sources.yaml

# Collection configuration.
#
# cadence_hours drives gap detection. Delta windows and daily snapshots are NOT retroactively
# acquirable, so an overdue capture is permanent data loss — cadence here is a data-integrity
# setting, not a preference.
#
# full_copy defaults to false everywhere. Full copies are large (GLEIF lei2 csv ~474 MB,
# json ~886 MB; the Companies House PSC snapshot and SEC companyfacts are larger still).
# Turning them on is a storage and retention decision that deserves an ADR first.

gleif:
  enabled: true
  cadence_hours: 8          # publisher issues golden copy 3x daily
  gap_tolerance: 2.0
  format: csv               # csv | json | xml
  full_copy: false          # ADR required before enabling
  delta_windows: [IntraDay, LastDay, LastWeek, LastMonth]

companies_house:
  enabled: true
  cadence_hours: 24         # PSC snapshot is replaced every morning
  gap_tolerance: 1.5
  products: [psc]           # add: company (monthly), accounts (daily)
  keep_latest: 1

sec:
  enabled: true
  cadence_hours: 24         # bulk archives rebuilt nightly
  gap_tolerance: 2.0
  artifacts: [tickers, tickers_exchange, submissions]   # add companyfacts when ready

sanctions:
  enabled: true
  cadence_hours: 6          # OFAC publishes ad hoc, with legal urgency
  gap_tolerance: 4.0
  artifacts: [ofac_sdn, ofac_consolidated]
  allow_mirror: true        # fall back to the OpenSanctions per-source mirror when blocked