Merge pull request #7892 from gmambro/allow-prefix

Optional paths in backend.reading.allow to restrict allowed targets

.changeset/nasty-impalas-travel.md

@@ -0,0 +1,5 @@
+---
+'@backstage/backend-common': patch
+---
+
+Paths can be specified in backend.reading.allow to further restrict allowed targets

docs/features/software-catalog/descriptor-format.md

@@ -144,7 +144,8 @@ spec:
 
 Note that to be able to read from targets that are outside of the normal
 integration points such as `github.com`, you'll need to explicitly allow it by
-adding an entry in the `backend.reading.allow` list. For example:
+adding an entry in the `backend.reading.allow` list. Paths can be specified to
+further restrict targets For example:
 
 ```yml
 backend:
@@ -153,6 +154,8 @@ backend:
     allow:
       - host: example.com
       - host: '*.examples.org'
+      - host: example.net
+        paths: ['/api/']
 ```
 
 ## Common to All Kinds: The Envelope

packages/backend-common/config.d.ts

@@ -130,6 +130,14 @@ export interface Config {
          * The host may also contain a port, for example `example.com:8080`.
          */
         host: string;
+
+        /**
+         * An optional list of paths. In case they are present only targets matching
+         * any of them will are allowed. You can use trailing slashes to make sure only
+         * subdirectories are allowed, for example `/mydir/` will allow targets with
+         * paths like `/mydir/a` but will block paths like `/mydir2`.
+         */
+        paths?: string[];
       }>;
     };
 

packages/backend-common/src/reading/FetchUrlReader.test.ts

@@ -77,6 +77,10 @@ describe('FetchUrlReader', () => {
               { host: 'example.com:700' },
               { host: '*.examples.org' },
               { host: '*.examples.org:700' },
+              {
+                host: 'foobar.org',
+                paths: ['/dir1/'],
+              },
             ],
           },
         },
@@ -106,6 +110,9 @@ describe('FetchUrlReader', () => {
     expect(predicate(new URL('https://examples.org:700/test'))).toBe(false);
     expect(predicate(new URL('https://a.examples.org:700/test'))).toBe(true);
     expect(predicate(new URL('https://a.b.examples.org:700/test'))).toBe(true);
+    expect(predicate(new URL('https://foobar.org/dir1/subpath'))).toBe(true);
+    expect(predicate(new URL('https://foobar.org/dir12'))).toBe(false);
+    expect(predicate(new URL('https://foobar.org/'))).toBe(false);
   });
 
   describe('read', () => {

packages/backend-common/src/reading/FetchUrlReader.ts

@@ -24,6 +24,7 @@ import {
   SearchResponse,
   UrlReader,
 } from './types';
+import path from 'path';
 
 /**
  * A UrlReader that does a plain fetch of the URL.
@@ -39,18 +40,30 @@ export class FetchUrlReader implements UrlReader {
    * `host`:
    *   Either full hostnames to match, or subdomain wildcard matchers with a leading `*`.
    *   For example `example.com` and `*.example.com` are valid values, `prod.*.example.com` is not.
+   *
+   * `paths`:
+   *   An optional list of paths which are allowed. If the list is omitted all paths are allowed.
    */
   static factory: ReaderFactory = ({ config }) => {
     const predicates =
       config
         .getOptionalConfigArray('backend.reading.allow')
         ?.map(allowConfig => {
+          const paths = allowConfig.getOptionalStringArray('paths');
+          const checkPath = paths
+            ? (url: URL) => {
+                const targetPath = path.posix.normalize(url.pathname);
+                return paths.some(allowedPath =>
+                  targetPath.startsWith(allowedPath),
+                );
+              }
+            : (_url: URL) => true;
           const host = allowConfig.getString('host');
           if (host.startsWith('*.')) {
             const suffix = host.slice(1);
-            return (url: URL) => url.host.endsWith(suffix);
+            return (url: URL) => url.host.endsWith(suffix) && checkPath(url);
           }
-          return (url: URL) => url.host === host;
+          return (url: URL) => url.host === host && checkPath(url);
         }) ?? [];
 
     const reader = new FetchUrlReader();