deleter.py

import shutil
import os
import codecs
import json


"""
Using the list of active pages generated by compile_active from the API,
deleter removes whatever isn't accounted for online.

Warning: anything in a category folder that isn't in the compile_active list will be deleted.
"""

MIRROR = 'archive'


class Deleter:
    """
        A Deleter class for deleting folders in the archive of pages that have been deleted on the OSF.
        A CLI is designed to work with this deleter.
        Basic Workflow:
            Init -> compare archive to mirror -> delete folders in archive
    """

    def __init__(self, json_filename):
        """
            Constructor for Deleter class
            Generates three attributes to the deleter
                1. A list of all active projects
                2. A list of all active registrations
                3. A list of all active users

            :param json_filename: Name of the json file generated from compile_active
        """
        with codecs.open(json_filename, mode='r', encoding='utf-8') as file:
            active_lists = json.load(file)
            self.active_node_guids = active_lists["list_of_active_nodes"]
            self.active_registration_guids = active_lists["list_of_active_registrations"]
            self.active_user_guids = active_lists["list_of_active_users"]

    def compare_to_mirror(self, osf_type, active_list):
        """
            Compares the archive to the active nodes from the OSF and deletes the folders of guids that are
            present in the archive but no longer active on the OSF.

            :param osf_type: Whether the list is of registrations, projects, or user profiles
            :param active_list: The list of active guids
        """
        mirror_list = os.listdir(MIRROR + '/' + osf_type)
        print("OSF type: " + osf_type)

        for subdir in mirror_list:
            print("Checking", subdir)
            if subdir not in active_list:
                print(subdir, "inactive. Deleting")
                subdir_path = '/'.join([MIRROR, osf_type, subdir])
                self.delete_directory(subdir_path)

    def delete_directory(self, directory_path):
        """
            Deletes a given directory

            :param directory_path: The path to the directory to be deleted
        """
        print(directory_path)
        if os.path.isdir(directory_path):
            shutil.rmtree(directory_path)
            print("Deleted", directory_path)

    def run(self):
        """
            CLI Endpoint for deleter
            Controls workflow
            runs deleter on projects, registrations, and user profiles
        """
        self.compare_to_mirror('project', self.active_node_guids)
        self.compare_to_mirror('registration', self.active_registration_guids)
        self.compare_to_mirror('profile', self.active_user_guids)