Coupang Products Crawler

Coupang public products crawler

This Actor extracts publicly visible Coupang product information over bounded direct HTTPS requests. It captures public product identifiers, names, product URLs, prices, original prices, currency, brand, category, seller, availability, delivery text, descriptions, ratings, review counts, SKUs, image URLs, source provenance, and scrape timestamps when those values are exposed by public JSON-LD or page state.

What it is useful for

Use it for small, respectful product snapshots, Korean e-commerce price monitoring, assortment research, promotion checks, and public competitor analysis. Provide official Coupang product or search URLs, or a query such as laptop or wireless keyboard; when no URL is supplied, the Actor creates a public Coupang search URL.

The crawler is sequential and bounded by page and product limits, timeouts, retries, response size, and an optional request delay. It does not access customer accounts, private order data, or purchasing flows.

Access boundary

Requests use direct HTTPS only. The Actor does not use a browser, stealth automation, proxy, login, CAPTCHA solving, or access-control bypass. Redirects must remain on a public Coupang host. Access-denied, bot-check, CAPTCHA, security-verification, or similar responses become minimal diagnostics rather than product rows.

Input

Use query or searchQuery for a public search phrase. Use startUrls, sourceUrls, urls, or productUrls for official public Coupang pages. maxPages defaults to 5 and is bounded from 1 to 25; maxItems defaults to 50 and is bounded from 1 to 500. requestTimeoutSecs, requestDelayMillis, and maxRetries provide bounded network controls. includeDiagnostics defaults to true.

Example:

{
  "query": "wireless headphones",
  "maxPages": 2,
  "maxItems": 25,
  "requestDelayMillis": 500
}

For deterministic local QA, pass fixture or a repository-relative fixturePath. Fixture files are only used when explicitly supplied and must stay inside the Actor directory and outside node_modules.

Dataset output

Normal rows contain business and public-source fields such as productId, name, url, price, originalPrice, currency, brand, category, seller, availability, delivery, description, imageUrls, rating, reviewCount, sku, sourceUrl, sourceDomain, and scrapedAt. Missing public fields are omitted rather than inferred.

Normal rows do not contain internal IDs, dataAvailable, HTTP status, retries, elapsed time, actor identity, or debug fields. When enabled, a diagnostic row has exactly recordType: "diagnostic", diagnosticType, url, error, errorCode, and scrapedAt. Operational counters and completion status are stored in OUTPUT_SUMMARY, including sourcesRequested, sourcesProcessed, and sourcesSkipped when the bounded maxPages or maxItems limit leaves supplied URLs unrequested.

You can download the dataset in various formats such as JSON, HTML, CSV, or Excel.

Local development

Run the contract suite with node --test test/contract.regression.test.js, syntax checks with npm run check when available, and dataset-boundary checks with node validate-output.js after setting APIFY_LOCAL_STORAGE_DIR. The Actor targets Node.js 20 or newer.

Responsible use and cost

Collect only public information you are permitted to use. Follow Coupang’s terms, applicable marketplace rules, privacy requirements, and reasonable request-rate limits. Do not use this Actor to evade access controls or collect sensitive personal information. Direct HTTPS requests generally avoid proxy charges, but Apify compute, storage, and other platform resources can still incur costs; begin with small limits and inspect output before expanding.