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