sanitize strings (#113)

Author: adamfweidman

packages/synthetics-sdk-broken-links/src/link_utils.ts

@@ -272,6 +272,37 @@ export function shuffleAndTruncate(
   return linksToFollow.slice(0, link_limit! - 1);
 }
 
+/**
+ * Sanitizes an object name string for safe use, ensuring compliance with
+ * naming restrictions.
+ *
+ * @param {string} inputString - The original object name string.
+ * @returns {string} The sanitized object name.
+ *
+ * **Sanitization Rules:**
+ * * Removes control characters ([\u007F-\u009F]).
+ * * Removes disallowed characters (#, [, ], *, ?, ", <, >, |, /).
+ * * Replaces the forbidden prefix ".well-known/acme-challenge/" with an underscore.
+ * * Replaces standalone occurrences of "." or ".." with an underscore.
+ */
+export function sanitizeObjectName(
+  inputString: string | null | undefined
+): string {
+  if (!inputString || inputString === '.' || inputString === '..') return '_';
+
+  // Regular expressions for:
+  /*eslint no-useless-escape: "off"*/
+  const invalidCharactersRegex = /[\r\n\u007F-\u009F#\[\]*?:"<>|/]/g; // Control characters, special characters, path separator
+  const wellKnownPrefixRegex = /^\.well-known\/acme-challenge\//;
+
+  // Core sanitization:
+  return inputString
+    .replace(wellKnownPrefixRegex, '_') // Replace forbidden prefix
+    .replace(invalidCharactersRegex, '_') // replace invalid characters
+    .trim() // Clean up any leading/trailing spaces
+    .replace(/\s+/g, '_'); // Replace one or more spaces with underscores
+}
+
 export function getTimeLimitPromise(
   startTime: string,
   totalTimeoutMillis: number,

packages/synthetics-sdk-broken-links/src/storage_func.ts

@@ -24,6 +24,7 @@ import {
   getExecutionRegion,
   resolveProjectId,
 } from '@google-cloud/synthetics-sdk-api';
+import { sanitizeObjectName } from './link_utils';
 
 // External Dependencies
 import { Storage, Bucket } from '@google-cloud/storage';
@@ -55,14 +56,14 @@ export async function getOrCreateStorageBucket(
   let bucketName = '';
 
   try {
-    const projectId = await resolveProjectId();
-    const region = await getExecutionRegion();
-
-    // if storageClient was not properly initialized OR the user chose to
-    // use/create the default bucket but we were not able to resolve projectId
-    // or cloudRegion
-    if (!storageClient || (!storageLocation && (!projectId || !region)))
-      return null;
+    if (!storageClient) return null;
+
+    const projectId = sanitizeObjectName(await resolveProjectId());
+    const region = sanitizeObjectName(await getExecutionRegion());
+
+    // if the user chose to use/create the default bucket but we were not able
+    // to resolve projectId or cloudRegion
+    if (!storageLocation && (!projectId || !region)) return null;
 
     bucketName = storageLocation
       ? storageLocation.split('/')[0]

packages/synthetics-sdk-broken-links/test/integration/integration.spec.ts

@@ -32,7 +32,7 @@ import {
 require('../../test/example_html_files/integration_server.js');
 const { getTestServer } = require('@google-cloud/functions-framework/testing');
 
-describe.only('CloudFunctionV2 Running Broken Link Synthetics', async () => {
+describe('CloudFunctionV2 Running Broken Link Synthetics', async () => {
   const status_class_2xx = {
     status_class: ResponseStatusCode_StatusClass.STATUS_CLASS_2XX,
   };

packages/synthetics-sdk-broken-links/test/unit/broken_links.spec.ts

@@ -35,7 +35,7 @@ import {
   BrokenLinkCheckerOptions,
 } from '../../src/broken_links';
 
-describe.only('runBrokenLinks', async () => {
+describe('runBrokenLinks', async () => {
   const status_class_2xx: ResponseStatusCode = {
     status_class: ResponseStatusCode_StatusClass.STATUS_CLASS_2XX,
   };

packages/synthetics-sdk-broken-links/test/unit/link_utils.spec.ts

@@ -30,6 +30,7 @@ import {
   createSyntheticResult,
   getGenericSyntheticResult,
   LinkIntermediate,
+  sanitizeObjectName,
   shuffleAndTruncate,
 } from '../../src/link_utils';
 import { setDefaultOptions } from '../../src/options_func';
@@ -210,4 +211,34 @@ describe('GCM Synthetics Broken Links Utilies', async () => {
     expect(startTime).to.be.lessThan(endTime);
     expect(milliDifference).to.be.greaterThan(0);
   });
+
+  describe('sanitizeObjectName', () => {
+    it('should remove invalid characters', () => {
+      const input = "test/\@#$%^&*()/_+\-=[]{};':\"\|,.<>/?\r\n\t";
+      const expectedOutput = "test_@_$%^&_()__+-=__{};'__\_,.______";
+      expect(sanitizeObjectName(input)).to.equal(expectedOutput);
+    });
+
+    it('should replace the forbidden prefix', () => {
+      const input = ".well-known/acme-challenge/test";
+      const expectedOutput = "_test";
+      expect(sanitizeObjectName(input)).to.equal(expectedOutput);
+    });
+
+    it('should handle standalone "." and ".."', () => {
+      expect(sanitizeObjectName(".")).to.equal("_");
+      expect(sanitizeObjectName("..")).to.equal("_");
+    });
+
+    it('should handle null and undefined', () => {
+      expect(sanitizeObjectName(null)).to.equal("_");
+      expect(sanitizeObjectName(undefined)).to.equal("_");
+    })
+
+    it('should trim leading and trailing whitespace', () => {
+      const input = "  test name  ";
+      const expectedOutput = "test_name";
+      expect(sanitizeObjectName(input)).to.equal(expectedOutput);
+    });
+  });
 });