dume-api 0.1.0

Tiny helper for downloading the “données essentielles” exposed at https://dume.chorus-pro.gouv.fr/#/accueil/donnees-essentielles/rechercher. Everything here reproduces the JSON calls the official SPA makes to https://dume.chorus-pro.gouv.fr/dumes/donneesEssentielles.

Setup

pip install dume-api

Quick start

import dume_api as dume

# Lightweight list view (same fields as the website table).
rows = dume.get_contracts(date="2024-01-11")

# Add filters the same way you would in the UI.
rows = dume.get_contracts(
    date_start="2024-01-01",
    date_end="2024-01-31",
    cpv_code="45210000",
    type_de="Marché public",
)

# Request the complete contract structure (one API call per reference).
contracts = dume.get_contracts(date="2024-01-05", complete=True)

# Single-contract lookup (used internally by the snippet above).
record = dume.get_contract("AIFEDUME-daoxup8c")

That’s it—you rarely need to instantiate anything else. If you do want to manage your own session (different base URL, manual iteration, etc.), instantiate dume_api.DumeApiClient.

Filters & options

Both helper functions accept the same keyword arguments as the website:

Supported filter keywords and their UI equivalents:

Behind the scenes

• The platform caps each query at 100 rows, so get_contracts() automatically splits large date ranges and toggles between MP/CC when needed. If the API still returns 100 results for a tiny filter the helper raises an error so you can narrow it manually.
• Each HTTP request retries on 429/5xx responses, backing off exponentially. Call dume.configure_client(pause=0.5, max_retries=5) to slow down or override the defaults.