JobKorea public jobs scraper

This Apify Actor extracts publicly available JobKorea (jobkorea.co.kr) job postings. It reads official search pages, verifies each posting on its official detail page, and writes only complete, identity-checked job records to the default dataset.

The implementation uses bounded native HTTPS requests and Cheerio. It does not use a proxy, browser automation, fingerprinting, or challenge bypass. This keeps runs fast and makes target blocking visible in RUN_DIAGNOSTICS instead of producing guessed records.

Modes

mode selects one of five workflows:

• search — search one query and enrich up to maxItems postings.
• searchMultiple — run up to five queries, taking up to maxItems unique postings per query.
• single — fetch one official /Recruit/GI_Read/<id> detail URL.
• multiple — fetch a bounded list of official detail URLs.
• startUrls — accept official search and/or detail URLs.

Search pagination is bounded by maxPages (1–3). Detail concurrency is bounded by concurrency (1–5). maxItems is limited to 10 per query/direct-input mode, and each request has a 5–45 second timeout.

Input

The complete input definition is .actor/input_schema.json. A normal search input is:

{
  "mode": "search",
  "query": "developer",
  "location": "Seoul",
  "maxItems": 3,
  "maxPages": 2,
  "concurrency": 4,
  "requestTimeoutSecs": 25
}

For multiple and startUrls, URL entries are objects:

{
  "mode": "multiple",
  "urls": [
    { "url": "https://www.jobkorea.co.kr/Recruit/GI_Read/49844042" },
    { "url": "https://www.jobkorea.co.kr/Recruit/GI_Read/49856065" }
  ],
  "maxItems": 2
}

Only HTTPS URLs on the official JobKorea host are accepted. Detail identifiers, canonical URLs, titles, employers, and search-query matches are checked before a record is emitted.

Dataset

The strict output definition is .actor/dataset_schema.json. Records include, when publicly present:

• stable posting ID, title, employer, company URL/logo, canonical detail URL, and source page;
• advertised location, structured address, remote indicator, employment type, experience, education, schedule, salary, categories, skills, qualifications, preferred qualifications, and benefits;
• normalized posting/start/deadline dates and original deadline text;
• JSON-LD summary, cleaned public detail-section HTML, a readable source-backed description, company information, map URL, and explicit application URL provenance;
• source receipts, identity/canonical verification, search rank/query, HTTP status, and quality counts.

JobKorea currently exposes the concise JSON-LD summary and selected server-rendered public sections to this local client. Some full advertisement-body content is client-only or access-controlled, so fullDescriptionAvailable is explicitly false when that body is not publicly available; the Actor never fabricates it or invents an application URL. Optional fields are omitted rather than emitted as null, blank, placeholder, or empty values.

Run state is kept outside the dataset in the default key-value store:

• RUN_SUMMARY — mode, counts, timing, requests, environment, and cloud build/run provenance when available;
• RUN_DIAGNOSTICS — target, transport, or parsing errors;
• RUN_SKIPS — candidates rejected after verification (for example, a query mismatch);
• RUN_HEALTH — dataset quality receipts.

Local development

From this directory:

npm install
npm test
npm run lint
npx --yes apify-cli validate-schema .actor/input_schema.json
npx --yes apify-cli run --purge --input-file INPUT.json
npm run validate

The local run writes to storage/. npm run validate checks required fields, official URL/ID consistency, duplicate IDs/URLs, recursive null/blank/placeholder/empty values, application provenance, and key-value-store consistency.

Cloud deployment

apify push
apify call jobsapi/jobkorea-jobs-search-scraper --memory 512 --timeout 300 --input-file INPUT.json

Keep cloud validation bounded with the provided input. The default dataset contains only persisted complete jobs; diagnostics remain in the default key-value store.

Responsible use

Collect only public job information, respect JobKorea’s terms and robots directives, and use modest bounded request limits. Do not use this Actor for authenticated pages, paywall or challenge bypassing, or collection of candidate or recruiter personal data.