git-history-data analyses a Git source code repository and dumps out data in a form that is easy to analyse. In its simplest form it prints out one line for every change to every file in history, and who made the change, like this:
$ cd myrepo
$ git-history-data
"Commit", "Date", "Author", "Added", "Removed", "File"
"e35a4f0", "2015-11-11", "Pete", 1, 1, "githistorydata/main.py"
"5a6172d", "2015-07-15", "Andy", 1, 0, "githistorydata/codeline.py"
"5a6172d", "2015-07-15", "Andy", 8, 2, "githistorydata/commitdetail.py"
"5a6172d", "2015-07-15", "Andy", 32, 0, "githistorydata/dataline.py"
"5a6172d", "2015-07-15", "Andy", 8, 13, "tests/git_parsing__test.py"
"12f2881", "2015-07-13", "Pete", 4, 4, "githistorydata/git.py"
"8fd2224", "2015-07-13", "Andy", 18, 0, "githistorydata/commitdetail.py"
It is intended to be easy to analyse the results, and to be convenient to manipulate with standard Unix command-line tools.
For example, if you want to see very large changes to a specific file (in the Git project itself):
$ cd git
$ git-history-data > git-git-history.csv
$ grep "diff.c" git-git-history.csv | awk -F',' '{print $4, $3, $1 }' | sort -n -r | head -5
4047 "Junio C Hamano" "3686aa1caf907d22fe318c28efe93f0e7870ba50"
1805 "Martin Langhoff" "e660e3997fbad830e5723336d61883f3a50dbc92"
1803 "Junio C Hamano" "c66b6c067e49c5ec80f1254daef79aa1c7f5ffce"
1795 "Junio C Hamano" "e9b5b75ca87f45292de8ecde5d4d0512ac9542cd"
1795 "Junio C Hamano" "b8ed7f0f40743dae6111c8950ba55051933298ca"
Or the files with terrifying numbers of authors:
$ awk -F', ' '{print $6, $3}' git-git-history.csv | sort | uniq | awk '{arr[$1]++}END{for (a in arr) print arr[a], a}' | sort -n -r | head -5
235 "Makefile"
198 "Documentation/config.txt"
137 "cache.h"
130 "git-svn.perl"
115 "diff.c"
You can find more examples here: Learning about the Git codebase using git-history-data
git-history-data requires Git, Python 3 and the Python DateUtil library. On Debian, Ubuntu and similar you can install these with:
sudo apt-get install git python3 python3-dateutil
Get the code:
cd
git clone https://github.com/andybalaam/git-history-data.git
Now add a line to your PATH by doing something like this:
echo 'export PATH="$PATH:${HOME}/git-history-data"' >> ~/.bashrc
(Log out and back in again, and use echo $PATH to check your PATH has been updated.)
cd into the working tree of a git repository, and then run git-history-data, redirecting the result to a file. For example:
cd git
git-history-data > hist.csv
Now hist.csv contains one line per file per commit in the entire history of the project, showing the commit ID, timestamp, author and the number of lines added and removed in that file in that commit.
That's it.
We welcome contributions!
Before we can accept your pull request you must confirm that all your contributions are written by you. To do this, create a pull request that adds a file inside the legal directory, creating a file called <yourusername>-DCO1.1.txt that is a copy of one of the other files in that directory. Once that pull request has been accepted, we can accept other pull requests as normal.
A good area to start contributing is to make a little script that analyses the data coming out of git-history-data and produces a little report (on the command line). We plan to make a library of scripts that report e.g. which developer creates the most commits, which files are most frequently modified.
git-history-data is written by Andy Balaam and the git-history-data contributors.
(c) Copyright 2015-2020 IBM Corporation, Andy Balaam and the git-history-data contributors. Distributed under the BSD 2-clause license.