import { createRouter, createCrawlContext, ConnectorMetaSchema } from '@rareindex/connectors'; import { saveFixture } from '@rareindex/connectors/testing'; import metaJson from './meta.json' with { type: 'json' }; import createConnector from './index.js'; const save = process.argv.includes('--save'); const meta = ConnectorMetaSchema.parse(metaJson); const connector = createConnector(meta); const router = createRouter({ firecrawlApiKey: process.env.FIRECRAWL_API_KEY, scrapflyApiKey: process.env.SCRAPFLY_API_KEY }); const ctx = createCrawlContext({ router, meta, options: { mode: 'probe', limit: 2 } }); let i = 0; for await (const raw of connector.crawl(ctx)) { const out = await connector.normalize({ ...raw, externalId: raw.externalId ?? null, fetchedAt: raw.fetchedAt ?? new Date() }); const sales = out.filter((r) => r.kind === 'sale'); console.log(`raw ${raw.externalId} (${raw.engine}) → ${out.length} records, ${sales.length} sales`); for (const r of sales.slice(0, 3)) console.log(' ', JSON.stringify(r).slice(0, 420)); if (save && i < 2 && sales.length) saveFixture(meta.id, `sale-${raw.externalId}`, { raw: { url: raw.url, externalId: raw.externalId ?? null, kind: raw.kind, engine: raw.engine, fetchedAt: raw.fetchedAt ?? new Date(), payload: raw.payload }, expect: { minCount: 10, kinds: ['sale', 'auction_lot'], requiredFields: ['attributes.brand', 'rawTitle'] }, note: 'Captured live via Firecrawl from phillips.com' }); i++; } console.log('engineStats', ctx.engineStats, 'anomalies', ctx.anomalies);