TypeScript 61.9%
HTML 37.2%
SQL 0.7%
1import { loadDotenv } from './lib/env.ts';2loadDotenv();3process.env.RI_SERVICE ??= 'rareindex-worker';45import { env, logger } from '@rareindex/shared';6import { createQueue, JOBS } from './lib/queue.ts';7import { closeDb, db } from './lib/db.ts';8import { flushCosts } from './lib/costs.ts';9import { runCrawl } from './crawler/run.ts';10import { scheduleDueCrawls } from './crawler/scheduler.ts';11import { normalizeBatch } from './normalizer/index.ts';12import { sql } from 'drizzle-orm';13import { pendingCount, resolveBatch } from './entity-resolution/index.ts';14import { assetsNeedingValuation, computePremiums, valueMany } from './valuation/run.ts';15import { runCategorySnapshots, runIndices, runRadar } from './indices/run.ts';16import { syncFx } from './fx.ts';17import { syncBenchmarks } from './benchmarks.ts';18import { computeHealth } from './health.ts';19import { expireListings } from './listings-expire.ts';20import { processImages } from './image-processing/index.ts';21import { verifyCertificates } from './certs-verify.ts';22import { assessLots } from './auctions/assess.ts';2324/**25 * RareIndex worker process (PM2: rareindex-worker). One process runs the queue handlers and the26 * scheduler; scale out by starting more processes against the same database.27 */28export async function startWorker(): Promise<() => Promise<void>> {29 const log = logger.child({ component: 'worker' });30 const queue = createQueue();31 await queue.start();32 // Single-worker deployment: jobs left 'active' by a previous process (restart/crash) would wait for33 // their expiry (hours). Re-queue them so crawls resume immediately.34 await db().execute(sql`update pgboss.job set state = 'retry', started_on = null where state = 'active' and started_on < now() - interval '90 seconds'`);35 // Runs left 'running' by a dead process can never finish: close them so the per-connector guard below works.36 await db().execute(sql`update connector_runs set status = 'failed', finished_at = now(), error = 'worker restarted' where status = 'running'`);37 // Collapse duplicate queued crawls per connector (keep the oldest).38 await db().execute(sql`delete from pgboss.job j using pgboss.job k where j.name = 'crawl.run' and k.name = 'crawl.run' and j.state in ('created','retry') and k.state in ('created','retry') and j.data->>'connectorId' = k.data->>'connectorId' and j.created_on > k.created_on`);39 const concurrency = env().WORKER_CONCURRENCY;4041 // ---- handlers ----42 await queue.work<{ connectorId: string; mode?: 'incremental' | 'backfill' | 'probe'; limit?: number; trigger?: string }>(JOBS.crawlRun, { concurrency: Math.max(1, Math.floor(concurrency / 2)), pollingIntervalSeconds: 5 }, async (data) => {43 // One crawl per connector at a time (duplicates waste engine credits and crawl slots).44 const [running] = (await db().execute(sql`select 1 from connector_runs where connector_id = ${data.connectorId} and status = 'running' and started_at > now() - interval '6 hours' limit 1`)) as unknown as unknown[];45 if (running) {46 log.info({ connector: data.connectorId }, 'crawl already running; skipping duplicate job');47 return;48 }49 const res = await runCrawl(data.connectorId, { mode: data.mode ?? 'incremental', limit: data.limit, trigger: data.trigger ?? 