SPB Git

spb/rareindex

Public
10.3 KB · 218 lines typescript
1#!/usr/bin/env tsx2/**3 * RareIndex operator CLI.4 *   ri connectors | ri crawl <id> [--mode probe|incremental|backfill] [--limit N]5 *   ri backfill <id…> [--start-only] [--reset] [--pause] [--status] [--start YYYY-MM-DD] [--end YYYY-MM-DD]6 *   ri certs [--limit N] [--grader psa] | ri certs --backfill   (verify cert numbers via cert_lookup connectors / register certs from existing sales+listings)7 *   ri normalize [--connector id] [--limit N] | ri resolve [--limit N]8 *   ri value [--asset id | --all] [--history] | ri premiums [--category slug] | ri regrade [--kind sale|listing|both] [--dry-run]9 *   ri index | ri snapshots | ri radar | ri health [--probe] | ri fx [--backfill] | ri benchmarks10 *   ri expire | ri stats | ri run-all [--limit N] | ri worker11 */12import { loadDotenv } from './lib/env.ts';13loadDotenv();14process.env.RI_SERVICE ??= 'ri-cli';1516import { sql } from 'drizzle-orm';17import { listConnectorMeta } from '@rareindex/connectors';18import { logger } from '@rareindex/shared';19import { closeDb, db } from './lib/db.ts';20import { flushCosts } from './lib/costs.ts';21import { activeBackfill, pauseBackfill, runCrawl, startBackfill } from './crawler/run.ts';22import { normalizeBatch } from './normalizer/index.ts';23import { pendingCount, resolveBatch } from './entity-resolution/index.ts';24import { regradeRecords } from './entity-resolution/regrade.ts';25import { assetsNeedingValuation, computePremiums, valueAsset, valueMany } from './valuation/run.ts';26import { runCategorySnapshots, runIndices, runRadar } from './indices/run.ts';27import { syncFx } from './fx.ts';28import { syncBenchmarks } from './benchmarks.ts';29import { computeHealth } from './health.ts';30import { expireListings } from './listings-expire.ts';31import { imageStats, processImages } from './image-processing/index.ts';32import { backfillCertificates, verifyCertificates } from './certs-verify.ts';3334const [cmd = 'help', ...rest] = process.argv.slice(2);35const flags: Record<string, string | boolean> = {};36const positional: string[] = [];37for (let i = 0; i < rest.length; i++) {38  const a = rest[i]!;39  if (a.startsWith('--')) {40    const k = a.slice(2);41    const next = rest[i + 1];42    if (next && !next.startsWith('--')) {43      flags[k] = next;44      i++;45    } else flags[k] = true;46  } else positional.push(a);47}48const num = (k: string, d?: number) => (flags[k] !== undefined ? Number(flags[k]) : d);49const str = (k: string) => (typeof flags[k] === 'string' ? (flags[k] as string) : undefined);50const print = (x: unknown) => console.log(typeof x === 'string' ? x : JSON.stringify(x, null, 2));5152async function drain(fn: () => Promise<{ processed: number }>, batch = 500): Promise<number> {53  let total = 0;54  for (let i = 0; i < 10_000; i++) {55    const r = await fn();56    total += r.processed;57    if (r.processed < batch) break;58  }59  return total;60}6162async function stats() {63  const tables = ['sources', 'connectors', 'connector_runs', 'raw_records', 'normalized_records', 'assets', 'asset_variants', 'sets', 'brands', 'sales', 'listings', 'price_observations', 'auction_lots', 'population_reports', 'valuations', 'asset_stats', 'price_snapshots', 'index_values', 'category_snapshots', 'radar_findings', 'fx_rates', 'benchmarks', 'costs', 'events', 'audit_log'];64  const out: Record<string, number> = {};65  for (const t of tables) {66    const r = (await db().execute(sql.raw(`select count(*)::int as n from ${t}`))) as unknown as Array<{ n: number }>;67    out[t] = r[0]?.n ?? 0;68  }69  const priced = (await db().execute(sql`select count(*)::int as n from asset_stats where riv_usd is not null`)) as unknown as Array<{ n: number }>;70  out['assets_priced'] = priced[0]?.n ?? 0;71  const idx = (await db().execute(sql`select i.ticker, v.date, v.value, v.constituents_count from index_values v join indices i on i.id = v.index_id where v.date = (select max(date) from index_values v2 where v2.index_id = v.index_id) order by i.ticker`)) as unknown as Array<Record<string, unknown>>;72  return { counts: out, indices: idx };73}7475async function main() {76  switch (cmd) {77    case 'connectors': {78      const metas = listConnectorMeta();79      print(metas.map((m) => ({ id: m.id, source: m.sourceName, type: m.sourceType, engines: m.enginePriority.join('>'), categories: m.categories.join(','), enabled: m.enabled, refreshMin: m.refreshFrequencyMinutes })));80      break;81    }82    case 'crawl': {83      const id = positional[0];84      if (!id) throw new Error('usage: ri crawl <connectorId> [--mode ...] [--limit N]');85      print(await runCrawl(id, { mode: (str('mode') as 'probe' | 'incremental' | 'backfill') ?? 