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, type PaperSummary } 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 // Build heuristic summaries in parallel
216 const heuristicResults = await Promise.all(newPosts.map(async (post) => {
217 try {
218 return { post, summary: await buildSummary(post) };
219 } catch (err: any) {
220 return { post, error: err?.message ?? err };
221 }
222 }));
223
224 // Enrich in parallel (batch of 5 to avoid rate limits)
225 const enrichedResults: Array<{ post: PaperPost; summary: PaperSummary; error?: string }> = [];
226 for (let i = 0; i < heuristicResults.length; i += 5) {
227 const batch = heuristicResults.slice(i, i + 5);
228 const batchResults = await Promise.all(batch.map(async ({ post, summary, error }) => {
229 if (error || !summary) return { post, summary: summary!, error };
230 if (!apiKey || !agentId) return { post, summary };
231 try {
232 const enrichment = await enrichWithLetta(
233 { title: summary.title, paperUrl: summary.paperUrl, abstract: summary.abstract, domains: summary.domains, venue: summary.venue },
234 apiKey,
235 agentId,
236 );
237 if (enrichment.title) summary.title = enrichment.title;
238 summary.summary = enrichment.summary;
239 summary.domains = enrichment.domains;
240 if (enrichment.takeaway) (summary as any).takeaway = enrichment.takeaway;
241 if (enrichment.year) summary.year = enrichment.year;
242 return { post, summary };
243 } catch (err: any) {
244 console.error(`Enrichment failed for ${post.paperUrl}: ${err?.message ?? err}, using heuristic`);
245 return { post, summary };
246 }
247 }));
248 enrichedResults.push(...batchResults);
249 }
250
251 // Write all to PDS and D1
252 for (const { post, summary, error } of enrichedResults) {
253 if (error || !summary) {
254 const msg = `${post.paperUrl}: ${error}`;
255 console.error(`Failed to build summary for ${msg}`);
256 if (errors < 3) errorDetails.push(msg.substring(0, 200));
257 errors++;
258 continue;
259 }
260 try {
261 const result = await writeSummaryToPds(post, summary);
262
263 const rkey = result.uri.split("/").pop() || "";
264 await db
265 .prepare("INSERT OR REPLACE INTO records_summary (uri, cid, did, rkey, record, time_us, indexed_at) VALUES (?, ?, ?, ?, ?, ?, ?)")
266 .bind(result.uri, result.cid, "did:plc:3kkhul7jznlb6ba7rprzawnj", rkey, JSON.stringify({
267 $type: "org.latha.papers.summary",
268 ...summary,
269 sourceUri: post.postUri,
270 posterDid: post.posterDid,
271 posterHandle: post.posterHandle,
272 postText: post.postText,
273 indexedAt: new Date().toISOString(),
274 }), Date.now() * 1000, Date.now() * 1000)
275 .run();
276 written++;
277 } catch (err: any) {
278 const msg = `${post.paperUrl}: ${err?.message ?? err}`;
279 console.error(`Failed to write summary for ${msg}`);
280 if (errors < 3) errorDetails.push(msg.substring(0, 200));
281 errors++;
282 }
283 }
284
285 // Save cursor
286 if (newCursor) {
287 await db
288 .prepare("INSERT OR REPLACE INTO instance_settings (key, value) VALUES ('feed_cursor', ?)")
289 .bind(newCursor)
290 .run();
291 }
292
293 return c.json({
294 polled: posts.length,
295 fromFeed: feedPosts.length,
296 fromFollows: followedPosts.length,
297 new: newPosts.length,
298 written,
299 errors,
300 lastError: errors > 0 ? errorDetails[0] : undefined,
301 });
302 });
303
304 // Health check
305 app.get("/api/health", async (c) => {
306 await ensureContrailReady(db);
307 let count = 0;
308 try {
309 const row = await db
310 .prepare("SELECT COUNT(*) as cnt FROM records_summary")
311 .first<{ cnt: number }>();
312 count = row?.cnt || 0;
313 } catch {
314 // table might not exist yet
315 }
316 const hasKey = !!process.env.RESEARCHER_KEY_PEM;
317 return c.json({ ok: true, summaries: count, hasPdsKey: hasKey });
318 });
319
320 // Enrich existing summaries using Letta API
321 app.post("/api/enrich", async (c) => {
322 const apiKey = process.env.LETTA_API_KEY;
323 const agentId = process.env.LETTA_AGENT_ID;
