add startup event in main.py

starter/main.py

@@ -17,6 +17,7 @@
 
 
 app = FastAPI()
+MODEL_CONFIGS = {}
 
 
 class InferenceRequest(BaseModel):
@@ -48,19 +49,28 @@ async def create_item(item: dict):
     return item
 
 
+@app.on_event("startup")
+async def startup_event():
+    cwd_p = os.getcwd()
+    MODEL_CONFIGS['trained_model'] = \
+        joblib.load(f"{cwd_p}/starter/model/model_trained.joblib")
+    MODEL_CONFIGS['encoder'] = \
+        joblib.load(f"{cwd_p}/starter/model/encoder.joblib")
+    MODEL_CONFIGS['labels'] = \
+        joblib.load(f"{cwd_p}/starter/model/lb.joblib")
+
+
 @app.post('/predict')
 async def get_prediction(request_data: InferenceRequest):
-    cwd_p = os.getcwd()
-    trained_model = joblib.load(f"{cwd_p}/starter/model/model_trained.joblib")
-    encoder = joblib.load(f"{cwd_p}/starter/model/encoder.joblib")
-    labels = joblib.load(f"{cwd_p}/starter/model/lb.joblib")
     request_dict = request_data.dict(by_alias=True)
     request_df = pd.DataFrame(request_dict, index=[0])
     processed_data, _, _, _ = process_data(
         request_df, categorical_features=CAT_FEATURES, label=None,
-        training=False, encoder=encoder, lb=labels
+        training=False, encoder=MODEL_CONFIGS['encoder'],
+        lb=MODEL_CONFIGS['labels']
     )
-    preds = inference(trained_model, np.array(processed_data))
+    preds = inference(MODEL_CONFIGS['trained_model'],
+                      np.array(processed_data))
     if preds[0]:
         pred_cat = '>50K'
     else:

starter/test_main.py

@@ -1,20 +1,26 @@
 from fastapi.testclient import TestClient
 import logging
+import pytest
 from main import app
 
 
 logging.basicConfig(level=logging.INFO, format="%(asctime)-15s %(message)s")
 logger = logging.getLogger()
-client = TestClient(app)
 
 
-def test_welcome():
+@pytest.fixture
+def client():
+    with TestClient(app) as clt:
+        yield clt
+
+
+def test_welcome(client):
     req = client.get('/')
     assert req.status_code == 200, "Status code is not 200"
     assert req.json() == "Welcome, this API returns predictions on Salary", "Wrong json output"
 
 
-def test_post():
+def test_post(client):
     sample_dict = {
         "age": 49,
         "workclass": "State-gov",
@@ -24,7 +30,7 @@ def test_post():
     assert response.json() == sample_dict
 
 
-def test_get_prediction_negative():
+def test_get_prediction_negative(client):
     input_dict = {
         "age": 49,
         "workclass": "State-gov",
@@ -47,7 +53,7 @@ def test_get_prediction_negative():
         "Wrong json output"
 
 
-def test_get_prediction_positive():
+def test_get_prediction_positive(client):
     input_dict = {
         "age": 41,
         "workclass": "Private",