'incremental', limit: num('limit'), trigger: 'manual' }));86      break;87    }88    case 'backfill': {89      // ri backfill <id> [--reset] [--pause] [--status] [--start YYYY-MM-DD] [--end YYYY-MM-DD] [--limit N]90      const id = positional[0];91      if (!id) throw new Error('usage: ri backfill <connectorId> [--reset|--pause|--status]');92      if (flags.status) {93        print(await activeBackfill(id));94        break;95      }96      if (flags.pause) {97        print({ paused: await pauseBackfill(id) });98        break;99      }100      const ids = positional.length > 1 ? positional : [id];101      for (const cid of ids) {102        const campaign = await startBackfill(cid, { reset: Boolean(flags.reset), startDate: str('start') ?? null, endDate: str('end') ?? null });103        print({ connector: cid, campaign: campaign.id, status: campaign.status, resumedFrom: campaign.lastCursor });104        // --start-only: register the campaign and let the worker scheduler run the slices (recommended for many sources)105        if (!flags['start-only']) {106          print(await runCrawl(cid, { mode: 'backfill', limit: num('limit'), trigger: 'manual' }));107          print(await activeBackfill(cid));108        }109      }110      break;111    }112    case 'certs':113      if (flags.backfill) print(await backfillCertificates());114      else print(await verifyCertificates({ limit: num('limit', 100), grader: str('grader') }));115      break;116    case 'normalize':117      print({ processed: await drain(() => normalizeBatch({ connectorId: str('connector'), limit: num('limit', 500) }), num('limit', 500)) });118      break;119    case 'resolve': {120      const processed = await drain(() => resolveBatch({ limit: num('limit', 500), connectorId: str('connector') }), num('limit', 500));121      print({ processed, stillPending: await pendingCount() });122      break;123    }124    case 'premiums':125      print({ written: await computePremiums(str('category')) });126      break;127    case 'regrade': {128      // ri regrade [--kind sale|listing|both] [--connector id] [--limit N] [--dry-run]129      const kind = (str('kind') ?? 'both') as 'sale' | 'listing' | 'both';130      print(await regradeRecords({ kind, connectorId: str('connector'), limit: num('limit', 50_000), dryRun: Boolean(flags['dry-run']), recheck: Boolean(flags.recheck) }));131      break;132    }133    case 'value': {134      const asset = str('asset');135      if (asset) print(await valueAsset(asset, { rebuildHistory: Boolean(flags.history) }));136      else {137        await computePremiums();138        const ids = await assetsNeedingValuation({ all: Boolean(flags.all), limit: num('limit', 200_000) });139        print(await valueMany(ids, { rebuildHistory: Boolean(flags.history) }));140      }141      break;142    }143    case 'index':144      print(await runIndices());145      break;146    case 'snapshots':147      print({ categories: await runCategorySnapshots() });148      break;149    case 'radar':150      print({ findings: await runRadar() });151      break;152    case 'health':153      print((await computeHealth({ connectorId: positional[0], probe: Boolean(flags.probe) })).map((h) => ({ connector: h.connector, status: h.status, success24h: h.success_rate_24h, records24h: h.records_24h, lastError: h.last_error, anomalies: h.anomalies })));154      break;155    case 'fx':156      print(await syncFx({ backfill: Boolean(flags.backfill) }));157      break;158    case 'benchmarks':159      print(await syncBenchmarks());160      break;161    case 'expire':162      print({ expired: await expireListings() });163      break;164    case 'stats':165      print(await stats());166      break;167    case 'images': {168      const r = await processImages({ limit: num('limit', 2000), recheck: Boolean(flags.recheck) });169      print({ ...r, totals: await imageStats() });170      break;171    }172    case 'run-all': {173      const limit = num('limit', 200);174      const [fx] = (await db().execute(sql`select count(*)::int as n from fx_rates`)) as unknown as Array<{ n: number }>;175      if (!fx || fx.n === 0) print({ fx: await syncFx({ backfill: true }) });176      const summary: Record<string, unknown> = {};177      for (const m of listConnectorMeta({ enabled: true })) {178        try {179          const r = await runCrawl(m.id, { mode: 'incremental', limit, trigger: 'manual' });180          summary[m.id] = { status: r.status, raw: r.recordsRaw, dupes: r.recordsDuplicate, error: r.error };181        } catch (err) {182          summary[m.id] = { status: 'failed', error: err instanceof Error ? err.message : String(err) };183        }184      }185      summary.normalized = await drain(() => normalizeBatch({ limit: 500 }));186      summary.resolved = await drain(() => resolveBatch({ limit: 500 }));187      summary.premiums = await computePremiums();188      const ids = await assetsNeedingValuation({ limit: 200_000 });189      summary.valuation = await valueMany(ids, { rebuildHistory: Boolean(flags.history) });190      summary.indices = await runIndices();191      summary.snapshots = await runCategorySnapshots();192      summary.radar = await runRadar();193      summary.health = (await computeHealth()).map((h) => `${h.connector}:${h.status}`);194      print(summary);195      break;196    }197    case 'worker': {198      const { startWorker } = await import('./main.ts');199      const stop = await startWorker();200      const shutdown = () => void stop().then(() => process.exit(0));201      process.on('SIGINT', shutdown);202      process.on('SIGTERM', shutdown);203      return; // keep running204    }205    default:206      print(`RareIndex CLI\n  ri connectors | crawl <id> [--mode probe|incremental|backfill] [--limit N] | normalize [--connector id] | resolve\n  ri value [--asset id|--all] [--history] | premiums | regrade [--dry-run] | index | snapshots | radar | health [--probe] | fx [--backfill] | benchmarks | expire\n  ri stats | run-all [--limit N] [--history] | worker`);207  }208  await flushCosts();209  await closeDb();210}211212main().catch(async (err) => {213  logger.error({ err }, 'ri failed');214  console.error(err);215  await closeDb();216  process.exit(1);217});218
Files · b5d6968a47f5d2546937866bdabc967739349047
Menu

Theme