324 if (!apiKey || !agentId) {
325 return c.json({ error: "Missing LETTA_API_KEY or LETTA_AGENT_ID" }, 500);
326 }
327
328 await ensureContrailReady(db);
329
330 // Get records that need enrichment (empty/placeholder summaries, no takeaway, Untitled, or missing year)
331 const rows = await db
332 .prepare(
333 `SELECT rkey, record FROM records_summary
334 WHERE json_extract(record, '$.summary') = ''
335 OR json_extract(record, '$.summary') LIKE 'Published in%'
336 OR json_extract(record, '$.title') = 'Untitled'
337 OR json_extract(record, '$.takeaway') IS NULL
338 OR json_extract(record, '$.takeaway') = ''
339 OR json_extract(record, '$.year') IS NULL
340 OR json_extract(record, '$.year') = 0
341 ORDER BY time_us DESC LIMIT 20`
342 )
343 .all<{ rkey: string; record: string }>();
344
345 if (!rows.results?.length) {
346 return c.json({ enriched: 0, message: "No records need enrichment" });
347 }
348
349 let enriched = 0;
350 let errors = 0;
351
352 // Enrich in parallel batches of 5
353 for (let i = 0; i < rows.results.length; i += 5) {
354 const batch = rows.results.slice(i, i + 5);
355 const results = await Promise.allSettled(batch.map(async (row) => {
356 const record = JSON.parse(row.record) as {
357 title: string;
358 paperUrl: string;
359 abstract: string;
360 domains: string[];
361 venue: string;
362 summary: string;
363 takeaway?: string;
364 year?: number;
365 };
366
367 const result = await enrichWithLetta(
368 {
369 title: record.title,
370 paperUrl: record.paperUrl,
371 abstract: record.abstract,
372 domains: record.domains || [],
373 venue: record.venue || "",
374 },
375 apiKey,
376 agentId,
377 );
378
379 record.summary = result.summary;
380 record.domains = result.domains;
381 if (result.takeaway) record.takeaway = result.takeaway;
382 if (result.title) record.title = result.title;
383 if (result.year) record.year = result.year;
384
385 await db
386 .prepare("UPDATE records_summary SET record = ? WHERE rkey = ?")
387 .bind(JSON.stringify(record), row.rkey)
388 .run();
389
390 return record.title;
391 }));
392
393 for (const r of results) {
394 if (r.status === "fulfilled") {
395 enriched++;
396 console.log(`Enriched: ${r.value?.substring(0, 50)}...`);
397 } else {
398 console.error(`Enrichment failed: ${r.reason}`);
399 errors++;
400 }
401 }
402 }
403
404 return c.json({ enriched, errors, total: rows.results.length });
405 });
406
407 // Crawl endpoint: register a DID and trigger immediate ingest
408 app.post("/xrpc/com.atproto.sync.requestCrawl", async (c) => {
409 let body: { did?: string; hostname?: string };
410 try {
411 body = await c.req.json();
412 } catch {
413 body = {};
414 }
415 const did = body.did;
416 if (!did) {
417 return c.json({ error: "BadRequest", message: "Provide did" }, 400);
418 }
419
420 await ensureContrailReady(db);
421 try {
422 // Ensure identities table has this DID
423 await db.exec(
424 "CREATE TABLE IF NOT EXISTS identities (did TEXT PRIMARY KEY, handle TEXT, time_us INTEGER)"
425 );
426 await db
427 .prepare("INSERT OR IGNORE INTO identities (did, handle, time_us) VALUES (?, ?, ?)")
428 .bind(did, "researcher.pds.latha.org", Date.now() * 1000)
429 .run();
430
431 // Ensure records_summary table exists
432 await db.exec(
433 "CREATE TABLE IF NOT EXISTS records_summary (uri TEXT PRIMARY KEY, cid TEXT, did TEXT, rkey TEXT, record TEXT, time_us INTEGER)"
434 );
435
436 // Fetch records from PDS and insert directly
437 const pdsUrl = `https://pds.latha.org/xrpc/com.atproto.repo.listRecords?repo=${did}&collection=org.latha.papers.summary&limit=100`;
438 const res = await fetch(pdsUrl);
439 if (res.ok) {
440 const data = await res.json() as {
441 records: Array<{ uri: string; cid: string; value: any }>;
442 };
443 for (const rec of data.records) {
444 const rkey = rec.uri.split("/").pop() || "";
445 await db
446 .prepare(
447 "INSERT OR REPLACE INTO records_summary (uri, cid, did, rkey, record, time_us) VALUES (?, ?, ?, ?, ?, ?)"
