nandi.uk
This repository has no description
1/**
2 * Papers Appview — watches the Paper Skygest feed, detects linked papers,
3 * generates summaries, and writes org.latha.papers.summary records to PDS.
4 *
5 * Hosted at papers.latha.org
6 */
7
8import { Hono } from "hono";
9import { cors } from "hono/cors";
10import { createWorker } from "@atmo-dev/contrail/worker";
11import { Contrail } from "@atmo-dev/contrail";
12import { lexicons } from "../lexicons/generated/index.js";
13import { config } from "./contrail.config.js";
14import { pollNewPapers, pollFollowedAccounts, type PaperPost } from "./feed-watcher.js";
15import { buildSummary, enrichWithLetta } from "./paper-summarizer.js";
16import { writeSummaryToPds } from "./pds-writer.js";
17import { LANDING_PAGE } from "./landing-page.js";
18import { SUMMARY_PAGE } from "./summary-page.js";
19
20interface Env {
21 DB: D1Database;
22 FEED_CURSOR?: string;
23 RESEARCHER_KEY_PEM?: string;
24 LETTA_API_KEY?: string;
25 LETTA_AGENT_ID?: string;
26}
27
28const contrailWorker = createWorker(config, { lexicons });
29const contrail = new Contrail(config);
30let contrailReady = false;
31
32async function ensureContrailReady(db: D1Database): Promise<void> {
33 if (contrailReady) return;
34 await contrail.init(db);
35 contrailReady = true;
36}
37
38let app: Hono<{ Bindings: Env }> | null = null;
39
40function buildApp(env: Env): Hono<{ Bindings: Env }> {
41 const app = new Hono<{ Bindings: Env }>();
42 const db = env.DB;
43
44 app.use("*", cors());
45
46 // Landing page — browse all indexed summaries
47 app.get("/", async (c) => {
48 await ensureContrailReady(db);
49
50 let summaries: any[] = [];
51 try {
52 const rows = await db
53 .prepare(
54 "SELECT uri, record, did, rkey, time_us FROM records_summary ORDER BY time_us DESC LIMIT 100"
55 )
56 .all();
57 summaries = (rows.results || []).map((r: any) => {
58 try {
59 const value = JSON.parse(r.record);
60 return {
61 uri: r.uri,
62 rkey: r.rkey,
63 title: value?.title || "Untitled",
64 paperUrl: value?.paperUrl || "",
65 venue: value?.venue || "",
66 year: value?.year || null,
67 authors: value?.authors || [],
68 summary: value?.summary || "",
69 domains: value?.domains || [],
70 posterHandle: value?.posterHandle || "",
71 indexedAt: value?.indexedAt || "",
72 };
73 } catch {
74 return null;
75 }
76 }).filter(Boolean);
77 } catch {
78 // empty list is fine
79 }
80
81 const page = LANDING_PAGE.replace(
82 "</head>",
83 `<script>window.__SUMMARIES__=${JSON.stringify(summaries).replace(/<\//g, "<\\/")};</script></head>`
84 );
85 return c.html(page);
86 });
87
88 // Summary detail page
89 app.get("/paper/:rkey", async (c) => {
90 const rkey = c.req.param("rkey");
91 await ensureContrailReady(db);
92
93 let summary: any = null;
94 try {
95 const row = await db
96 .prepare("SELECT uri, record, did, rkey FROM records_summary WHERE rkey = ? LIMIT 1")
97 .bind(rkey)
98 .first<{ uri: string; record: string; did: string; rkey: string }>();
99 if (row) {
100 const value = JSON.parse(row.record);
101 summary = {
102 uri: row.uri,
103 rkey: row.rkey,
104 ...value,
105 };
106 }
107 } catch {
108 // not found
109 }
110
111 if (!summary) {
112 return c.html("<h1>Not Found</h1>", 404);
113 }
114
115 const page = SUMMARY_PAGE.replace(
116 "</head>",
117 `<script>window.__SUMMARY__=${JSON.stringify(summary)};</script></head>`
118 );
119 return c.html(page);
120 });
121
122 // API: list summaries as JSON
123 app.get("/api/summaries", async (c) => {
124 await ensureContrailReady(db);
125 const limit = Math.min(Number(c.req.query("limit") || 50), 200);
126 const cursor = c.req.query("cursor");
127
128 let rows: D1Result<any>;
129 if (cursor) {
130 rows = await db
131 .prepare("SELECT uri, record, did, rkey, time_us FROM records_summary WHERE time_us < ? ORDER BY time_us DESC LIMIT ?")
