MatchCode approximate code search tests

This repository is a test suite for approximate code search including AI-generated code search.

• Homepage: https://github.com/aboutcode-org/matchcode-tests/
• Related repos:
  • https://github.com/aboutcode-org/purldb
  • https://github.com/aboutcode-org/matchcode-toolkit
  • https://github.com/aboutcode-org/ai-gen-code-search

Usage

• Clone this repository

• In the clone, run make dev

• run . venv/bin/activate

• run the full test suite with:

  pytest -vvs tests
  

This is designed to run only on Linux.

Tests

test_matchcode.py uses the dataset "Analyzing the Dependability of Large Language Models for Code Clone Generation" (https://zenodo.org/records/11398703). This dataset contains code solutions to problems from LeetCode that have been generated by AI from an original solution. The tests in test_matchcode.py compare the original solutions to the different variations of AI generated solutions, where we compare Hamming distances and detected ngrams from the different solutions.

License

• the data is under a CC-BY-4.0 license
• the code is under the Apache-2.0 license