'schedule' });50 if (res.recordsRaw > 0) await queue.send(JOBS.normalizeBatch, { connectorId: data.connectorId }, { singletonKey: `normalize:${data.connectorId}`, singletonSeconds: 30 });51 });52 await queue.work<{ connectorId?: string }>(JOBS.normalizeBatch, { concurrency: 2, pollingIntervalSeconds: 5 }, async (data) => {53 let total = 0;54 let more = false;55 for (let i = 0; i < 40; i++) {56 const r = await normalizeBatch({ connectorId: data.connectorId, limit: 500 });57 total += r.processed;58 more = r.processed >= 500;59 if (!more) break;60 }61 // Large crawls (100k+ raw rows) exceed one job's budget: chain another job instead of leaving rows unprocessed.62 if (more) await queue.send(JOBS.normalizeBatch, { connectorId: data.connectorId }, { singletonKey: `normalize:${data.connectorId ?? 'all'}:${Date.now()}`, startAfterSeconds: 1 });63 if (total > 0) for (let k = 0; k < Number(process.env.RESOLVE_CONCURRENCY ?? 3); k++) await queue.send(JOBS.resolveBatch, {}, { singletonKey: `resolve:${k}`, singletonSeconds: 30 });64 });65 const RESOLVE_WORKERS = Number(process.env.RESOLVE_CONCURRENCY ?? 3);66 // Release rows claimed by a previous worker process that died mid-batch.67 await db().execute(sql`update normalized_records set status = 'pending' where status = 'processing'`);68 await queue.work<Record<string, never>>(JOBS.resolveBatch, { concurrency: RESOLVE_WORKERS, pollingIntervalSeconds: 5 }, async () => {69 const touched = new Set<string>();70 for (let i = 0; i < 40; i++) {71 const r = await resolveBatch({ limit: 500 });72 for (const a of r.touchedAssets) touched.add(a);73 if (r.processed < 500) break;74 }75 if ((await pendingCount()) > 0) for (let k = 0; k < RESOLVE_WORKERS; k++) await queue.send(JOBS.resolveBatch, {}, { singletonKey: `resolve:${k}`, singletonSeconds: 30, startAfterSeconds: 5 });76 if (touched.size) await queue.send(JOBS.valuationAsset, { assetIds: [...touched].slice(0, 5000) }, { singletonKey: `value:${Date.now()}` });77 if (touched.size) await queue.send(JOBS.imagesProcess, { assetIds: [...touched].slice(0, 5000), limit: 5000 }, { singletonKey: `images:${Date.now()}` });78 });79 await queue.work<{ assetIds: string[] }>(JOBS.valuationAsset, { concurrency: 1, pollingIntervalSeconds: 5 }, async (data) => {80 await valueMany(data.assetIds, { concurrency: 4 });81 });82 await queue.work<{ all?: boolean; rebuildHistory?: boolean }>(JOBS.valuationRebuild, { concurrency: 1, pollingIntervalSeconds: 30 }, async (data) => {83 await computePremiums();84 const ids = await assetsNeedingValuation({ all: data.all, limit: 200_000 });85 await valueMany(ids, { rebuildHistory: data.rebuildHistory, concurrency: 4 });86 });87 await queue.work(JOBS.indicesDaily, { concurrency: 1, pollingIntervalSeconds: 30 }, async () => {88 await runIndices();89 });90 await queue.work(JOBS.snapshotsDaily, { concurrency: 1, pollingIntervalSeconds: 30 }, async () => {91 await runCategorySnapshots();92 });93 await queue.work(JOBS.radarScan, { concurrency: 1, pollingIntervalSeconds: 30 }, async () => {94 await runRadar();95 });96 await queue.work(JOBS.healthCompute, { concurrency: 1, pollingIntervalSeconds: 30 }, async () => {97 await computeHealth();98 });99 await queue.work<{ backfill?: boolean }>(JOBS.fxSync, { concurrency: 1, pollingIntervalSeconds: 30 }, async (data) => {100 await syncFx({ backfill: data.backfill });101 });102 await queue.work(JOBS.benchmarksSync, { concurrency: 1, pollingIntervalSeconds: 60 }, async () => {103 await syncBenchmarks();104 });105 await queue.work<{ assetIds?: string[]; limit?: number; recheck?: boolean }>(JOBS.imagesProcess, { concurrency: 1, pollingIntervalSeconds: 30 }, async (data) => {106 await processImages({ assetIds: data.assetIds, limit: data.limit ?? 