Extending PyTest Suite

.github/workflows/ci.yml

@@ -42,4 +42,4 @@ jobs:
         shell: bash -l {0}  # Ensures the Conda environment is properly loaded
         run: |
           export PYTHONPATH=$PYTHONPATH:$(pwd)
-          pytest -v tests/tests_encoding.py
+          pytest

data/hooks/encoding.py

@@ -396,8 +396,10 @@ def fit(self, x):
         # Iterate through the columns in the OHE in our specified order to identity clusters of infrequent groups
         col_group = []
         for c in self.order:
-            # If this column doesn't exist in the OHE, skip it entirely
+            # If this column doesn't exist in the OHE, treat it as frequency 0 and add a homogenous column to the ohe_df
             if c not in ohe_df.columns:
+                col_group.append(c)
+                ohe_df[c] = 0
                 continue
 
             # Otherwise, append it to our current column group
@@ -475,7 +477,7 @@ def ohe_to_ladder(self, x_df: pd.DataFrame):
             rung_val = rung_df.any(axis="columns")
 
             # Generate a string representing the combination of columns which will be grouped
-            group_str = "|".join(g)
+            group_str = "|".join([str(x) for x in g])
 
             # Generate the corresponding column's name, but only if we had a prior group (the first col will be dropped)
             col_name = ""

tests/README.md

@@ -0,0 +1,19 @@
+This package uses PyTest to run its tests, but due to the way its structured there is a little bit of setup required:
+
+1. Ensure all required packages are installed (see `environment.yml` in the directory above this one)
+2. Activate the environment you installed the packages into it (if installing to environment which is not your "base")
+3. Add the path of Modular Optuna ML to the Python path, allowing PyTest to see its contents.
+
+    ```bash
+    cd ..
+    export PYTHONPATH=$PYTHONPATH:"$PWD"
+    cd tests
+    ```
+
+4. Run the tests you want. For example, to run all tests:
+
+    ```bash
+    pytest
+    ```
+
+If you are running PyTest through an IDE (such as PyCharm), you may also need to mark the `tests` directory as a testing directory to do in-place runs.

tests/conftest.py

@@ -0,0 +1,81 @@
+import json
+import tempfile
+from pathlib import Path
+from typing import Self
+
+import numpy as np
+import pandas as pd
+import pytest
+
+from config.data import DataConfig
+from data.base import BaseDataManager
+
+
+class DummyDataManager(BaseDataManager):
+
+    def __init__(self, new_data: pd.DataFrame, **kwargs):
+        # Initiate superclass stuff
+        super().__init__(**kwargs)
+
+        # Set the DataFrame managed by this object to be the dataframe provided
+        self._data = new_data
+
+    @property
+    def data(self) -> pd.DataFrame:
+        return self._data
+
+    def _replace_data(self, new_df: pd.DataFrame):
+        self._data = new_df
+
+    def shallow_copy(self) -> Self:
+        return DummyDataManager(self._data)
+
+    @classmethod
+    def from_config(cls, config: dict) -> Self:
+        raise NotImplementedError(
+            f"This is a dummy data manager meant for automated testing; it cannot be built from a config file."
+        )
+
+    def pre_split(self, is_cross: bool, targets: Self = None) -> Self:
+        raise NotImplementedError(
+            f"This is a dummy data manager meant for automated testing; it should not be split."
+        )
+
+    def split(self, train_idx: np.ndarray, test_idx: np.ndarray, train_target: Self, test_target: Self,
+              is_cross: bool = True) -> (Self, Self):
+        raise NotImplementedError(
+            f"This is a dummy data manager meant for automated testing; it should not be split."
+        )
+
+    def __len__(self):
+        return self._data.shape[0]
+
+
+@pytest.fixture
+def iris_data_config():
+    """
+    Fixture that returns a DataConfig instance for the Iris dataset.
+    The tsv file is loaded as a TabularDataManager instance: data_config.data_manager
+    Loaded data is stored as a pandas DataFrame: data_config.data_manager.data
+    """
+    config_dict = {
+        "label": "IrisTesting",
+        "format": "tabular",
+        "data_source": str(Path(__file__).parent.resolve() / 'testing_files' / 'iris_data' / 'iris_testing.tsv'),
+        "separator": "\t",
+        "index": "id",
+    }
+
+    # Create a temporary file for the config
+    with tempfile.NamedTemporaryFile(mode='w', delete=False, suffix=".json") as tmp_file:
+        json.dump(config_dict, tmp_file)
+        tmp_config_path = Path(tmp_file.name)  # Convert string path to Path object
+
+    # Ensure the file exists before using it
+    assert tmp_config_path.exists(), f"Temporary config file {tmp_config_path} was not created."
+
+    data_config = DataConfig.from_json_file(tmp_config_path)
+
+    return data_config
+
+