review

Dockerfile

@@ -1,23 +1,3 @@
-# FROM python:3.12.4
-#
-# # Set working directory
-# WORKDIR /app
-#
-# # Copy dependencies
-# COPY requirements.txt .
-#
-# # Install dependencies
-# RUN pip install --upgrade pip && pip install --no-cache-dir -r requirements.txt
-#
-# # Copy the rest of the app
-# COPY . .
-#
-# # Expose API port
-# EXPOSE 7860
-#
-# # Run the app with uvicorn
-# CMD ["uvicorn", "app:app", "--host", "0.0.0.0", "--port", "7860"]
-
 FROM python:3.12-slim
 
 # (Optional but useful for xgboost/lightgbm/numba, etc.)

app.py

@@ -12,16 +12,12 @@ from pydantic import BaseModel, Field, field_validator
 from unidecode import unidecode
 
 
-# =======================
 # Configuration
-# =======================
 PORT = int(os.getenv("PORT", 7860))
 LOCAL_MODEL_PATH = os.getenv("MODEL_PATH", "model_bundle/model")
 
 
-# =======================
 # Helper functions
-# =======================
 def load_features_from_artifacts(model_dir: str) -> list[str]:
     """
     Attempt to load `features_used.json` generated during training.
@@ -58,26 +54,29 @@ def load_features_from_artifacts(model_dir: str) -> list[str]:
     ]
 
 
-# =======================
 # FastAPI initialization
-# =======================
 app = FastAPI(
     title="🚗 Getaround Pricing API",
     description=(
         "Prédiction du prix journalier de location.\n\n"
         "• Dashboard : https://flodussart-getaroundcertifter.hf.space\n"
-        "• Endpoint ML : POST /predict — body: {\"rows\": [...] } (recommandé) "
-        "ou {\"input\": [[...], ...]} (ordre strict des features).\n"
+        "• Endpoint ML : POST /predict — body: {\"input\": [[...], ...]} (ordre strict des features)"
+        "ou {\"rows\": [...] } .\n"
     ),
     version="1.0",
     docs_url="/docs",
     redoc_url="/redoc",
 )
 
-# Enable CORS (useful when the Streamlit client calls this API from the browser)
+# Authorized origins — Streamlit app and local dev
+origins = [
+    "https://flodussart-getaroundcertifter.hf.space",  # Streamlit dashboard on Hugging Face
+    "http://localhost:8501",                           # local Streamlit testing
+]
+
 app.add_middleware(
     CORSMiddleware,
-    allow_origins=["*"],       # set a specific origin list in production if needed
+    allow_origins=origins,
     allow_credentials=True,
     allow_methods=["*"],
     allow_headers=["*"],
@@ -88,15 +87,13 @@ try:
     model = mlflow.pyfunc.load_model(LOCAL_MODEL_PATH)
 except Exception as e:
     raise RuntimeError(
-        f"❌ Unable to load local MLflow model '{LOCAL_MODEL_PATH}': {e}"
+        f"Unable to load local MLflow model '{LOCAL_MODEL_PATH}': {e}"
     )
 
 FEATURES: list[str] = load_features_from_artifacts(LOCAL_MODEL_PATH)
 
 
-# =======================
-# Pydantic schemas (v2)
-# =======================
+# Pydantic schemas 
 ALLOWED_FUEL = {"diesel", "petrol", "other"}
 ALLOWED_PAINT = {
     "black",

model_bundle/model/artifacts/features_used.json

@@ -0,0 +1,21 @@
+{
+  "numeric": [
+    "mileage",
+    "engine_power"
+  ],
+  "categorical": [
+    "model_key",
+    "fuel_grouped",
+    "paint_color",
+    "car_type"
+  ],
+  "boolean": [
+    "private_parking_available",
+    "has_gps",
+    "has_air_conditioning",
+    "automatic_car",
+    "has_getaround_connect",
+    "has_speed_regulator",
+    "winter_tires"
+  ]
+}