SPB Git forge

spb/rareindex

Public
54commits 1branches 0releases
7.1 MBsize
maindefault branch
11 days agolast push
TypeScript 61.9% HTML 37.2% SQL 0.7%
1.6 KB · 24 lines typescript
Raw Blame History
1/** Live smoke: pnpm tsx connectors/firecrawl/brickeconomy/_smoke.ts [seedUrl…]  (CAPTURE=1 to save fixtures) */2import { createRouter, createCrawlContext, getConnectorMeta } from '@rareindex/connectors';3import { saveFixture } from '@rareindex/connectors/testing';4import { env } from '@rareindex/shared';5import createConnector from './index.js';67const e = env();8const meta = getConnectorMeta('brickeconomy');9const connector = createConnector(meta);10const router = createRouter({ firecrawlApiKey: e.FIRECRAWL_API_KEY, scrapflyApiKey: e.SCRAPFLY_API_KEY });11const seeds = process.argv.slice(2);12const ctx = createCrawlContext({ router, meta: { ...meta, config: { ...meta.config, setsPerSeed: 3 } }, options: { mode: 'probe', limit: Number(process.env.LIMIT ?? 3), seeds: seeds.length ? seeds : undefined } });1314for await (const raw of connector.crawl(ctx)) {15  const recs = await connector.normalize({ ...raw, externalId: raw.externalId ?? null, fetchedAt: raw.fetchedAt ?? new Date() });16  console.log(`\n# ${raw.url} — ${recs.length} records`);17  for (const r of recs) console.log(JSON.stringify(r).slice(0, 700));18  if (process.env.CAPTURE) {19    const slug = raw.url.split('/').slice(-2).join('__');20    saveFixture('brickeconomy', slug, { raw: { url: raw.url, externalId: raw.externalId ?? null, kind: raw.kind, engine: raw.engine, fetchedAt: raw.fetchedAt ?? new Date(), payload: raw.payload }, expect: { minCount: 2, kinds: ['catalog_item', 'price_observation'] }, note: `captured live from ${raw.url}` });21  }22}23console.log('\nengine stats', ctx.engineStats, 'anomalies', ctx.anomalies);24