Merge branch 'feat-search-mime-type' into 'main'

feat: add mime_type column in nostr_events, add the NIP 50 search extensions: exact_mime_type, partial_mime_type, only_media

See merge request soapbox-pub/ditto!623

deno.json

@@ -45,7 +45,7 @@
     "@lambdalisue/async": "jsr:@lambdalisue/async@^2.1.1",
     "@negrel/webpush": "jsr:@negrel/webpush@^0.3.0",
     "@noble/secp256k1": "npm:@noble/secp256k1@^2.0.0",
-    "@nostrify/db": "jsr:@nostrify/db@^0.36.2",
+    "@nostrify/db": "jsr:@nostrify/db@^0.37.0",
     "@nostrify/nostrify": "jsr:@nostrify/nostrify@^0.38.0",
     "@nostrify/policies": "jsr:@nostrify/policies@^0.36.1",
     "@nostrify/types": "jsr:@nostrify/types@^0.36.0",

deno.lock

@@ -30,11 +30,10 @@
     "jsr:@lambdalisue/async@^2.1.1": "2.1.1",
     "jsr:@negrel/http-ece@0.6.0": "0.6.0",
     "jsr:@negrel/webpush@0.3": "0.3.0",
-    "jsr:@nostrify/db@~0.36.2": "0.36.2",
+    "jsr:@nostrify/db@0.37": "0.37.0",
     "jsr:@nostrify/nostrify@0.31": "0.31.0",
     "jsr:@nostrify/nostrify@0.32": "0.32.0",
     "jsr:@nostrify/nostrify@0.36": "0.36.2",
-    "jsr:@nostrify/nostrify@0.37": "0.37.0",
     "jsr:@nostrify/nostrify@0.38": "0.38.0",
     "jsr:@nostrify/nostrify@~0.22.1": "0.22.5",
     "jsr:@nostrify/nostrify@~0.22.4": "0.22.4",
@@ -349,10 +348,10 @@
         "jsr:@std/path@0.224.0"
       ]
     },
-    "@nostrify/db@0.36.2": {
+    "@nostrify/db@0.37.0": {
-      "integrity": "6bf079b44fcb3ff5a85eadf9a9d4eb677fc770f1c80ad966602aa3d9dd8c88e8",
+      "integrity": "77398757ff52b1cf29ad3f610a8d3fcb8da37dd3300264baa4c318b4036684ab",
       "dependencies": [
-        "jsr:@nostrify/nostrify@0.37",
+        "jsr:@nostrify/nostrify@0.38",
         "jsr:@nostrify/types@0.36",
         "npm:kysely@~0.27.3",
         "npm:nostr-tools@^2.10.4"
@@ -456,21 +455,6 @@
         "npm:zod"
       ]
     },
-    "@nostrify/nostrify@0.37.0": {
-      "integrity": "fa1439cc5e9a74986c4fb799a38a9ed7bd8663c62ae2a9363ca9b987548e27a0",
-      "dependencies": [
-        "jsr:@nostrify/types@0.36",
-        "jsr:@std/crypto",
-        "jsr:@std/encoding@~0.224.1",
-        "npm:@scure/base",
-        "npm:@scure/bip32",
-        "npm:@scure/bip39",
-        "npm:lru-cache@^10.2.0",
-        "npm:nostr-tools@^2.7.0",
-        "npm:websocket-ts",
-        "npm:zod"
-      ]
-    },
     "@nostrify/nostrify@0.38.0": {
       "integrity": "9ec7920057ee3a4dcbaef7e706dedea622bfdfdf0f6aac11047443f88d953deb",
       "dependencies": [
@@ -2364,7 +2348,7 @@
       "jsr:@hono/hono@^4.4.6",
       "jsr:@lambdalisue/async@^2.1.1",
       "jsr:@negrel/webpush@0.3",
-      "jsr:@nostrify/db@~0.36.2",
+      "jsr:@nostrify/db@0.37",
       "jsr:@nostrify/nostrify@0.38",
       "jsr:@nostrify/policies@~0.36.1",
       "jsr:@nostrify/types@0.36",

