CVbankas.lt Job Search Scraper

Discover current public CVbankas.lt listings and enrich every published record from its canonical detail page. Output includes stable identity, title, employer, full source-backed description, location, employment and salary data when disclosed, dates, structured source data, request receipts, and runtime provenance.

The Actor accepts a public CVbankas listing/search url, maxItems, maxPages, a bounded request timeout, and bounded retries for transient non-barrier failures. It uses the ordinary official origin only: proxy configuration and concurrent detail requests are intentionally unavailable so a detected access barrier stops subsequent requests immediately.

{ "url": "https://www.cvbankas.lt/", "maxItems": 3, "maxPages": 1 }

Only complete, identity-verified job records with official CVbankas URLs and successful listing/detail receipts enter the dataset; diagnostics, fixtures, and incomplete records are never written as dataset rows. Run-level diagnostics and request receipts are stored in KVS; each eligible job row also carries its own listing/detail receipts as provenance. The platform status is only SUCCEEDED or FAILED; the extraction outcome is reported separately as resultStatus (COMPLETE, LIMITED, SKIPPED, DEFERRED, or FAILED). This Actor does not currently recognize affirmative source-authored empty-result evidence, so zero rows without a confirmed barrier or fatal invalid-input/runtime/storage error are SUCCEEDED/DEFERRED, never NO_DATA. HTTP 401/403/429/451 or a clear source-authored verification/security challenge stops the crawl without retry or redirect following, discards all staged rows, and records SKIPPED. Generic transport failures remain DEFERRED. Fatal zero-row invalid-input/runtime/storage failures call Actor.fail(). If a later non-block issue follows successfully stored rows, those rows remain and the result is SUCCEEDED/LIMITED.

OUTPUT and RUN_HEALTH contain status and row counts. RUN_DIAGNOSTICS and REQUEST_RECEIPTS preserve bounded evidence for each run. Existing historical storage is not rewritten by validation.