FR
Copied
Modules

Public-registry enrichment

Public-registry enrichment

Fills the gaps in a list (phone, email and website) by matching each business against official public registries loaded locally. No crawling, no AI, no network call at job time: the answer is either in the registry or it is not.

Purpose

Most enrichment modules read a business website, so a list without websites is a dead end. This module works the other way round: it identifies the business (company ID, or name plus postcode) and returns the contact details published in open government registries. It is the natural step right after sirene, which returns official company records but never contact details.

Matching is deterministic and ordered from safest to loosest: company ID (SIRET), then parent company ID (SIREN), then normalised name + postcode, then normalised name + town. When two same-name records in the same town disagree, nothing is filled: a blank is better than a wrong number.

Registries covered

Registry What it covers What it brings
Annuaire Santé (RPPS) Doctors, dentists, pharmacists, midwives Phone, email
DATAtourisme Hotels, restaurants, campsites, venues and activities Website
Public administration directory (DILA) Town halls, prefectures, public bodies Phone, email, website
FINESS Hospitals, clinics, care homes, social care Phone
Education directory Schools and high schools Phone, website
ADEME, RGE companies Construction and energy retrofit Phone, email, website
Agence Bio Organic producers, processors and distributors Phone, website
Atout France, classified accommodation Hotels, campsites, tourist residences Website
Vehicle inspection centres Automotive Phone, website
Qualité Tourisme and Tourisme et Handicap (DGE) Certified tourism businesses Phone, email, website
OpenStreetMap, shops and craftspeople Bakeries, hairdressers, garages, offices, local shops Phone, email, website
OpenStreetMap, food service Restaurants, cafés, bars Phone, website
OpenStreetMap, healthcare, public services and accommodation Practices, public facilities, hotels and guesthouses Phone, email, website

That is roughly 1,160,000 establishments, of which 854,000 have a phone number, 287,000 an email and 570,000 a website. Coverage is nationwide (France) and the registries are refreshed every month, with nothing to do on your side.

Licensing and attribution

The State registries are published under the Licence Ouverte. The last three come from OpenStreetMap and are published under the ODbL, which requires attribution and share-alike. That is what the opendata_source column is for: every value written names the registry it came from. If you redistribute a file containing this data, keep that column.

Inputs

Field Required Notes
nom yes (unless siret/siren) Business name. Aliases accepted (name, raison sociale, entreprise…).
siret no 14 digits. Best match key: used first when present.
siren no 9 digits. Used when no SIRET matches.
code_postal no Strongly recommended: without a postcode or town, name-based matching is skipped.
commune no Fallback when there is no postcode.

This is the only enrichment module that accepts rows with no contact detail at all: finding one is its job. Rows without a name and without a company ID are kept in the output file untouched.

Outputs

Column Type Description
site_web string Filled only if empty in the input.
telephone string Filled only if empty in the input. National format.
email string Filled only if empty in the input. Generic institutional mailboxes are excluded.
opendata_champs string Which columns this module filled, |-separated. Empty when nothing was filled.
opendata_match string How the row was matched: siret, siren, nom+cp or nom+commune. Empty when no match.
opendata_source string Human-readable registry the values came from (attribution required by the open licence).

An existing value is never overwritten. An empty opendata_champs means the module ran and found nothing usable for that row, never that the row was dropped.

Lifecycle

Standard job lifecycle, see Jobs lifecycle. Progress is reported in businesses and final volume in rows completed.

Pipeline

opendata is an enrichment module: it augments an existing list rather than generating one.

needs:    poi_list
produces: enriched_list

Typical chain:

sirene → opendata → emails → verify_emails → filter

Placing it right after sirene is what makes the rest of the chain possible: modules such as emails, socials, legal_mentions and techstack all require site_web.

Endpoints

POST /api/jobs/opendata

Field Type Required Description
items array of objects no Rows to enrich, max 10000. Optional when source_job_id is set.
source_job_id string no Reuse the rows of one of your completed jobs instead of resending them.

Response: the standard JobPublic envelope.

Example:

POST /api/jobs/opendata
{ "items": [ { "nom": "Dupont Couverture", "code_postal": "35000", "siret": "48123456700019" } ] }

{ "id": "6b1f…", "job_type": "opendata", "status": "pending", "results_count": 0 }

Global quotas and per-job ceilings: see Limits.

Errors

Condition Response
No row has a name, a SIRET or a SIREN 400 with the exact composition of the list submitted
More than 10000 rows 422 validation error
source_job_id unknown, not yours, unfinished or over 10000 rows 400

What's next