whackthejacker/DataHubHub

1"""2Simple test file for the ML Dataset & Code Generation Manager application.3This script checks basic aspects of the application structure and setup.4"""5 6import os7import sys8import pandas as pd9import numpy as np10from pathlib import Path11 12def test_directory_structure():13    """Test if the required directories exist"""14    # Ensure necessary directories exist15    os.makedirs('database/data', exist_ok=True)16    os.makedirs('assets', exist_ok=True)17    os.makedirs('fine_tuned_models', exist_ok=True)18    19    # Check if directories exist20    assert Path("database").exists() and Path("database").is_dir(), "Database directory not found"21    assert Path("assets").exists() and Path("assets").is_dir(), "Assets directory not found"22    assert Path("fine_tuned_models").exists() and Path("fine_tuned_models").is_dir(), "Fine-tuned models directory not found"23    24    print("✅ Directory structure test passed")25 26def test_css_file():27    """Test if the CSS file exists"""28    css_file = Path("assets/custom.css")29    assert css_file.exists() and css_file.is_file(), "CSS file not found in assets directory"30    31    print("✅ CSS file test passed")32 33def test_huggingface_config():34    """Test if Hugging Face configuration file exists"""35    config_file = Path("huggingface-spacefile")36    assert config_file.exists() and config_file.is_file(), "Hugging Face configuration file not found"37    38    print("✅ Hugging Face configuration test passed")39 40def test_streamlit_config():41    """Test if Streamlit configuration exists"""42    config_dir = Path(".streamlit")43    config_file = config_dir / "config.toml"44    assert config_dir.exists() and config_dir.is_dir(), ".streamlit directory not found"45    assert config_file.exists() and config_file.is_file(), "config.toml file not found in .streamlit directory"46    47    print("✅ Streamlit configuration test passed")48 49def test_sample_dataframe():50    """Test creation of sample dataframes"""51    # Create a sample dataframe52    df = pd.DataFrame({53        "code": ["def hello():", "import numpy as np", "print('Hello')"],54        "comment": ["Function greeting", "Import numpy library", "Print hello message"]55    })56    57    # Test dataframe properties58    assert len(df) == 359    assert list(df.columns) == ["code", "comment"]60    61    print("✅ Sample dataframe test passed")62 63def test_database_initialization():64    """Test if database can be initialized"""65    try:66        from database import init_db67        init_db()68        assert Path("database/data/mlmanager.db").exists(), "Database file was not created"69        print("✅ Database initialization test passed")70    except ImportError:71        print("⚠️ Could not import database module")72        assert False, "Database module not found"73 74def run_tests():75    """Run all tests"""76    print("Running tests for ML Dataset & Code Generation Manager...")77    78    test_directory_structure()79    test_css_file()80    test_huggingface_config()81    test_streamlit_config()82    test_sample_dataframe()83    test_database_initialization()84    85    print("\nAll tests passed! ✅")86 87def test_components_existence():88    """Test if core components directories exist"""89    # Check for components directory90    components_dir = Path("components")91    assert components_dir.exists() and components_dir.is_dir(), "Components directory not found"92    93    # Check for fine_tuning subdirectory94    fine_tuning_dir = components_dir / "fine_tuning"95    assert fine_tuning_dir.exists() and fine_tuning_dir.is_dir(), "Fine-tuning components directory not found"96    97    # Check for essential component files98    assert (components_dir / "code_quality.py").exists(), "Code quality component not found"99    assert (components_dir / "dataset_uploader.py").exists(), "Dataset uploader component not found"100    101    print("✅ Components existence test passed")102 103# Run the tests if executed directly104if __name__ == '__main__':105    run_tests()