132 .bind(cursor, limit)
133 .all();
134 } else {
135 rows = await db
136 .prepare("SELECT uri, record, did, rkey, time_us FROM records_summary ORDER BY time_us DESC LIMIT ?")
137 .bind(limit)
138 .all();
139 }
140
141 const summaries = (rows.results || []).map((r: any) => {
142 try {
143 const value = JSON.parse(r.record);
144 return { uri: r.uri, rkey: r.rkey, ...value };
145 } catch {
146 return null;
147 }
148 }).filter(Boolean);
149
150 const lastRow = rows.results?.[rows.results.length - 1];
151 const nextCursor = lastRow?.time_us || undefined;
152
153 return c.json({ summaries, cursor: nextCursor });
154 });
155
156 // API: trigger feed poll + summary generation
157 app.post("/api/poll", async (c) => {
158 await ensureContrailReady(db);
159
160 // Get last cursor from D1
161 let lastCursor: string | undefined;
162 try {
163 const row = await db
164 .prepare("SELECT value FROM instance_settings WHERE key = 'feed_cursor' LIMIT 1")
165 .first<{ value: string }>();
166 lastCursor = row?.value || undefined;
167 } catch {
168 // first run
169 }
170
171 const { posts: feedPosts, newCursor } = await pollNewPapers(lastCursor);
172
173 // Also poll followed accounts for paper links
174 let followedPosts: PaperPost[] = [];
175 try {
176 followedPosts = await pollFollowedAccounts(5);
177 } catch (err: any) {
178 console.error(`Followed accounts poll failed: ${err?.message ?? err}`);
179 }
180
181 // Merge and deduplicate
182 const seenPaperUrls = new Set<string>();
183 const posts: PaperPost[] = [];
184 for (const p of [...feedPosts, ...followedPosts]) {
185 if (!seenPaperUrls.has(p.paperUrl)) {
186 seenPaperUrls.add(p.paperUrl);
187 posts.push(p);
188 }
189 }
190
191 // Deduplicate against existing records
192 const existingUrls = new Set<string>();
193 if (posts.length > 0) {
194 const placeholders = posts.map(() => "?").join(",");
195 const urls = posts.map(p => p.paperUrl);
196 const rows = await db
197 .prepare(
198 `SELECT DISTINCT json_extract(record, '$.paperUrl') as paperUrl FROM records_summary WHERE paperUrl IN (${placeholders})`
199 )
200 .bind(...urls)
201 .all<{ paperUrl: string }>();
202 for (const r of rows.results || []) {
203 if (r.paperUrl) existingUrls.add(r.paperUrl);
204 }
205 }
206
207 const newPosts = posts.filter(p => !existingUrls.has(p.paperUrl));
208 let written = 0;
209 let errors = 0;
210 const errorDetails: string[] = [];
211
212 const apiKey = process.env.LETTA_API_KEY;
213 const agentId = process.env.LETTA_AGENT_ID;
214
215 for (const post of newPosts) {
216 try {
217 let summary = await buildSummary(post);
218
219 // Enrich with LLM if available
220 if (apiKey && agentId) {
221 try {
222 const enrichment = await enrichWithLetta(
223 { title: summary.title, paperUrl: summary.paperUrl, abstract: summary.abstract, domains: summary.domains, venue: summary.venue },
224 apiKey,
225 agentId,
226 );
227 if (enrichment.title) summary.title = enrichment.title;
228 summary.summary = enrichment.summary;
229 summary.domains = enrichment.domains;
230 if (enrichment.takeaway) (summary as any).takeaway = enrichment.takeaway;
231 } catch (err: any) {
232 console.error(`Enrichment failed for ${post.paperUrl}: ${err?.message ?? err}, using heuristic`);
233 }
234 }
235
236 const result = await writeSummaryToPds(post, summary);
237
238 // Index the new record into D1 directly
239 const rkey = result.uri.split("/").pop() || "";
240 await db
241 .prepare("INSERT OR REPLACE INTO records_summary (uri, cid, did, rkey, record, time_us, indexed_at) VALUES (?, ?, ?, ?, ?, ?, ?)")
