Analysis-of-Socio-Technical-Congruence

Create interactive collaboration graph for the N most active contributors in given project based on number of common files from commits between them.

Usage

Make sure to install all needed dependencies and files.

Dependencies

• numpy
• pygithub
• pathlib
• progressbar2
• pygit2
• pyvis

Install dependencies

Install dependencies via requirements.txt

pip install -r requirements.txt

Github token

Project works with Github API. Before start make sure you created needed token for work. Instruction you can find below.

How to create Github token

Github got a good instruction for it. link to instruction

At the moment you don't need any Select scopes options. Save created token for later usage.

Example (facebook/react)

At root directory you can find ready to go main.py file.

1. Simply run command after installing all dependencies:

   python main.py

2. Enter token
3. Wait for result. Script will open result graph in default browser. If it didn't happened you can find result in assets/data/graphs/Graph.html

As the result you'll get interactive graph as below.

Choose node to highlight all edges from it. By holding cursor on node you'll get its neighbors and number of common files in DESC order.

Template

In main.py you can find detailed realisation in template function and change it for your needs.

How it works

At the moment here 3 main classes.

FilesGetter

Class for getting all changed files by N top contributors.
Workflow:

1. Init remote and local repositories. We can't work only with remote repository, because Github API got limitation up to 5000 requests per hour for user. Also working with local repository faster than with remote.
2. Get top N users and theirs commits from remote repository.
3. Iterate over users and theirs commits while getting SHA of each commit. Each SHA is used to get changed files by using local repository. The main idea here is to encode each user and changed files and use that information further in FilesCongruence.

FilesConguence

Class for calculating file-congruence between users.
Workflow:

1. Get encoded users and files from FilesGetter

2. Creates boolean 2D matrix of users and encoded files. "True" value if file appeared in commit, otherwise "False".

   	fileId1	fileId2	...
   userId1	True	False	...
   userId2	True	True	...
   ...	...	...	...

3. Iterate over all rows by using each row as mask on whole matrix and sum all values in row. It'll create columns of common files between users.

4. Merge columns

As the result we got 2D matrix with users ID's on both axes where each cell [i, j] represents number of common files between users.

	userId1	userId2	userId3	...
userId1	10	5	3	...
userId2	5	8	0	...
userId3	3	0	3	...
...	...	...	...	...

FileCongruenceGraph

Class for creating file congruence graph. Nodes representing users. Edge exists between nodes if users got common changed files. Edge color and width depends on number of common files between users. Graph is formed based on result matrix from FilesConguence. Result will be saved as *.html file and opened in default browser.