Analyze all contributions in the issues area of a GitHub repository

analyze-github-issues.py

#!/usr/bin/env python
#
# Analyze all contributions in the issues area of a GitHub repository.
# Output a summary of all users and their contributions (issues, pull requests and comments).
# The summary is sorted in decreasing order of number of contributions.
#

import requests  # pip install requests, see https://docs.python-requests.org/

# TO BE COMPLETED: fill with the right values for your repo
gh_owner = '...'   # GitHub repository owner or organization
gh_repo  = '...'   # GitHub repository name
gh_user  = '...'   # GitHub user name for authenticated access to GitHub API (optional)
gh_token = '...'   # GitHub personal access token for above user

# Get a paginated GitHub API output as one single big list.
def gh_get_paginated(url, params = {}):
    result = []
    params['per_page'] = '100'
    while not url is None:
        resp = requests.get(url,
                            params = params,
                            headers = {'Accept': 'application/vnd.github.v3+json'},
                            auth = (gh_user, gh_token))
        result.extend(resp.json())
        url = resp.links['next']['url'] if ('next' in resp.links and 'url' in resp.links['next']) else None
        params = None
    return result

# User context:
class user_context:
    def __init__(self, name):
        self.name = name
        self.issues = 0
        self.prs = 0
        self.comments = 0
        self.characters = 0
    def total(self):
        return self.issues + self.prs + self.comments
    def __lt__(self, other):
        return self.total() < other.total()
    def __str__(self):
        return '%-20s %8d %8d %8d %8d %8d' % (self.name, self.issues, self.prs, self.comments, self.total(), self.characters)
    header = '%-20s %8s %8s %8s %8s %8s' % ('User name', 'Issues', 'PR', 'Comments', 'Total', 'Size')

# A dictionary of all users
users = {}

# Total of all users in one single instance.
total = user_context('Total')

# Get all issues and all comments.
base_url = 'https://api.github.com/repos/' + gh_owner + '/' + gh_repo
issues = gh_get_paginated(base_url + '/issues', {'state': 'all'})
comments = gh_get_paginated(base_url + '/issues/comments', {'state': 'all'})

# Analyze all issues.
for item in issues:
    if 'user' in item and 'login' in item['user']:
        user_name = item['user']['login']
        if not user_name in users:
            users[user_name] = user_context(user_name)
        if 'pull_request' in item:
            users[user_name].prs += 1
            total.prs += 1
        else:
            users[user_name].issues += 1
            total.issues += 1
        if 'body' in item:
            size = len(item['body'])
            users[user_name].characters += size
            total.characters += size

# Analyze all comments
for item in comments:
    if 'user' in item and 'login' in item['user']:
        user_name = item['user']['login']
        if not user_name in users:
            users[user_name] = user_context(user_name)
        users[user_name].comments += 1
        total.comments += 1
        if 'body' in item:
            size = len(item['body'])
            users[user_name].characters += size
            total.characters += size

# Print final summary
total.name += ' (%d users)' % len(users)
print(user_context.header)
print(str(total))
for user in sorted(users.values(), reverse = True):
    print(str(user))