Initial Commit

Check which projects are affected by the cleanup and prevent deletion by commiting to them. 

1. run script gitlab_check_projects.py

- checks which projects are affected by deletions

- creates an Excel file (might contain projects that do not exist anymore in the UI)

2. modify excel

- modify column "keep_project" to define which projects to keep (use WAHR or FALSCH)

3. run script commit_to_keep.py

- creates commits for project that are marked with "keep_project"

Check which projects are affected by the cleanup and prevent deletion by commiting to them. 

1. run script gitlab_check_projects.py

- checks which projects are affected by deletions

- creates an Excel file (might contain projects that do not exist anymore in the UI)

2. modify excel

- modify column "keep_project" to define which projects to keep (use WAHR or FALSCH)

3. run script commit_to_keep.py

- creates commits for project that are marked with "keep_project"

 No newline at end of file

import pandas as pd

import requests

import os

# Constants

GITLAB_API_URL = "https://aidaho-edu.uni-hohenheim.de/gitlab/api/v4"

PERSONAL_ACCESS_TOKEN = ""

SUBFOLDER = "data"

# Headers for authentication

headers = {

    "Private-Token": PERSONAL_ACCESS_TOKEN,

    "Content-Type": "application/json"

}

def create_commit(project_id,branch="main"):

    url = f"{GITLAB_API_URL}/projects/{project_id}/repository/commits"

    data = {

        "branch": branch,

        "commit_message": "Automated commit",

        "actions": [

            {

                "action": "create",

                "file_path": "prevent_project_deletion.txt",

                "content": "This file was created to prevent the automated deletion of the project during GitLab clean up at the end of the semester. This file can be deleted. "

            }

        ]

    }

    response = requests.post(url, headers=headers, json=data)

    response.raise_for_status()

    return response.json()

def main():

    # Read the Excel file

    script_directory = os.path.dirname(os.path.abspath("__file__"))

    excel_path = os.path.join(script_directory,SUBFOLDER,'gitlab_projects_latest_commits.xlsx')

    df = pd.read_excel(excel_path)

    # Filter projects where "Keep Project" is True

    projects_to_keep = df[df['keep_project'] == True]

    # Iterate through filtered projects and create a commit

    for index, row in projects_to_keep.iterrows():

        project_id = row['Project ID'] 

        branch = row['Default Branch'] 

        try:

            commit_response = create_commit(project_id=project_id, branch=branch)

            print(f"Commit created for project {project_id}: {commit_response}")

        except Exception as e:

            print(f"Failed to create commit for project {project_id}: {e}")

if __name__ == "__main__":

    main()

*.xlsx

import requests

import pandas as pd

from datetime import datetime

from dateutil.parser import parse # to convert times

import os

# Configuration

GITLAB_API_URL = "https://aidaho-edu.uni-hohenheim.de/gitlab/api/v4"

# List of owners to check for

PRIVATE_TOKEN = "<MY PRIVATE TOKEN>"

OWNER_LIST = ['aidaho','<MY USERNAME>']

CUTOFF_DATE = datetime.strptime("2024-01-01", "%Y-%m-%d")

SUBFOLDER = "data"

# Headers for authentication

headers = {

    "Private-Token": PRIVATE_TOKEN

}

def get_projects():

    projects = []

    page = 1

    while True:

        url = f"{GITLAB_API_URL}/projects?membership=true&per_page=100&page={page}"

        response = requests.get(url, headers=headers)

        response.raise_for_status()

        project_batch = response.json()

        if not project_batch:

            break

        projects.extend(project_batch)

        page += 1

    return projects

def get_latest_commit(project,owner_list):

    project_id = project['id']

    try:

        project_owner = project['owner']['username'] 

    except KeyError:

        return None

    if project_owner not in owner_list:

        return None

    url = f"{GITLAB_API_URL}/projects/{project_id}/repository/commits?per_page=1"

    response = requests.get(url, headers=headers)

    response.raise_for_status()

    commits = response.json()

    if commits:

        return commits[0]

    return None

def get_default_branch(project,owner_list):

    try:

        project_owner = project['owner']['username'] 

        project_branch = project['default_branch']

    except KeyError:

        return None

    if project_owner not in owner_list:

        return None

    return(project_branch)

def main():

    projects = get_projects()

    data = []

    for project in projects:

        project_id = project['id']

        print(project_id)

        project_name = project['name']

        branch = get_default_branch(project,OWNER_LIST)

        latest_commit = get_latest_commit(project,OWNER_LIST)

        if latest_commit:

            commit_date = parse(latest_commit['created_at']).replace(tzinfo=None)

            is_before_cutoff = commit_date < CUTOFF_DATE

            data.append({

                'Project ID' : project_id,

                'Project Name': project_name,

                'Latest Commit': latest_commit['id'],

                'Default Branch': branch,

                'Commit Date': latest_commit['created_at'],

                'Affected by Delete': is_before_cutoff,

                'keep_project' :False 

            })

    df = pd.DataFrame(data)

    script_directory = os.path.dirname(os.path.abspath(__file__))

    subfolder = script_directory +"/"+ SUBFOLDER

    if not os.path.exists(subfolder):

        os.makedirs(subfolder)

    excel_path = os.path.join(subfolder,'gitlab_projects_latest_commits.xlsx')

    df.to_excel(excel_path, index=False)

if __name__ == "__main__":

    main()