Do our best to get the user to the right place and try joining `via` derived server name (#168)

Split out from https://github.com/matrix-org/matrix-public-archive/pull/167

server/lib/matrix-utils/ensure-room-joined.js

@@ -4,15 +4,29 @@ const assert = require('assert');
 const urlJoin = require('url-join');
 
 const { fetchEndpointAsJson } = require('../fetch-endpoint');
+const getServerNameFromMatrixRoomIdOrAlias = require('./get-server-name-from-matrix-room-id-or-alias');
 
 const config = require('../config');
 const StatusError = require('../status-error');
 const matrixServerUrl = config.get('matrixServerUrl');
 assert(matrixServerUrl);
 
-async function ensureRoomJoined(accessToken, roomIdOrAlias, viaServers = []) {
+async function ensureRoomJoined(accessToken, roomIdOrAlias, viaServers = new Set()) {
+  // We use a `Set` to ensure that we don't have duplicate servers in the list
+  assert(viaServers instanceof Set);
+
+  // Let's do our best for the user to join the room. Since room ID's are
+  // unroutable on their own and won't be found if the server doesn't already
+  // know about the room, we'll try to join the room via the server name that
+  // we derived from the room ID or alias.
+  const viaServersWithAssumptions = new Set(viaServers);
+  const derivedServerName = getServerNameFromMatrixRoomIdOrAlias(roomIdOrAlias);
+  if (derivedServerName) {
+    viaServersWithAssumptions.add(derivedServerName);
+  }
+
   let qs = new URLSearchParams();
-  [].concat(viaServers).forEach((viaServer) => {
+  Array.from(viaServersWithAssumptions).forEach((viaServer) => {
     qs.append('server_name', viaServer);
   });
 

server/lib/matrix-utils/get-server-name-from-matrix-room-id-or-alias.js

@@ -0,0 +1,22 @@
+'use strict';
+
+const assert = require('assert');
+
+// See https://spec.matrix.org/v1.5/appendices/#server-name
+function getServerNameFromMatrixRoomIdOrAlias(roomIdOrAlias) {
+  assert(roomIdOrAlias);
+
+  const pieces = roomIdOrAlias.split(':');
+  // We can only derive the server name if there is a colon in the string. Since room
+  // IDs are supposed to be treated as opaque strings, there is a future possibility
+  // that they will not contain a colon.
+  if (pieces.length < 2) {
+    return null;
+  }
+
+  const servername = pieces.slice(1).join(':');
+
+  return servername;
+}
+
+module.exports = getServerNameFromMatrixRoomIdOrAlias;

server/lib/parse-via-servers-from-user-input.js

@@ -0,0 +1,30 @@
+'use strict';
+
+const StatusError = require('../lib/status-error');
+
+function parseViaServersFromUserInput(rawViaServers) {
+  // `rawViaServers` could be an array, a single string, or undefined. Turn it into an
+  // array no matter what
+  const rawViaServerList = [].concat(rawViaServers || []);
+  if (rawViaServerList.length === 0) {
+    return new Set();
+  }
+
+  const viaServerList = rawViaServerList.map((viaServer) => {
+    // Sanity check to ensure that the via servers are strings (valid enough looking
+    // host names)
+    if (typeof viaServer !== 'string') {
+      throw new StatusError(
+        400,
+        `?via server must be a string, got ${viaServer} (${typeof viaServer})`
+      );
+    }
+
+    return viaServer;
+  });
+
+  // We use a `Set` to ensure that we don't have duplicate servers in the list
+  return new Set(viaServerList);
+}
+
+module.exports = parseViaServersFromUserInput;

server/routes/room-routes.js

@@ -11,6 +11,7 @@ const timeoutMiddleware = require('./timeout-middleware');
 const redirectToCorrectArchiveUrlIfBadSigil = require('./redirect-to-correct-archive-url-if-bad-sigil-middleware');
 
 const { HTTPResponseError } = require('../lib/fetch-endpoint');
+const parseViaServersFromUserInput = require('../lib/parse-via-servers-from-user-input');
 const fetchRoomData = require('../lib/matrix-utils/fetch-room-data');
 const fetchEventsFromTimestampBackwards = require('../lib/matrix-utils/fetch-events-from-timestamp-backwards');
 const ensureRoomJoined = require('../lib/matrix-utils/ensure-room-joined');
@@ -173,7 +174,11 @@ router.get(
 
