Find orphant stubnitz media archive dirs not linked to any event in the db.

find_orphants.py

#!/usr/bin/env python

import os, re
from mysql.connector import MySQLConnection, Error

__author__ = 'Patrick <space> Meyer'

ARCHIVE_MOUNT = '/var/www/'
EVENT_DIR_RE = ARCHIVE_MOUNT + r'current-archive\/(audio|video|bilder)\/[0-9]+\/[0-9]+\/[0-9]+\/[0-9]+'

def get_db_events():
    try:
        dbconfig = {
            'database': '<db>',
            'host': '<host>',
            'user': '<user>',
            'password': '<pass>',
        }
        conn = MySQLConnection(**dbconfig)
        cursor = conn.cursor()
        cursor.execute("SELECT id, datum, zeit FROM veranstaltung")
        rows = cursor.fetchall()

        # Return a dict with YYYYMMDDHHHH as key and the id as the value for all events.
        return {
            str(row[1]) + str(row[2]).zfill(4): row[0]
            for row in rows
        }

    except Error as e:
        print(e)
 
    finally:
        cursor.close()
        conn.close()

def get_fs_event_dirs():
    # Example: /mnt/stubnitz/cl/bilder/2016/10/02/2200/
    event_dirs = {}

    event_dir_cre = re.compile(EVENT_DIR_RE)
    for dirpath, _, _ in os.walk(ARCHIVE_MOUNT, followlinks=True):
        if event_dir_cre.search(dirpath):
            disk, media_type, year, month, day, hour = dirpath[len(ARCHIVE_MOUNT):].split('/')
            event_dirs[dirpath] = year + month + day + hour.zfill(4)

    # Return a dict with the path as key and YYYYMMDDHHHH as value.
    return event_dirs

def main():
    print("Querying FS for events..")
    fs_event_dirs = get_fs_event_dirs()
    print("{} event dirs found in the FS.".format(len(fs_event_dirs)))

    print("Querying DB for events..")
    db_events = get_db_events()
    print("{} events found in the DB.".format(len(db_events)))

    for path, date in fs_event_dirs.items():
        if date not in db_events.keys():
            print("Error: No db event matches dir: {}".format(path))

if __name__ == '__main__':
    main()