Merge pull request from GHSA-hwvq-6gjx-j797

* sanitizer fix

* Pass sanitizer options explicitly

Author: afshin

bower.json

@@ -9,7 +9,6 @@
     "create-react-class": "https://cdn.jsdelivr.net/npm/[email protected]/create-react-class.min.js",
     "es6-promise": "~1.0",
     "font-awesome": "components/font-awesome#~4.7.0",
-    "google-caja": "5669",
     "jed": "~1.1.1",
     "jquery": "components/jquery#~3.4.1",
     "jquery-typeahead": "~2.0.0",

notebook/static/base/js/namespace.js

@@ -32,7 +32,7 @@ define(function(){
     // expose modules
 
     jprop('utils','base/js/utils')
-    
+
     //Jupyter.load_extensions = Jupyter.utils.load_extensions;
     // 
     jprop('security','base/js/security');

notebook/static/base/js/security.js

@@ -3,124 +3,24 @@
 
 define([
     'jquery',
-    'components/google-caja/html-css-sanitizer-minified',
-], function($, sanitize) {
+    'components/sanitizer/index',
+], function($, sanitizer) {
     "use strict";
-    
+
     var noop = function (x) { return x; };
-    
-    var caja;
-    if (window && window.html) {
-        caja = window.html;
-        caja.html4 = window.html4;
-        caja.sanitizeStylesheet = window.sanitizeStylesheet;
-    }
-    
-    var sanitizeAttribs = function (tagName, attribs, opt_naiveUriRewriter, opt_nmTokenPolicy, opt_logger) {
-        /**
-         * add trusting data-attributes to the default sanitizeAttribs from caja
-         * this function is mostly copied from the caja source
-         */
-        var ATTRIBS = caja.html4.ATTRIBS;
-        for (var i = 0; i < attribs.length; i += 2) {
-            var attribName = attribs[i];
-            if (attribName.substr(0,5) == 'data-') {
-                var attribKey = '*::' + attribName;
-                if (!ATTRIBS.hasOwnProperty(attribKey)) {
-                    ATTRIBS[attribKey] = 0;
-                }
-            }
-        }
-        // Caja doesn't allow data uri for img::src, see
-        // https://github.com/google/caja/issues/1558
-        // This is not a security issue for browser post ie6 though, so we
-        // disable the check
-        // https://www.owasp.org/index.php/Script_in_IMG_tags
-        ATTRIBS['img::src'] = 0;
-        return caja.sanitizeAttribs(tagName, attribs, opt_naiveUriRewriter, opt_nmTokenPolicy, opt_logger);
-    };
-    
-    var sanitize_css = function (css, tagPolicy) {
-        /**
-         * sanitize CSS
-         * like sanitize_html, but for CSS
-         * called by sanitize_stylesheets
-         */
-        return caja.sanitizeStylesheet(
-            window.location.pathname,
-            css,
-            {
-                containerClass: null,
-                idSuffix: '',
-                tagPolicy: tagPolicy,
-                virtualizeAttrName: noop
-            },
-            noop
-        );
-    };
-    
-    var sanitize_stylesheets = function (html, tagPolicy) {
-        /**
-         * sanitize just the css in style tags in a block of html
-         * called by sanitize_html, if allow_css is true
-         */
-        var h = $("<div/>").append(html);
-        var style_tags = h.find("style");
-        if (!style_tags.length) {
-            // no style tags to sanitize
-            return html;
-        }
-        style_tags.each(function(i, style) {
-            style.innerHTML = sanitize_css(style.innerHTML, tagPolicy);
-        });
-        return h.html();
-    };
-    
+    var defaultSanitizer = sanitizer.defaultSanitizer;
+
     var sanitize_html = function (html, allow_css) {
         /**
          * sanitize HTML
          * if allow_css is true (default: false), CSS is sanitized as well.
          * otherwise, CSS elements and attributes are simply removed.
          */
-        var html4 = caja.html4;
-
-        if (allow_css) {
-            // allow sanitization of style tags,
-            // not just scrubbing
-            html4.ELEMENTS.style &= ~html4.eflags.UNSAFE;
-            html4.ATTRIBS.style = html4.atype.STYLE;
-        } else {
-            // scrub all CSS
-            html4.ELEMENTS.style |= html4.eflags.UNSAFE;
-            html4.ATTRIBS.style = html4.atype.SCRIPT;
-        }
-        
-        var record_messages = function (msg, opts) {
-            console.log("HTML Sanitizer", msg, opts);
-        };
-        
-        var policy = function (tagName, attribs) {
-            if (!(html4.ELEMENTS[tagName] & html4.eflags.UNSAFE)) {
-                return {
-                    'attribs': sanitizeAttribs(tagName, attribs,
-                        noop, noop, record_messages)
-                    };
-            } else {
-                record_messages(tagName + " removed", {
-                  change: "removed",
-                  tagName: tagName
-                });
-            }
-        };
-        
-        var sanitized = caja.sanitizeWithPolicy(html, policy);
-        
-        if (allow_css) {
-            // sanitize style tags as stylesheets
-            sanitized = sanitize_stylesheets(sanitized, policy);
-        }
-        
-        return sanitized;
+         const options = {};
+         if (!allow_css) {
+             options.allowedStyles = {};
+         }
+        return defaultSanitizer.sanitize(html, options);
     };
 
