fix(isURL): improve protocol detection. Resolves CVE-2025-56200

.github/workflows/ci.yml

@@ -9,7 +9,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        node-version: [22, 20, 18, 16, 14, 12, 10, 8, 6]
+        node-version: [22, 20, 18, 16, 14, 12, 10, 8]
     name: Run tests on Node.js ${{ matrix.node-version }}
     steps:
     - name: Setup Node.js ${{ matrix.node-version }}

README.md

@@ -167,7 +167,7 @@ Validator                               | Description
 **isStrongPassword(str [, options])**   | check if the string can be considered a strong password or not. Allows for custom requirements or scoring rules. If `returnScore` is true, then the function returns an integer score for the password rather than a boolean.<br/>Default options: <br/>`{ minLength: 8, minLowercase: 1, minUppercase: 1, minNumbers: 1, minSymbols: 1, returnScore: false, pointsPerUnique: 1, pointsPerRepeat: 0.5, pointsForContainingLower: 10, pointsForContainingUpper: 10, pointsForContainingNumber: 10, pointsForContainingSymbol: 10 }`
 **isTime(str [, options])**             | check if the string is a valid time e.g. [`23:01:59`, new Date().toLocaleTimeString()].<br/><br/> `options` is an object which can contain the keys `hourFormat` or `mode`.<br/><br/>`hourFormat` is a key and defaults to `'hour24'`.<br/><br/>`mode` is a key and defaults to `'default'`. <br/><br/>`hourFormat` can contain the values `'hour12'` or `'hour24'`, `'hour24'` will validate hours in 24 format and `'hour12'` will validate hours in 12 format. <br/><br/>`mode` can contain the values `'default', 'withSeconds', withOptionalSeconds`, `'default'` will validate `HH:MM` format, `'withSeconds'` will validate the `HH:MM:SS` format, `'withOptionalSeconds'` will validate `'HH:MM'` and `'HH:MM:SS'` formats.
 **isTaxID(str, locale)**                | check if the string is a valid Tax Identification Number. Default locale is `en-US`.<br/><br/>More info about exact TIN support can be found in `src/lib/isTaxID.js`.<br/><br/>Supported locales: `[ 'bg-BG', 'cs-CZ', 'de-AT', 'de-DE', 'dk-DK', 'el-CY', 'el-GR', 'en-CA', 'en-GB', 'en-IE', 'en-US', 'es-AR', 'es-ES', 'et-EE', 'fi-FI', 'fr-BE', 'fr-CA', 'fr-FR', 'fr-LU', 'hr-HR', 'hu-HU', 'it-IT', 'lb-LU', 'lt-LT', 'lv-LV', 'mt-MT', 'nl-BE', 'nl-NL', 'pl-PL', 'pt-BR', 'pt-PT', 'ro-RO', 'sk-SK', 'sl-SI', 'sv-SE', 'uk-UA']`.
-**isURL(str [, options])**              | check if the string is a URL.<br/><br/>`options` is an object which defaults to `{ protocols: ['http','https','ftp'], require_tld: true, require_protocol: false, require_host: true, require_port: false, require_valid_protocol: true, allow_underscores: false, host_whitelist: false, host_blacklist: false, allow_trailing_dot: false, allow_protocol_relative_urls: false, allow_fragments: true, allow_query_components: true, disallow_auth: false, validate_length: true }`.<br/><br/>`protocols` - valid protocols can be modified with this option.<br/>`require_tld` - If set to false isURL will not check if the URL's host includes a top-level domain.<br/>`require_protocol` - if set to true isURL will return false if protocol is not present in the URL.<br/>`require_host` - if set to false isURL will not check if host is present in the URL.<br/>`require_port` - if set to true isURL will check if port is present in the URL.<br/>`require_valid_protocol` - isURL will check if the URL's protocol is present in the protocols option.<br/>`allow_underscores` - if set to true, the validator will allow underscores in the URL.<br/>`host_whitelist` - if set to an array of strings or regexp, and the domain matches none of the strings defined in it, the validation fails.<br/>`host_blacklist` - if set to an array of strings or regexp, and the domain matches any of the strings defined in it, the validation fails.