feat(jobs): ingest.karakeep worker

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

lib/jobs/index.js

@@ -3,8 +3,9 @@ import * as echo from './workers/echo.js';
 import * as url from './workers/url.js';
 import * as blob from './workers/blob.js';
 import * as embed from './workers/embed.js';
+import * as karakeep from './workers/karakeep.js';
 
-const WORKERS = [echo, url, blob, embed];
+const WORKERS = [echo, url, blob, embed, karakeep];
 
 export async function registerWorkers() {
   for (const w of WORKERS) {

lib/jobs/workers/karakeep.js

@@ -0,0 +1,50 @@
+import crypto from 'node:crypto';
+import { getBookmark } from '../../karakeep/client.js';
+import { safeFetch } from '../../ingest/safe_fetch.js';
+import { extract } from '../../ingest/readability.js';
+import * as refs from '../../db/repos/refs.js';
+import { pool } from '../../db/pool.js';
+
+export const NAME = 'ingest.karakeep';
+
+function key(space_id, bookmark_id) {
+  return crypto.createHash('sha256')
+    .update(space_id + '\x00karakeep:' + bookmark_id).digest('hex');
+}
+
+export async function handler(job) {
+  const { bookmark_id, space_id } = job.data;
+  const bm = await getBookmark(bookmark_id);
+  if (!bm) return { skipped: 'gone' };
+
+  const idem = key(space_id, bookmark_id);
+  const { rows: [existing] } = await pool.query(
+    `SELECT id FROM refs WHERE source_kind='karakeep' AND external_id=$1 LIMIT 1`,
+    [idem]
+  );
+  if (existing) return { ref_id: existing.id, idempotent: true };
+
+  let html = bm.html_content;
+  if (!html && bm.url) {
+    const res = await safeFetch(bm.url, {
+      headers: { 'User-Agent': 'void-ingest/2.0' },
+      signal: AbortSignal.timeout(15_000)
+    });
+    if (res.ok) html = await res.text();
+  }
+  const parsed = html ? extract(html, bm.url) : { title: null, textContent: '', excerpt: null };
+
+  const row = await refs.create({
+    space_id,
+    kind: 'url',
+    source_url: bm.url,
+    title: bm.title || parsed.title || bm.url,
+    summary: parsed.excerpt,
+    body_text: (parsed.textContent || '').slice(0, 200_000),
+    source_kind: 'karakeep',
+    external_id: idem,
+    metadata: { karakeep_id: bookmark_id, tags: (bm.tags || []).map(t => t.name) }
+  }, { kind: 'system', id: null });
+
+  return { ref_id: row.id };
+}