src/controllers/api/accounts.ts

@@ -197,12 +197,13 @@ const accountStatusesQuerySchema = z.object({
   limit: z.coerce.number().nonnegative().transform((v) => Math.min(v, 40)).catch(20),
   exclude_replies: booleanParamSchema.optional(),
   tagged: z.string().optional(),
+  only_media: z.boolean().optional(),
 });
 
 const accountStatusesController: AppController = async (c) => {
   const pubkey = c.req.param('pubkey');
   const { since, until } = c.get('pagination');
-  const { pinned, limit, exclude_replies, tagged } = accountStatusesQuerySchema.parse(c.req.query());
+  const { pinned, limit, exclude_replies, tagged, only_media } = accountStatusesQuerySchema.parse(c.req.query());
   const { signal } = c.req.raw;
 
   const store = await Storages.db();
@@ -240,6 +241,10 @@ const accountStatusesController: AppController = async (c) => {
     limit,
   };
 
+  if (only_media) {
+    filter.search = 'media:true';
+  }
+
   if (tagged) {
     filter['#t'] = [tagged];
   }

src/db/DittoTables.ts

@@ -3,7 +3,6 @@ import { Generated } from 'kysely';
 import { NPostgresSchema } from '@nostrify/db';
 
 export interface DittoTables extends NPostgresSchema {
-  nostr_events: NostrEventsRow;
   auth_tokens: AuthTokenRow;
   author_stats: AuthorStatsRow;
   event_stats: EventStatsRow;
@@ -12,10 +11,6 @@ export interface DittoTables extends NPostgresSchema {
   push_subscriptions: PushSubscriptionRow;
 }
 
-type NostrEventsRow = NPostgresSchema['nostr_events'] & {
-  language: string | null;
-};
-
 interface AuthorStatsRow {
   pubkey: string;
   followers_count: number;

src/db/migrations/042_add_search_ext.ts

@@ -0,0 +1,38 @@
+import { Kysely, sql } from 'kysely';
+
+export async function up(db: Kysely<any>): Promise<void> {
+  await db.schema
+    .alterTable('nostr_events')
+    .addColumn('search_ext', 'jsonb', (col) => col.notNull().defaultTo(sql`'{}'::jsonb`))
+    .execute();
+
+  await db.schema
+    .alterTable('nostr_events')
+    .addCheckConstraint('nostr_events_search_ext_chk', sql`jsonb_typeof(search_ext) = 'object'`)
+    .execute();
+
+  await db.schema
+    .createIndex('nostr_events_search_ext_idx').using('gin')
+    .on('nostr_events')
+    .column('search_ext')
+    .ifNotExists()
+    .execute();
+}
+
+export async function down(db: Kysely<any>): Promise<void> {
+  await db.schema
+    .dropIndex('nostr_events_search_ext_idx')
+    .on('nostr_events')
+    .ifExists()
+    .execute();
+
+  await db.schema
+    .alterTable('nostr_events')
+    .dropConstraint('nostr_events_search_ext_chk')
+    .execute();
+
+  await db.schema
+    .alterTable('nostr_events')
+    .dropColumn('search_ext')
+    .execute();
+}

src/pipeline.ts

@@ -15,7 +15,6 @@ import { hydrateEvents } from '@/storages/hydrate.ts';
 import { Storages } from '@/storages.ts';
 import { eventAge, parseNip05, Time } from '@/utils.ts';
 import { getAmount } from '@/utils/bolt11.ts';
-import { detectLanguage } from '@/utils/language.ts';
 import { errorJson } from '@/utils/log.ts';
 import { nip05Cache } from '@/utils/nip05.ts';
 import { purifyEvent } from '@/utils/purify.ts';
@@ -121,7 +120,6 @@ async function handleEvent(event: DittoEvent, opts: PipelineOpts): Promise<void>
     Promise.allSettled([
       handleZaps(kysely, event),
       parseMetadata(event, opts.signal),
-      setLanguage(event),
       generateSetEvents(event),
     ])
       .then(() =>
@@ -237,24 +235,6 @@ async function parseMetadata(event: NostrEvent, signal: AbortSignal): Promise<vo
   }
 }
 
