Skip to content

Commit a751649

Browse files
petebacondarwinMiško Hevery
petebacondarwin
authored and
Miško Hevery
committed
fix(core): use appropriate inert document strategy for Firefox & Safari (#17019)
Both Firefox and Safari are vulnerable to XSS if we use an inert document created via `document.implementation.createHTMLDocument()`. Now we check for those vulnerabilities and then use a DOMParser or XHR strategy if needed. Further the platform-server has its own library for parsing HTML, so we sniff for that (by checking whether DOMParser exists) and fall back to the standard strategy. Thanks to @cure53 for the heads up on this issue. PR Close #17019
1 parent 3f5a3d6 commit a751649

File tree

4 files changed

+250
-82
lines changed

4 files changed

+250
-82
lines changed

‎integration/_payload-limits.json

+2-2
Original file line numberDiff line numberDiff line change
@@ -3,15 +3,15 @@
33
"master": {
44
"uncompressed": {
55
"inline": 1447,
6-
"main": 151639,
6+
"main": 154185,
77
"polyfills": 59179
88
}
99
}
1010
},
1111
"hello_world__closure": {
1212
"master": {
1313
"uncompressed": {
14-
"bundle": 100661
14+
"bundle": 101744
1515
}
1616
}
1717
},

‎packages/platform-browser/src/security/html_sanitizer.ts

+36-80
Original file line numberDiff line numberDiff line change
@@ -10,35 +10,9 @@ import {isDevMode} from '@angular/core';
1010

1111
import {DomAdapter, getDOM} from '../dom/dom_adapter';
1212

13+
import {InertBodyHelper} from './inert_body';
1314
import {sanitizeSrcset, sanitizeUrl} from './url_sanitizer';
1415

15-
/** A <body> element that can be safely used to parse untrusted HTML. Lazily initialized below. */
16-
let inertElement: HTMLElement|null = null;
17-
/** Lazily initialized to make sure the DOM adapter gets set before use. */
18-
let DOM: DomAdapter = null !;
19-
20-
/** Returns an HTML element that is guaranteed to not execute code when creating elements in it. */
21-
function getInertElement() {
22-
if (inertElement) return inertElement;
23-
DOM = getDOM();
24-
25-
// Prefer using <template> element if supported.
26-
const templateEl = DOM.createElement('template');
27-
if ('content' in templateEl) return templateEl;
28-
29-
const doc = DOM.createHtmlDocument();
30-
inertElement = DOM.querySelector(doc, 'body');
31-
if (inertElement == null) {
32-
// usually there should be only one body element in the document, but IE doesn't have any, so we
33-
// need to create one.
34-
const html = DOM.createElement('html', doc);
35-
inertElement = DOM.createElement('body', doc);
36-
DOM.appendChild(html, inertElement);
37-
DOM.appendChild(doc, html);
38-
}
39-
return inertElement;
40-
}
41-
4216
function tagSet(tags: string): {[k: string]: boolean} {
4317
const res: {[k: string]: boolean} = {};
4418
for (const t of tags.split(',')) res[t] = true;
@@ -121,53 +95,54 @@ class SanitizingHtmlSerializer {
12195
// because characters were re-encoded.
12296
public sanitizedSomething = false;
12397
private buf: string[] = [];
98+
private DOM = getDOM();
12499

125100
sanitizeChildren(el: Element): string {
126101
// This cannot use a TreeWalker, as it has to run on Angular's various DOM adapters.
127102
// However this code never accesses properties off of `document` before deleting its contents
128103
// again, so it shouldn't be vulnerable to DOM clobbering.
129-
let current: Node = el.firstChild !;
104+
let current: Node = this.DOM.firstChild(el) !;
130105
while (current) {
131-
if (DOM.isElementNode(current)) {
106+
if (this.DOM.isElementNode(current)) {
132107
this.startElement(current as Element);
133-
} else if (DOM.isTextNode(current)) {
134-
this.chars(DOM.nodeValue(current) !);
108+
} else if (this.DOM.isTextNode(current)) {
109+
this.chars(this.DOM.nodeValue(current) !);
135110
} else {
136111
// Strip non-element, non-text nodes.
137112
this.sanitizedSomething = true;
138113
}
139-
if (DOM.firstChild(current)) {
140-
current = DOM.firstChild(current) !;
114+
if (this.DOM.firstChild(current)) {
115+
current = this.DOM.firstChild(current) !;
141116
continue;
142117
}
143118
while (current) {
144119
// Leaving the element. Walk up and to the right, closing tags as we go.
145-
if (DOM.isElementNode(current)) {
120+
if (this.DOM.isElementNode(current)) {
146121
this.endElement(current as Element);
147122
}
148123

149-
let next = checkClobberedElement(current, DOM.nextSibling(current) !);
124+
let next = this.checkClobberedElement(current, this.DOM.nextSibling(current) !);
150125

151126
if (next) {
152127
current = next;
153128
break;
154129
}
155130

156-
current = checkClobberedElement(current, DOM.parentElement(current) !);
131+
current = this.checkClobberedElement(current, this.DOM.parentElement(current) !);
157132
}
158133
}
159134
return this.buf.join('');
160135
}
161136

162137
private startElement(element: Element) {
163-
const tagName = DOM.nodeName(element).toLowerCase();
138+
const tagName = this.DOM.nodeName(element).toLowerCase();
164139
if (!VALID_ELEMENTS.hasOwnProperty(tagName)) {
165140
this.sanitizedSomething = true;
166141
return;
167142
}
168143
this.buf.push('<');
169144
this.buf.push(tagName);
170-
DOM.attributeMap(element).forEach((value: string, attrName: string) => {
145+
this.DOM.attributeMap(element).forEach((value: string, attrName: string) => {
171146
const lower = attrName.toLowerCase();
172147
if (!VALID_ATTRS.hasOwnProperty(lower)) {
173148
this.sanitizedSomething = true;
@@ -186,7 +161,7 @@ class SanitizingHtmlSerializer {
186161
}
187162

