Randstad public jobs scraper

This Apify Actor retrieves complete, source-verified job records from Randstad's public global jobs pages. It uses ordinary direct HTTPS requests and parses the official page's embedded route data, JobPosting JSON-LD, semantic HTML, metadata, and public application link.

The Actor is local-test friendly and fail-closed: it never fabricates missing fields or treats a blocked or failed page as a dataset record; no proxy or alternate access route is used.

Supported modes

• search: query the official jobs collection, follow detail pages, and return complete records.
• single: fetch one exact official Randstad job URL.
• multiple: fetch several exact official Randstad job URLs.
• jobUrls: compatibility alias that maps one URL to single and several URLs to multiple.

Search defaults to:

https://www.randstad.com/jobs/q-developer/

Direct URLs must use www.randstad.com or randstad.com and end with a numeric job ID. The Actor accepts query, keyword, location, startUrls, jobUrl, jobUrls, jobs, maxItems, maxCandidates, maxPages, timeoutMs, timeoutSecs, and maxRequests. Public retry, proxy, cookie, browser, concurrency, fixture, and temporary-context controls are not exposed.

Dataset

Every emitted item is a complete record with a stable recordId and verified canonical job URL. Required identity and quality fields include:

• recordId, jobId, title, jobUrl, canonicalUrl
• companyName, companyId, accountId
• location, city, region, country, countryCode, postalCode, coordinates, remote
• employmentType, workingHours, role, industry, sector, specialism
• salary, salaryMin, salaryMax, salaryCurrency, salaryPeriod, benefits
• description, descriptionHtml, headings, sections, bullets, links, responsibilities, requirements, qualifications, skills
• published, modified, and expiration dates
• contact details are recursively redacted from descriptions and embedded source records
• sourceRecord, request receipt, field coverage, data quality, and verification metadata

Unavailable optional values are omitted. Empty strings, nulls, placeholder text, empty arrays, and empty objects are removed. Records are written only after identity, detail, source, HTTP, and minimum-description checks pass.

Only complete randstad-job records are written to the Dataset; diagnostics and run-state records are KVS-only. Run state is written under RUN_SUMMARY, RUN_DIAGNOSTICS, RUN_METADATA, RUN_HEALTH, OUTPUT, and OUTPUT_SUMMARY. The platform status is SUCCEEDED or FAILED; resultStatus separately reports COMPLETE, LIMITED, SKIPPED, DEFERRED, NO_DATA, or FAILED. A source-authored access block stops the crawl, discards staged rows, and returns SUCCEEDED/SKIPPED; a generic transport failure is DEFERRED, not treated as a site block. If rows were stored before a later processing failure, they are preserved and the run reports SUCCEEDED/LIMITED. A fatal error before any usable row reports FAILED/FAILED and invokes Actor.fail(). Complete records must contain more than 20 canonical meaningful source fields.

Local commands

Install from the lockfile:

npm ci --ignore-scripts --no-audit --no-fund

Run static checks and unit tests:

npm test
npm run check

Run the bounded primary fixture:

$env:APIFY_LOCAL_STORAGE_DIR='storage-randstad-validation'
apify run --resurrect --input-file INPUT.json
npm run validate -- storage-randstad-validation

The repository also includes INPUT-single.json and INPUT-multiple.json for direct-detail mode checks. Local validation inspects the selected isolated storage datasets and key-value store state files.

Responsible use

Use only publicly accessible Randstad pages and respect the site's terms, robots guidance, rate limits, and applicable law. This Actor does not bypass authentication, CAPTCHA, paywalls, or access controls.

Cloud execution and privacy

The Actor uses ordinary direct HTTPS with one attempt per request, caps response bodies and the request budget, stops on HTTP 401/403/429/451 or source-authored access challenges as SKIPPED, and records generic transport failures as DEFERRED. It recursively redacts email addresses and phone numbers from emitted records. No alternate route, identity, cookie, proxy, fingerprint, or challenge bypass is used.