[\s\S]*?<\/div>\s*<\/div>\s*<\/div>\s*<\/div>/g) ?? []; const listHtml = `

${m.slice(0, 4).join('\n')}

`; saveFixture(id, 'auctions-list', { raw: { url: 'https://whisky.auction/auctions', externalId: 'auctions-list', kind: 'sale', engine: 'api', fetchedAt: list.fetchedAt, payload: { kind: 'lot_page', url: 'https://whisky.auction/auctions', auctionId: entries[0]!.id, auctionTitle: null, seed: { category: '', slug: 'whisky' }, pageIndex: 0, pageSize: 90, total: 0, lots: [], snapshot: listHtml } }, expect: { count: 0 }, note: `Live capture ${today}: /auctions past-auction list trimmed to 4 entries (parser test only; no lots).` }); const seeds = meta.config.seeds as Array<{ category: string; slug: 'whisky' | 'rum' | 'cognac' | 'wine' }>; const auctionId = '201'; for (const [name, seed, pick] of [ ['auction-201-whisky-p0', seeds[0]!, (lots: unknown[]) => lots.slice(0, 6)], ['auction-201-rum-p0', seeds[1]!, (lots: unknown[]) => [...lots.slice(0, 4), ...lots.slice(-3)]], ] as const) { const url = pageUrl(auctionId, seed, 0, 90); const res = await ctx.fetch(url, { engines: ['api'], responseType: 'text', minQuality: 0 }); const parsed = parseLotPage(res.html!); const payload = { kind: 'lot_page', url, auctionId, auctionTitle: parsed.auctionTitle, seed, pageIndex: 0, pageSize: 90, total: parsed.total, lots: pick(parsed.lots), snapshot: trimLotHtml(res.html!, 3) }; saveFixture(id, name, { raw: { url, externalId: `auction:${auctionId}:${seed.slug}:p0`, kind: 'sale', engine: 'api', fetchedAt: res.fetchedAt, payload }, expect: { minCount: 1, kinds: ['sale'], requiredFields: ['price', 'currency', 'saleDate', 'attributes.identifiers.whisky_auction_lot'] }, note: `Live capture ${today} of whisky.auction auction 201 (May 2026), ${seed.slug} filter, first grid page sorted by price desc; lots trimmed (${name.includes('rum') ? 'first 4 + last 3, the tail includes unsold NotMet lots' : 'first 6'}), HTML snapshot trimmed to 3 lot cards.` }); } continue; } if (id === 'chairish' || id === 'pamono' || id === 'peyton-street-pens' || id === '1stdibs') { const seeds = meta.config.seeds as Array<{ path: string }>; const picks: Array<[string, string, number]> = id === 'chairish' ? [['vintage-furniture-p2', '/collection/vintage-furniture', 2], ['decor-p1', '/collection/decor', 1]] : id === 'pamono' ? [['furniture-p1', '/furniture', 1], ['jewelry-watches-p1', '/jewelry-watches', 1]] : id === 'peyton-street-pens' ? [['parker-p1', '/pens-by-brand/parker/', 1], ['montblanc-p1', '/pen-makers-europe/montblanc/', 1]] : [['seating-chairs-p1', '/furniture/seating/chairs/', 1], ['wrist-watches-p1', '/jewelry/watches/wrist-watches/', 1]]; void seeds; const pageUrl = mod.pageUrl as (p: string, n: number) => string; const ctx = ctxFor(meta); for (const [name, seedPath, page] of picks) { const url = pageUrl(seedPath, page); const res = await ctx.fetch(url, { engines: ['api'], responseType: 'text', minQuality: 0, timeoutMs: 90_000 }); if (!res.success || !res.html) throw new Error(`${url}: ${res.error ?? res.httpStatus}`); const seedCfg = (meta.config.seeds as Array>).find((s) => s.path === seedPath) ?? { path: seedPath }; let payload: Record; let snapshot: string | undefined; if (id === 'chairish') { const items = (mod.parseBrowseHtml as (h: string) => unknown[])(res.html).slice(0, 6); snapshot = (mod.trimBrowseHtml as (h: string, n: number) => string)(res.html, 3); payload = { kind: 'listing_page', url, seed: seedCfg, page, items, snapshot }; } else if (id === 'pamono') { const cards = (mod.parseCategoryHtml as (h: string) => unknown[])(res.html).slice(0, 6); snapshot = (mod.trimCategoryHtml as (h: string, n: