J&jk &ddlmZGddeZy)) AppConfigceZdZdZdZddZy)ServicesConfigzdjango.db.models.BigAutoFieldservicesNcJddlm}m} ddlm}|y#||f$rYywxYw)u Load the country snapshot into memory at Django startup. This runs once per process: - gunicorn pre-fork: master process loads, workers inherit via fork. - gunicorn post-fork: each worker calls ready() independently. - Kubernetes replicas: each pod loads independently from the same DB. Failure policy -------------- We distinguish two categories of startup failure: 1. MIGRATION STATE (table does not exist yet): django.db.OperationalError / ProgrammingError are caught and silently ignored. The snapshot will NOT be loaded, and any attempt to call resolve_istat_country_code() before the table exists will raise SnapshotNotLoadedError — which is the correct behavior during `manage.py migrate`. 2. ALL OTHER FAILURES (DB reachable, table exists, data invalid): RuntimeError from _validate_snapshot() is NOT caught here. The process will fail to start with a clear error message. This is intentional — a partially-initialized country reference is a fatal misconfiguration, not a recoverable runtime error. Multi-worker safety ------------------- The snapshot is loaded per process. No shared memory is assumed across nodes or Kubernetes replicas. Each worker/pod loads independently from the same DB at startup. r)OperationalErrorProgrammingError)load_country_snapshotN) django.dbrr services.country_snapshotr )selfrr r s /backend/services/apps.pyreadyzServicesConfig.readys1B A  G ! # "23    s "")returnN)__name__ __module__ __qualname__default_auto_fieldnamerrrrs8 D+rrN) django.appsrrrrrrs!/Y/r