@data-fair/processing-deaths

Data-fair processing plugin: imports the INSEE file of deceased persons (French death records since 1970, published on data.gouv.fr) into a REST dataset, enriched with the current commune, department and region of the places of birth and death.

Modes

• Create: creates the dataset and imports every yearly file from the start year (1970 by default), then switches the processing to update mode.
• Update: imports again the files of the year of the last update (yearly file, then the monthly files of the latest year). Lines are upserted on nom, prenom, numero_acte_deces.
• Inconsistencies and oppositions: only removes incoherent ages and dates, and deletes the people listed in the opposition file. Create and update modes run this step too, after the import.

Reference data

• Communes (datasetInsee): a dataset of current communes with the concepts "Code commune INSEE", "Commune", "Code département INSEE", "Code région INSEE" and the columns nom_departement and nom_region (e.g. "Communes de France").

• Former commune codes (datasetCodesActuels, optional): columns COM_AV → COM_AP (e.g. "Anciens codes commune").

• Countries: shipped with the plugin in lib/countries.json, built from the INSEE COG country files (current list and list since 1943). A death abroad is named after the country as it was at the date of death (99123 is "URSS" before 1991-12-25, "Russie" after). Refresh it once a year when INSEE publishes a new vintage:

  npm run update-countries -- <"Pays et territoires étrangers au 01/01/YYYY" url> <"... depuis 1943" url>

Development

• Node 24 (nvm use), native TypeScript (no build step).
• npm install
• npm run build-types: generates the types from the JSON schemas.
• npm test: node:test tests.
• npm run lint / npm run lint-fix.

Publication

Publication to the registry is done by GitHub Actions: push on main → staging, tag v* → production.