Freelancermap IT Freelance Projects Scraper (Europe)

Scrapes IT freelance/contract project listings from Freelancermap — a European freelance project board.

Public data only. No login, no cookies, no browser. No bot challenge of any kind — fully open robots.txt, no WAF observed on any surface touched during recon.

The one thing you need to know before using this

Each search URL returns one page (~22 projects) — there is no working pagination. Freelancermap's own search page believes it has dozens of pages, but every way of asking for page 2 (the exact link the site itself generates, alternate parameter names, XHR-style headers) returned page 1's content again — a real, verified constraint (see CRAWLING_METHOD.md), not a missing feature. To get broader coverage, pass multiple search URLs with different keywords, countries or categories rather than expecting one URL to page through everything.

What you get

Two record types share one dataset, told apart by recordType.

PROJECT — one row per listing

Rows are the upstream's own result objects, almost verbatim — title, a full HTML description (not an excerpt — no detail page needed at all), company/poster (contact info), country, city, contractType, projectContractType.remoteInPercent, budget, industry, beginningText/durationText, created/updated/expires timestamps, and a computed projectUrl. Two internal ML-only fields (embedding, a ~500-float recommendation vector, and matching, relevance-scoring metadata) are dropped — everything else is passed through as-is.

SEARCH_SUMMARY — one row per search URL

How many projects came back, plus (informational only) upstreamBelievedTotalPages — what the site's own pagination widget claims, even though this actor cannot reach past page 1 (see above).

ERROR — one row per input that could not be processed

Every input maps to at least one row, so nothing disappears silently.

Input

Field	What it does
Search URLs	full freelancermap.com/projects... URLs, copied from the site — one fetch each
Max concurrent requests	absorbs latency variance — not the throttle
Min interval between requests	the actual speed control — 0.5s default

Example

{
  "searchUrls": [
    "https://www.freelancermap.com/projects?query=python",
    "https://www.freelancermap.com/projects?remoteInPercent=100",
    "https://www.freelancermap.com/projects/web-and-software-development"
  ]
}

Notes on reliability

• Query filters are genuinely applied server-side — verified with query=, countries=, remoteInPercent=, and a gibberish query (honest zero results, not a silent fallback to the unfiltered baseline).
• No WAF, no rate limit observed. Residential proxy is still the cloud default, as elsewhere in this portfolio.

Known limits

• No pagination past page 1 of a given search URL (see above) — this is the main thing to plan around when using this actor.
• No detail-page enrichment offered or needed — the search row already carries the full HTML description and every field the detail page would.