Merge branch 'search-perf' into 'main'

Improve performance of account search

See merge request soapbox-pub/ditto!653

src/controllers/api/accounts.ts

@@ -112,13 +112,14 @@ const accountLookupController: AppController = async (c) => {
 
 const accountSearchQuerySchema = z.object({
   q: z.string().transform(decodeURIComponent),
-  resolve: booleanParamSchema.optional().transform(Boolean),
+  resolve: booleanParamSchema.optional(),
   following: z.boolean().default(false),
 });
 
 const accountSearchController: AppController = async (c) => {
   const { signal } = c.req.raw;
   const { limit } = c.get('pagination');
+
   const kysely = await Storages.kysely();
   const viewerPubkey = await c.get('signer')?.getPublicKey();
 
@@ -136,27 +137,28 @@ const accountSearchController: AppController = async (c) => {
 
   if (!event && lookup) {
     const pubkey = await lookupPubkey(lookup);
-    return c.json(pubkey ? [await accountFromPubkey(pubkey)] : []);
+    return c.json(pubkey ? [accountFromPubkey(pubkey)] : []);
   }
 
-  const followedPubkeys: Set<string> = viewerPubkey ? await getFollowedPubkeys(viewerPubkey) : new Set();
-  const pubkeys = Array.from(await getPubkeysBySearch(kysely, { q: query, limit, offset: 0, followedPubkeys }));
+  const events: NostrEvent[] = [];
 
-  let events = event ? [event] : await store.query([{ kinds: [0], authors: pubkeys, limit }], {
-    signal,
-  });
+  if (event) {
+    events.push(event);
+  } else {
+    const following = viewerPubkey ? await getFollowedPubkeys(viewerPubkey) : new Set<string>();
+    const authors = [...await getPubkeysBySearch(kysely, { q: query, limit, offset: 0, following })];
+    const profiles = await store.query([{ kinds: [0], authors, limit }], { signal });
 
-  if (!event) {
-    events = pubkeys
-      .map((pubkey) => events.find((event) => event.pubkey === pubkey))
-      .filter((event) => !!event);
+    for (const pubkey of authors) {
+      const profile = profiles.find((event) => event.pubkey === pubkey);
+      if (profile) {
+        events.push(profile);
+      }
+    }
   }
-  const accounts = await hydrateEvents({ events, store, signal }).then(
-    (events) =>
-      Promise.all(
-        events.map((event) => renderAccount(event)),
-      ),
-  );
+
+  const accounts = await hydrateEvents({ events, store, signal })
+    .then((events) => events.map((event) => renderAccount(event)));
 
   return c.json(accounts);
 };

src/controllers/api/search.ts

@@ -104,8 +104,8 @@ async function searchEvents(
 
   // For account search, use a special index, and prioritize followed accounts.
   if (type === 'accounts') {
-    const followedPubkeys = viewerPubkey ? await getFollowedPubkeys(viewerPubkey) : new Set<string>();
-    const searchPubkeys = await getPubkeysBySearch(kysely, { q, limit, offset, followedPubkeys });
+    const following = viewerPubkey ? await getFollowedPubkeys(viewerPubkey) : new Set<string>();
+    const searchPubkeys = await getPubkeysBySearch(kysely, { q, limit, offset, following });
 
     filter.authors = [...searchPubkeys];
     filter.search = undefined;

src/cron.ts

@@ -1,6 +1,13 @@
-import { updateTrendingLinks } from '@/trends.ts';
-import { updateTrendingHashtags } from '@/trends.ts';
-import { updateTrendingEvents, updateTrendingPubkeys, updateTrendingZappedEvents } from '@/trends.ts';
+import { sql } from 'kysely';
+
+import { Storages } from '@/storages.ts';
+import {
+  updateTrendingEvents,
+  updateTrendingHashtags,
+  updateTrendingLinks,
+  updateTrendingPubkeys,
+  updateTrendingZappedEvents,
+} from '@/trends.ts';
 
 /** Start cron jobs for the application. */
 export function cron() {
@@ -9,4 +16,9 @@ export function cron() {
   Deno.cron('update trending events', '15 * * * *', updateTrendingEvents);
   Deno.cron('update trending hashtags', '30 * * * *', updateTrendingHashtags);
   Deno.cron('update trending links', '45 * * * *', updateTrendingLinks);
+
+  Deno.cron('refresh top authors', '20 * * * *', async () => {
+    const kysely = await Storages.kysely();
+    await sql`refresh materialized view top_authors`.execute(kysely);
+  });
 }

src/db/DittoTables.ts

@@ -9,6 +9,8 @@ export interface DittoTables extends NPostgresSchema {
   event_stats: EventStatsRow;
   event_zaps: EventZapRow;
   push_subscriptions: PushSubscriptionRow;
+  /** This is a materialized view of `author_stats` pre-sorted by followers_count. */
+  top_authors: Pick<AuthorStatsRow, 'pubkey' | 'followers_count' | 'search'>;
 }
 
