Arg for on/off PR comments export (#62)

* add --comments args

* log PR comments only if flag is set

* Update commits_parser.py (fix author.login error)

* update name of pr_comments flag & add info to readme

* fix "Possibly using variable 'start' / 'finish' before assignment"

Author: HadronCollider

README.md

@@ -32,7 +32,7 @@ python3 main.py [-i, --issues] [-t, --token] token (github токен вмест
 ```
 3. Логирование pull requests
 ```commandline
-python3 main.py [-p, --pull_requests] [-t, --token] token (github токен вместо token) [-l, --list]  list (list - строка пути к txt файлу со списком репозиториев) [-o, --out] out (out - название csv файла, в который будут помещены все логи)
+python3 main.py [-p, --pull_requests] [-t, --token] token (github токен вместо token) [-l, --list]  list (list - строка пути к txt файлу со списком репозиториев) [-o, --out] out (out - название csv файла, в который будут помещены все логи) [--pr_comments] (если установлен - также выгружаются комментарии к PR)
 ```
 4. Логирование непринятых приглашений в репо
 ```commandline

commits_parser.py

@@ -39,7 +39,7 @@ def log_repository_commits(repository: Repository, csv_name, start, finish, bran
                 continue
             if commit.commit is not None:
                 nvl = lambda val: val or EMPTY_FIELD
-                commit_data = [repository.full_name, commit.commit.author.name, nvl(commit.author.login), nvl(commit.commit.author.email),
+                commit_data = [repository.full_name, commit.commit.author.name, nvl(commit.author.login if commit.author else None), nvl(commit.commit.author.email),
                                commit.commit.author.date, '; '.join([file.filename for file in commit.files]), commit.commit.sha, branch]
                 info = dict(zip(FIELDNAMES, commit_data))
 

main.py

@@ -23,6 +23,7 @@ def parse_args():
     parser.add_argument('-l', '--list', type=str, required=True, help='Path to the file containing the list of repositories. Repositories should be separated by a line break. Names should be in the format <organization or owner>/<name> ')
     parser.add_argument("--download_repos", type=str, help="path to downloaded repositories", default='./')
     parser.add_argument('-o', '--out', type=str, required=True, help='output filename')
+    parser.add_argument("--pr_comments", help="log comments for PR", action="store_true")
     parser.add_argument('-s', '--start', type=str, required=False, help='start time', default='2000/01/01-00:00:00')
     parser.add_argument('-f', '--finish', type=str, required=False, help='finish time', default='2400/01/01-00:00:00')
     parser.add_argument('-b', '--branch', type=str, required=False, help='branch to select commits, by default use "default" repository branch, use "all" to get all commits from all branches', default=None)
@@ -60,6 +61,8 @@ def main():
     csv_name = args.out
     path_drepo = args.download_repos
     fork_flag = args.forks_include
+    log_pr_comments = args.pr_comments
+    start, finish = None, None
 
     try:
         client = git_logger.login(token=token)
@@ -74,7 +77,7 @@ def main():
         if args.commits:
             commits_parser.log_commits(client, working_repos, csv_name, start, finish, args.branch, fork_flag)
         if args.pull_requests:
-            pull_requests_parser.log_pull_requests(client, working_repos, csv_name, token, start, finish, fork_flag)
+            pull_requests_parser.log_pull_requests(client, working_repos, csv_name, token, start, finish, fork_flag, log_pr_comments)
         if args.issues:
             issues_parser.log_issues(client, working_repos, csv_name, token, start, finish, fork_flag)
         if args.invites:

pull_requests_parser.py

@@ -71,7 +71,7 @@ def get_related_issues(pull_request_number, repo_owner, repo_name, token):
     return ';'.join(list_issues_url)
 
 
-def log_repositories_pr(repository: Repository, csv_name, token, start, finish):
+def log_repositories_pr(repository: Repository, csv_name, token, start, finish, log_comments=False):
     for pull in repository.get_pulls(state='all'):
         if pull.created_at.astimezone(pytz.timezone(TIMEZONE)) < start or pull.created_at.astimezone(
                 pytz.timezone(TIMEZONE)) > finish:
@@ -106,23 +106,25 @@ def log_repositories_pr(repository: Repository, csv_name, token, start, finish):
             'milestone': get_info(pull.milestone, 'title')
         }
 
-        if pull.get_comments().totalCount > 0:
-            for comment in pull.get_comments():
-                info = info_tmp
-                info['comment body'] = comment.body
-                info['comment created at'] = comment.created_at
-                info['comment author name'] = comment.user.name
-                info['comment author login'] = comment.user.login
-                info['comment author email'] = nvl(comment.user.email)
-                log_pr_to_csv(info, csv_name)
-                log_pr_to_stdout(info)
+        if log_comments:
+            comments = pull.get_comments()
+            if comments.totalCount > 0:
+                for comment in comments:
+                    info = info_tmp
+                    info['comment body'] = comment.body
+                    info['comment created at'] = comment.created_at
+                    info['comment author name'] = comment.user.name
+                    info['comment author login'] = comment.user.login
+                    info['comment author email'] = nvl(comment.user.email)
+                    log_pr_to_csv(info, csv_name)
+                    log_pr_to_stdout(info)
         else:
             log_pr_to_csv(info_tmp, csv_name)
             log_pr_to_stdout(info_tmp)
         sleep(TIMEDELTA)
 
 
-def log_pull_requests(client: Github, working_repos, csv_name, token, start, finish, fork_flag):
+def log_pull_requests(client: Github, working_repos, csv_name, token, start, finish, fork_flag, log_comments=False):
     with open(csv_name, 'w', newline='') as file:
         writer = csv.writer(file)
         writer.writerow(FIELDNAMES)
@@ -134,7 +136,7 @@ def log_pull_requests(client: Github, working_repos, csv_name, token, start, fin
             if fork_flag:
                 for forked_repo in repo.get_forks():
                     print('=' * 20, "FORKED:", forked_repo.full_name, '=' * 20)
-                    log_repositories_pr(forked_repo, csv_name, token, start, finish)
+                    log_repositories_pr(forked_repo, csv_name, token, start, finish, log_comments)
                     sleep(TIMEDELTA)
             sleep(TIMEDELTA)
         except Exception as e: