# External Metering with a File-Backed Internal Auth Token
#
# Same pre-request balance check and post-response usage reporting as
# external-metering.yaml, but the internal bearer credential used to
# authenticate the metering callouts is read from a file rather than
# embedded in the config or image.
#
# This suits a Kubernetes projected Secret mounted at a fixed path: the
# platform rotates the file in place and `external_metering` picks up the
# new token without a restart or image rebuild. The token value never
# appears in logs, tracing, or the rendered config. If the file is
# missing, unreadable, empty, or oversized at startup the proxy fails to
# start; if a post-startup rotation produces no valid token the metering
# callouts fail closed (never dialed unauthenticated) and the existing
# `fail_open` policy then decides whether the request proceeds.
#
# Usage:
#   printf 'my-internal-metering-token' > /etc/praxis/metering-token
#   cargo run -p praxis-ai-proxy -- -c examples/configs/external-metering-file-key.yaml
#   curl -X POST http://localhost:8080/v1/chat/completions \
#     -H "Content-Type: application/json" \
#     -H "x-tenant-username: alice" \
#     -H "x-tenant-group: engineering" \
#     -d '{"model":"gpt-4","messages":[{"role":"user","content":"hi"}]}'
#
# The metering service then receives both callouts with an
# `Authorization: Bearer <token>` header read live from the file:
#   GET  {metering_url}/api/v1/customers/{username}/entitlements/{feature_key}/value?model={model}
#   POST {metering_url}/api/v1/events  (CloudEvents 1.0 JSON)

listeners:
  - name: default
    address: "127.0.0.1:8080"
    filter_chains:
      - main

filter_chains:
  - name: main
    filters:
      - filter: router
        routes:
          - path_prefix: "/"
            cluster: backend

      - filter: external_metering
        metering_url: "http://127.0.0.1:9090"
        # The metering service in this example listens on loopback, a
        # non-public address, so callouts must be explicitly opted in.
        allow_private_endpoint: true
        timeout_seconds: 5
        feature_key: "inference-tokens"
        source: "ai-gateway"
        fail_open: true
        identity_header_prefix: "x-tenant-"
        # Internal bearer token for the metering callouts, read from a file
        # and watched for rotation. One of `value | env_var | file`.
        internal_auth_token:
          file: /etc/praxis/metering-token

      - filter: token_count
        provider: openai

      - filter: load_balancer
        clusters:
          - name: backend
            endpoints:
              - "127.0.0.1:3000"

insecure_options:
  allow_private_endpoints: true # example proxies to local backends