Aller au contenu
Toutes les missions
Veille marché · Dev-to Publiée il y a 1 mois

How to Scrape jobs.ge Postings in Georgia Without a Paid API

The problem Jobs.ge is Georgia's oldest job board, and for anyone tracking the Georgian labor market — a recruiter sourcing candidates, an analyst watching hiring trends, an aggregator build...

Entreprise es
Mode Remote

Résumé

The problem Jobs.ge is Georgia's oldest job board, and for anyone tracking the Georgian labor market — a recruiter sourcing candidates, an analyst watching hiring trends, an aggregator building a job feed — it is the primary public source. But there is no API. The manual alternative is opening listing pages in a browser, clicking into every posting, and copying titles, employers and dates into a spreadsheet, re-doing the whole thing the next day to catch new postings. Two details make hand-collection worse than it looks. Jobs.ge prints its "Published" and "Deadline" dates without a year — a human has to infer it, and gets it wrong around year boundaries. And the board mixes Georgian and English interface labels with predominantly Georgian posting text, so ad-hoc parsing scripts silently misread dates and filters. A scraper that handles the site's quirks correctly is a small project of its own. What the actor does The Jobs.ge Georgia Jobs Scraper searches Jobs.ge's public, server-rendered listing pages — no login, no API key — and returns one structured row per job found. You can narrow the search with: Category and location filters using Jobs.ge's own classification — categories such as it_programming, sales_procurement or medicine_pharmacy, and location groups including Tbilisi, regions, remote and abroad; Title keywords (up to 10 per run) matched against real posting titles — each row reports which of your terms matched; Publication date window (posted_within_days, 1–365 UTC calendar dates), applied before the result cap. For every job the row carries title, company and company_url (null when the poster hides them), the direct posting url and job_id, location as the explicit source label beside the title (never inferred from description text), and both timestamps — posted_date and deadline_date — resolved into proper ISO 8601. That last point matters: the source never prints a year, and the actor reconstructs it from its own clock rather than leaving you to guess. The actor follows pagination up to a 20-page safety limit, deduplicates job IDs, and confirms your selected filters on every scanned page. Each result states listing_pages_fetched, listing_pages_available, listing_scan_complete and listing_stop_reason, so you can tell a complete scan from a partial one. If a later page fails or filters are lost, useful rows are marked partial and a free diagnostic row explains the incomplete search. An optional fetch_full_description setting visits each matched job's own page for the free-text posting body (off by default — the listing page itself carries no snippet, not even a teaser, so description is null without it). An optional fetch_job_category lookup confirms category membership against Jobs.ge's own category listings within bounded extra requests. "Nothing found" is honest: a keyword that matched nothing pushes a found: false row with an empty error, while a request the source failed to answer pushes found: false with a real error message — the two never look the same, and neither carries a result fee. Example: input and output { "keywords": [], "language": "en", "max_items": 5, "fetch_full_description": false } A real row from a real run with default settings: { "found": true, "url": "https://www.jobs.ge/en/?view=jobs&id=741288", "job_id": "741288", "title": "Cashier-Consultant", "company": "Way Mart", "company_url": "https://www.jobs.ge/en/?view=client&client=way-mart", "location": null, "posted_date": "2026-07-30T00:00:00.000Z", "salary_raw": null, "description": null, "source_board": "jobs-ge", "scraped_at": "2026-07-30T14:22:28.264Z", "deadline_date": "2026-08-30T00:00:00.000Z", "matched_keywords": [] } Note salary_raw: null — Jobs.ge essentially never publishes a structured salary field, and the actor never estimates or invents a figure. Pricing and the free limit Pay per event: $0.005 per run start plus $0.002 per job found. No-match notices, source errors and unconfirmed filter diagnostics produce found: false rows with no result fee; only the separate start fee applies. Apify's free plan gives $5 of usage credits per month. At this tariff, $5 covers up to 2,497 job rows in a single run (0.005 + 0.002 × 2,497 ≈ $5.00) — or about 24 runs of 100 jobs each at $0.205 per run. Try it Pick a category or just press Start with the prefilled input: Jobs.ge Georgia Jobs Scraper For AI agents and MCP The actor takes JSON in and returns structured JSON dataset rows via the Apify API, so an agent can start a run, read the dataset, and branch on the found flag and scan metadata (listing_scan_complete, listing_stop_reason) before trusting a zero-result answer. The README notes it is a standard Apify Actor callable from the Apify API, the SDK, or the Apify MCP server, and documents the MCP server setup alongside JavaScript, Python and CLI examples.