Dataset Automation

Titimangsa uses automation-assisted curation. Automation can discover official sources and extract candidate dates, but it must never publish verified data without human review.

Review Flow

Official source candidate
        |
        v
Automated extraction
        |
        v
Review-required draft
        |
        v
Human verification against the original document
        |
        v
Curated source YAML, generated JSON, and API deployment

Automation output belongs in data/drafts. It must use status: draft and requiresReview: true. Run:

pnpm automation:check

Import a manually selected official HTML page:

pnpm import:official-html -- \
  --year 2027 \
  --url "https://example.go.id/official-holiday-page" \
  --publisher "Official Publisher"

The manual Import Official HTML Draft GitHub Actions workflow runs the same importer and uploads a review artifact. It does not commit or publish data.

Candidate dates may remain unknown when classification is uncertain. A maintainer must verify whether each date is a national holiday, collective leave, or a different work arrangement before updating final data.

The project will add PDF extraction, source discovery, and automated pull requests incrementally. Automatic merge or direct publication of extracted dates is intentionally out of scope.

Titimangsa is not an official government API. Always refer to the original documents for legal or administrative certainty.