testgen command

The testgen command is the primary feature of testcode.ai (BETA), allowing you to generate comprehensive unit tests for your Python code. This guide covers all aspects of using this command effectively.

Basic usage

The most basic usage of the testgen command requires only a source file:

probe testgen --source path/to/your/file.py

This will:

1. Analyze the source file using static code analysis
2. Display a list of methods found in the file
3. Let you select a method to test
4. Generate a comprehensive unit test for the selected method
5. Save the test to the default test directory (tests/)

Example

probe testgen --source src/utils/calculator.py

Command options

The testgen command supports several options to customize its behavior:

Required options

Optional options

Example with options

probe testgen --source src/utils/calculator.py --test-dir custom_tests --root-dir /path/to/project --debug

Method selection

When you run the testgen command, testcode.ai will analyze your source file and display a list of available methods. You can navigate this list using:

• Up/Down arrow keys: Move through the list
• Enter: Select the highlighted method

If your file contains only one method, it will be automatically selected.

Method selection interface

The method selection interface looks like this:

Test generation process

After selecting a method, testcode.ai will:

1. Analyze the method: Examine its signature, body, and dependencies using static code analysis
2. Extract dependencies: Identify imports and external dependencies
3. Generate a test template: Create a structure for the test
4. Use LLM to generate tests: Send the template to an LLM for completion
5. Process the response: Format and validate the generated test
6. Save the test file: Write the test to the specified directory

This process typically takes a few seconds, depending on the complexity of the method and the LLM response time.

Privacy during test generation

When generating tests, testcode.ai:

• Performs all code analysis locally on your machine
• Sends code directly to OpenAI's API (using the o3-mini model by default)
• Never stores or transmits your code through testcode.ai servers
• Maintains a strict zero data retention policy

Progress indicators

testcode.ai displays progress indicators during test generation:

Output files

The generated test files follow standard pytest naming conventions:

• For a file named calculator.py, the test file will be test_calculator.py
• For a method named add, the test function will be test_add

Test file structure

The generated test files include:

• Proper imports for pytest and the module being tested
• Mock objects for dependencies
• Test cases for normal operation
• Test cases for edge cases and error conditions
• Appropriate assertions

Example test file:

import pytest
from unittest.mock import patch, MagicMock
from src.utils.calculator import add

def test_add():
    # Test normal operation
    assert add(1, 2) == 3
    assert add(-1, 1) == 0
    
    # Test edge cases
    assert add(0, 0) == 0
    assert add(999999, 1) == 1000000
    
    # Test type handling
    with pytest.raises(TypeError):
        add("1", 2)

Best practices

To get the most out of the testgen command:

Organize your code

• Keep methods focused on a single responsibility
• Use clear naming for methods and parameters
• Document your code with docstrings

Select appropriate methods

• Start with simpler methods before complex ones
• Focus on methods with clear inputs and outputs
• Test utility functions before testing code that depends on them

Review and refine

• Always review generated tests
• Add additional test cases if needed
• Refine mocks for complex dependencies

Troubleshooting

Common issues

Method not found

If testcode.ai doesn't find any methods in your file:

• Ensure the file contains Python functions or methods
• Check that functions are properly defined (not just imported)
• Verify the file path is correct

Test generation fails

If test generation fails:

• Check your internet connection (required for LLM access)
• Verify your license is valid
• Try running with --debug for more detailed logs

Generated tests don't run

If the generated tests don't run:

• Ensure pytest is installed (pip install pytest)
• Check that the imports in the test file are correct
• Verify that dependencies are available

Next steps

After generating tests with the testgen command, you might want to:

• Run your tests with pytest: pytest tests/
• Learn more about how testcode.ai works