langunittest 0.1.1

LLM agent to facilitate the generation of unit test cases.

langunittest: Your Command-line LLM agent for unit test cases.

Problem

From clippy, you can generate unit test cases for Python modules with ease.

Take the following math.py module as an example:

def add(a: float, b: float) -> float:
    return a + b

def div(a: float, b: float) -> float:
    if b == 0.0:
        raise ValueError('Invalid divisor!')
    return a / b

You can generate unit test cases using the following command:

$ pack math.py | clippy 'Create unit test cases for "math.py".'

Example AI response:

import unittest
from math import add, div

class TestMathFunctions(unittest.TestCase):

    # Test cases for add()
    def test_add_positive_numbers(self):
        self.assertAlmostEqual(add(1, 2), 3)

    def test_add_negative_numbers(self):
        self.assertAlmostEqual(add(-1, -2), -3)

    def test_add_mixed_numbers(self):
        self.assertAlmostEqual(add(-1, 2), 1)
        self.assertAlmostEqual(add(1, -2), -1)

    def test_add_with_zero(self):
        self.assertAlmostEqual(add(5, 0), 5)
        self.assertAlmostEqual(add(0, 0), 0)

    def test_add_float_numbers(self):
        self.assertAlmostEqual(add(0.1, 0.2), 0.3)
        self.assertAlmostEqual(add(1.5, 2.5), 4.0)

    # Test cases for div()
    def test_div_positive_numbers(self):
        self.assertAlmostEqual(div(6, 2), 3)

    def test_div_negative_numbers(self):
        self.assertAlmostEqual(div(-6, -2), 3)

    def test_div_by_zero(self):
        with self.assertRaisesRegex(ValueError, 'Invalid divisor!'):
            div(10, 0)
        with self.assertRaisesRegex(ValueError, 'Invalid divisor!'):
            div(0, 0)

if __name__ == '__main__':
    unittest.main(argv=['first-arg-is-ignored'], exit=False)

Current Limitations: Despite the convenience, this approach comes with several drawbacks:

• ⚠️ The generated test cases are printed to the console and must be copied manually into a file.
• ❌ The test cases may contain errors and are not always executable out of the box.
• 🧠 There’s no memory or iterative refinement; you cannot adjust or fine-tune the generated tests interactively.
• 📉 Test coverage of the generated code is not reported or optimized.

Introducing langunittest

To address these limitations, we propose langunittest, a module designed to provide an interactive agent that works with you to generate, refine, and validate unit tests more efficiently.

Key benefits include:

• 💾 Output saved directly to files—no copy-paste needed.
• ✅ Automatically verified executable test cases.
• 🔄 Interactive back-and-forth refinement with memory and state.
• 📈 Built-in test coverage analysis to ensure sufficient coverage.

With langunittest, the goal is to generate reliable, high-quality, and maintainable unit test suites through an iterative and intelligent workflow.

Usage

Below we will demo how to use langunittest to help use create unit test cases.

Firstly, let's use below command to launch Python REPL:

$ python -m asyncio
>>>

Then we enter below line to launch agent to write test cases:

# Import langunittest:
>>> from langunittest.agents import agent

# Instantiated an agent:
>>> tc_agent = await agent.TCAgent.create()

# Start to interact with agent:
>>> await tc_agent.interact_async()

Not It is ready interact with agent to write test cases:

1. Setup project root path:

[User] $ Set project root path as /tmp
...
[Agent] # OK. The project root path has been set to /tmp.

2. Show the source code of target module:

[User] $ Show me the source code of utils/my_math.py
...
[Agent] # The source code is as:
'''python
def add(a: int, b: int) -> int:
  """Sums up the input `a` and `b`.

  Args:
    a: First value to add
    b: Second value to add

  Returns:
    Return value of `a + b`.
  """
  return a + b
'''

3. Ask agent to create tests:

[User] $ Please create test cases for it.
...
----------------------------------------------------------------------
Ran 4 tests in 0.000s

OK
[Agent] # Great! All test cases passed. Here is the test code I generated:

'''python
import unittest
from utils.my_math import add

class TestMyMath(unittest.TestCase):

    def test_add_positive_numbers(self):
        self.assertEqual(add(2, 3), 5)

    def test_add_negative_numbers(self):
        self.assertEqual(add(-2, -3), -5)

    def test_add_mixed_numbers(self):
        self.assertEqual(add(2, -3), -1)

    def test_add_zero(self):
        self.assertEqual(add(2, 0), 2)

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

4. Ask agent to modify tests:

[User] $ Pleaes add one line comment to each test cases.
...
----------------------------------------------------------------------
Ran 4 tests in 0.000s

OK
[Agent] # Excellent, all tests passed! Here's the updated test code with comments:

'''python
import unittest
from utils.my_math import add

class TestMyMath(unittest.TestCase):

    def test_add_positive_numbers(self):
        # Test adding two positive numbers
        self.assertEqual(add(2, 3), 5)

    def test_add_negative_numbers(self):
        # Test adding two negative numbers
        self.assertEqual(add(-2, -3), -5)

    def test_add_mixed_numbers(self):
        # Test adding a positive and a negative number
        self.assertEqual(add(2, -3), -1)

    def test_add_zero(self):
        # Test adding a number with zero
        self.assertEqual(add(2, 0), 2)

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