Fix/pydantic validation settings

pyproject.toml

@@ -95,14 +95,12 @@ omit = ["__main__.py"]
 
 [tool.mypy]
 pretty = true
-strict = false
+strict = true
 python_version = "3.12"
-check_untyped_defs = false
+check_untyped_defs = true
 ignore_missing_imports = true
 plugins = ["pandera.mypy", "pydantic.mypy"]
-exclude = ["src/regression_model_template/jobs/evaluations.py",
-"src/regression_model_template/jobs/training.py",
-"src/regression_model_template/io/registries.py"]
+exclude = []
 
 [tool.pytest.ini_options]
 addopts = "--verbosity=2"

src/regression_model_template/controller/kafka_app.py

@@ -13,7 +13,7 @@
 from fastapi import FastAPI, HTTPException
 from pydantic import BaseModel
 
-from confluent_kafka import Producer, Consumer, KafkaError
+from confluent_kafka import Producer, Consumer, KafkaError, Message
 
 from regression_model_template.core.schemas import InputsSchema, Outputs
 from regression_model_template.io import services, registries
@@ -65,9 +65,9 @@ class PredictionRequest(BaseModel):
         "registered": [0, 50, 100, 150],
     }
 
-    def model_validate(self):
+    def validate_schema(self) -> pd.DataFrame:
         """Validates the input data against InputsSchema."""
-        return InputsSchema.validate(pd.DataFrame([self.input_data]))
+        return InputsSchema.validate(pd.DataFrame([self.input_data]))  # type: ignore[return-value]
 
 
 class PredictionResponse(BaseModel):
@@ -96,7 +96,7 @@ def __init__(
         self.producer: Producer | None = None
         self.consumer: Consumer | None = None
 
-    def delivery_report(self, err, msg):
+    def delivery_report(self, err: KafkaError | None, msg: Message) -> None:
         """Called once for each message produced to indicate delivery result."""
         if err is not None:
             logger.error(f"Message delivery failed: {err}")
@@ -155,15 +155,15 @@ def _consume_messages(self) -> None:
             self._process_message(msg)
         self._close_consumer()
 
-    def _poll_message(self):
+    def _poll_message(self) -> Message | None:
         """Poll message from Kafka consumer."""
         if self.consumer:
             return self.consumer.poll(1.0)
         else:
             logger.error("Kafka consumer is not initialized.")
             return None
 
-    def _handle_message_error(self, msg) -> bool:
+    def _handle_message_error(self, msg: Message) -> bool:
         """Handle errors in polled messages."""
         if msg.error().code() == KafkaError._PARTITION_EOF:
             logger.debug("Reached end of partition.")
@@ -172,7 +172,7 @@ def _handle_message_error(self, msg) -> bool:
             logger.error(f"Consumer error: {msg.error()}")
             return False
 
-    def _process_message(self, msg) -> None:
+    def _process_message(self, msg: Message) -> None:
         """Process a valid Kafka message."""
         predictionresponse: PredictionResponse = PredictionResponse()
         try:
@@ -251,12 +251,12 @@ async def predict(request: PredictionRequest) -> PredictionResponse:  # Use glob
 
 
 @app.get("/health", summary="Health Check", tags=["System"])
-async def health_check():
+async def health_check() -> dict[str, str]:
     """Simple health check endpoint to verify that the service is running."""
     return {"status": "healthy"}
 
 
-def main():
+def main() -> None:
     global fastapi_kafka_service
     # Configuration
     alias_or_version: str | int = "Champion"

src/regression_model_template/core/metrics.py

@@ -90,7 +90,9 @@ def eval_fn(predictions: pd.Series[int], targets: pd.Series[int]) -> MlflowMetri
             score = self.score(targets=score_targets, outputs=score_outputs)
             return MlflowMetric(aggregate_results={self.name: score * sign})
 
-        return mlflow.metrics.make_metric(eval_fn=eval_fn, name=self.name, greater_is_better=self.greater_is_better)
+        return mlflow.metrics.make_metric(  # type: ignore[no-any-return]
+            eval_fn=eval_fn, name=self.name, greater_is_better=self.greater_is_better
+        )
 
 
 class SklearnMetric(Metric):
@@ -141,4 +143,6 @@ def to_mlflow(self) -> MlflowThreshold:
         Returns:
             MlflowThreshold: the mlflow threshold.
         """
-        return MlflowThreshold(threshold=self.threshold, greater_is_better=self.greater_is_better)
+        return MlflowThreshold(  # type: ignore[no-untyped-call]
+            threshold=self.threshold, greater_is_better=self.greater_is_better
+        )

src/regression_model_template/io/osvariables.py

@@ -23,3 +23,4 @@ class Config:
         case_sensitive: bool = False  # Optional: make env var lookup case-insensitive
         env_file: str = ".env"  # Enable reading from .env file
         env_file_encoding: str = "utf-8"
+        extra = "ignore"

src/regression_model_template/io/registries.py

@@ -102,7 +102,7 @@ class CustomSaver(Saver):
 
