Defines and URL sanitizer that only blocks javascript: URLs.

RELNOTES: Defines and URL sanitizer that only blocks javascript: URLs.

PiperOrigin-RevId: 500960110
Change-Id: I1491fb817418cffc7687457fac13f8ae131f684d

closure/goog/html/BUILD

@@ -117,7 +117,10 @@ closure_js_library(
 closure_js_library(
     name = "safeurl_test_vectors",
     testonly = 1,
-    srcs = ["safeurl_test_vectors.js"],
+    srcs = [
+        "javascript_url_test_vectors.js",
+        "safeurl_test_vectors.js",
+    ],
     lenient = True,
 )
 

closure/goog/html/javascript_url_test_vectors.js

@@ -0,0 +1,70 @@
+/**
+ * @license
+ * Copyright The Closure Library Authors.
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+// AUTOGENERATED. DO NOT EDIT.
+// clang-format off
+
+goog.module('goog.html.javascriptUrlTestVectors');
+goog.setTestOnly('goog.html.javascriptUrlTestVectors');
+
+/** @typedef {{input: string, expected: string, safe: boolean }} */
+let TestVector;
+
+/** @const {!Array<!TestVector>} */
+const BASE_VECTORS = [
+      {input: '', expected: '', safe: true},
+      {input: 'http://example.com/', expected: 'http://example.com/', safe: true},
+      {input: 'https://example.com', expected: 'https://example.com', safe: true},
+      {input: 'mailto:foo@example.com', expected: 'mailto:foo@example.com', safe: true},
+      {input: 'ftp://example.com', expected: 'ftp://example.com', safe: true},
+      {input: 'ftp://username@example.com', expected: 'ftp://username@example.com', safe: true},
+      {input: 'ftp://username:password@example.com', expected: 'ftp://username:password@example.com', safe: true},
+      {input: 'HTtp://example.com/', expected: 'HTtp://example.com/', safe: true},
+      {input: 'https://example.com/path?foo\u003Dbar#baz', expected: 'https://example.com/path?foo\u003Dbar#baz', safe: true},
+      {input: 'https://example.com:123/path?foo\u003Dbar\u0026abc\u003Ddef#baz', expected: 'https://example.com:123/path?foo\u003Dbar\u0026abc\u003Ddef#baz', safe: true},
+      {input: '//example.com/path', expected: '//example.com/path', safe: true},
+      {input: '/path', expected: '/path', safe: true},
+      {input: '/path?foo\u003Dbar#baz', expected: '/path?foo\u003Dbar#baz', safe: true},
+      {input: 'path', expected: 'path', safe: true},
+      {input: 'path?foo\u003Dbar#baz', expected: 'path?foo\u003Dbar#baz', safe: true},
+      {input: 'p//ath', expected: 'p//ath', safe: true},
+      {input: 'p//ath?foo\u003Dbar#baz', expected: 'p//ath?foo\u003Dbar#baz', safe: true},
+      {input: '#baz', expected: '#baz', safe: true},
+      {input: '?:', expected: '?:', safe: true},
+      {input: 'not-data:image/png;base64,z\u003D', expected: 'not-data:image/png;base64,z\u003D', safe: true},
+      {input: ' data:image/png;base64,z\u003D', expected: ' data:image/png;base64,z\u003D', safe: true},
+      {input: 'tel:+1234567890', expected: 'tel:+1234567890', safe: true},
+      {input: 'sms:+1234567890', expected: 'sms:+1234567890', safe: true},
+      {input: 'callto:+1234567890', expected: 'callto:+1234567890', safe: true},
+      {input: 'wtai://wp/mc;+1234567890', expected: 'wtai://wp/mc;+1234567890', safe: true},
+      {input: 'rtsp://example.org/', expected: 'rtsp://example.org/', safe: true},
+      {input: 'market://details?id\u003Dapp', expected: 'market://details?id\u003Dapp', safe: true},
+      {input: 'itms://itunes.apple.com/us', expected: 'itms://itunes.apple.com/us', safe: true},
+      {input: 'javascript:evil(1);', expected: 'about:invalid#zClosurez', safe: false},
+      {input: 'javascript:evil(2);//\u000Ahttp://good.com/', expected: 'about:invalid#zClosurez', safe: false},
+      {input: ' javascript:evil(3);', expected: 'about:invalid#zClosurez', safe: false},
+      {input: '\u0009javascript:evil(4);', expected: 'about:invalid#zClosurez', safe: false},
+      {input: '\u000Bjavascript:evil(5);', expected: 'about:invalid#zClosurez', safe: false},
+      {input: 'JaVasCriPT:evil(6);', expected: 'about:invalid#zClosurez', safe: false},
+      {input: 'javascript:evil(8);', expected: 'about:invalid#zClosurez', safe: false},
+      {input: 'javascript:evil(9);', expected: 'about:invalid#zClosurez', safe: false},
+      {input: 'javasc\u0009ript:evil(10);', expected: 'about:invalid#zClosurez', safe: false},
+      {input: 'javasc\u0009ript:evil(11);', expected: 'about:invalid#zClosurez', safe: false}
+];
+
+/** @const {!Array<!TestVector>} */
+const TEL_VECTORS = [
+];
+
+/** @const {!Array<!TestVector>} */
+const SMS_VECTORS = [
+];
+
+/** @const {!Array<!TestVector>} */
+const SSH_VECTORS = [
+];
+
+exports = {BASE_VECTORS, TEL_VECTORS, SMS_VECTORS, SSH_VECTORS};

