From ac43ca8f80c042f0256fb551ea5203863dec4481 Mon Sep 17 00:00:00 2001
From: Rodney Rehm <mail+github@rodneyrehm.de>
Date: Sun, 11 Jul 2021 20:16:43 +0200
Subject: [PATCH] fix(parse): more backslash galore #410

issue reported privately by @ready-research via https://huntr.dev/
---
 src/URI.js   |   3 +
 test/urls.js | 343 +++++++++++++++++++++++++++++++++++++++++++++++++++
 2 files changed, 346 insertions(+)

diff --git a/src/URI.js b/src/URI.js
index 0a5e77e..9a601a3 100644
--- a/src/URI.js
+++ b/src/URI.js
@@ -512,6 +512,9 @@
       string = string.substring(0, pos);
     }
 
+    // slashes and backslashes have lost all meaning for the web protocols (https, http, wss, ws)
+    string = string.replace(/^(https?|ftp|wss?)?:[/\\]*/, '$1://');
+
     // extract protocol
     if (string.substring(0, 2) === '//') {
       // relative-scheme
diff --git a/test/urls.js b/test/urls.js
index 30547e6..efb192f 100644
--- a/test/urls.js
+++ b/test/urls.js
@@ -2131,6 +2131,349 @@ var urls = [{
         idn: false,
         punycode: false
       }
+    }, {
+      name: 'backslashes protocol excessive',
+      url: 'https:/\/\/\attacker.com',
+      _url: 'https://attacker.com/',
+      parts: {
+        protocol: 'https',
+        username: null,
+        password: null,
+        hostname: 'attacker.com',
+        port: null,
+        path: '/',
+        query: null,
+        fragment: null
+      },
+      accessors: {
+        protocol: 'https',
+        username: '',
+        password: '',
+        port: '',
+        path: '/',
+        query: '',
+        fragment: '',
+        resource: '/',
+        authority: 'attacker.com',
+        origin: 'https://attacker.com',
+        userinfo: '',
+        subdomain: '',
+        domain: 'attacker.com',
+        tld: 'com',
+        directory: '/',
+        filename: '',
+        suffix: '',
+        hash: '',
+        search: '',
+        host: 'attacker.com',
+        hostname: 'attacker.com'
+      },
+      is: {
+        urn: false,
+        url: true,
+        relative: false,
+        name: true,
+        sld: false,
+        ip: false,
+        ip4: false,
+        ip6: false,
+        idn: false,
+        punycode: false
+      }
+    }, {
+      name: 'no slash protocol https',
+      url: 'https:attacker.com',
+      _url: 'https://attacker.com/',
+      parts: {
+        protocol: 'https',
+        username: null,
+        password: null,
+        hostname: 'attacker.com',
+        port: null,
+        path: '/',
+        query: null,
+        fragment: null
+      },
+      accessors: {
+        protocol: 'https',
+        username: '',
+        password: '',
+        port: '',
+        path: '/',
+        query: '',
+        fragment: '',
+        resource: '/',
+        authority: 'attacker.com',
+        origin: 'https://attacker.com',
+        userinfo: '',
+        subdomain: '',
+        domain: 'attacker.com',
+        tld: 'com',
+        directory: '/',
+        filename: '',
+        suffix: '',
+        hash: '',
+        search: '',
+        host: 'attacker.com',
+        hostname: 'attacker.com'
+      },
+      is: {
+        urn: false,
+        url: true,
+        relative: false,
+        name: true,
+        sld: false,
+        ip: false,
+        ip4: false,
+        ip6: false,
+        idn: false,
+        punycode: false
+      }
+    }, {
+      name: 'single slash protocol https',
+      url: 'https:/attacker.com',
+      _url: 'https://attacker.com/',
+      parts: {
+        protocol: 'https',
+        username: null,
+        password: null,
+        hostname: 'attacker.com',
+        port: null,
+        path: '/',
+        query: null,
+        fragment: null
+      },
+      accessors: {
+        protocol: 'https',
+        username: '',
+        password: '',
+        port: '',
+        path: '/',
+        query: '',
+        fragment: '',
+        resource: '/',
+        authority: 'attacker.com',
+        origin: 'https://attacker.com',
+        userinfo: '',
+        subdomain: '',
+        domain: 'attacker.com',
+        tld: 'com',
+        directory: '/',
+        filename: '',
+        suffix: '',
+        hash: '',
+        search: '',
+        host: 'attacker.com',
+        hostname: 'attacker.com'
+      },
+      is: {
+        urn: false,
+        url: true,
+        relative: false,
+        name: true,
+        sld: false,
+        ip: false,
+        ip4: false,
+        ip6: false,
+        idn: false,
+        punycode: false
+      }
+    }, {
+      name: 'excessive slash protocol https',
+      url: 'https://////attacker.com',
