TCG - Top (Git) Contributors Graph

Builds a weighted graph representing top git contributors in a project. (Internship task)

Tecnhnologies

• Python 3
• NumPy
• graphviz (as a python lib)
• Various Unix utilities

Usage

To build TopN git contributors graph of a repo, clone this repo inside the repo you want to work with and launch main.py.

$ cd /your_repo/
$ git clone *this_repo*
$ cd /TCG/
$ python main.py [-n N_contributors] [--dry] [--verbose]

As a result, a graph will be shown. This graph will be saved as .gv and .pdf in the /TCG/ folder.

How it works

1. Top N contributors list is retrieved from git log and saved to /data/contributors_N.txt;
2. For each contributor from the list a list of all files changed by this contributor is retrived and saved to /data/contributor_name.txt;
3. For every possible pair of contributors files changed by both contributors are counted (using comm);
4. Graph is built with contributors as nodes and edges' length inversely proportional to the number of "shared" files from step 3. Edges representing links above 75th percentile are colored red.

Problems

• Python is redundant here, but it was faster/easier to do it this way;
• Graph becomes incomprehensible even with n~30. graphviz graphs have rank attributes for nodes and other things that might help prettify the graph;
• Solution should be lazier. If data for top 50 was already retrieved, then on the next run the data should be re-used, not re-downloaded. If data was retrieved for top 30, then only data for 31-50th contributors should be retrived on -n 50 launch. (Of course, then an additional flag --force-redownload should be introduced);
• Possible problem with the React repo: "Dan" and "Dan Abramov" are probably one person. Maybe close nodes with similar names should be merged;