remove hard condition on /predict route and added /

.github/workflows/build-container.yaml

@@ -2,8 +2,11 @@ name: "Build applications images"
 
 on:
   push:
-    # paths:
-    #   - "app/**"
+    branches:
+      - main
+    paths:
+      - "src/**"
+      - "dockerfiles/**"
   workflow_dispatch:
 
 concurrency:

README.md

@@ -43,7 +43,7 @@ docker run -ti -p 5000:5000 -e HF_MODEL_ID=distilbert-base-uncased-distilled-squ
 
 ```bash
 curl --request POST \
-  --url http://localhost:5000/predict \
+  --url http://localhost:5000 \
   --header 'Content-Type: application/json' \
   --data '{
 	"inputs": {

src/huggingface_inference_toolkit/webservice_starlette.py

@@ -73,7 +73,7 @@ async def predict(request):
         pred = inference_handler(deserialized_body)
         # log request time
         # TODO: repalce with middleware
-        logger.info(f"POST /predict |  Duration: {(perf_counter()-start_time) *1000:.2f} ms")
+        logger.info(f"POST {request.url.path} |  Duration: {(perf_counter()-start_time) *1000:.2f} ms")
         # deserialized and resonds with json
         return Response(Jsoner.serialize(pred))
     except Exception as e:
@@ -86,6 +86,7 @@ async def predict(request):
     routes=[
         Route("/", health, methods=["GET"]),
         Route("/health", health, methods=["GET"]),
+        Route("/", predict, methods=["POST"]),
         Route("/predict", predict, methods=["POST"]),
     ],
     on_startup=[some_startup_task],

tests/integ/test_container.py

@@ -49,14 +49,14 @@ def verify_task(container: DockerClient, task: str, port: int = 5000, framework:
         or task == "zero-shot-image-classification"
     ):
         prediction = requests.post(
-            f"{BASE_URL}/predict", data=task2input[task], headers={"content-type": "image/x-image"}
+            f"{BASE_URL}", data=task2input[task], headers={"content-type": "image/x-image"}
         ).json()
     elif task == "automatic-speech-recognition" or task == "audio-classification":
         prediction = requests.post(
-            f"{BASE_URL}/predict", data=task2input[task], headers={"content-type": "audio/x-audio"}
+            f"{BASE_URL}", data=task2input[task], headers={"content-type": "audio/x-audio"}
         ).json()
     else:
-        prediction = requests.post(f"{BASE_URL}/predict", json=input).json()
+        prediction = requests.post(f"{BASE_URL}", json=input).json()
     assert task2validation[task](result=prediction, snapshot=task2output[task]) is True
 
 
@@ -193,7 +193,7 @@ def test_cpu_container_custom_pipeline(repository_id) -> None:
         BASE_URL = "http://localhost:5000"
         wait_for_container_to_be_ready(BASE_URL)
         payload = {"inputs": "this is a test"}
-        prediction = requests.post(f"{BASE_URL}/predict", json=payload).json()
+        prediction = requests.post(f"{BASE_URL}", json=payload).json()
         assert prediction == payload
         # time.sleep(5)
         container.stop()