24_deb_pkg_gov/pr_data_get.py

import csv
from perceval.backends.core.git import Git
import os
import datetime as dt
import time
import shutil 
import pandas as pd
import dateutil
from tqdm import tqdm
import math

key = os.environ.get('KKEXKEY')

early_cutoff = dt.datetime(2008,2, 8)
temp_dir = "/data/users/mgaughan/tmp/"

'''
- rate of change, rate of all/day
'''
def file_get_pr(upstream_vcs_link, me_read):
    # if we're looking at readmes me_read is true and if not, if we're looking at contributing files, it's false
    #this is the window of days on either side of the event that we're looking at
    window = 182
    #print(upstream_vcs_link.split('/')[4])
    project_dict = {}
    project_dict['upstream_vcs_link'] = upstream_vcs_link
    upstream_vcs_link = upstream_vcs_link.strip()
    if "github" in upstream_vcs_link or "gitlab" in upstream_vcs_link:
        #making an evaluation that sub branches aren't being used and that people would fork if needed
        #this only looks at main
        upstream_vcs_link = "/".join(upstream_vcs_link.split("/")[0:5])
        print(upstream_vcs_link)
        full_temp_path = temp_dir + upstream_vcs_link.split('/')[4] + ".git"
    else:
        full_temp_path = temp_dir + upstream_vcs_link.split('/')[- 1] + ".git"
    print(upstream_vcs_link)
    if upstream_vcs_link == "https://gitlab.com/ubports/core" or upstream_vcs_link == "https://gitlab.freedesktop.org/xorg/lib":
        shutil.rmtree(full_temp_path, ignore_errors=True)
        return {}
    repo = Git(uri=upstream_vcs_link, gitpath=full_temp_path)
    try:
        commits = repo.fetch()
    except:
        print("perceval issue")
        return
    has_readme = False
    has_contributing = False
    merge_pre_rm, merge_post_rm, merge_pre_cont, merge_post_cont = 0, 0, 0, 0
    #list of tuples which has date and whether it was a merge
    commit_list = []
    first_date_readme = ""
    for commit in commits:
        #print(commit['data'])
        if "Merge" in commit['data'].keys():
            commit_list.append([commit['data']['CommitDate'], True, commit['data']['Author'], commit['data']['Commit']])
            if has_contributing:
                merge_post_cont += 1
            else:
                merge_pre_cont += 1
        else: 
            commit_list.append([commit['data']['CommitDate'], False, commit['data']['Author'], commit['data']['Commit']])
        files = commit['data']['files']
        #print(commit['data']['CommitDate'])
        #print(type(dateutil.parser.parse(commit['data']['CommitDate'])))
        for file in files:
            if "CONTRIBUTING" in file['file'] and has_contributing == False:
                has_contributing = True
                first_date_contributing = dateutil.parser.parse(commit['data']['CommitDate'])
            if "README" in file['file'] and has_readme == False:
                has_readme = True
                first_date_readme = dateutil.parser.parse(commit['data']['CommitDate'])
                project_dict['readme_commit_hash'] = commit['data']['commit']
    shutil.rmtree(full_temp_path, ignore_errors=True)
    if first_date_readme == "":
        return {}
    if me_read:
        project_dict['first_readme'] = first_date_readme
        before_read = pr_count(first_date_readme+ dt.timedelta(days=-window, hours=0), first_date_readme, commit_list, [], [])
        if before_read != None:
            project_dict['before_allcom_read'] = before_read[0]
            project_dict['before_mrg_read'] = before_read[1]
            project_dict['before_auth_new'] = before_read[2]
            project_dict['before_commit_new'] = before_read[3]
        else:
            return {}
        after_read = pr_count(first_date_readme, first_date_readme + dt.timedelta(days=window, hours=0), commit_list, before_read[4], before_read[5])
        if after_read != None:
            project_dict['after_allcom_read'] = after_read[0]
            project_dict['after_mrg_read'] = after_read[1]
            project_dict['after_auth_new'] = after_read[2]
            project_dict['after_commit_new'] = after_read[3]
        else: 
            return {}
    else:
        project_dict['first_contributing'] = first_date_contributing
        before_cont = pr_count(first_date_contributing + dt.timedelta(days=-window, hours=0), first_date_contributing, commit_list, [], [])
        if before_cont != None:
            project_dict['before_allcom_cont'] = before_cont[0]
            project_dict['before_mrg_cont'] = before_cont[1]
            project_dict['before_auth_new'] = before_cont[2]
            project_dict['before_commit_new'] = before_cont[3]
        else:
            return {}
        after_cont = pr_count(first_date_contributing, first_date_contributing + dt.timedelta(days=window, hours=0), commit_list, before_cont[4], before_cont[5])
        if after_cont != None:
            project_dict['after_allcom_cont'] = after_cont[0]
            project_dict['after_mrg_cont'] = after_cont[1]
            project_dict['after_auth_new'] = after_cont[2]
            project_dict['after_commit_new'] = after_cont[3]
        else: 
            return {}
    print(project_dict)
    return project_dict


