scraper-js — JS scraper template

The Node.js half of the Brewtality-3-16 self-healing job-scraper template for peviitor.ro. node-fetch + Cheerio, ESM, Jest.

This is a template. scraper/config/*.json, docs/, ai/ and the workflows ship {{PLACEHOLDER}} values. To make a real scraper, copy this folder into a new repo and replace them — see the placeholder list.

What it does (once configured)

1. Validate the company via the public ANAF API (demoanaf.ro) by CIF — name, active/inactive status, address. Cached in tmp/company.json (7-day TTL) with a stale-cache fallback.
2. Scrape jobs from the company's own careers listing (HTTP + Cheerio, no browser), reconciled against its job sitemap, plus ANOFM by CIF.
3. Self-heal every field through a selector cascade (primary CSS → fallback CSS → structural / JSON-LD → regex). See ai/AGENTS.md.
4. Validate + canary — drop jobs with a bad URL / empty title; abort before any write if the scrape produced nothing.
5. Upsert to the Peviitor API (retry + backoff on transient failures).
6. Generate docs/jobs.md and refresh docs/company.json for GitHub Pages.

Quick start

npm install
npm run test:unit        # 151 tests — pass with placeholders in place
npm run scrape           # runs the full pipeline (no-op until configured)

Placeholders

Placeholder	Fill with
SOF MEDICA S.A.	legal name, uppercase (e.g. EXAMPLE COMPANY SRL)
Sofmedica	commercial brand
6719715	fiscal code (CUI), no RO prefix
https://sofmedica.com	https://www.example.com
https://apply.workable.com/sofmedica/	the open-positions listing page
``	the job sitemap URL (or "" if the site has none)
https://apply.workable.com/j/	canonical job-permalink prefix, e.g. https://www.example.com/jobs/
Otopeni	HQ city (falls back to România in the transform)
/ / ``	primary CSS selectors for the listing (keep the generic fallbacks after them)
peviitor-scrapers / sof-medica-nodejs-scraper	the derived repo's owner / name

Then adapt parseListing and scrapeCareers in scraper/index.js to the site, and add a test per new cascade level.

Testing

npm run test:unit          # always runs (placeholder-safe)
npm run test:integration   # self-skips until a company is configured + ANAF reachable
npm run test:e2e           # self-skips until configured; live careers-site + API otherwise
npm run test:consistency   # needs GITHUB_REPOSITORY + GITHUB_TOKEN (for a derived repo)

JavaScript-rendered career sites (Playwright)

If the job list is built client-side, node-fetch only sees an empty shell. Set rendering in scraper/config/scraper.json:

"rendering": { "mode": "browser", "waitSelector": ".job-card", "timeoutMs": 30000 }

static (default) never launches a browser; browser always renders; auto renders only when the static HTML parses to zero jobs. The scrape workflow installs Chromium only when the mode is not static. Locally: npx playwright install chromium. Code: scraper/browser.js; tests: tests/unit/browser.test.js (fake Playwright) and tests/integration/browser-live.test.js (real Chromium, skipped if absent).

License

MIT — see LICENSE. Managed by ASOCIATIA OPORTUNITATI SI CARIERE for the peviitor.ro job board.