Finn.no Jobs Search Scraper

This Actor searches public FINN job result cards and sequentially verifies each selected public detail page before emitting a record. It uses ordinary direct HTTPS requests only. It does not use proxies, stealth, fingerprint changes, custom identities, retries, or raw source payloads.

Input

The public input contains only:

• keyword — search keyword or employer.
• jobType — fulltime, parttime, management, or it.
• maxItems — maximum complete detail-verified records.
• maxPages — maximum result pages inspected sequentially.

Temporary fixture, context, raw-payload, timeout, retry, proxy, and user-agent controls are intentionally not part of the input schema.

Dataset

Each emitted record is source-grounded and keeps listing/detail/canonical URLs and HTTP receipts. It includes the public title, heading, employer, location, employment, sector, industries, occupations, remote options, languages, keywords, skills, description HTML/text/paragraphs/sections, responsibilities, requirements, benefits, compensation when published, application metadata, public media/branding, dates, and run-quality evidence. Contact objects and raw FINN payloads are omitted; contact details found in public text are redacted.

Records are buffered until the requested complete details are verified, so partial jobs are not published. A successful record must have a readable public description, employer, and application link and must contain more than 20 meaningful source-grounded fields.

Access policy

The platform run status is always SUCCEEDED or FAILED; the separate resultStatus reports COMPLETE, LIMITED, SKIPPED, DEFERRED, NO_DATA, or FAILED. Only HTTP 401, 403, 429, 451, visible source-authored denial/challenge text, or a challenge-platform marker without expected public content is treated as SKIPPED; this stops requests and discards staged rows. Generic transport/body failures, HTTP 407, unverified redirects, and server unavailability remain DEFERRED, not evidence that FINN blocked access. A non-block fatal failure with no stored rows calls Actor.fail() and is FAILED/FAILED; successfully pushed rows survive later non-block failures as SUCCEEDED/LIMITED. If Actor finalization rejects, the run records KVS evidence and retries exit once; a zero-row finalization failure is FAILED/FAILED, while stored rows remain SUCCEEDED/LIMITED. Diagnostics and request evidence are stored in the key-value store, never as dataset rows. The Actor never attempts to bypass a site barrier.

Local validation

npm install --ignore-scripts --no-audit --no-fund
npm test
npm run check
npm run lint
apify validate-schema
apify run --purge --input-file ./INPUT.json
npm run validate:dataset -- storage --require-data