Skip to main content

Architecture

RecallRx has two static outputs: the recall search app and this documentation site.

Official sourcesCountry adaptersNormalized JSONStatic search UIGitHub Pages

Collector pipeline​

The Python CLI loads enabled source adapters from the registry. Each adapter discovers candidate records, fetches source pages, parses source-specific fields, normalizes them into the shared schema, and returns build diagnostics.

Static search app​

The frontend reads:

  • data/metadata.json
  • data/countries/<country>/metadata.json
  • data/countries/<country>/recalls-summary.json

The app uses exact matching for product codes, lots, and recall ids before falling back to fuzzy text search.

Documentation site​

The Docusaurus site lives in docs-site/ and builds independently from the recall search app. This keeps product documentation testable without changing the data deployment path.