CodegrainHouse

Basic Information

CodegrainHouse is a multi-linguistic dataset consisting of 5 popular programming languages, mined from more than 5000 GitHub repositories containing 371483 methods extracted from code from 250090 bug-fixing commits. Most importantly, the dataset includes the buggy, and the fixed version of methods extracted from the bugfixing commit. It also contains some additional information like the commit message, or the commit diff information.

Data Format

project: the project's name containing the bug

hexsha: hex sha identification of bugfixing commit

parent_hexsha: sha of the bugfixing commit's parent

message: the commit message

summary: the commit summary

file: modified file by the commit (location in repo)

total: metadata on changes - eg.: addition, or removal of lines

language: main language of the repository

size: size of the bugfix

diff: commit diff information

repo_url: the URL of the repository

part: makes distincion between different bugfixing parts in case the bugfix affects multiple functions

parts: the number of function affected by the bugfixing commit

a: fixed code segment

b: buggy code segment