+      _url: 'https://attacker.com/',
+      parts: {
+        protocol: 'https',
+        username: null,
+        password: null,
+        hostname: 'attacker.com',
+        port: null,
+        path: '/',
+        query: null,
+        fragment: null
+      },
+      accessors: {
+        protocol: 'https',
+        username: '',
+        password: '',
+        port: '',
+        path: '/',
+        query: '',
+        fragment: '',
+        resource: '/',
+        authority: 'attacker.com',
+        origin: 'https://attacker.com',
+        userinfo: '',
+        subdomain: '',
+        domain: 'attacker.com',
+        tld: 'com',
+        directory: '/',
+        filename: '',
+        suffix: '',
+        hash: '',
+        search: '',
+        host: 'attacker.com',
+        hostname: 'attacker.com'
+      },
+      is: {
+        urn: false,
+        url: true,
+        relative: false,
+        name: true,
+        sld: false,
+        ip: false,
+        ip4: false,
+        ip6: false,
+        idn: false,
+        punycode: false
+      }
+    }, {
+      name: 'no slash protocol ftp',
+      url: 'ftp:attacker.com',
+      _url: 'ftp://attacker.com/',
+      parts: {
+        protocol: 'ftp',
+        username: null,
+        password: null,
+        hostname: 'attacker.com',
+        port: null,
+        path: '/',
+        query: null,
+        fragment: null
+      },
+      accessors: {
+        protocol: 'ftp',
+        username: '',
+        password: '',
+        port: '',
+        path: '/',
+        query: '',
+        fragment: '',
+        resource: '/',
+        authority: 'attacker.com',
+        origin: 'ftp://attacker.com',
+        userinfo: '',
+        subdomain: '',
+        domain: 'attacker.com',
+        tld: 'com',
+        directory: '/',
+        filename: '',
+        suffix: '',
+        hash: '',
+        search: '',
+        host: 'attacker.com',
+        hostname: 'attacker.com'
+      },
+      is: {
+        urn: false,
+        url: true,
+        relative: false,
+        name: true,
+        sld: false,
+        ip: false,
+        ip4: false,
+        ip6: false,
+        idn: false,
+        punycode: false
+      }
+    }, {
+      name: 'single slash protocol ftp',
+      url: 'ftp:/attacker.com',
+      _url: 'ftp://attacker.com/',
+      parts: {
+        protocol: 'ftp',
+        username: null,
+        password: null,
+        hostname: 'attacker.com',
+        port: null,
+        path: '/',
+        query: null,
+        fragment: null
+      },
+      accessors: {
+        protocol: 'ftp',
+        username: '',
+        password: '',
+        port: '',
+        path: '/',
+        query: '',
+        fragment: '',
+        resource: '/',
+        authority: 'attacker.com',
+        origin: 'ftp://attacker.com',
+        userinfo: '',
+        subdomain: '',
+        domain: 'attacker.com',
+        tld: 'com',
+        directory: '/',
+        filename: '',
+        suffix: '',
+        hash: '',
+        search: '',
+        host: 'attacker.com',
+        hostname: 'attacker.com'
+      },
+      is: {
+        urn: false,
+        url: true,
+        relative: false,
+        name: true,
+        sld: false,
+        ip: false,
+        ip4: false,
+        ip6: false,
+        idn: false,
+        punycode: false
+      }
+    }, {
+      name: 'excessive slash protocol ftp',
+      url: 'ftp://////attacker.com',
+      _url: 'ftp://attacker.com/',
+      parts: {
+        protocol: 'ftp',
+        username: null,
+        password: null,
+        hostname: 'attacker.com',
+        port: null,
+        path: '/',
+        query: null,
+        fragment: null
+      },
+      accessors: {
+        protocol: 'ftp',
+        username: '',
+        password: '',
+        port: '',
+        path: '/',
+        query: '',
+        fragment: '',
+        resource: '/',
+        authority: 'attacker.com',
+        origin: 'ftp://attacker.com',
+        userinfo: '',
+        subdomain: '',
+        domain: 'attacker.com',
+        tld: 'com',
+        directory: '/',
+        filename: '',
+        suffix: '',
+        hash: '',
+        search: '',
+        host: 'attacker.com',
+        hostname: 'attacker.com'
+      },
+      is: {
+        urn: false,
+        url: true,
+        relative: false,
+        name: true,
+        sld: false,
+        ip: false,
+        ip4: false,
+        ip6: false,
+        idn: false,
+        punycode: false
+      }
     }, {
       name: '__proto__ in query',
       url: 'http://www.example.org/?__proto__=hasOwnProperty&__proto__=eviltwin&uuid',