24_deb_pkg_gov/perceval_tasks.py

import datetime as dt
from perceval.backends.core.git import Git
import argparse

#globals
#repo_dir = '/tmp/'

#main function for all subsequent tasks using perceval 
def main(vcs_path, begin_date):
    perceval_info = {}
    perceval_info['list_of_commits'] = get_perceval_log(vcs_path, begin_date)
    if len(perceval_info['list_of_commits']) > 0:
        perceval_info['age_of_project'] = get_repo_age(perceval_info['list_of_commits'])
        perceval_info['contributors'], perceval_info['collaborators'] = get_all_actors(perceval_info['list_of_commits'])
        del perceval_info['list_of_commits']
        return perceval_info
    else:
        print('error, no commits found?')
        return {}



# this is the primary function for getting the list of commits from perceval
def get_perceval_log(vcs_path, begin_date):
    print(vcs_path)
    try:
        repo_dir = '/data/users/mgaughan/tmp/' + str(vcs_path[0].split('/')[-1])
        #gitpath=repo_dir
        repo = Git(uri=vcs_path[0], gitpath=repo_dir)
        # this is a temporary date_from, will need to be more inclusive in the future
        fetched_commits = repo.fetch(from_date=begin_date)
        return list(fetched_commits)
    except: 
        print("error, cannot fetch repo data?")
        return {}

#this function is just to evaluate the repository age, as defined by Tamburri and used by van Meijel
def get_repo_age(all_commits):
    first_commit = all_commits[0]
    last_commit = all_commits[-1]
    first_date = dt.datetime.strptime(first_commit['data']["CommitDate"], '%c %z')
    last_date = dt.datetime.strptime(last_commit['data']["CommitDate"], '%c %z')
    #project life, as defined in YOSHI, unit is days
    project_life = last_date - first_date
    return project_life.total_seconds() / 86400


#attempt at getting the rosters, though need to make sure that we can get the MR
def get_all_actors(all_commits):
    #collaborators are more senior than contributors, doing it by author/commit
    authors = []
    committers = []
    for commit in all_commits:
        author = commit['data']['Author']
        committer = commit['data']['Commit']
        if committer not in committers:
            committers.append(committer)
        if author not in authors:
            authors.append(author)
    # now cleaning lists
    for committer in committers:
        if committer in authors:
            authors.remove(committer)
    return len(authors), len(committers)