188163
private endElement(current: Element) {
189-
const tagName = DOM.nodeName(current).toLowerCase();
164+
const tagName = this.DOM.nodeName(current).toLowerCase();
190165
if (VALID_ELEMENTS.hasOwnProperty(tagName) && !VOID_ELEMENTS.hasOwnProperty(tagName)) {
191166
this.buf.push('</');
192167
this.buf.push(tagName);
@@ -195,14 +170,14 @@ class SanitizingHtmlSerializer {
195170
}
196171

197172
private chars(chars: string) { this.buf.push(encodeEntities(chars)); }
198-
}
199173

200-
function checkClobberedElement(node: Node, nextNode: Node): Node {
201-
if (nextNode && DOM.contains(node, nextNode)) {
202-
throw new Error(
203-
`Failed to sanitize html because the element is clobbered: ${DOM.getOuterHTML(node)}`);
174+
checkClobberedElement(node: Node, nextNode: Node): Node {
175+
if (nextNode && this.DOM.contains(node, nextNode)) {
176+
throw new Error(
177+
`Failed to sanitize html because the element is clobbered: ${this.DOM.getOuterHTML(node)}`);
178+
}
179+
return nextNode;
204180
}
205-
return nextNode;
206181
}
207182

208183
// Regular Expressions for parsing tags and attributes
@@ -232,33 +207,20 @@ function encodeEntities(value: string) {
232207
.replace(/>/g, '&gt;');
233208
}
234209

235-
/**
236-
* When IE9-11 comes across an unknown namespaced attribute e.g. 'xlink:foo' it adds 'xmlns:ns1'
237-
* attribute to declare ns1 namespace and prefixes the attribute with 'ns1' (e.g. 'ns1:xlink:foo').
238-
*
239-
* This is undesirable since we don't want to allow any of these custom attributes. This method
240-
* strips them all.
241-
*/
242-
function stripCustomNsAttrs(el: Element) {
243-
DOM.attributeMap(el).forEach((_, attrName) => {
244-
if (attrName === 'xmlns:ns1' || attrName.indexOf('ns1:') === 0) {
245-
DOM.removeAttribute(el, attrName);
246-
}
247-
});
248-
for (const n of DOM.childNodesAsList(el)) {
249-
if (DOM.isElementNode(n)) stripCustomNsAttrs(n as Element);
250-
}
251-
}
210+
let inertBodyHelper: InertBodyHelper;
252211

253212
/**
254213
* Sanitizes the given unsafe, untrusted HTML fragment, and returns HTML text that is safe to add to
255214
* the DOM in a browser environment.
256215
*/
257216
export function sanitizeHtml(defaultDoc: any, unsafeHtmlInput: string): string {
217+
const DOM = getDOM();
218+
let inertBodyElement: HTMLElement|null = null;
258219
try {
259-
const containerEl = getInertElement();
220+
inertBodyHelper = inertBodyHelper || new InertBodyHelper(defaultDoc, DOM);
260221
// Make sure unsafeHtml is actually a string (TypeScript types are not enforced at runtime).
261222
let unsafeHtml = unsafeHtmlInput ? String(unsafeHtmlInput) : '';
223+
inertBodyElement = inertBodyHelper.getInertBodyElement(unsafeHtml);
262224

263225
// mXSS protection. Repeatedly parse the document to make sure it stabilizes, so that a browser
264226
// trying to auto-correct incorrect HTML cannot cause formerly inert HTML to become dangerous.
@@ -272,31 +234,25 @@ export function sanitizeHtml(defaultDoc: any, unsafeHtmlInput: string): string {
272234
mXSSAttempts--;
273235

274236
unsafeHtml = parsedHtml;
275-
DOM.setInnerHTML(containerEl, unsafeHtml);
276-
if (defaultDoc.documentMode) {
277-
// strip custom-namespaced attributes on IE<=11
278-
stripCustomNsAttrs(containerEl);
279-
}
280-
parsedHtml = DOM.getInnerHTML(containerEl);
237+
parsedHtml = DOM.getInnerHTML(inertBodyElement);
238+
inertBodyElement = inertBodyHelper.getInertBodyElement(unsafeHtml);
281239
} while (unsafeHtml !== parsedHtml);
282240

283241
const sanitizer = new SanitizingHtmlSerializer();
284-
const safeHtml = sanitizer.sanitizeChildren(DOM.getTemplateContent(containerEl) || containerEl);
285-
286-
// Clear out the body element.
287-
const parent = DOM.getTemplateContent(containerEl) || containerEl;
288-
for (const child of DOM.childNodesAsList(parent)) {
289-
DOM.removeChild(parent, child);
290-
}
291-
242+
const safeHtml =
243+
sanitizer.sanitizeChildren(DOM.getTemplateContent(inertBodyElement) || inertBodyElement);
292244
if (isDevMode() && sanitizer.sanitizedSomething) {
293245
DOM.log('WARNING: sanitizing HTML stripped some content (see http://g.co/ng/security#xss).');
294246
}
295247

296248
return safeHtml;
297-
} catch (e) {
249+
} finally {
298250
// In case anything goes wrong, clear out inertElement to reset the entire DOM structure.
299-
inertElement = null;
300-
throw e;
251+
if (inertBodyElement) {
252+
const parent = DOM.getTemplateContent(inertBodyElement) || inertBodyElement;
253+
for (const child of DOM.childNodesAsList(parent)) {
254+
DOM.removeChild(parent, child);
255+
}
256+
}
301257
}
302258
}

0 commit comments

Comments
 (0)