detect non-contributors in an organization

lazy_users.py

#!/usr/bin/env python

# you will need http://github3py.readthedocs.org/en/latest/#installation
# I'm not sure if this works on Windows due to the use of strptime()
# Based on https://gist.github.com/morido/9817399
# Improvement : take user, pass and orga from command line + use the new github3 API
from github3 import login
import datetime
import json
import pprint
import logging
import sys # to use sys.stdout
import os
import argparse

if hasattr(__builtins__, 'raw_input'): 
    input = raw_input

# Turn off console output buffering
buf_arg = 0
if sys.version_info[0] == 3:
    os.environ['PYTHONUNBUFFERED'] = '1'
    buf_arg = 1
try:
    sys.stdout = os.fdopen(sys.stdout.fileno(), 'a+', buf_arg)
    sys.stderr = os.fdopen(sys.stderr.fileno(), 'a+', buf_arg)
except:
    pass

parser = argparse.ArgumentParser(description='''
List all active and inactive user of a GitHub organization.''')
parser.add_argument('-u',metavar='USER', dest='github_user', action='store', required=False,
    help='GitHub user',
    default=None)
parser.add_argument('-p',metavar='PASSWORD', dest='github_pass', action='store', required=False,
    help='GitHub password',
    default=None)
parser.add_argument('-o',metavar='ORG', dest='github_org', action='store', required=False,
    help='GitHub organization to list. Default %(default)s',
    default='D4UDigitalPlatform')
parser.add_argument('-w',metavar='WEEKS', dest='weeks', action='store', required=False,
    help='Time of interests, in weeks. Default %(default)s',
    default='52')
parser.add_argument('--url',metavar='URL', dest='github_url', action='store', required=False,
    help='GitHub API url. Default %(default)s',
    default='https://api.github.com')

args = parser.parse_args()
if (args.github_user == None):
    args.github_user = input("GitHub Username :")
if (args.github_pass == None):
    print "GitHub Password (sorry, it will be displayed in the console. Use -p instead"
    args.github_pass = input(":")
	

time_of_interest_in_weeks = int(args.weeks)

# Attention: There is an hourly limit of 5000 requests to the GitHub-API. This script will use a lot of these...

token = login(args.github_user, args.github_pass)
organization = token.organization(args.github_org)

time_of_interest_in_weeks_absolute = datetime.datetime.today() - datetime.timedelta(weeks = time_of_interest_in_weeks)
openETCSusers = []
inactiveopenETCSusers = []

def removeFromInactiveList(loginname):
	if loginname in inactiveopenETCSusers:			
		inactiveopenETCSusers.remove(loginname)

# Setup: Get all users in the organization
for current_repo in organization.repositories():	
	print current_repo.name + " : "
	try:
		for current_contributor in current_repo.contributors():
			print " - " + current_contributor.login
			openETCSusers.append(current_contributor.login)
	except Exception as e:
		print e
			
openETCSusers = list(set(openETCSusers)) # remove duplicates
inactiveopenETCSusers = list(openETCSusers) # initialize the inactive users

print "Found " + str(len(openETCSusers)) + " unique users"

# Main part: Remove all users who 'did something'
for current_repo in organization.repositories():
	print "Processing Repository",
	print current_repo.name
	# Step1: remove all users who made on a commit within time_of_interest_in_weeks
	try:
		#print "1/3 : Get commits"
		commitShasToInvestigate = []
		for current_commit in current_repo.commits(since = time_of_interest_in_weeks_absolute):
			current_commit = current_commit.as_dict()
			if (current_commit.get('author') <> None):
				removeFromInactiveList(current_commit.get('author').get('login'))
			# usually author and committer are equal, but there are very few exceptions
			if (current_commit.get('commiter') <> None):
				removeFromInactiveList(current_commit.get('commiter').get('login'))
				
			commitShasToInvestigate.append(current_commit.get('sha'))
		# Note: This will probably fail to recognize contributions for commits opened in the distant past (before 'since') and updated very recently
	except Exception as e:
		print e
	#print " - removing users from " + str(len(commitShasToInvestigate)) + " commits"
	
	# Step2: remove all users who commented on a commit within time_of_interest_in_weeks
	''' remove this part, seems to be very slow and not working as expected
	for commit_to_investigate in commitShasToInvestigate:
		print commit_to_investigate
		
		for current_commit_comments in current_repo.commits(sha = commit_to_investigate):
			current_commit_comments = current_commit_comments.as_dict()
			if (current_commit_comments.get('author') <> None):
				removeFromInactiveList(str(current_commit_comments.get('author').get('login')))
    '''
	#print "3/3 : Need to inspect issues"
	# Step3: remove all users who commented on something within time_of_interest_in_weeks
	nbissue = 0
	try:
		for current_issue in current_repo.issues():
			nbissue = nbissue + 1
			for current_comment in current_issue.comments():			  
				json_data = current_comment.to_json()			
				json_parsed = json.loads(json.dumps(json_data))
				JSON_TIMEFORMAT = '%Y-%m-%dT%H:%M:%SZ'
				updated_at = datetime.datetime.strptime(json_parsed["updated_at"], JSON_TIMEFORMAT)
				if updated_at > time_of_interest_in_weeks_absolute: 
					removeFromInactiveList(json_parsed["user"]["login"])		
		print " - removing users from " + nbissue + " issues"
	except Exception as e:
		# no idea why this happens occassionally, some access rights issue?
		print e


	
	print "Inactive user remaining : " + str(len(inactiveopenETCSusers)) + " on " + str(len(openETCSusers))
###
### OUTPUT
###

print '\n'
print 'Considered all contributions after:',
print str(time_of_interest_in_weeks_absolute)
print '\n'
print '---------------'
print 'Inactive users:'
print '---------------\n'
for current_user in inactiveopenETCSusers:
	print current_user

print '\n'
print 'Summary:'
print '--------\n'

print '# All users: ',
print len(openETCSusers)
print '# Inactive users: ',
print len(inactiveopenETCSusers),
print ' since ' + str(time_of_interest_in_weeks) + ' weeks'

A new approach, written from scratch, with direct GitHub API (no need for an extra lib) is here : https://gist.github.com/psa-jforestier/28a5de5d262ac2af5b00595e651696d0