     KIND: T.Literal["CustomSaver"] = "CustomSaver"
 
-    class Adapter(mlflow.pyfunc.PythonModel):  # type: ignore[misc]
+    class Adapter(mlflow.pyfunc.PythonModel):  # type: ignore[misc, name-defined]
         """Adapt a custom model to the Mlflow PyFunc flavor for saving operations.
 
         https://mlflow.org/docs/latest/python_api/mlflow.pyfunc.html?#mlflow.pyfunc.PythonModel
@@ -118,7 +118,7 @@ def __init__(self, model: models.Model):
 
         def predict(
             self,
-            context: mlflow.pyfunc.PythonModelContext,
+            context: mlflow.pyfunc.PythonModelContext,  # type: ignore[name-defined]
             model_input: schemas.Inputs,
             params: dict[str, T.Any] | None = None,
         ) -> schemas.Outputs:
@@ -137,7 +137,7 @@ def predict(
 
     def save(self, model: models.Model, signature: signers.Signature, input_example: schemas.Inputs) -> Info:
         adapter = CustomSaver.Adapter(model=model)
-        return mlflow.pyfunc.log_model(
+        return mlflow.pyfunc.log_model(  # type: ignore[no-any-return]
             python_model=adapter,
             signature=signature,
             artifact_path=self.path,
@@ -163,10 +163,10 @@ def save(
         model: models.Model,
         signature: signers.Signature,
         input_example: schemas.Inputs | None = None,
-    ) -> mlflow.entities.model_registry.ModelVersion:
+    ) -> Info:
         builtin_model = model.get_internal_model()
         module = getattr(mlflow, self.flavor)
-        return module.log_model(
+        return module.log_model(  # type: ignore[no-any-return]
             builtin_model, artifact_path=self.path, signature=signature, input_example=input_example
         )
 
@@ -222,7 +222,7 @@ class CustomLoader(Loader):
     class Adapter(Loader.Adapter):
         """Adapt a custom model for the project inference."""
 
-        def __init__(self, model: mlflow.pyfunc.PyFuncModel) -> None:
+        def __init__(self, model: mlflow.pyfunc.PyFuncModel) -> None:  # type: ignore[name-defined]
             """Initialize the adapter from an mlflow pyfunc model.
 
             Args:
@@ -254,7 +254,7 @@ class BuiltinLoader(Loader):
     class Adapter(Loader.Adapter):
         """Adapt a builtin model for the project inference."""
 
-        def __init__(self, model: mlflow.pyfunc.PyFuncModel) -> None:
+        def __init__(self, model: mlflow.pyfunc.PyFuncModel) -> None:  # type: ignore[name-defined]
             """Initialize the adapter from an mlflow pyfunc model.
 
             Args:

src/regression_model_template/jobs/evaluations.py

@@ -81,10 +81,10 @@ def run(self) -> base.Locals:
             logger.debug("- Targets lineage: {}", targets_lineage.to_dict())
             # dataset
             logger.info("Create dataset: inputs & targets")
-            dataset = mlflow.data.from_pandas(
+            dataset = mlflow.data.from_pandas(  # type: ignore[attr-defined]
                 df=pd.concat([inputs, targets], axis="columns"),
                 name="evaluation",
-                source=f"{inputs_lineage.source.uri} & {targets_lineage.source.uri}",
+                source=f"{inputs_lineage.source.uri} & {targets_lineage.source.uri}",  # type: ignore[attr-defined]
                 targets=schemas.TargetsSchema.cnt,
             )
             logger.debug("- Dataset: {}", dataset.to_dict())
@@ -104,7 +104,7 @@ def run(self) -> base.Locals:
             logger.debug("- Validation thresholds: {}", validation_thresholds)
             # evaluations
             logger.info("Compute evaluations: {}", self.model_type)
-            evaluations = mlflow.evaluate(
+            evaluations = mlflow.evaluate(  # type: ignore[no-untyped-call]
                 data=dataset,
                 model=model_uri,
                 model_type=self.model_type,

src/regression_model_template/settings.py

@@ -10,7 +10,7 @@
 # %% SETTINGS
 
 
-class Settings(pdts.BaseSettings, strict=True, frozen=True, extra="forbid"):
+class Settings(pdts.BaseSettings, strict=True, frozen=True, extra="ignore"):
     """Base class for application settings.
 
     Use settings to provide high-level preferences.