Add API authentication to YOLOv4 backend

app/api/model_control/views.py

@@ -15,6 +15,7 @@
 
 from .serializers import InferenceSerializer, ModelWeightsSerializer
 from model_control.model_utils import (
+    encoded_jwt,
     get_deploy_cache,
     stream_response_from_external_api,
     health_check,
@@ -123,7 +124,16 @@ def post(self, request, *args, **kwargs):
             )
             byte_im = buf.getvalue()
             file = {"file": byte_im}
-            inference_data = requests.post(internal_url, files=file, timeout=5)
+            try:
+                headers = {"Authorization": f"Bearer {encoded_jwt}"}
+                inference_data = requests.post(internal_url, files=file, headers=headers, timeout=5)
+                inference_data.raise_for_status()
+            except requests.exceptions.HTTPError as http_err:
+                if inference_data.status_code == status.HTTP_401_UNAUTHORIZED:
+                    return Response(status=status.HTTP_401_UNAUTHORIZED)
+                else:
+                    return Response(status=status.HTTP_500_INTERNAL_SERVER_ERROR)
+
             return Response(inference_data.json(), status=status.HTTP_200_OK)
         else:
             return Response(serializer.errors, status=status.HTTP_400_BAD_REQUEST)

app/docker-compose.yml

@@ -21,7 +21,7 @@ services:
       - "8000:8000"
     # command: bash
     # dev server can be used for breakpoint debugging, does not support streaming
-    # command: ./manage.py runserver 0.0.0.0:8000
+    # command: python ./manage.py runserver 0.0.0.0:8000
     # gunicorn is used from production, supports streaming
 
     command: gunicorn --workers 3 --bind 0.0.0.0:8000 --preload --timeout 1200 api.wsgi:application
@@ -52,7 +52,7 @@ services:
       # On first application load resources for transformers/etc
       # are downloaded. The UI should not start until these resources
       # have been downloaded. Adjust timeout if on a very slow connection
-      test: ["CMD", "curl", "-f", "http://localhost:8000/up/"]
+      test: [ "CMD", "curl", "-f", "http://localhost:8000/up/" ]
       timeout: 120s
       interval: 10s
       retries: 5
@@ -93,7 +93,7 @@ services:
       - "8111:8111"
     healthcheck:
       # Adjust below to match your container port
-      test: ["CMD", "curl", "-f", "http://localhost:8111/api/v1/heartbeat"]
+      test: [ "CMD", "curl", "-f", "http://localhost:8111/api/v1/heartbeat" ]
       interval: 10s
       timeout: 10s
       retries: 3