<br/>`allow_trailing_dot` - if set to true, the validator will allow the domain to end with a `.` character.<br/>`allow_protocol_relative_urls` - if set to true protocol relative URLs will be allowed.<br/>`allow_fragments` - if set to false isURL will return false if fragments are present.<br/>`allow_query_components` - if set to false isURL will return false if query components are present.<br/>`disallow_auth` - if set to true, the validator will fail if the URL contains an authentication component, e.g. `http://username:[email protected]`.<br/>`validate_length` - if set to false isURL will skip string length validation. `max_allowed_length` will be ignored if this is set as `false`.<br/>`max_allowed_length` - if set, isURL will not allow URLs longer than the specified value (default is 2084 that IE maximum URL length).<br/>
+**isURL(str [, options])**              | check if the string is a URL.<br/><br/>`options` is an object which defaults to `{ protocols: ['http','https','ftp'], require_tld: true, require_protocol: false, require_host: true, require_port: false, require_valid_protocol: true, allow_underscores: false, host_whitelist: false, host_blacklist: false, allow_trailing_dot: false, allow_protocol_relative_urls: false, allow_fragments: true, allow_query_components: true, disallow_auth: false, validate_length: true }`.<br/><br/>`protocols` - valid protocols can be modified with this option.<br/>`require_tld` - If set to false isURL will not check if the URL's host includes a top-level domain.<br/>`require_protocol` - **RECOMMENDED** if set to true isURL will return false if protocol is not present in the URL. Without this setting, some malicious URLs cannot be distinguishable from a valid URL with authentication information.<br/>`require_host` - if set to false isURL will not check if host is present in the URL.<br/>`require_port` - if set to true isURL will check if port is present in the URL.<br/>`require_valid_protocol` - isURL will check if the URL's protocol is present in the protocols option.<br/>`allow_underscores` - if set to true, the validator will allow underscores in the URL.<br/>`host_whitelist` - if set to an array of strings or regexp, and the domain matches none of the strings defined in it, the validation fails.<br/>`host_blacklist` - if set to an array of strings or regexp, and the domain matches any of the strings defined in it, the validation fails.<br/>`allow_trailing_dot` - if set to true, the validator will allow the domain to end with a `.` character.<br/>`allow_protocol_relative_urls` - if set to true protocol relative URLs will be allowed.<br/>`allow_fragments` - if set to false isURL will return false if fragments are present.<br/>`allow_query_components` - if set to false isURL will return false if query components are present.<br/>`disallow_auth` - if set to true, the validator will fail if the URL contains an authentication component, e.g. `http://username:[email protected]`.<br/>`validate_length` - if set to false isURL will skip string length validation. `max_allowed_length` will be ignored if this is set as `false`.<br/>`max_allowed_length` - if set, isURL will not allow URLs longer than the specified value (default is 2084 that IE maximum URL length).<br/>
 **isULID(str)**                         | check if the string is a [ULID](https://github.com/ulid/spec).
 **isUUID(str [, version])**             | check if the string is an RFC9562 UUID.<br/>`version` is one of `'1'`-`'8'`, `'nil'`, `'max'`, `'all'` or `'loose'`. The `'loose'` option checks if the string is a UUID-like string with hexadecimal values, ignoring RFC9565.
 **isVariableWidth(str)**                | check if the string contains a mixture of full and half-width chars.

src/lib/isURL.js

@@ -83,21 +83,94 @@ export default function isURL(url, options) {
   split = url.split('?');
   url = split.shift();
 