     var sanitize_html_and_parse = function (html, allow_css) {
@@ -141,9 +41,8 @@ define([
             $.htmlPrefilter = prev_htmlPrefilter;  // Set it back again
         }
     };
-    
+
     var security = {
-        caja: caja,
         sanitize_html_and_parse: sanitize_html_and_parse,
         sanitize_html: sanitize_html
     };

package.json

@@ -11,13 +11,18 @@
   "scripts": {
     "bower": "bower install",
     "build": "python setup.py js css",
-    "build:watch": "set -e; while true; do npm run build; sleep 3; done"
+    "build:webpack": "webpack --mode development",
+    "build:watch": "npm run watch",
+    "watch": "onchange 'notebook/static/**/!(*.min).js' 'notebook/static/**/*.less' 'bower.json' -- npm run build"
   },
   "devDependencies": {
-    "bower": "*",
+    "@jupyterlab/apputils": "^3.1.3",
+    "bower": "^1.8.8",
     "less": "~2",
-    "requirejs": "^2.1.17",
-    "po2json": "^0.4.5"
-  },
-  "dependencies": {}
+    "onchange": "^6.0.0",
+    "po2json": "^0.4.5",
+    "requirejs": "^2.3.6",
+    "webpack": "^5.46.0",
+    "webpack-cli": "^4.7.2"
+  }
 }

setupbase.py

@@ -140,7 +140,6 @@ def find_package_data():
         pjoin(components, "font-awesome", "css", "*.css"),
         pjoin(components, "es6-promise", "*.js"),
         pjoin(components, "font-awesome", "fonts", "*.*"),
-        pjoin(components, "google-caja", "html-css-sanitizer-minified.js"),
         pjoin(components, "jed", "jed.js"),
         pjoin(components, "jquery", "jquery.min.js"),
         pjoin(components, "jquery-typeahead", "dist", "jquery.typeahead.min.js"),
@@ -153,6 +152,7 @@ def find_package_data():
         pjoin(components, "requirejs", "require.js"),
         pjoin(components, "requirejs-plugins", "src", "json.js"),
         pjoin(components, "requirejs-text", "text.js"),
+        pjoin(components, "sanitizer", "index.js"),
         pjoin(components, "underscore", "underscore-min.js"),
         pjoin(components, "moment", "moment.js"),
         pjoin(components, "moment", "min", "*.js"),
@@ -376,14 +376,21 @@ def finalize_options(self):
 
     bower_dir = pjoin(static, 'components')
     node_modules = pjoin(repo_root, 'node_modules')
+    sanitizer_dir = pjoin(bower_dir, 'sanitizer')
 
     def should_run(self):
         if self.force:
             return True
         if not os.path.exists(self.bower_dir):
             return True
-        
-        return mtime(self.bower_dir) < mtime(pjoin(repo_root, 'bower.json'))
+        if not os.path.exists(self.sanitizer_dir):
+            return True
+
+        bower_stale = mtime(self.bower_dir) < mtime(pjoin(repo_root, 'bower.json'))
+        if bower_stale:
+            return True
+
+        return mtime(self.sanitizer_dir) < mtime(pjoin(repo_root, 'webpack.config.js'))
 
     def should_run_npm(self):
         if not which('npm'):
@@ -417,6 +424,8 @@ def run(self):
             print("You can install js dependencies with `npm install`", file=sys.stderr)
             raise
         # self.npm_components()
+        if not os.path.exists(self.sanitizer_dir):
+            run(['npm', 'run', 'build:webpack'], cwd=repo_root, env=env)
         os.utime(self.bower_dir, None)
         # update package data in case this created new files
         update_package_data(self.distribution)

webpack.config.js

@@ -0,0 +1,10 @@
+const path = require('path');
+
+module.exports = {
+  entry: '@jupyterlab/apputils/lib/sanitizer',
+  output: {
+    filename: 'index.js',
+    path: path.resolve(__dirname, 'notebook/static/components/sanitizer'),
+    libraryTarget: "amd"
+  }
+}