number) => string)(res.html, 3); payload = { kind: 'listing_page', url, seed: seedCfg, page, currency: (mod.parseStoreCurrency as (h: string) => string | null)(res.html), cards, snapshot }; } else if (id === 'peyton-street-pens') { const parsed = (mod.parseCategoryHtml as (h: string) => { cards: unknown[]; totalPages: number | null })(res.html); snapshot = (mod.trimCategoryHtml as (h: string, n: number) => string)(res.html, 3); payload = { kind: 'listing_page', url, seed: seedCfg, page, totalPages: parsed.totalPages, cards: parsed.cards.slice(0, 6), snapshot }; } else { const parsed = (mod.parseBrowseHtml as (h: string) => { items: unknown[]; totalResults: number | null; maxPages: number | null })(res.html)!; snapshot = (mod.trimStoreHtml as (h: string, n: number) => string)(res.html, 3); payload = { kind: 'listing_page', url, seed: seedCfg, page, totalResults: parsed.totalResults, maxPages: parsed.maxPages, items: parsed.items.slice(0, 6) }; // the Relay snapshot is stored as a separate .html next to the fixture const fx = fixture({ url, externalId: `${seedPath}:p${page}`, kind: 'listing', engine: 'api', payload: { ...payload, snapshot }, fetchedAt: res.fetchedAt }, { minCount: 1, kinds: ['listing'], requiredFields: ['price', 'currency', 'attributes.identifiers.firstdibs_item_id'] }, `Live capture ${today} of ${url}; items trimmed to 6, Relay store snapshot reduced to the records reachable from the first 3 items.`); saveFixture(id, name, fx); continue; } const req = id === 'chairish' ? 'attributes.identifiers.chairish_product_id' : id === 'pamono' ? 'attributes.identifiers.pamono_sku' : 'seller'; saveFixture(id, name, fixture({ url, externalId: `${seedPath}:p${page}`, kind: 'listing', engine: 'api', payload, fetchedAt: res.fetchedAt }, { minCount: 1, kinds: ['listing'], requiredFields: ['price', 'currency', req] }, `Live capture ${today} of ${url}; items trimmed to 6, HTML snapshot trimmed to 3 cards.`)); } continue; } if (id === 'chatterley-luxuries') { for (const [name, handle, wantSku] of [['pens-p1', 'pens', true], ['lighters-p1', 'lighters', false]] as const) { const m2 = { ...meta, config: { ...meta.config, collections: (meta.config.collections as Array<{ handle: string }>).filter((c) => c.handle === handle) } }; const c2 = (mod.default as (m: ConnectorMeta) => RareIndexConnector)(m2); let chosen: RawRecordInput | null = null; for await (const raw of c2.crawl(ctxFor(m2, undefined, 12))) { const sku = ((raw.payload as { product: { sku?: string | null; prices?: { price?: string } } }).product.sku ?? '').trim(); const priced = Number((raw.payload as { product: { prices?: { price?: string } } }).product.prices?.price ?? 0) > 0; if (priced && (!wantSku || sku)) { chosen = raw; break; } } if (!chosen) throw new Error(`no product captured for ${handle}`); const p = chosen.payload as { product: Record }; p.product = { ...p.product, description: String(p.product.description ?? '').slice(0, 1500), images: (p.product.images as unknown[]).slice(0, 2) }; saveFixture(id, name, fixture(chosen, { count: 1, kinds: ['listing'], requiredFields: ['price', 'currency'] }, `Live capture ${today}: first priced product of the '${handle}' category via the WooCommerce Store API (description/images trimmed).`)); } continue; } // gated official APIs: capture only when credentials exist const raw = await firstRaw(connector, meta); if (!raw) { console.log(` [${id}] nothing captured (gated / no credentials)`); continue; } saveFixture(id, `live-${today}`, fixture(raw, { minCount: 1 }, `Live capture ${today}.`)); } catch (err) { console.error(` [${id}] capture failed:`, err instanceof Error ? err.message : err); } }