Catch and format prediction result errors

Catch and format as a JSON response, exceptions that occur when
converting the result of model.predict() to the expected Response type
(as documented in the mlflow model metadata)

Fixes #12

fastapi_mlflow/applications.py

@@ -13,7 +13,9 @@
 from fastapi.responses import JSONResponse
 
 from mlflow.pyfunc import PyFuncModel  # type: ignore
-from fastapi_mlflow.predictors import build_predictor, PyFuncModelPredictError
+
+from fastapi_mlflow.exceptions import DictSerialisableException
+from fastapi_mlflow.predictors import build_predictor
 
 
 def build_app(pyfunc_model: PyFuncModel) -> FastAPI:
@@ -29,7 +31,7 @@ def build_app(pyfunc_model: PyFuncModel) -> FastAPI:
     )
 
     @app.exception_handler(Exception)
-    def handle_exception(_: Request, exc: PyFuncModelPredictError):
+    def handle_exception(_: Request, exc: DictSerialisableException):
         return JSONResponse(
             status_code=500,
             content=exc.to_dict(),

fastapi_mlflow/exceptions.py

@@ -0,0 +1,22 @@
+# -*- coding: utf-8 -*-
+"""Exceptions.
+
+Copyright (C) 2023, Auto Trader UK
+Created 17. Apr 2023
+
+"""
+
+
+class DictSerialisableException(Exception):
+    """An Exception wrapper for formatting."""
+
+    def __init__(self, name: str, message: str):
+        self.name = name
+        self.message = message
+
+    @classmethod
+    def from_exception(cls, exc: Exception):
+        return cls(name=exc.__class__.__name__, message=str(exc))
+
+    def to_dict(self):
+        return {"name": self.name, "message": self.message}

fastapi_mlflow/predictors.py

@@ -21,16 +21,7 @@
     build_model_fields,
     MLFLOW_SIGNATURE_TO_PYTHON_TYPE_MAP,
 )
-
-
-class PyFuncModelPredictError(Exception):
-    def __init__(self, exc: Exception):
-        super().__init__()
-        self.error_type_name = exc.__class__.__name__
-        self.message = str(exc)
-
-    def to_dict(self):
-        return {"name": self.error_type_name, "message": self.message}
+from fastapi_mlflow.exceptions import DictSerialisableException
 
 
 @no_type_check  # Some types here are too dynamic for type checking
@@ -80,11 +71,10 @@ def request_to_dataframe(request: Request) -> pd.DataFrame:
     def predictor(request: Request) -> Response:
         try:
             predictions = model.predict(request_to_dataframe(request))
+            response_data = convert_predictions_to_python(predictions)
+            return Response(data=response_data)
         except Exception as exc:
-            raise PyFuncModelPredictError(exc) from exc
-
-        response_data = convert_predictions_to_python(predictions)
-        return Response(data=response_data)
+            raise DictSerialisableException.from_exception(exc) from exc
 
     return predictor  # type: ignore
 

tests/test_predictors.py

@@ -7,6 +7,7 @@
 from datetime import datetime
 from inspect import signature
 from typing import Union
+from unittest.mock import patch
 
 import numpy as np
 import numpy.typing as npt
@@ -15,11 +16,8 @@
 import pytest
 from mlflow.pyfunc import PyFuncModel  # type: ignore
 
-from fastapi_mlflow.predictors import (
-    build_predictor,
-    convert_predictions_to_python,
-    PyFuncModelPredictError,
-)
+from fastapi_mlflow.exceptions import DictSerialisableException
+from fastapi_mlflow.predictors import build_predictor, convert_predictions_to_python
 
 
 def test_build_predictor_returns_callable(
@@ -152,18 +150,33 @@ def test_predictor_handles_model_returning_nan(
             assert item.b is None
 
 
-def test_predictor_exception_from_model_raised_from(
+def test_predictor_raises_custom_wrapped_exception_on_model_error(
     model_input: pd.DataFrame, pyfunc_model_value_error: PyFuncModel
 ):
     predictor = build_predictor(pyfunc_model_value_error)
 
     request_type = signature(predictor).parameters["request"].annotation
     request_obj = request_type(data=model_input.to_dict(orient="records"))
 
-    with pytest.raises(PyFuncModelPredictError):
+    with pytest.raises(DictSerialisableException):
         predictor(request_obj)
 
 
+def test_predictor_raises_custom_wrapped_exception_on_model_output_conversion(
+    model_input: pd.DataFrame,
+    pyfunc_model_ndarray: PyFuncModel,
+):
+    predictor = build_predictor(pyfunc_model_ndarray)
+
+    request_type = signature(predictor).parameters["request"].annotation
+    request_obj = request_type(data=model_input.to_dict(orient="records"))
+
+    with patch.object(pyfunc_model_ndarray, "predict") as predict:
+        predict.return_value = ["Fail!"]
+        with pytest.raises(DictSerialisableException):
+            predictor(request_obj) * len(model_input)
+
+
 def test_convert_predictions_to_python_ndarray():
     predictions = np.array([1, 2, 3, np.nan])
     response_data = convert_predictions_to_python(predictions)