242 .bind(result.uri, result.cid, "did:plc:3kkhul7jznlb6ba7rprzawnj", rkey, JSON.stringify({
243 $type: "org.latha.papers.summary",
244 ...summary,
245 sourceUri: post.postUri,
246 posterDid: post.posterDid,
247 posterHandle: post.posterHandle,
248 postText: post.postText,
249 indexedAt: new Date().toISOString(),
250 }), Date.now() * 1000, Date.now() * 1000)
251 .run();
252 written++;
253 } catch (err: any) {
254 const msg = `${post.paperUrl}: ${err?.message ?? err}`;
255 console.error(`Failed to write summary for ${msg}`);
256 if (errors < 3) errorDetails.push(msg.substring(0, 200));
257 errors++;
258 }
259 }
260
261 // Save cursor
262 if (newCursor) {
263 await db
264 .prepare("INSERT OR REPLACE INTO instance_settings (key, value) VALUES ('feed_cursor', ?)")
265 .bind(newCursor)
266 .run();
267 }
268
269 return c.json({
270 polled: posts.length,
271 fromFeed: feedPosts.length,
272 fromFollows: followedPosts.length,
273 new: newPosts.length,
274 written,
275 errors,
276 lastError: errors > 0 ? errorDetails[0] : undefined,
277 });
278 });
279
280 // Health check
281 app.get("/api/health", async (c) => {
282 await ensureContrailReady(db);
283 let count = 0;
284 try {
285 const row = await db
286 .prepare("SELECT COUNT(*) as cnt FROM records_summary")
287 .first<{ cnt: number }>();
288 count = row?.cnt || 0;
289 } catch {
290 // table might not exist yet
291 }
292 const hasKey = !!process.env.RESEARCHER_KEY_PEM;
293 return c.json({ ok: true, summaries: count, hasPdsKey: hasKey });
294 });
295
296 // Enrich existing summaries using Letta API
297 app.post("/api/enrich", async (c) => {
298 const apiKey = process.env.LETTA_API_KEY;
299 const agentId = process.env.LETTA_AGENT_ID;
300 if (!apiKey || !agentId) {
301 return c.json({ error: "Missing LETTA_API_KEY or LETTA_AGENT_ID" }, 500);
302 }
303
304 await ensureContrailReady(db);
305
306 // Get records that need enrichment (empty/placeholder summaries, no takeaway, or Untitled)
307 const rows = await db
308 .prepare(
309 `SELECT rkey, record FROM records_summary
310 WHERE json_extract(record, '$.summary') = ''
311 OR json_extract(record, '$.summary') LIKE 'Published in%'
312 OR json_extract(record, '$.title') = 'Untitled'
313 OR json_extract(record, '$.takeaway') IS NULL
314 OR json_extract(record, '$.takeaway') = ''
315 ORDER BY time_us DESC LIMIT 20`
316 )
317 .all<{ rkey: string; record: string }>();
318
319 if (!rows.results?.length) {
320 return c.json({ enriched: 0, message: "No records need enrichment" });
321 }
322
323 let enriched = 0;
324 let errors = 0;
325
326 for (const row of rows.results) {
327 try {
328 const record = JSON.parse(row.record) as {
329 title: string;
330 paperUrl: string;
331 abstract: string;
332 domains: string[];
333 venue: string;
334 summary: string;
335 takeaway?: string;
336 };
337
338 const result = await enrichWithLetta(
339 {
340 title: record.title,
341 paperUrl: record.paperUrl,
342 abstract: record.abstract,
343 domains: record.domains || [],
344 venue: record.venue || "",
345 },
346 apiKey,
347 agentId,
348 );
349
350 // Update the record in D1
351 record.summary = result.summary;
352 record.domains = result.domains;
353 if (result.takeaway) record.takeaway = result.takeaway;
354 if (result.title) record.title = result.title;
355
356 await db
357 .prepare("UPDATE records_summary SET record = ? WHERE rkey = ?")