-  split = url.split('://');
-  if (split.length > 1) {
-    protocol = split.shift().toLowerCase();
+  // Replaced the 'split("://")' logic with a regex to match the protocol.
+  // This correctly identifies schemes like `javascript:` which don't use `//`.
+  // However, we need to be careful not to confuse authentication credentials (user:password@host)
+  // with protocols. A colon before an @ symbol might be part of auth, not a protocol separator.
+  const protocol_match = url.match(/^([a-z][a-z0-9+\-.]*):/i);
+  let had_explicit_protocol = false;
+
+  const cleanUpProtocol = (potential_protocol) => {
+    had_explicit_protocol = true;
+    protocol = potential_protocol.toLowerCase();
+
     if (options.require_valid_protocol && options.protocols.indexOf(protocol) === -1) {
+      // The identified protocol is not in the allowed list.
       return false;
     }
+
+    // Remove the protocol from the URL string.
+    return url.substring(protocol_match[0].length);
+  };
+
+  if (protocol_match) {
+    const potential_protocol = protocol_match[1];
+    const after_colon = url.substring(protocol_match[0].length);
+
+    // Check if what follows looks like authentication credentials (user:password@host)
+    // rather than a protocol. This happens when:
+    // 1. There's no `//` after the colon (protocols like `http://` have this)
+    // 2. There's an `@` symbol before any `/`
+    // 3. The part before `@` contains only valid auth characters (alphanumeric, -, _, ., %, :)
+    const starts_with_slashes = after_colon.slice(0, 2) === '//';
+
+    if (!starts_with_slashes) {
+      const first_slash_position = after_colon.indexOf('/');
+      const before_slash = first_slash_position === -1
+        ? after_colon
+        : after_colon.substring(0, first_slash_position);
+      const at_position = before_slash.indexOf('@');
+
+      if (at_position !== -1) {
+        const before_at = before_slash.substring(0, at_position);
+        const valid_auth_regex = /^[a-zA-Z0-9\-_.%:]*$/;
+        const is_valid_auth = valid_auth_regex.test(before_at);
+
+        if (is_valid_auth) {
+          // This looks like authentication (e.g., user:password@host), not a protocol
+          if (options.require_protocol) {
+            return false;
+          }
+
+          // Don't consume the colon; let the auth parsing handle it later
+        } else {
+          // This looks like a malicious protocol (e.g., javascript:alert();@host)
+          url = cleanUpProtocol(potential_protocol);
+
+          if (url === false) {
+            return false;
+          }
+        }
+      } else {
+        // No @ symbol, this is definitely a protocol
+        url = cleanUpProtocol(potential_protocol);
+
+        if (url === false) {
+          return false;
+        }
+      }
+    } else {
+      // Starts with '//', this is definitely a protocol like http://
+      url = cleanUpProtocol(potential_protocol);
+
+      if (url === false) {
+        return false;
+      }
+    }
   } else if (options.require_protocol) {
     return false;
-  } else if (url.slice(0, 2) === '//') {
-    if (!options.allow_protocol_relative_urls) {
+  }
+
+  // Handle leading '//' only as protocol-relative when there was NO explicit protocol.
+  // If there was an explicit protocol, '//' is the normal separator
+  // and should be stripped unconditionally.
+  if (url.slice(0, 2) === '//') {
+    if (!had_explicit_protocol && !options.allow_protocol_relative_urls) {
       return false;
     }
-    split[0] = url.slice(2);
+
+    url = url.slice(2);
   }
-  url = split.join('://');
 
   if (url === '') {
     return false;