     // We have to wait for the room join to happen first before we can fetch
     // any of the additional room info or messages.
-    const roomId = await ensureRoomJoined(matrixAccessToken, roomIdOrAlias, req.query.via);
+    const roomId = await ensureRoomJoined(
+      matrixAccessToken,
+      roomIdOrAlias,
+      parseViaServersFromUserInput(req.query.via)
+    );
 
     // Find the closest day to the current time with messages
     const { originServerTs } = await timestampToEvent({
@@ -192,7 +197,7 @@ router.get(
         // We can avoid passing along the `via` query parameter because we already
         // joined the room above (see `ensureRoomJoined`).
         //
-        //viaServers: req.query.via,
+        //viaServers: parseViaServersFromUserInput(req.query.via),
       })
     );
   })
@@ -245,7 +250,11 @@ router.get(
 
     // We have to wait for the room join to happen first before we can use the jump to
     // date endpoint
-    const roomId = await ensureRoomJoined(matrixAccessToken, roomIdOrAlias, req.query.via);
+    const roomId = await ensureRoomJoined(
+      matrixAccessToken,
+      roomIdOrAlias,
+      parseViaServersFromUserInput(req.query.via)
+    );
 
     let eventIdForClosestEvent;
     let tsForClosestEvent;
@@ -536,7 +545,11 @@ router.get(
 
     // We have to wait for the room join to happen first before we can fetch
     // any of the additional room info or messages.
-    const roomId = await ensureRoomJoined(matrixAccessToken, roomIdOrAlias, req.query.via);
+    const roomId = await ensureRoomJoined(
+      matrixAccessToken,
+      roomIdOrAlias,
+      parseViaServersFromUserInput(req.query.via)
+    );
 
     // Do these in parallel to avoid the extra time in sequential round-trips
     // (we want to display the archive page faster)

test/server/matrix-utils/get-server-name-from-matrix-room-id-or-alias-tests.js

@@ -0,0 +1,58 @@
+'use strict';
+
+const assert = require('assert');
+
+const getServerNameFromMatrixRoomIdOrAlias = require('../../../server/lib/matrix-utils/get-server-name-from-matrix-room-id-or-alias');
+
+describe('getServerNameFromMatrixRoomIdOrAlias', () => {
+  // Some examples from https://spec.matrix.org/v1.5/appendices/#server-name
+  const testCases = [
+    {
+      name: 'can parse normal looking domain name',
+      input: '!foo:matrix.org',
+      expected: 'matrix.org',
+    },
+    {
+      name: 'can parse sub-domain',
+      input: '!foo:archive.matrix.org',
+      expected: 'archive.matrix.org',
+    },
+    {
+      name: 'can parse domain with port',
+      input: '!foo:matrix.org:8888',
+      expected: 'matrix.org:8888',
+    },
+    {
+      name: 'can parse IPv4 address',
+      input: '!foo:192.168.1.1',
+      expected: '192.168.1.1',
+    },
+    {
+      name: 'can parse IPv4 address with port',
+      input: '!foo:192.168.1.1:1234',
+      expected: '192.168.1.1:1234',
+    },
+    {
+      name: 'can parse IPv6 address',
+      input: '!foo:[1234:5678::abcd]',
+      expected: '[1234:5678::abcd]',
+    },
+    {
+      name: 'can parse IPv6 address with port',
+      input: '!foo:[1234:5678::abcd]:1234',
+      expected: '[1234:5678::abcd]:1234',
+    },
+    {
+      name: `opaque room ID is *NOT* parsed and we can't derive a server name`,
+      input: '!foobarbaz',
+      expected: null,
+    },
+  ];
+
+  testCases.forEach((testCaseMeta) => {
+    it(testCaseMeta.name, () => {
+      const actual = getServerNameFromMatrixRoomIdOrAlias(testCaseMeta.input);
+      assert.strictEqual(actual, testCaseMeta.expected);
+    });
+  });
+});