closure/goog/html/safeurl.js

@@ -741,6 +741,58 @@ goog.html.SafeUrl.sanitizeAssertUnchanged = function(url, opt_allowDataUrl) {
   return goog.html.SafeUrl.createSafeUrlSecurityPrivateDoNotAccessOrElse(url);
 };
 
+/**
+ * Extracts the scheme from the given URL. If the URL is relative, https: is
+ * assumed.
+ * @param {string} url The URL to extract the scheme from.
+ * @return {string|undefined} the URL scheme.
+ */
+goog.html.SafeUrl.extractScheme = function(url) {
+  let parsedUrl;
+  try {
+    parsedUrl = new URL(url);
+  } catch (e) {
+    // According to https://url.spec.whatwg.org/#constructors, the URL
+    // constructor with one parameter throws if `url` is not absolute. In this
+    // case, we are sure that no explicit scheme (javascript: ) is set.
+    // This can also be a URL parsing error, but in this case the URL won't be
+    // run anyway.
+    return 'https:';
+  }
+  return parsedUrl.protocol;
+};
+
+/**
+ * Creates a SafeUrl object from `url`. If `url` is a
+ * `goog.html.SafeUrl` then it is simply returned. Otherwise javascript: URLs
+ * are rejected.
+ *
+ * This function asserts (using goog.asserts) that the URL scheme is not
+ * javascript. If it is, in addition to failing the assert, an innocuous URL
+ * will be returned.
+ *
+ * @see http://url.spec.whatwg.org/#concept-relative-url
+ * @param {string|!goog.string.TypedString} url The URL to validate.
+ * @return {!goog.html.SafeUrl} The validated URL, wrapped as a SafeUrl.
+ */
+goog.html.SafeUrl.sanitizeJavascriptUrlAssertUnchanged = function(url) {
+  'use strict';
+  if (url instanceof goog.html.SafeUrl) {
+    return url;
+  } else if (typeof url == 'object' && url.implementsGoogStringTypedString) {
+    url = /** @type {!goog.string.TypedString} */ (url).getTypedStringValue();
+  } else {
+    url = String(url);
+  }
+  // We don't rely on goog.url here to prevent a dependency cycle.
+  const parsedScheme = goog.html.SafeUrl.extractScheme(url);
+  if (!goog.asserts.assert(
+          parsedScheme !== 'javascript:', '%s is a javascript: URL', url)) {
+    url = goog.html.SafeUrl.INNOCUOUS_STRING;
+  }
+  return goog.html.SafeUrl.createSafeUrlSecurityPrivateDoNotAccessOrElse(url);
+};
+
 /**
  * Token used to ensure that object is created only from this file. No code
  * outside of this file can access this token.

closure/goog/html/safeurl_test.js

@@ -15,6 +15,7 @@ const SafeUrl = goog.require('goog.html.SafeUrl');
 const TrustedResourceUrl = goog.require('goog.html.TrustedResourceUrl');
 const fsUrl = goog.require('goog.fs.url');
 const googObject = goog.require('goog.object');
+const javascriptUrlTestVectors = goog.require('goog.html.javascriptUrlTestVectors');
 const safeUrlTestVectors = goog.require('goog.html.safeUrlTestVectors');
 const testSuite = goog.require('goog.testing.testSuite');
 const {assertExists} = goog.require('goog.asserts');
@@ -382,6 +383,23 @@ testSuite({
     }
   },
 
+  /**
+     @suppress {missingProperties,checkTypes} suppression added to enable type
+     checking
+   */
+  testSafeUrlSanitize_sanitizeJavascriptUrlAssertUnchanged() {
+    for (const v of javascriptUrlTestVectors.BASE_VECTORS) {
+      if (v.safe) {
+        const asserted = SafeUrl.sanitizeJavascriptUrlAssertUnchanged(v.input);
+        assertEquals(v.expected, SafeUrl.unwrap(asserted));
+      } else {
+        assertThrows(() => {
+          SafeUrl.sanitizeJavascriptUrlAssertUnchanged(v.input);
+        });
+      }
+    }
+  },
+
   testSafeUrlSanitize_sanitizeProgramConstants() {
     // .sanitize() works on program constants.
     const good = Const.from('http://example.com/');