feat: added html sanitizer for remote rendering (#1128)

Co-authored-by: Joe Pea <[email protected]>

Author: anikethsaha

Diff for: .eslintrc.js

@@ -19,6 +19,7 @@ module.exports = {
   rules: {
     'prettier/prettier': ['error'],
     camelcase: ['warn'],
+    'no-useless-escape': ['warn'],
     curly: ['error', 'all'],
     'dot-notation': ['error'],
     eqeqeq: ['error'],

Diff for: package-lock.json

@@ -57,6 +57,7 @@
     "*.js": "eslint --fix"
   },
   "dependencies": {
+    "dompurify": "^2.0.8",
     "marked": "^0.7.0",
     "medium-zoom": "^1.0.5",
     "opencollective-postinstall": "^2.0.2",
@@ -82,7 +83,7 @@
     "esm": "^3.1.4",
     "husky": "^3.1.0",
     "jsdom": "^16.2.2",
-    "lerna": "^3.17.0",
+    "lerna": "^3.22.1",
     "lint-staged": "^10.1.2",
     "live-server": "^1.2.1",
     "mkdirp": "^0.5.1",

Diff for: package.json

@@ -3,6 +3,7 @@ import { resolve, basename } from 'path';
 import resolvePathname from 'resolve-pathname';
 import fetch from 'node-fetch';
 import debug from 'debug';
+import DOMPurify from 'dompurify';
 import { AbstractHistory } from '../../src/core/router/history/abstract';
 import { Compiler } from '../../src/core/render/compiler';
 import { isAbsolutePath } from '../../src/core/router/util';
@@ -13,6 +14,32 @@ function cwd(...args) {
   return resolve(process.cwd(), ...args);
 }
 
+function isExternal(url) {
+  let match = url.match(
+    /^([^:\/?#]+:)?(?:\/\/([^\/?#]*))?([^?#]+)?(\?[^#]*)?(#.*)?/
+  );
+  if (
+    typeof match[1] === 'string' &&
+    match[1].length > 0 &&
+    match[1].toLowerCase() !== location.protocol
+  ) {
+    return true;
+  }
+  if (
+    typeof match[2] === 'string' &&
+    match[2].length > 0 &&
+    match[2].replace(
+      new RegExp(
+        ':(' + { 'http:': 80, 'https:': 443 }[location.protocol] + ')?$'
+      ),
+      ''
+    ) !== location.host
+  ) {
+    return true;
+  }
+  return false;
+}
+
 function mainTpl(config) {
   let html = `<nav class="app-nav${
     config.repo ? '' : ' no-badge'
@@ -60,6 +87,7 @@ export default class Renderer {
 
   async renderToString(url) {
     this.url = url = this.router.parse(url).path;
+    this.isRemoteUrl = isExternal(this.url);
     const { loadSidebar, loadNavbar, coverpage } = this.config;
 
     const mainFile = this._getPath(url);
@@ -95,9 +123,8 @@ export default class Renderer {
       this._renderHtml('cover', await this._render(coverFile), 'cover');
     }
 
-    const html = this.html;
+    const html = this.isRemoteUrl ? DOMPurify.sanitize(this.html) : this.html;
     this.html = this.template;
-
     return html;
   }
 

Diff for: packages/docsify-server-renderer/index.js

@@ -17,6 +17,7 @@
   "dependencies": {
     "debug": "^4.1.1",
     "docsify": "^4.11.2",
+    "dompurify": "^2.0.8",
     "node-fetch": "^2.6.0",
     "resolve-pathname": "^3.0.0"
   }

Diff for: packages/docsify-server-renderer/package-lock.json

@@ -20,6 +20,32 @@ function loadNested(path, qs, file, next, vm, first) {
   ).then(next, _ => loadNested(path, qs, file, next, vm));
 }
 
+function isExternal(url) {
+  let match = url.match(
+    /^([^:\/?#]+:)?(?:\/\/([^\/?#]*))?([^?#]+)?(\?[^#]*)?(#.*)?/
+  );
+  if (
+    typeof match[1] === 'string' &&
+    match[1].length > 0 &&
+    match[1].toLowerCase() !== location.protocol
+  ) {
+    return true;
+  }
+  if (
+    typeof match[2] === 'string' &&
+    match[2].length > 0 &&
+    match[2].replace(
+      new RegExp(
+        ':(' + { 'http:': 80, 'https:': 443 }[location.protocol] + ')?$'
+      ),
+      ''
+    ) !== location.host
+  ) {
+    return true;
+  }
+  return false;
+}
+
 export function fetchMixin(proto) {
   let last;
 
@@ -84,6 +110,7 @@ export function fetchMixin(proto) {
     const file = this.router.getFile(path);
     const req = request(file + qs, true, requestHeaders);
 
+    this.isRemoteUrl = isExternal(file);
     // Current page is html
     this.isHTML = /\.html$/g.test(file);
 

Diff for: packages/docsify-server-renderer/package.json

@@ -1,5 +1,6 @@
 /* eslint-disable no-unused-vars */
 import tinydate from 'tinydate';
+import DOMPurify from 'dompurify';
 import * as dom from '../util/dom';
 import cssVars from '../util/polyfill/css-vars';
 import { callHook } from '../init/lifecycle';
@@ -172,6 +173,7 @@ export function renderMixin(proto) {
           },
           tokens => {
             html = this.compiler.compile(tokens);
+            html = this.isRemoteUrl ? DOMPurify.sanitize(html) : html;
             callback();
             next();
           }