FixKit

FixKit is a collection of various automated program repair approaches and techniques. It is designed to be a flexible and extensible framework for experimenting with different repair strategies. The framework is written in Python for Python and is designed to be easy to use and extend.

Installation

To install FixKit, simply get the package from PyPI:

pip install fixkit

To install the latest version from the repository, use:

pip install git+https://github.com/smythi93/fixkit.git

Usage

FixKit is a library and can be used directly in your Python code.

from fixkit.repair.pygenprog import PyGenProg
from fixkit.localization.coverage import CoverageLocalization

repair = PyGenProg.from_source(
  src= "subjects/middle",
  excludes=["tests.py"],
  localization=CoverageLocalization(
    "subjects/middle",
    cov="middle",
    metric="Ochiai",
    tests=["tests.py"],
  ),
  population_size=40,
  max_generations=10,
)
patches = repair.repair()

Components

Individual components can be swapped out and replaced with custom implementations. The following components are currently available:

• Fault Localization: Determines the location of the fault in the source code.
• Repair Approach: Generates patches to fix the fault in the source code.
• Search Strategy: Searches for patches in the search space.
• Fitness Function: Evaluates the fitness of patches based on a given metric.
• Selection Strategy: Selects patches for the next generation based on their fitness.
• Crossover Strategy: Combines patches to create new patches.
• Mutation Operators: Mutates patches to create new patches.
• Minimization Strategy: Minimizes patches to reduce their size.