358 .bind(JSON.stringify(record), row.rkey)
359 .run();
360
361 enriched++;
362 console.log(`Enriched: ${record.title?.substring(0, 50)}...`);
363 } catch (err: any) {
364 console.error(`Enrichment failed for ${row.rkey}: ${err?.message ?? err}`);
365 errors++;
366 }
367 }
368
369 return c.json({ enriched, errors, total: rows.results.length });
370 });
371
372 // Crawl endpoint: register a DID and trigger immediate ingest
373 app.post("/xrpc/com.atproto.sync.requestCrawl", async (c) => {
374 let body: { did?: string; hostname?: string };
375 try {
376 body = await c.req.json();
377 } catch {
378 body = {};
379 }
380 const did = body.did;
381 if (!did) {
382 return c.json({ error: "BadRequest", message: "Provide did" }, 400);
383 }
384
385 await ensureContrailReady(db);
386 try {
387 // Ensure identities table has this DID
388 await db.exec(
389 "CREATE TABLE IF NOT EXISTS identities (did TEXT PRIMARY KEY, handle TEXT, time_us INTEGER)"
390 );
391 await db
392 .prepare("INSERT OR IGNORE INTO identities (did, handle, time_us) VALUES (?, ?, ?)")
393 .bind(did, "researcher.pds.latha.org", Date.now() * 1000)
394 .run();
395
396 // Ensure records_summary table exists
397 await db.exec(
398 "CREATE TABLE IF NOT EXISTS records_summary (uri TEXT PRIMARY KEY, cid TEXT, did TEXT, rkey TEXT, record TEXT, time_us INTEGER)"
399 );
400
401 // Fetch records from PDS and insert directly
402 const pdsUrl = `https://pds.latha.org/xrpc/com.atproto.repo.listRecords?repo=${did}&collection=org.latha.papers.summary&limit=100`;
403 const res = await fetch(pdsUrl);
404 if (res.ok) {
405 const data = await res.json() as {
406 records: Array<{ uri: string; cid: string; value: any }>;
407 };
408 for (const rec of data.records) {
409 const rkey = rec.uri.split("/").pop() || "";
410 await db
411 .prepare(
412 "INSERT OR REPLACE INTO records_summary (uri, cid, did, rkey, record, time_us) VALUES (?, ?, ?, ?, ?, ?)"