test/validators.test.js

@@ -424,6 +424,12 @@ describe('Validators', () => {
         'http://[2010:836B:4179::836B:4179]',
         'http://example.com/example.json#/foo/bar',
         'http://1337.com',
+        // TODO: those probably should not be marked as valid URLs; CVE-2025-56200
+        /* eslint-disable no-script-url */
+        'javascript:%61%6c%65%72%74%28%31%[email protected]',
+        'http://[email protected]/',
+        'ｊａｖａｓｃｒｉｐｔ:alert(1)@example.com',
+        /* eslint-enable no-script-url */
       ],
       invalid: [
         'http://localhost:3000/',
@@ -466,6 +472,18 @@ describe('Validators', () => {
         '////foobar.com',
         'http:////foobar.com',
         'https://example.com/foo/<script>alert(\'XSS\')</script>/',
+        // the following tests are because of CVE-2025-56200
+        /* eslint-disable no-script-url */
+        "javascript:alert(1);a=';@example.com/alert(1)'",
+        'JaVaScRiPt:alert(1)@example.com',
+        'javascript:/* comment */alert(1)@example.com',
+        'javascript:var a=1; alert(a);@example.com',
+        'javascript:alert(1)@[email protected]',
+        'javascript:alert(1)@example.com?q=safe',
+        'data:text/html,<script>alert(1)</script>@example.com',
+        'vbscript:msgbox("XSS")@example.com',
+        '//evil-site.com/[email protected]',
+        /* eslint-enable no-script-url */
       ],
     });
   });
@@ -478,9 +496,11 @@ describe('Validators', () => {
       }],
       valid: [
         'rtmp://foobar.com',
+        'rtmp:foobar.com',
       ],
       invalid: [
         'http://foobar.com',
+        'tel:+15551234567',
       ],
     });
   });
@@ -533,6 +553,9 @@ describe('Validators', () => {
         'rtmp://foobar.com',
         'http://foobar.com',
         'test://foobar.com',
+        // Dangerous! This allows to mark malicious URLs as a valid URL (CVE-2025-56200)
+        // eslint-disable-next-line no-script-url
+        'javascript:alert(1);@example.com',
       ],
       invalid: [
         'mailto:[email protected]',
@@ -704,6 +727,61 @@ describe('Validators', () => {
     });
   });
 
+  it('should validate authentication strings if a protocol is not required', () => {
+    test({
+      validator: 'isURL',
+      args: [{
+        require_protocol: false,
+      }],
+      valid: [
+        'user:[email protected]/',
+      ],
+      invalid: [
+        'user:pw,@foobar.com/',
+      ],
+    });
+  });
+
+  it('should reject authentication strings if a protocol is required', () => {
+    test({
+      validator: 'isURL',
+      args: [{
+        require_protocol: true,
+      }],
+      valid: [
+        'http://user:[email protected]/',
+        'https://user:[email protected]',
+        'ftp://admin:[email protected]/',
+      ],
+      invalid: [
+        'user:[email protected]/',
+        'user:[email protected]',
+        'admin:[email protected]/',
+      ],
+    });
+  });
+
+  it('should reject invalid protocols when require_valid_protocol is enabled', () => {
+    test({
+      validator: 'isURL',
+      args: [{
+        require_valid_protocol: true,
+        protocols: ['http', 'https', 'ftp'],
+      }],
+      valid: [
+        'http://example.com',
+        'https://example.com',
+        'ftp://example.com',
+      ],
+      invalid: [
+        // eslint-disable-next-line no-script-url
+        'javascript:alert(1);@example.com',
+        'data:text/html,<script>alert(1)</script>@example.com',
+        'file:///etc/[email protected]',
+      ],
+    });
+  });
+
   it('should let users specify a host whitelist', () => {
     test({
       validator: 'isURL',
@@ -782,6 +860,24 @@ describe('Validators', () => {
     });
   });
 
+  it('GHSA-9965-vmph-33xx vulnerability - protocol delimiter parsing difference', () => {
+    const DOMAIN_WHITELIST = ['example.com'];
+
+    test({
+      validator: 'isURL',
+      args: [{
+        protocols: ['https'],
+        host_whitelist: DOMAIN_WHITELIST,
+        require_host: false,
+      }],
+      valid: [],
+      invalid: [
+        // eslint-disable-next-line no-script-url
+        "javascript:alert(1);a=';@example.com/alert(1)",
+      ],
+    });
+  });
+
   it('should allow rejecting urls containing authentication information', () => {
     test({
       validator: 'isURL',