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