Follow tombstone and predecessor history

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 +1,21 @@
+'use strict';
+
+const assert = require('assert');
+
+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/routes/room-routes.js

@@ -22,6 +22,7 @@ const {
   MS_LOOKUP,
   TIME_PRECISION_VALUES,
   DIRECTION,
+  VALID_ENTITY_DESCRIPTOR_TO_SIGIL_MAP,
 } = require('matrix-public-archive-shared/lib/reference-values');
 const { ONE_DAY_IN_MS, ONE_HOUR_IN_MS, ONE_MINUTE_IN_MS, ONE_SECOND_IN_MS } = MS_LOOKUP;
 const {
@@ -56,10 +57,6 @@ const router = express.Router({
   mergeParams: true,
 });
 
-const VALID_ENTITY_DESCRIPTOR_TO_SIGIL_MAP = {
-  r: '#',
-  roomid: '!',
-};
 const validSigilList = Object.values(VALID_ENTITY_DESCRIPTOR_TO_SIGIL_MAP);
 const sigilRe = new RegExp(`^(${validSigilList.join('|')})`);
 
@@ -159,6 +156,29 @@ function parseArchiveRangeFromReq(req) {
   };
 }
 
+function parseViaServersFromReq(req) {
+  const rawViaServers = [].concat(req.query.via || []);
+  if (rawViaServers.length === 0) {
+    return new Set();
+  }
+
+  const viaServerList = rawViaServers.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);
+}
+
 router.use(redirectToCorrectArchiveUrlIfBadSigil);
 
 router.get(
@@ -173,7 +193,11 @@ 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,
+      parseViaServersFromReq(req)
+    );
 
     // Find the closest day to the current time with messages
     const { originServerTs } = await timestampToEvent({
@@ -192,7 +216,7 @@ 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: parseViaServersFromReq(req),
       })
     );
   })
@@ -245,7 +269,11 @@ 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,
+      parseViaServersFromReq(req)
+    );
 
     let eventIdForClosestEvent;
     let tsForClosestEvent;
@@ -536,7 +564,11 @@ 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,
+      parseViaServersFromReq(req)
+    );
 
     // Do these in parallel to avoid the extra time in sequential round-trips
     // (we want to display the archive page faster)
@@ -587,6 +619,7 @@ router.get(
     const stylesUrl = urlJoin(basePath, '/css/styles.css');
     const jsBundleUrl = urlJoin(basePath, '/js/entry-client-hydrogen.es.js');
 
+    // XXX: The `renderHydrogenVmRenderScriptToPageHtml` API surface is pretty awkward
     const pageHtml = await renderHydrogenVmRenderScriptToPageHtml(
       path.resolve(__dirname, '../../shared/hydrogen-vm-render-script.js'),
       {

shared/lib/reference-values.js

@@ -28,9 +28,15 @@ const VALID_SIGIL_TO_ENTITY_DESCRIPTOR_MAP = {
   '!': 'roomid',
 };
 
+const VALID_ENTITY_DESCRIPTOR_TO_SIGIL_MAP = {
+  r: '#',
+  roomid: '!',
+};
+
 module.exports = {
   MS_LOOKUP,
   TIME_PRECISION_VALUES,
   DIRECTION,
   VALID_SIGIL_TO_ENTITY_DESCRIPTOR_MAP,
+  VALID_ENTITY_DESCRIPTOR_TO_SIGIL_MAP,
 };

shared/lib/timestamp-utilities.js

@@ -32,6 +32,7 @@ function roundUpTimestampToUtcSecond(ts) {
   return dateRountedUp.getTime();
 }
 
+// XXX: Should these just be renamed to `roundDownTimestampToUtcDay`?
 function getUtcStartOfDayTs(ts) {
   assert(typeof ts === 'number' || ts instanceof Date);
   const date = new Date(ts);