# Parses JSON access logs from NGINX fronting a Docker Registry.
# Emits Prometheus counters for pull/push bytes and seconds, plus request counts.

# -------- Metrics
counter registry_requests_total by method, path_class, status, repo, user, node
counter registry_pull_blob_bytes_total by repo, user, node
counter registry_pull_blob_seconds_total by repo, user, node
counter registry_push_upload_bytes_total by repo, user, node
counter registry_push_upload_seconds_total by repo, user, node

# -------- Field extraction (order-independent)
/"method":"([A-Z]+)"/                         { method = $1 }
/"path":"([^"]+)"/                            { path = $1 }
/"remote_address":"([^"]*)"/                  { node = $1 }
/"remote_user":"([^"]*)"/                     { user = $1 }
/"status":([0-9]{3})/                         { status = $1 }
/"request_time":([0-9.]+)/                    { request_time = $1 }
/"body_bytes_sent":([0-9]+)/                  { body_bytes_sent = $1 }
/"upstream_range":"([^"]*)"/                  { up_range = $1 }
/"docker_upload_uuid":"([^"]*)"/              { upload_uuid = $1 }
/"docker_content_digest":"([^"]*)"/           { digest = $1 }

# Capture namespace/repo between /v2/ and the next section (blobs|manifests|tags|_catalog)
/"path":"\/v2\/(.+?)\/(blobs|manifests|tags|_catalog)/ {
  repo = $1
}

# Defaults per line (and a tiny reset)
/$/ {
  if repo == "" { repo = "unknown" }
  if user == "" { user = "anonymous" }
  if node == "" { node = "unknown" }
  if uploaded_bytes == 0 { uploaded_bytes = 0 }  # explicit init per line
}

# Helper: parse upstream_range "a-b" => bytes = b+1
/"upstream_range":"([0-9]+)-([0-9]+)"/ {
  uploaded_bytes = int($2) + 1
}

# -------- Classify & emit

# Downloaded blob bytes/time: GET /v2/<repo>/blobs/sha256:...
/"method":"GET".*"\/v2\/.+\/blobs\/sha256:[a-f0-9]+"/ {
  registry_pull_blob_bytes_total[repo=repo, user=user, node=node] += body_bytes_sent
  registry_pull_blob_seconds_total[repo=repo, user=user, node=node] += request_time
  registry_requests_total[method=method, path_class="pull_blob", status=string(status), repo=repo, user=user, node=node]++
}

# Upload data transfer: PATCH /v2/<repo>/blobs/uploads/<uuid> 202
/"method":"PATCH".*"\/v2\/.+\/blobs\/uploads\/[a-f0-9-]+"/ {
  if uploaded_bytes > 0 {
    registry_push_upload_bytes_total[repo=repo, user=user, node=node] += uploaded_bytes
  }
  registry_push_upload_seconds_total[repo=repo, user=user, node=node] += request_time
  registry_requests_total[method=method, path_class="push_upload", status=string(status), repo=repo, user=user, node=node]++
  uploaded_bytes = 0
}

# Upload session start/end (bookkeeping)
/"method":"POST".*"\/v2\/.+\/blobs\/uploads\/"/ {
  registry_requests_total[method=method, path_class="push_start", status=string(status), repo=repo, user=user, node=node]++
}
/"method":"PUT".*"\/v2\/.+\/blobs\/uploads\/[a-f0-9-]+"/ {
  registry_requests_total[method=method, path_class="push_commit", status=string(status), repo=repo, user=user, node=node]++
}

# Manifest pulls
/"method":"GET".*"\/v2\/.+\/manifests\/[^"]+"/ {
  registry_requests_total[method=method, path_class="pull_manifest", status=string(status), repo=repo, user=user, node=node]++
}

# Manifest writes
/"method":"PUT".*"\/v2\/.+\/manifests\/[^"]+"/ {
  registry_requests_total[method=method, path_class="push_manifest", status=string(status), repo=repo, user=user, node=node]++
}

# Root auth probe
/"path":"\/v2\/"/ {
  registry_requests_total[method=method, path_class="v2_root", status=string(status), repo="none", user=user, node=node]++
}