448 )
449 .bind(rec.uri, rec.cid, did, rkey, JSON.stringify(rec.value), Date.now() * 1000)
450 .run();
451 }
452 }
453 } catch (err: any) {
454 console.error(`Crawl error: ${err?.message ?? err}`);
455 }
456
457 return c.json({ ok: true, did });
458 });
459
460 // All other routes pass through to contrail
461 app.all("*", async (c) => {
462 const response = await contrailWorker.fetch(
463 c.req.raw,
464 c.env as unknown as Record<string, unknown>
465 );
466 return response;
467 });
468
469 return app;
470}
471
472export default {
473 fetch(request: Request, env: Env): Response | Promise<Response> {
474 app ??= buildApp(env);
475 return app.fetch(request, env);
476 },
477 async scheduled(event: ScheduledEvent, env: Env, ctx: ExecutionContext) {
478 // Cron: poll feed and generate summaries
479 const db = env.DB;
480 await ensureContrailReady(db);
481
482 let lastCursor: string | undefined;
483 try {
484 const row = await db
485 .prepare("SELECT value FROM instance_settings WHERE key = 'feed_cursor' LIMIT 1")
486 .first<{ value: string }>();
487 lastCursor = row?.value || undefined;
488 } catch {
489 // first run
490 }
491
492 const { posts: feedPosts, newCursor } = await pollNewPapers(lastCursor);
493
494 // Also poll followed accounts
495 let followedPosts: PaperPost[] = [];
496 try {
497 followedPosts = await pollFollowedAccounts(5);
498 } catch {}
499
500 // Merge and deduplicate
501 const seenPaperUrls = new Set<string>();
502 const posts: PaperPost[] = [];
503 for (const p of [...feedPosts, ...followedPosts]) {
504 if (!seenPaperUrls.has(p.paperUrl)) {
505 seenPaperUrls.add(p.paperUrl);
506 posts.push(p);
507 }
508 }
509
510 // Deduplicate against existing records
511 const existingUrls = new Set<string>();
512 if (posts.length > 0) {
513 const placeholders = posts.map(() => "?").join(",");
514 const urls = posts.map(p => p.paperUrl);
515 const rows = await db
516 .prepare(
517 `SELECT DISTINCT json_extract(record, '$.paperUrl') as paperUrl FROM records_summary WHERE paperUrl IN (${placeholders})`
518 )
519 .bind(...urls)
520 .all<{ paperUrl: string }>();
521 for (const r of rows.results || []) {
522 if (r.paperUrl) existingUrls.add(r.paperUrl);
523 }
524 }
525
526 const newPosts = posts.filter(p => !existingUrls.has(p.paperUrl));
527 const apiKey = process.env.LETTA_API_KEY;
528 const agentId = process.env.LETTA_AGENT_ID;
529
530 // Build heuristic summaries in parallel
531 const heuristicResults = await Promise.all(newPosts.map(async (post) => {
532 try {
533 return { post, summary: await buildSummary(post) };
534 } catch (err: any) {
535 return { post, error: err?.message ?? err };
536 }
537 }));
538
539 // Enrich in parallel (batch of 5)
540 const enrichedResults: Array<{ post: PaperPost; summary: PaperSummary; error?: string }> = [];
541 for (let i = 0; i < heuristicResults.length; i += 5) {
542 const batch = heuristicResults.slice(i, i + 5);
543 const batchResults = await Promise.all(batch.map(async ({ post, summary, error }) => {
544 if (error || !summary) return { post, summary: summary!, error };
545 if (!apiKey || !agentId) return { post, summary };
546 try {
547 const enrichment = await enrichWithLetta(
548 { title: summary.title, paperUrl: summary.paperUrl, abstract: summary.abstract, domains: summary.domains, venue: summary.venue },
549 apiKey,
550 agentId,
551 );
552 if (enrichment.title) summary.title = enrichment.title;
553 summary.summary = enrichment.summary;
554 summary.domains = enrichment.domains;
555 if (enrichment.takeaway) (summary as any).takeaway = enrichment.takeaway;
556 if (enrichment.year) summary.year = enrichment.year;
557 return { post, summary };
558 } catch {
559 return { post, summary };
560 }
561 }));
562 enrichedResults.push(...batchResults);
563 }
564
565 // Write all to PDS and D1
566 for (const { post, summary, error } of enrichedResults) {
567 if (error || !summary) {
568 console.error(`Cron: failed to build summary for ${post.paperUrl}: ${error}`);
569 continue;
570 }
571 try {
572 const result = await writeSummaryToPds(post, summary);
573
574 const rkey = result.uri.split("/").pop() || "";
575 await db
576 .prepare("INSERT OR REPLACE INTO records_summary (uri, cid, did, rkey, record, time_us, indexed_at) VALUES (?, ?, ?, ?, ?, ?, ?)")
577 .bind(result.uri, result.cid, "did:plc:3kkhul7jznlb6ba7rprzawnj", rkey, JSON.stringify({
578 $type: "org.latha.papers.summary",
579 ...summary,
580 sourceUri: post.postUri,
581 posterDid: post.posterDid,
582 posterHandle: post.posterHandle,
583 postText: post.postText,
584 indexedAt: new Date().toISOString(),
585 }), Date.now() * 1000, Date.now() * 1000)
586 .run();
587 } catch (err: any) {
588 console.error(`Cron: failed to write summary for ${post.paperUrl}: ${err?.message ?? err}`);
589 }
590 }
591
592 if (newCursor) {
593 await db
594 .prepare("INSERT OR REPLACE INTO instance_settings (key, value) VALUES ('feed_cursor', ?)")
595 .bind(newCursor)
596 .run();
597 }
598 },
599};