2000, recheck: data.recheck ?? false });107 });108 await queue.work(JOBS.listingsExpire, { concurrency: 1, pollingIntervalSeconds: 60 }, async () => {109 await expireListings();110 });111 await queue.work<{ limit?: number; grader?: string }>(JOBS.certsVerify, { concurrency: 1, pollingIntervalSeconds: 60 }, async (data) => {112 await verifyCertificates({ limit: data.limit ?? 100, grader: data.grader });113 });114 await queue.work<{ limit?: number; all?: boolean }>(JOBS.auctionsAssess, { concurrency: 1, pollingIntervalSeconds: 60 }, async (data) => {115 await assessLots({ limit: data.limit ?? 20_000, all: data.all ?? false });116 });117 await queue.work(JOBS.accountJobs, { concurrency: 1, pollingIntervalSeconds: 60 }, async () => {118 try {119 const accountModule = './account/index.ts'; // provided by the account module when present120 const mod = (await import(accountModule)) as { runAccountJobs?: (d: ReturnType<typeof db>) => Promise<unknown> };121 if (mod.runAccountJobs) await mod.runAccountJobs(db());122 } catch (err) {123 if (!/Cannot find module|ERR_MODULE_NOT_FOUND/.test(String(err))) log.error({ err }, 'account jobs failed');124 }125 });126127 // ---- schedules (UTC) ----128 await queue.schedule(JOBS.fxSync, '15 16 * * 1-5', {}); // after ECB 16:00 CET publication129 await queue.schedule(JOBS.benchmarksSync, '30 22 * * *', {});130 await queue.schedule(JOBS.valuationRebuild, '0 3 * * *', { all: false });131 await queue.schedule(JOBS.snapshotsDaily, '20 4 * * *', {});132 await queue.schedule(JOBS.indicesDaily, '40 4 * * *', {});133 await queue.schedule(JOBS.radarScan, '0 5 * * *', {});134 await queue.schedule(JOBS.healthCompute, '*/30 * * * *', {});135 await queue.schedule(JOBS.listingsExpire, '10 * * * *', {});136 await queue.schedule(JOBS.certsVerify, '25 * * * *', { limit: 100 }); // SPEC §22: verify cert numbers seen on marketplaces against public grader lookups137 await queue.schedule(JOBS.imagesProcess, '50 5 * * *', { limit: 20000, recheck: true });138 await queue.schedule(JOBS.imagesProcess, '*/15 * * * *', { limit: 3000 });139 await queue.schedule(JOBS.accountJobs, '*/10 * * * *', {});140 await queue.schedule(JOBS.auctionsAssess, '*/20 * * * *', { limit: 20000 }); // §33–§35: all-in bid vs RIV on live lots141142 // ---- crawl scheduler loop ----143 const tick = async () => {144 try {145 await scheduleDueCrawls(queue);146 // Normalize whatever raw rows exist even while long crawls are still running (sales show up progressively).147 await queue.send(JOBS.normalizeBatch, {}, { singletonKey: 'normalize:periodic', singletonSeconds: 120 });148 } catch (err) {149 log.error({ err }, 'scheduler tick failed');150 }151 };152 await tick();153 if ((await pendingCount()) > 0) for (let k = 0; k < RESOLVE_WORKERS; k++) await queue.send(JOBS.resolveBatch, {}, { singletonKey: `resolve:${k}`, singletonSeconds: 30 });154 const timer = setInterval(tick, 60_000);155 // hourly incremental valuation for assets with fresh evidence156 const hourly = setInterval(() => void queue.send(JOBS.valuationRebuild, { all: false }, { singletonKey: 'valuation:hourly', singletonSeconds: 3000 }), 3600_000);157 log.info({ concurrency }, 'worker started');158159 return async () => {160 clearInterval(timer);161 clearInterval(hourly);162 await queue.stop();163 await flushCosts();164 await closeDb();165 log.info('worker stopped');166 };167}168169const isMain = process.argv[1] && /workers\/main\.ts$/.test(process.argv[1]);170if (isMain) {171 startWorker()172 .then((stop) => {173 const shutdown = () => void stop().then(() => process.exit(0));174 process.on('SIGINT', shutdown);175 process.on('SIGTERM', shutdown);176 })177 .catch((err) => {178 logger.error({ err }, 'worker failed to start');179 process.exit(1);180 });181}182