测试类中使用pd.read_csv()读取命令行传入路径的CSV文件失败

Key Issues in Original Code #

• Pytest misinterprets positional arguments: Passing extra positional arguments to pytest makes it treat them as test files/names, causing the "not found" error.
• Early dataset loading: The dataset is loaded when the test class is defined—before pytest parses command line arguments—so the path isn't correctly retrieved at that stage.
• Incorrect sys.path usage: The CSV file path doesn't belong in sys.path (this list is for module imports, not data files).

Corrected Test Code #

import sys
import numpy as np
import pandas as pd
import pytest

def pytest_addoption(parser):
    # Define custom command line options for module and CSV paths
    parser.addoption("--module-path", action="store", required=True, 
                     help="Path to the directory containing training.py")
    parser.addoption("--csv-path", action="store", required=True, 
                     help="Full path to the customer data CSV file")

@pytest.fixture(scope="session", autouse=True)
def setup_module_path(request):
    # Add the module path to sys.path to enable importing from training.py
    module_path = request.config.getoption("--module-path")
    sys.path.insert(0, module_path)

@pytest.fixture(scope="class")
def dataset(request):
    # Load dataset once per test class using the command line-specified CSV path
    csv_path = request.config.getoption("--csv-path")
    return pd.read_csv(csv_path, parse_dates=["InvoiceDate"])

# Import after adding module path to sys.path
from training import CustomerSegmentation

class TestCustomerSegmentation:
    def test_dataset_loaded(self, dataset):
        # Verify dataset is successfully loaded
        assert not dataset.empty, "Dataset should not be empty"
    
    # Add your other test methods here, using the dataset fixture as needed

Correct Command to Run #

python3 -m pytest test_training.py --module-path "customer-segmentation-v1" --csv-path "customer-segmentation-v1/customer_data.csv"

Explanation #

• Custom pytest options: The pytest_addoption hook defines --module-path and --csv-path so pytest recognizes these as configuration parameters, not test items.
• Autouse fixture: setup_module_path adds the module directory to sys.path once at the start of the test session, ensuring the CustomerSegmentation import works.
• Dataset fixture: The dataset fixture loads the CSV file after pytest parses command line arguments, guaranteeing the correct path is used. It runs once per test class for efficiency.

内容的提问来源于stack exchange，提问作者MahrukhHassan