From 6dfe2134888682b745330887f03ae8504470e2e7 Mon Sep 17 00:00:00 2001
From: prashantjagasia <108553844+prashantjagasia@users.noreply.github.com>
Date: Tue, 17 Oct 2023 11:02:28 -0700
Subject: [PATCH] feat: added url-sanitization function (#110)

* feat: added url-sanitization function

* chore: fixed regex const

* chore: added constants and updated code

* chore: added tests, fixed build issues

* chore: removed modified dist files
---
 src/constants.js               | 12 ++++++++
 src/util.js                    | 54 +++++++++++++++++++++++++++++++++-
 test/tests/util/index.js       |  1 +
 test/tests/util/sanitizeUrl.js | 31 +++++++++++++++++++
 4 files changed, 97 insertions(+), 1 deletion(-)
 create mode 100644 test/tests/util/sanitizeUrl.js

diff --git a/src/constants.js b/src/constants.js
index b2d3144..752e0e1 100644
--- a/src/constants.js
+++ b/src/constants.js
@@ -10,3 +10,15 @@ export const ATTRIBUTES = {
 };
 
 export const UID_HASH_LENGTH = 30;
+
+/* eslint-disable no-control-regex*/
+export const invalidProtocolRegex: RegExp =
+  /([^\w]*)(javascript|data|vbscript)/im;
+export const htmlEntitiesRegex: RegExp = /&#(\w+)(^\w|;)?/g;
+export const htmlCtrlEntityRegex: RegExp = /&(newline|tab);/gi;
+export const ctrlCharactersRegex: RegExp =
+  /[\u0000-\u001F\u007F-\u009F\u2000-\u200D\uFEFF]/gim;
+export const urlSchemeRegex: RegExp = /^.+(:|&colon;)/gim;
+export const relativeFirstCharacters = [".", "/"];
+export const BLANK_URL = "about:blank";
+/* eslint-enable no-control-regex*/
diff --git a/src/util.js b/src/util.js
index d490755..27c462d 100644
--- a/src/util.js
+++ b/src/util.js
@@ -1,9 +1,17 @@
 /* @flow */
 /* eslint max-lines: 0 */
-
 import { ZalgoPromise } from "@krakenjs/zalgo-promise/src";
 import { WeakMap } from "@krakenjs/cross-domain-safe-weakmap/src";
 
+import {
+  BLANK_URL,
+  ctrlCharactersRegex,
+  htmlCtrlEntityRegex,
+  htmlEntitiesRegex,
+  invalidProtocolRegex,
+  relativeFirstCharacters,
+  urlSchemeRegex,
+} from "./constants";
 import type { CancelableType } from "./types";
 
 export function isElement(element: mixed): boolean {
@@ -1370,3 +1378,47 @@ export class ExtendableError extends Error {
     }
   }
 }
+
+function isRelativeUrlWithoutProtocol(url: string): boolean {
+  return relativeFirstCharacters.indexOf(url[0]) > -1;
+}
+
+function decodeHtmlCharacters(str: string): string {
+  const removedNullByte: string = str.replace(ctrlCharactersRegex, "");
+  return removedNullByte.replace(htmlEntitiesRegex, (matchRegex, dec) => {
+    return String.fromCharCode(dec);
+  });
+}
+
+export function sanitizeUrl(url?: string): string {
+  if (!url) {
+    return BLANK_URL;
+  }
+
+  const sanitizedUrl = decodeHtmlCharacters(url)
+    .replace(htmlCtrlEntityRegex, "")
+    .replace(ctrlCharactersRegex, "")
+    .trim();
+
+  if (!sanitizedUrl) {
+    return BLANK_URL;
+  }
+
+  if (isRelativeUrlWithoutProtocol(sanitizedUrl)) {
+    return sanitizedUrl;
+  }
+
+  const urlSchemeParseResults = sanitizedUrl.match(urlSchemeRegex);
+
+  if (!urlSchemeParseResults) {
+    return sanitizedUrl;
+  }
+
+  const urlScheme = urlSchemeParseResults[0];
+
+  if (invalidProtocolRegex.test(urlScheme)) {
+    return BLANK_URL;
+  }
+
+  return sanitizedUrl;
+}
diff --git a/test/tests/util/index.js b/test/tests/util/index.js
index 61858d9..19e0f71 100644
--- a/test/tests/util/index.js
+++ b/test/tests/util/index.js
@@ -11,3 +11,4 @@ import "./stringifyErrorMessage";
 import "./isRegex";
 import "./isDefined";
 import "./base64encode";
+import "./sanitizeUrl";
diff --git a/test/tests/util/sanitizeUrl.js b/test/tests/util/sanitizeUrl.js
new file mode 100644
index 0000000..0f1e207
--- /dev/null
+++ b/test/tests/util/sanitizeUrl.js
@@ -0,0 +1,31 @@
+/* @flow */
+
+import { sanitizeUrl } from "../../../src/util";
+
+describe("sanitizeUrl", () => {
+  it("should return about:blank for malicious URLs", () => {
+    const testURL =
+      'https://www.paypal.com/smart/checkout/venmo/popup?parentDomain=www.paypal.com&venmoWebEnabled=true&venmoWebUrl=javascript:alert(document["cookie"])//';
+    const sanitizedUrl = sanitizeUrl(testURL);
+
+    if (sanitizedUrl !== "about:blank") {
+      throw new Error(
+        `Does not match. Original data:\n\n${testURL}
+          \n\nSanitized:\n\n${"about:blank"}\n`
+      );
+    }
+  });
+
+  it("should return original URL when URL is clean", () => {
+    const testURL =
+      "https://www.paypal.com/smart/checkout/venmo/popup?parentDomain=www.paypal.com&venmoWebEnabled=true&venmoWebUrl=www.venmo.com";
+    const sanitizedUrl = sanitizeUrl(testURL);
+
+    if (sanitizedUrl !== testURL) {
+      throw new Error(
+        `Does not match. Original data:\n\n${testURL}
+          \n\nSanitized:\n\n${sanitizedUrl}\n`
+      );
+    }
+  });
+});