#TODO: pr_count should return an array of values for weekly/6mo
def pr_count(start, end, commits, author_roster, commit_roster):
    count = 0
    merge_count = 0
    by_week = [0] * 27
    by_week_merge =[0] * 27
    current_week = 0
    new_authors = 0
    new_committers = 0
    for commit in tqdm(commits):
        if dateutil.parser.parse(commit[0]) <= start:
            if commit[2] not in author_roster:
                author_roster.append(commit[2])
            if commit[1] and commit[3] not in commit_roster:
                commit_roster.append(commit[3])
        if dateutil.parser.parse(commit[0]) > start:
            if math.floor((dateutil.parser.parse(commit[0]) - start).days / 7) <= 26:
                by_week[math.floor((dateutil.parser.parse(commit[0]) - start).days / 7)] += 1
                if commit[1]:
                    by_week_merge[math.floor((dateutil.parser.parse(commit[0]) - start).days / 7)] += 1
                    if commit[3] not in commit_roster:
                        new_committers += 1
                        #remaining question of whether to make this the author of the merge commit[2] or the committer of the merge commit[3]
                        commit_roster.append(commit[3])
                if commit[2] not in author_roster:
                    new_authors += 1
                    author_roster.append(commit[2])
        if dateutil.parser.parse(commit[0]) > end:
            print(len(by_week))
            return [by_week, by_week_merge, new_authors, new_committers, author_roster, commit_roster]

def for_files():
    #csv_path = "final_data/deb_contribfile_roster.csv"
    csv_path = "final_data/deb_readme_roster.csv"
    count = 0 
    with open(csv_path, 'r') as file:
        csv_reader = csv.DictReader(file)
        with open('kk_031624_pr_did.csv', "w") as writing_file:
            # this would also have to get switched fro the cont dataset
            keys = ['upstream_vcs_link', "first_readme", "readme_commit_hash", "before_allcom_read", "before_mrg_read", "after_allcom_read", "after_mrg_read", 'before_auth_new', 'after_commit_new', 'after_auth_new', 'before_commit_new']
            dict_writer = csv.DictWriter(writing_file, keys)
            dict_writer.writeheader()
            for row in csv_reader:
                count += 1
                print(row['upstream_vcs_link'])
                # this would have to get switched to false for the cont dataset
                try:
                    dict_row = file_get_pr(row['upstream_vcs_link'].strip(), True)
                except:
                    dict_row = {}
                dict_writer.writerow(dict_row)
    

if __name__ == "__main__":
    for_files()
    #file_get_pr("https://github.com/tqdm/tqdm", True)
    #file_get_pr("https://github.com/GameServerManagers/LinuxGSM", True)
    #file_get_pr("https://github.com/walling/unorm/issues/new/", True)
    #file_get_pr("https://github.com/krahets/hello-algo/tree/dev1", True)