413 )
414 .bind(rec.uri, rec.cid, did, rkey, JSON.stringify(rec.value), Date.now() * 1000)
415 .run();
416 }
417 }
418 } catch (err: any) {
419 console.error(`Crawl error: ${err?.message ?? err}`);
420 }
421
422 return c.json({ ok: true, did });
423 });
424
425 // All other routes pass through to contrail
426 app.all("*", async (c) => {
427 const response = await contrailWorker.fetch(
428 c.req.raw,
429 c.env as unknown as Record<string, unknown>
430 );
431 return response;
432 });
433
434 return app;
435}
436
437export default {
438 fetch(request: Request, env: Env): Response | Promise<Response> {
439 app ??= buildApp(env);
440 return app.fetch(request, env);
441 },
442 async scheduled(event: ScheduledEvent, env: Env, ctx: ExecutionContext) {
443 // Cron: poll feed and generate summaries
444 const db = env.DB;
445 await ensureContrailReady(db);
446
447 let lastCursor: string | undefined;
448 try {
449 const row = await db
450 .prepare("SELECT value FROM instance_settings WHERE key = 'feed_cursor' LIMIT 1")
451 .first<{ value: string }>();
452 lastCursor = row?.value || undefined;
453 } catch {
454 // first run
455 }
456
457 const { posts: feedPosts, newCursor } = await pollNewPapers(lastCursor);
458
459 // Also poll followed accounts
460 let followedPosts: PaperPost[] = [];
461 try {
462 followedPosts = await pollFollowedAccounts(5);
463 } catch {}
464
465 // Merge and deduplicate
466 const seenPaperUrls = new Set<string>();
467 const posts: PaperPost[] = [];
468 for (const p of [...feedPosts, ...followedPosts]) {
469 if (!seenPaperUrls.has(p.paperUrl)) {
470 seenPaperUrls.add(p.paperUrl);
471 posts.push(p);
472 }
473 }
474
475 // Deduplicate against existing records
476 const existingUrls = new Set<string>();
477 if (posts.length > 0) {
478 const placeholders = posts.map(() => "?").join(",");
479 const urls = posts.map(p => p.paperUrl);
480 const rows = await db
481 .prepare(
482 `SELECT DISTINCT json_extract(record, '$.paperUrl') as paperUrl FROM records_summary WHERE paperUrl IN (${placeholders})`
483 )
484 .bind(...urls)
485 .all<{ paperUrl: string }>();
486 for (const r of rows.results || []) {
487 if (r.paperUrl) existingUrls.add(r.paperUrl);
488 }
489 }
490
491 const newPosts = posts.filter(p => !existingUrls.has(p.paperUrl));
492 const apiKey = process.env.LETTA_API_KEY;
493 const agentId = process.env.LETTA_AGENT_ID;
494
495 for (const post of newPosts) {
496 try {
497 let summary = await buildSummary(post);
498
499 // Enrich with LLM if available
500 if (apiKey && agentId) {
501 try {
502 const enrichment = await enrichWithLetta(
503 { title: summary.title, paperUrl: summary.paperUrl, abstract: summary.abstract, domains: summary.domains, venue: summary.venue },
504 apiKey,
505 agentId,
506 );
507 if (enrichment.title) summary.title = enrichment.title;
508 summary.summary = enrichment.summary;
509 summary.domains = enrichment.domains;
510 if (enrichment.takeaway) (summary as any).takeaway = enrichment.takeaway;
511 } catch {
512 // Fall back to heuristic summary
513 }
514 }
515
516 const result = await writeSummaryToPds(post, summary);
517
518 // Index into D1
519 const rkey = result.uri.split("/").pop() || "";
520 await db
521 .prepare("INSERT OR REPLACE INTO records_summary (uri, cid, did, rkey, record, time_us, indexed_at) VALUES (?, ?, ?, ?, ?, ?, ?)")
522 .bind(result.uri, result.cid, "did:plc:3kkhul7jznlb6ba7rprzawnj", rkey, JSON.stringify({
523 $type: "org.latha.papers.summary",
524 ...summary,
525 sourceUri: post.postUri,
526 posterDid: post.posterDid,
527 posterHandle: post.posterHandle,
528 postText: post.postText,
529 indexedAt: new Date().toISOString(),
530 }), Date.now() * 1000, Date.now() * 1000)
531 .run();
532 } catch (err: any) {
533 console.error(`Cron: failed to write summary for ${post.paperUrl}: ${err?.message ?? err}`);
534 }
535 }
536
537 if (newCursor) {
538 await db
539 .prepare("INSERT OR REPLACE INTO instance_settings (key, value) VALUES ('feed_cursor', ?)")
540 .bind(newCursor)
541 .run();
542 }
543 },
544};