 interface AuthorStatsRow {

src/db/migrations/049_author_stats_sorted.ts

@@ -0,0 +1,20 @@
+import { Kysely, sql } from 'kysely';
+
+export async function up(db: Kysely<any>): Promise<void> {
+  await db.schema
+    .createView('top_authors')
+    .materialized()
+    .as(db.selectFrom('author_stats').select(['pubkey', 'followers_count', 'search']).orderBy('followers_count desc'))
+    .execute();
+
+  await sql`CREATE INDEX top_authors_search_idx ON top_authors USING GIN (search gin_trgm_ops)`.execute(db);
+
+  await db.schema.createIndex('top_authors_pubkey_idx').on('top_authors').column('pubkey').execute();
+
+  await db.schema.dropIndex('author_stats_search_idx').execute();
+}
+
+export async function down(db: Kysely<unknown>): Promise<void> {
+  await db.schema.dropView('top_authors').execute();
+  await sql`CREATE INDEX author_stats_search_idx ON author_stats USING GIN (search gin_trgm_ops)`.execute(db);
+}

src/utils/search.test.ts

@@ -1,4 +1,5 @@
 import { assertEquals } from '@std/assert';
+import { sql } from 'kysely';
 
 import { createTestDB } from '@/test.ts';
 import { getPubkeysBySearch } from '@/utils/search.ts';
@@ -14,18 +15,20 @@ Deno.test('fuzzy search works', async () => {
     following_count: 0,
   }).execute();
 
+  await sql`REFRESH MATERIALIZED VIEW top_authors`.execute(db.kysely);
+
   assertEquals(
-    await getPubkeysBySearch(db.kysely, { q: 'pat rick', limit: 1, offset: 0, followedPubkeys: new Set() }),
+    await getPubkeysBySearch(db.kysely, { q: 'pat rick', limit: 1, offset: 0, following: new Set() }),
     new Set(),
   );
   assertEquals(
-    await getPubkeysBySearch(db.kysely, { q: 'patrick dosreis', limit: 1, offset: 0, followedPubkeys: new Set() }),
+    await getPubkeysBySearch(db.kysely, { q: 'patrick dosreis', limit: 1, offset: 0, following: new Set() }),
     new Set([
       '47259076c85f9240e852420d7213c95e95102f1de929fb60f33a2c32570c98c4',
     ]),
   );
   assertEquals(
-    await getPubkeysBySearch(db.kysely, { q: 'dosreis.com', limit: 1, offset: 0, followedPubkeys: new Set() }),
+    await getPubkeysBySearch(db.kysely, { q: 'dosreis.com', limit: 1, offset: 0, following: new Set() }),
     new Set([
       '47259076c85f9240e852420d7213c95e95102f1de929fb60f33a2c32570c98c4',
     ]),
@@ -43,8 +46,10 @@ Deno.test('fuzzy search works with offset', async () => {
     following_count: 0,
   }).execute();
 
+  await sql`REFRESH MATERIALIZED VIEW top_authors`.execute(db.kysely);
+
   assertEquals(
-    await getPubkeysBySearch(db.kysely, { q: 'dosreis.com', limit: 1, offset: 1, followedPubkeys: new Set() }),
+    await getPubkeysBySearch(db.kysely, { q: 'dosreis.com', limit: 1, offset: 1, following: new Set() }),
     new Set(),
   );
 });

src/utils/search.ts

@@ -5,30 +5,34 @@ import { DittoTables } from '@/db/DittoTables.ts';
 /** Get pubkeys whose name and NIP-05 is similar to 'q' */
 export async function getPubkeysBySearch(
   kysely: Kysely<DittoTables>,
-  opts: { q: string; limit: number; offset: number; followedPubkeys: Set<string> },
+  opts: { q: string; limit: number; offset: number; following: Set<string> },
 ): Promise<Set<string>> {
-  const { q, limit, followedPubkeys, offset } = opts;
+  const { q, limit, following, offset } = opts;
 
-  let query = kysely
-    .selectFrom('author_stats')
-    .select((eb) => [
-      'pubkey',
-      'search',
-      eb.fn('word_similarity', [sql`${q}`, 'search']).as('sml'),
-    ])
-    .where(() => sql`${q} <% search`)
-    .orderBy(['followers_count desc'])
-    .orderBy(['sml desc', 'search'])
+  const pubkeys = new Set<string>();
+
+  const query = kysely
+    .selectFrom('top_authors')
+    .select('pubkey')
+    .where('search', sql`%>`, q)
     .limit(limit)
     .offset(offset);
 
-  const pubkeys = new Set((await query.execute()).map(({ pubkey }) => pubkey));
+  if (following.size) {
+    const authorsQuery = query.where('pubkey', 'in', [...following]);
 
-  if (followedPubkeys.size > 0) {
-    query = query.where('pubkey', 'in', [...followedPubkeys]);
+    for (const { pubkey } of await authorsQuery.execute()) {
+      pubkeys.add(pubkey);
+    }
   }
 
-  const followingPubkeys = new Set((await query.execute()).map(({ pubkey }) => pubkey));
+  if (pubkeys.size >= limit) {
+    return pubkeys;
+  }
 
-  return new Set(Array.from(followingPubkeys.union(pubkeys)));
+  for (const { pubkey } of await query.limit(limit - pubkeys.size).execute()) {
+    pubkeys.add(pubkey);
+  }
+
+  return pubkeys;
 }