-/** Update the event in the database and set its language. */
-async function setLanguage(event: NostrEvent): Promise<void> {
-  if (event.kind !== 1) return;
-
-  const language = detectLanguage(event.content, 0.90);
-  if (!language) return;
-
-  const kysely = await Storages.kysely();
-  try {
-    await kysely.updateTable('nostr_events')
-      .set('language', language)
-      .where('id', '=', event.id)
-      .execute();
-  } catch {
-    // do nothing
-  }
-}
-
 /** Determine if the event is being received in a timely manner. */
 function isFresh(event: NostrEvent): boolean {
   return eventAge(event) < Time.minutes(1);

src/storages/EventsDB.test.ts

@@ -65,8 +65,8 @@ Deno.test('query events with language search filter', async () => {
   await store.event(en);
   await store.event(es);
 
-  await kysely.updateTable('nostr_events').set('language', 'en').where('id', '=', en.id).execute();
+  await kysely.updateTable('nostr_events').set('search_ext', { language: 'en' }).where('id', '=', en.id).execute();
-  await kysely.updateTable('nostr_events').set('language', 'es').where('id', '=', es.id).execute();
+  await kysely.updateTable('nostr_events').set('search_ext', { language: 'es' }).where('id', '=', es.id).execute();
 
   assertEquals(await store.query([{ search: 'language:en' }]), [en]);
   assertEquals(await store.query([{ search: 'language:es' }]), [es]);

src/storages/EventsDB.ts

@@ -1,11 +1,10 @@
 // deno-lint-ignore-file require-await
 
-import { LanguageCode } from 'iso-639-1';
+import { NPostgres } from '@nostrify/db';
-import { NPostgres, NPostgresSchema } from '@nostrify/db';
 import { NIP50, NKinds, NostrEvent, NostrFilter, NSchema as n } from '@nostrify/nostrify';
 import { logi } from '@soapbox/logi';
 import { JsonValue } from '@std/json';
-import { Kysely, SelectQueryBuilder } from 'kysely';
+import { Kysely } from 'kysely';
 import { nip27 } from 'nostr-tools';
 
 import { DittoTables } from '@/db/DittoTables.ts';
@@ -15,6 +14,7 @@ import { isNostrId } from '@/utils.ts';
 import { abortError } from '@/utils/abort.ts';
 import { purifyEvent } from '@/utils/purify.ts';
 import { DittoEvent } from '@/interfaces/DittoEvent.ts';
+import { detectLanguage } from '@/utils/language.ts';
 
 /** Function to decide whether or not to index a tag. */
 type TagCondition = (opts: TagConditionOpts) => boolean;
@@ -62,10 +62,44 @@ class EventsDB extends NPostgres {
     't': ({ event, count, value }) => (event.kind === 1985 ? count < 20 : count < 5) && value.length < 50,
   };
 
+  static indexExtensions(event: NostrEvent): Record<string, string> {
+    const ext: Record<string, string> = {};
+
+    if (event.kind === 1) {
+      ext.reply = event.tags.some(([name]) => name === 'e').toString();
+
+      const language = detectLanguage(event.content, 0.90);
+
+      if (language) {
+        ext.language = language;
+      }
+    }
+
+    const imeta: string[][][] = event.tags
+      .filter(([name]) => name === 'imeta')
+      .map(([_, ...entries]) =>
+        entries.map((entry) => {
+          const split = entry.split(' ');
+          return [split[0], split.splice(1).join(' ')];
+        })
+      );
+
+    if (imeta.length) {
+      ext.media = 'true';
+    }
+
+    if (imeta.every((tags) => tags.some(([name, value]) => name === 'm' && value.startsWith('video/')))) {
+      ext.video = 'true';
+    }
+
+    return ext;
+  }
+
   constructor(private opts: EventsDBOpts) {
     super(opts.kysely, {
       indexTags: EventsDB.indexTags,
       indexSearch: EventsDB.searchText,
+      indexExtensions: EventsDB.indexExtensions,
     });
   }
 
