TestGenie

Accelerate your test driven development by using the power of AI to automatically write unit tests for your code.

Getting Started:

Assuming that you have a supported Python installation (3.x), you can start by cloning the repository:

git clone https://github.com/tsbrandon1010/LLM-Unit-Tests

Or by downloading the repository as a ZIP file:

Next, we can get the required packages:

pip install -r requirements.txt

Using the tool

After downloading the tool and package dependencies, you must create an API key with OpenAI.
More information can be found here.

Within the ~/LLM-Unit-Tests/main directory, create a file called .env. This is where you will put your OpenAI API key (your key will go within the quotations).

We should now be able to use the tool:
python auto_test.py -i <input> -o <output> (-r <max retries>) (-s) -l <language>

Alternatively, you can use auto_test.exe in the /main/ directory, and skip installing Python and the dependencies.
auto_test -i <input> -o <output> (-r <max retries>) (-s) -l <language>

-i: The path of the file for which the tool will generate a unit test.
-o: A path to the file where the generated unit test will be saved.
-r: The maximum number of times to re-generate the unit test if the test fails (optional).
-s: Sparse tests. By default, the program will generate a unit test for the entire file that is passed. Include this flag if you would like to pick and choose the functions to generate tests for (optional). NOTE: CURRENTLY ONLY SUPPORTS PYTHON
-l: Specify whether your language (currently only supports Python and C#).

Running through examples

There is an example program we are going to use called is-prime.py that is located in ~/programs/prime-checker/main

import math
def is_prime(num: int) -> bool:
    '''Check if a number (num) is prime or not.'''
    if num > 1:
        for i in range(2, int(num/2)+1):
            if (num % i) == 0:
                return False
        else:
            return True
    else:
        return False

To create a unit test for this simple example, we can do the following:
python auto_test.py -i ~/programs/prime-checker/main/is-prime.py -o ~/programs/prime-checker -r 0

The generated test might look like the following:

import unittest
import math

import math
def is_prime(num: int) -> bool:
    '''Check if a number (num) is prime or not.'''
    if num > 1:
        for i in range(2, int(num/2)+1):
            if (num % i) == 0:
                return False
        else:
            return True
    else:
        return False

class TestIsPrime(unittest.TestCase):
    # Test for prime numbers
    def test_prime_numbers(self):
        self.assertTrue(is_prime(2))
    
    # Test for non-prime numbers
    def test_non_prime_numbers(self):
        self.assertFalse(is_prime(0))

if __name__ == '__main__':
    unittest.main()

Common Problems:

TLDR: If you want the highest level of accuracy with your generated unit tests, it is recommended that you specify at least 1 retry attempt (-r 1).

A common problem with the current version of the test generator is that it will forget to write the function you are testing in the unit test, or attempt to import the function from an invalid path. Setting -r to a value > 1 will normally catch this, but you might have to import or define the function in the generated unit test yourself.