Add tests to ensure we don't show shared rooms in the archive

Tests for https://github.com/matrix-org/matrix-public-archive/pull/239

server/routes/room-directory-routes.js

@@ -22,7 +22,6 @@ const matrixServerName = config.get('matrixServerName');
 assert(matrixServerName);
 const matrixAccessToken = config.get('matrixAccessToken');
 assert(matrixAccessToken);
-const stopSearchEngineIndexing = config.get('stopSearchEngineIndexing');
 
 const router = express.Router({
   caseSensitive: true,
@@ -71,7 +70,8 @@ router.get(
     }
 
     // We index the room directory unless the config says we shouldn't index anything
-    const shouldIndex = !stopSearchEngineIndexing;
+    const stopSearchEngineIndexingFromConfig = config.get('stopSearchEngineIndexing');
+    const shouldIndex = !stopSearchEngineIndexingFromConfig;
 
     const pageOptions = {
       title: `Matrix Public Archive`,

server/routes/room-routes.js

@@ -57,7 +57,6 @@ const matrixServerUrl = config.get('matrixServerUrl');
 assert(matrixServerUrl);
 const matrixAccessToken = config.get('matrixAccessToken');
 assert(matrixAccessToken);
-const stopSearchEngineIndexing = config.get('stopSearchEngineIndexing');
 
 const matrixPublicArchiveURLCreator = new MatrixPublicArchiveURLCreator(basePath);
 
@@ -889,7 +888,8 @@ router.get(
 
     // Default to no indexing (safe default)
     let shouldIndex = false;
-    if (stopSearchEngineIndexing) {
+    const stopSearchEngineIndexingFromConfig = config.get('stopSearchEngineIndexing');
+    if (stopSearchEngineIndexingFromConfig) {
       shouldIndex = false;
     } else {
       // Otherwise we only allow search engines to index `world_readable` rooms

test/e2e-tests.js

@@ -2686,15 +2686,50 @@ describe('matrix-public-archive', () => {
         assert.strictEqual(dom.document.querySelector(`meta[name="robots"]`), null);
       });
 
-      it('search engines not allowed to index `public` room', async () => {
+      it('search engines not allowed to access public room with non-`world_readable` history visibility', async () => {
         const client = await getTestClientForHs(testMatrixServerUrl1);
         const roomId = await createTestRoom(client, {
-          // The default options for the test rooms adds a
+          // Set as `shared` since it's the next most permissive history visibility
-          // `m.room.history_visiblity` state event so we override that here so
+          // after `world_readable` but still not allowed to be accesible in the
-          // it's only a public room.
+          // archive.
-          initial_state: [],
+          initial_state: [
+            {
+              type: 'm.room.history_visibility',
+              state_key: '',
+              content: {
+                history_visibility: 'shared',
+              },
+            },
+          ],
         });
 
+        try {
+          archiveUrl = matrixPublicArchiveURLCreator.archiveUrlForRoom(roomId);
+          await fetchEndpointAsText(archiveUrl);
+          assert.fail(
+            new TestError(
+              'We expect the request to fail with a 403 since the archive should not be able to view a non-world_readable room but it succeeded'
+            )
+          );
+        } catch (err) {
+          if (err instanceof TestError) {
+            throw err;
+          }
+          assert.strictEqual(
+            err.response.status,
+            403,
+            `Expected err.response.status=${err?.response?.status} to be 403 but error was: ${err.stack}`
+          );
+        }
+      });
+
+      it('Configuring `stopSearchEngineIndexing` will stop search engine indexing', async () => {
+        // Disable search engine indexing across the entire instance
+        config.set('stopSearchEngineIndexing', true);
+
+        const client = await getTestClientForHs(testMatrixServerUrl1);
+        const roomId = await createTestRoom(client);
+
         archiveUrl = matrixPublicArchiveURLCreator.archiveUrlForRoom(roomId);
         const { data: archivePageHtml } = await fetchEndpointAsText(archiveUrl);