@@ -155,33 +189,6 @@ class EventsDB extends NPostgres {
     }
   }
 
-  protected override getFilterQuery(trx: Kysely<NPostgresSchema>, filter: NostrFilter) {
-    if (filter.search) {
-      const tokens = NIP50.parseInput(filter.search);
-
-      let query = super.getFilterQuery(trx, {
-        ...filter,
-        search: tokens.filter((t) => typeof t === 'string').join(' '),
-      }) as SelectQueryBuilder<DittoTables, 'nostr_events', DittoTables['nostr_events']>;
-
-      const languages = new Set<string>();
-
-      for (const token of tokens) {
-        if (typeof token === 'object' && token.key === 'language') {
-          languages.add(token.value);
-        }
-      }
-
-      if (languages.size) {
-        query = query.where('language', 'in', [...languages]);
-      }
-
-      return query;
-    }
-
-    return super.getFilterQuery(trx, filter);
-  }
-
   /** Get events for filters from the database. */
   override async query(
     filters: NostrFilter[],
@@ -210,29 +217,6 @@ class EventsDB extends NPostgres {
     return super.query(filters, { ...opts, timeout: opts.timeout ?? this.opts.timeout });
   }
 
-  /** Parse an event row from the database. */
-  protected override parseEventRow(row: DittoTables['nostr_events']): DittoEvent {
-    const event: DittoEvent = {
-      id: row.id,
-      kind: row.kind,
-      pubkey: row.pubkey,
-      content: row.content,
-      created_at: Number(row.created_at),
-      tags: row.tags,
-      sig: row.sig,
-    };
-
-    if (this.opts.pure) {
-      return event;
-    }
-
-    if (row.language) {
-      event.language = row.language as LanguageCode;
-    }
-
-    return event;
-  }
-
   /** Delete events based on filters from the database. */
   override async remove(filters: NostrFilter[], opts: { signal?: AbortSignal; timeout?: number } = {}): Promise<void> {
     logi({ level: 'debug', ns: 'ditto.remove', source: 'db', filters: filters as JsonValue });

src/trends.test.ts

@@ -93,12 +93,12 @@ Deno.test("getTrendingTagValues(): 'e' tag and WITH language parameter", async (
   }
 
   await db.kysely.updateTable('nostr_events')
-    .set('language', 'pt')
+    .set('search_ext', { language: 'pt' })
     .where('id', '=', post1.id)
     .execute();
 
   await db.kysely.updateTable('nostr_events')
-    .set('language', 'en')
+    .set('search_ext', { language: 'en' })
     .where('id', '=', post2.id)
     .execute();
 

src/trends.ts

@@ -145,7 +145,7 @@ export async function updateTrendingEvents(): Promise<void> {
     const rows = await kysely
       .selectFrom('nostr_events')
       .select('nostr_events.id')
-      .where('nostr_events.language', '=', language)
+      .where(sql`nostr_events.search_ext->>'language'`, '=', language)
       .where('nostr_events.created_at', '>=', yesterday)
       .where('nostr_events.created_at', '<=', now)
       .execute();

src/utils/search.test.ts

@@ -54,11 +54,13 @@ Deno.test('Searching for posts work', async () => {
 
   const event = genEvent({ content: "I'm not an orphan. Death is my importance", kind: 1 });
   await db.store.event(event);
-  await db.kysely.updateTable('nostr_events').set('language', 'en').where('id', '=', event.id).execute();
+  await db.kysely.updateTable('nostr_events').set('search_ext', { language: 'en' }).where('id', '=', event.id)
+    .execute();
 
   const event2 = genEvent({ content: 'The more I explore is the more I fall in love with the music I make.', kind: 1 });
   await db.store.event(event2);
-  await db.kysely.updateTable('nostr_events').set('language', 'en').where('id', '=', event2.id).execute();
+  await db.kysely.updateTable('nostr_events').set('search_ext', { language: 'en' }).where('id', '=', event2.id)
+    .execute();
 
   assertEquals(
     await getIdsBySearch(db.kysely, { q: 'Death is my importance', limit: 1, offset: 0 }), // ordered words

src/utils/search.ts

@@ -75,7 +75,7 @@ export async function getIdsBySearch(
   }
 
   if (languages.size) {
-    query = query.where('language', 'in', [...languages]);
+    query = query.where(sql`search_ext->>'language'`, 'in', [...languages]);
   }
 
   if (domains.size) {