agency_registry.py

# ingest/agency_registry.py

# Domains are PUBLIC FOIA / reading room hosts only.
# Labels may include units if (and only if) a public FOIA page exists
# and the user provides its URL.

ALLOWED_FOIA_SOURCES = {
    # Core
    "vault.fbi.gov": "FBI",
    "www.cia.gov": "CIA",
    "www.archives.gov": "NARA",
    "foia.state.gov": "State Dept",
    "www.nsa.gov": "NSA",
    "www.defense.gov": "DoD",
    "www.esd.whs.mil": "DoD FOIA",
    "www.whitehouse.gov": "White House",

    # Military (public FOIA pages)
    "www.af.mil": "USAF",
    "www.navy.mil": "US Navy",
    "www.marines.mil": "USMC",
    "www.army.mil": "US Army",
    "www.spaceforce.mil": "US Space Force",

    # Intelligence / defense components (public FOIA pages only)
    "www.dia.mil": "DIA",
    "www.nro.gov": "NRO",

    # Law enforcement / protective services (public FOIA pages)
    "www.secretservice.gov": "US Secret Service",
    "www.dea.gov": "DEA",

    # Labels for historical / organizational references
    # (NO claim of dedicated public repositories)
    # These are ONLY labels applied if a public FOIA URL is supplied.
    "label:SAC": "CIA Special Activities Center (label only)",
    "label:SAD": "CIA Special Activities Division (label only)",
    "label:NIA": "National Intelligence Authority (historical)"
}