Delete unspam.py

unspam.py

-import gitlab
-import os
-import datetime
-import json
-
-def log(s):
-    print(s)
-
-gitServerUrl = os.environ['CI_SERVER_URL']
-gl = gitlab.Gitlab(url=gitServerUrl, private_token=os.environ['GITLAB_ACCESS_TOKEN'])
-ci_project = gl.projects.get(os.environ['CI_PROJECT_ID'])
-gl.auth()
-
-log(f'Hello I am {gl.user.username}.')
-
-log(f'Starting at {datetime.datetime.now()}.')
-
-spam_users = []
-new_spam_users = {}
-log('Loading spam users from `spam_users.txt`...')
-if os.path.exists('spam_users.txt'):
-    f = open('spam_users.txt')
-    for l in f:
-        spam_users.append(l.strip())    
-else:
-    log('  file not found fallback to fixed value,')
-    spam_users = ['ghost', 'leadohlead', 'nhjoshua58', 'williamjvmarlow']
-log(f'  loaded {len(spam_users)} spam users.')
-
-log('Checking spam user status...')
-for u in spam_users:
-    [user] = gl.users.list(username=u) or [None]
-    if user :
-        log(f'  user {user.username} (id: {user.id}) does not seem to be blocked but is in list.')
-        log(f'  go to {gitServerUrl}/{user.username} to report.')
-
-spam_titles = []
-new_spam_titles = {}
-log('Loading spam titles from `spam_titles.txt`...')
-if os.path.exists('spam_titles.txt'):
-    f = open('spam_titles.txt')
-    for l in f:
-        spam_titles.append(l.strip())
-log(f'  loaded {len(spam_titles)} spam titles.')
-
-log('Checking for spam issues in projects...')
-issueBuckets = {}
-maxIssueAge = datetime.timedelta(days=5)
-projects = gl.projects.list(last_activity_after=datetime.datetime.now()-maxIssueAge, min_access_level=40, get_all=True)
-log(f'  {len(projects)} have recent activity.')
-for p in projects :
-    log(f'  checking project `{p.path_with_namespace}`.')
-    issues = p.issues.list(updated_after=datetime.datetime.now()-maxIssueAge, get_all=True)
-    log(f'    project has {len(issues)} recent issues.')
-    for i in issues:
-        issueBucketId = f'{i.created_at[:13]}_{i.author["username"]}'
-        issueBuckets.setdefault(issueBucketId, 0)
-        issueBuckets[issueBucketId] += 1
-
-        if i.author['username'] in spam_users:
-            log(f'    issue {i.web_url} identified as spam due to author. Taking action.')
-            new_spam_titles[i.title.strip()] = 1
-            backPath = f'./backup/{i.references["full"]}'.replace('#','_')
-            log(f'      writing backup to {backPath}.')
-            os.makedirs(backPath, exist_ok=True)
-            with open(f'{backPath}/issue.json', 'w') as f:
-                f.write(i.to_json())
-            try:                
-                log(f'      deleting issue.')
-                i.delete()
-            except Exception as e:
-                log(f'      error deleting issue: `{e}`. Closing instead.')
-                try:
-                    pass
-                    i.state_event = 'close'
-                    i.save()
-                except Exception as e:
-                    log(f'    error closing issue: `{e}`. Skipping.')
-        elif i.title in spam_titles:
-            log(f'    issue {i.web_url} identified as spam due to title.')
-            new_spam_users[i.author['username']] = 1
-
-log(f'Checking for users with abnormal activity.')
-for k,v in issueBuckets.items():
-    if v > 1:
-        author = k[14:]
-        time = k[:13]
-        log(f'  `{author}` created {v} issues at {time}, identified as spam due to frequency.')
-        new_spam_users[author] = 1
-
-log('Saving spam telemetry...')
-for t in spam_titles:
-    new_spam_titles.pop(t.strip(), 1)
-
-for u in spam_users:
-    new_spam_users.pop(u.strip(), 1)
-
-log(f'  found {len(new_spam_titles)} new titles and {len(new_spam_users)} new users.')
-
-if len(new_spam_titles) or len(new_spam_users):
-    commitData = {
-        'branch': 'autoupdate',
-        'commit_message': f'Update after job {os.environ["CI_JOB_ID"]}',
-        'actions': []
-    }
-
-    if len(new_spam_titles):
-        log(f'    add new titles to commit.')
-        commitData['actions'].append({
-            'action': 'update',
-            'file_path': 'spam_titles.txt',
-            'content': '\n'.join(spam_titles + list(new_spam_titles.keys()))
-        })
-
-    if len(new_spam_users):
-        log(f'    add new users to commit.')
-        commitData['actions'].append({
-            'action': 'update',
-            'file_path': 'spam_users.txt',
-            'content': '\n'.join(spam_users + list(new_spam_users.keys()))
-        })
-
-    branchName = 'autoupdate'
-    [b] = ci_project.branches.list(search=f'^{branchName}$') or [None]
-    if not b:
-        log(f'  will create new branch `{branchName}`.')
-        ci_project.branches.create({
-            'branch': branchName,
-            'ref': os.environ['CI_COMMIT_SHA']
-        })
-
-    log(f'  will commit {len(commitData["actions"])} files.')  
-    ci_project.commits.create(commitData)
-
-    [m] = ci_project.mergerequests.list(source_branch=branchName, state='opened') or [None]
-    if not m:
-        log('  creating new merge request.')
-        m = ci_project.mergerequests.create({
-            'source_branch': branchName,
-            'target_branch': ci_project.default_branch,
-            'title': 'Autoupdate'
-        })
-
-    log(f'  merge {m.web_url} to update.')
-
-
-log(f'Done at {datetime.datetime.now()}.')