python-git-info

A very simple project to get information from the git repository of your project. This package does not have any dependencies; it reads directly the data from the .git repository.

Installation

Just do a pip install python-git-info. This project should work with both python 2.7 and 3.x.

Usage

This app will search the current directory for a .git directory (which is always contained inside the root directory of a project). If one is found it will be used; else it will search the parent directory recursively until a .git is found.

There's a single function name get_git_info() with an optional dir parameter. If you leave it empty it will start the .git directory search from the current directory, if you provide a value for dir it will start from that directory. The get_git_info will return a dictionary with the following structure if everything works ok or None if something fishy happend or no .git folder was found:

  {
    'parent_commit': 'd54743b6e7cf9dc36354fe2907f2f415b9988198', 
    'message': 'commit: Small restructuring\n', 
    'commiter': 'Serafeim <email@email.com>', 
    'commit_date': '2018-11-14 13:52:34', 
    'author': 'Serafeim <email@email.com>', 
    'author_date': '2018-11-14 13:52:34', 
    'commit': '9e1eec364ad24df153ca36d1da8405bb6379e03b'
  }

How it works

This project will return the info from the latest commit of your current branch. To do this, it will read the .git/HEAD file which contains your current branch (i.e something like ref: refs/heads/master). It will then read the file it found there (i.e .git/refs/heads/master) to retrieve the actual sha of the latest commit, something like 8f6223c849d4bba75f037aeeb8660d9e6e306862.

This object is located in.git/objects/8f/6223c849d4bba75f037aeeb8660d9e6e306862 (notice the first two characters are a directory name and the rest is the actual filename). This is a zlib compressed folder. After it is uncompressed it has a simple format; I'm copying from the git internals manual:

The format for a commit object is simple: it specifies the top-level tree for the snapshot of the project at that point; the parent commits if any (the commit object described above does not have any parents); the author/committer information (which uses your user.name and user.email configuration settings and a timestamp); a blank line, and then the commit message.

So a sample commit message file would be something like this:

tree fa077d18fe3309aa12791dad2f733bfbb50fdee6
parent 943f6e8e3641ea38a9d9db3256944b46bcfc1f77
author Serafeim Papastefanos <spapas@example.com> 1562836041 +0300
committer Serafeim Papastefanos <spapas@example.com> 1562836041 +0300

prep new ver

The "pack" of snakes

Until now everything seems like sunshine; unfortunately there's a can of snakes in this process or better a pack of snakes: Non trivial git repositories will compress the contents of their .git/objects folder to save network (and disk) space to a file ending with .pack. This file is a dump of all the (zlib compressed) object your repository contains (including the commit messages of course) and is accompanied by a .idx object which says where each object can be found in the pack file. You can find these files in .git/objects/pack folder (if your repository has them of course).

In any case, the process of reading the .idx file to find the index of your commit and then reading that from the .pack file is not trivial; if you want to learn more about it you can check out this excellent resource: https://codewords.recurse.com/issues/three/unpacking-git-packfiles

Or you can take a look at my code at the pack_reader module which I tried to heavily comment to improve understanding.

Rationale

This project may seem useless or very useful, depending on the way you deploy to your servers. If you, like me, push every changeset to your VCS before deploying and then pull the changes from the remote server to actually deploy then you'll find this project priceless: You can easily add the latest commit information to somewhere in your web application so you'll be able to see immediately which changeset is deployed to each server without the need to actually login to the server and do a git log.

Also it is important to add here that this project is pure python and does not have any external dependencies (not even git); making it very easy to install and use in any project.

Changes

0.7.1

• Improve parsing of multi-line messages

0.7

• Various fixes to support more git repositories

0.6.1

• Remove non-needed print stmts

0.6

• It now parses the pack file to retrieve the commit object if it is packed!

0.5

• Change the parsing algorithm from using .git/logs to parse the real commit object inside the .git/objects folder.

0.4

• Add more error checks

0.3

• Make it work with '..'

0.2

• Initial