterExternalReferences(ExternalReferenceRegistry* registry) {
   registry->Register(GuessHandleType);
   registry->Register(FastGuessHandleType);
   registry->Register(fast_guess_handle_type_.GetTypeInfo());
-  registry->Register(ToUSVString);
 }
 
 void Initialize(Local<Object> target,
@@ -403,8 +365,6 @@ void Initialize(Local<Object> target,
                             "guessHandleType",
                             GuessHandleType,
                             &fast_guess_handle_type_);
-
-  SetMethodNoSideEffect(context, target, "toUSVString", ToUSVString);
 }
 
 }  // namespace util

typings/internalBinding/util.d.ts

@@ -43,5 +43,4 @@ export interface UtilBinding {
   shouldAbortOnUncaughtToggle: [shouldAbort: 0 | 1];
   WeakReference: typeof InternalUtilBinding.WeakReference;
   guessHandleType(fd: number): 'TCP' | 'TTY' | 'UDP' | 'FILE' | 'PIPE' | 'UNKNOWN';
-  toUSVString(str: string, start: number): string;
 }

typings/primordials.d.ts

@@ -420,6 +420,7 @@ declare namespace primordials {
   export const StringPrototypeToLocaleUpperCase: UncurryThis<typeof String.prototype.toLocaleUpperCase>
   export const StringPrototypeToLowerCase: UncurryThis<typeof String.prototype.toLowerCase>
   export const StringPrototypeToUpperCase: UncurryThis<typeof String.prototype.toUpperCase>
+  export const StringPrototypeToWellFormed: UncurryThis<typeof String.prototype.toWellFormed>
   export const StringPrototypeValueOf: UncurryThis<typeof String.prototype.valueOf>
   export const StringPrototypeReplaceAll: UncurryThis<typeof String.prototype.replaceAll>
   export import Symbol = globalThis.Symbol;