util: expose toUSVString

Expose toUSVString so it can be used by user libraries.

PR-URL: #39814
Refs: nodejs/undici#986
Reviewed-By: James M Snell <[email protected]>
Reviewed-By: Luigi Pinca <[email protected]>

Author: ronag

doc/api/util.md

@@ -2490,6 +2490,18 @@ const util = require('util');
 util.log('Timestamped message.');
 ```
 
+
+### `util.toUSVString(string)`
+<!-- YAML
+added: REPLACEME
+-->
+
+* `string` {string}
+
+Returns the `string` after replacing any surrogate code points
+(or equivalently, any unpaired surrogate code units) with the
+Unicode "replacement character" U+FFFD.
+
 [Common System Errors]: errors.md#errors_common_system_errors
 [Custom inspection functions on objects]: #util_custom_inspection_functions_on_objects
 [Custom promisified functions]: #util_custom_promisified_functions

lib/internal/url.js

@@ -19,7 +19,6 @@ const {
   ReflectApply,
   ReflectGetOwnPropertyDescriptor,
   ReflectOwnKeys,
-  RegExpPrototypeExec,
   String,
   StringPrototypeCharCodeAt,
   StringPrototypeIncludes,
@@ -40,7 +39,12 @@ const {
   isHexTable
 } = require('internal/querystring');
 
-const { getConstructorOf, removeColors } = require('internal/util');
+const {
+  getConstructorOf,
+  removeColors,
+  toUSVString,
+} = require('internal/util');
+
 const {
   codes: {
     ERR_ARG_NOT_ITERABLE,
@@ -82,7 +86,6 @@ const {
   domainToASCII: _domainToASCII,
   domainToUnicode: _domainToUnicode,
   encodeAuth,
-  toUSVString: _toUSVString,
   parse,
   setURLConstructor,
   URL_FLAGS_CANNOT_BE_BASE,
@@ -139,18 +142,6 @@ const IteratorPrototype = ObjectGetPrototypeOf(
   ObjectGetPrototypeOf([][SymbolIterator]())
 );
 
-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);
-}
-
 // Refs: https://html.spec.whatwg.org/multipage/browsers.html#concept-origin-opaque
 const kOpaqueOrigin = 'null';
 

lib/internal/util.js

@@ -17,6 +17,7 @@ const {
   Promise,
   ReflectApply,
   ReflectConstruct,
+  RegExpPrototypeExec,
   RegExpPrototypeTest,
   SafeMap,
   SafeSet,
@@ -27,6 +28,10 @@ const {
   SymbolFor,
 } = primordials;
 
+const {
+  toUSVString: _toUSVString,
+} = internalBinding('url');
+
 const {
   hideStackFrames,
   codes: {
@@ -52,6 +57,18 @@ 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() {
@@ -497,6 +514,7 @@ module.exports = {
   sleep,
   spliceOne,
   structuredClone,
+  toUSVString,
   removeColors,
 
   // Symbol used to customize promisify conversion

lib/util.js

@@ -72,7 +72,8 @@ const {
   deprecate,
   getSystemErrorMap,
   getSystemErrorName: internalErrorName,
-  promisify
+  promisify,
+  toUSVString,
 } = require('internal/util');
 
 let internalDeepEqual;
@@ -368,6 +369,7 @@ module.exports = {
   isPrimitive,
   log,
   promisify,
+  toUSVString,
   TextDecoder,
   TextEncoder,
   types

test/parallel/test-util.js

@@ -148,6 +148,8 @@ assert.strictEqual(util.isFunction(function() {}), true);
 assert.strictEqual(util.isFunction(), false);
 assert.strictEqual(util.isFunction('string'), false);
 
+assert.strictEqual(util.toUSVString('string\ud801'), 'string\ufffd');
+
 {
   assert.strictEqual(util.types.isNativeError(new Error()), true);
   assert.strictEqual(util.types.isNativeError(new TypeError()), true);