kontor/scripts/check_kontor.py

"""
Checks the database kontor
"""
from enum import Enum, auto

import mariadb
from pathlib import Path
from argparse import ArgumentParser, ArgumentDefaultsHelpFormatter
from setup import get_database_cursors, get_logger

parser = ArgumentParser(formatter_class=ArgumentDefaultsHelpFormatter)
parser.add_argument('--verbose', '-v', action='count', default=0)
parser.add_argument('--dir', '-d', default='/media/tpeetz/Media')
parser.add_argument('--dry-run', '-m', action='store_true')
parser.add_argument('--reset-cloud-link', '-r', action='store_true')
args = parser.parse_args()

class StatusType(Enum):
    UNKNOWN = auto()
    FILE_NAME = auto()
    FILE_ID = auto()
    DUPLICATE = auto()
    CLOUD_LINK = auto()
    CLOUD_LINK_ID = auto()


def get_status_of_file(found_file, cursor):
    status = StatusType.UNKNOWN
    file_id = ''
    try:
        cursor.execute(f'SELECT id, cloud_link FROM media_file WHERE file_name="{found_file.name}"')
        rows = cursor.fetchall()
        if len(rows) == 1:
            status = StatusType.FILE_NAME
            file_id = rows[0][0]
    except mariadb.Error as error:
        logger.debug(f'select failed with {error}')
    try:
        cursor.execute(f'SELECT id FROM media_file WHERE id="{found_file.stem}"')
        rows = cursor.fetchall()
        if len(rows) == 1:
            status = StatusType.FILE_ID
            file_id = rows[0][0]
        if len(rows) > 1:
            status = StatusType.DUPLICATE
            for row in rows:
                logger.info(f"found {row[0]} with {found_file}")
    except mariadb.Error as error:
        logger.debug(f'select failed with {error}')
    try:
        cursor.execute(f'SELECT id FROM media_file WHERE cloud_link LIKE "%{found_file.stem}%"')
        rows = cursor.fetchall()
        if len(rows) == 1:
            file_id = rows[0][0]
            if rows[0][0] == found_file.stem:
                status = StatusType.CLOUD_LINK_ID
            else:
                status = StatusType.CLOUD_LINK
    except mariadb.Error as error:
        logger.debug(f'select failed with {error}')
    return status, file_id

def rename_files_to_id(media_dir, conn, dry_run):
    media_path = Path(media_dir)
    cursor = conn.cursor()
    for file in media_path.iterdir():
        logger.debug('found file: {}'.format(file.name))
        (status, file_id) = get_status_of_file(file, cursor)
        new_file_path = file.with_name(f"{file_id}{file.suffix}")
        match status:
            case StatusType.FILE_NAME:
                logger.info(f'status of {file.name} is file_name')
                rename_file(file, new_file_path, dry_run)
                update_cloud_link(file_id, new_file_path, conn, dry_run)
            case StatusType.FILE_ID:
                logger.info(f'status of {file.name} is file_id')
                update_cloud_link(file_id, new_file_path, conn, dry_run)
            case StatusType.CLOUD_LINK:
                logger.info(f'status of {file.name} is cloud_link')
                rename_file(file, new_file_path, dry_run)
                update_cloud_link(file_id, new_file_path, conn, dry_run)
            case StatusType.CLOUD_LINK_ID:
                logger.debug(f'status of {file.name} is cloud_link_id')
                update_cloud_link(file_id, new_file_path, conn, dry_run)
            case StatusType.DUPLICATE:
                logger.info(f'status of {file.name} is duplicate')
            case StatusType.UNKNOWN:
                logger.info(f'status of {file.name} is unknown')
            case _:
                logger.info(f'status of {file.name} is not defined')

def rename_file(current_file, new_file_path, dry_run):
    if dry_run:
        logger.info('rename file {} to {}'.format(current_file.name, new_file_path.name))
    else:
        current_file.rename(Path(new_file_path))

def update_cloud_link(file_id, file_path, conn, dry_run):
    cursor = conn.cursor()
    logger.debug(f'update entry {file_id} with {file_path.absolute()}')
    if dry_run:
        logger.info(f'UPDATE media_file: cloud_link={file_path.absolute()}')
    else:
        cursor.execute('UPDATE media_file SET cloud_link="{}" WHERE id="{}"'.format(file_path.absolute(), file_id))
        conn.commit()

def reset_cloud_link(conn, dry_run):
    cursor = conn.cursor()
    if dry_run:
        logger.info('UPDATE media_file SET cloud_link=""')
    else:
        cursor.execute('UPDATE media_file SET cloud_link="" WHERE id is NOT NULL')
        conn.commit()


if __name__ == '__main__':
    logger = get_logger(args.verbose)
    logger.info("kontor.check_kontor started")
    _, mariadb_conn = get_database_cursors(logger)
    mariadb_cursor = mariadb_conn.cursor()
    if args.reset_cloud_link:
        reset_cloud_link(mariadb_conn, args.dry_run)
    link_list = []
    data_dir = args.dir
    logger.info("kontor.check_kontor.rename_files_to_id")
    rename_files_to_id(data_dir, mariadb_conn, args.dry_run)
    #logger.info("kontor.check_kontor.update_cloud_link_with_found_files")
    #update_cloud_link_with_found_files(data_dir, mariadb_conn, args.dry_run)
    #logger.info("kontor.check_kontor.get_ids_from_column_cloud_link")
    #get_ids_from_column_cloud_link(link_list, mariadb_cursor)
    #logger.info('found {} ids in column cloud_link'.format(len(link_list)))
    #logger.info("kontor.check_kontor.checking_ids_from_cloud_link")
    #checking_ids_from_cloud_link(link_list, mariadb_cursor)
    mariadb_conn.close()
    logger.info("kontor.check_kontor finished")