[Mastodon] Force redownload missing remote media

fix_acc_duplicated.py

import glob

import os
from os import path

import psycopg2

DATA_DIR = '[MEDIA_PATH]/accounts'

URL = '[DATABASE_URI]'

conn = psycopg2.connect(URL)


def format_name(id_):
    full = '{:09d}'.format(id_)
    return '{}/{}/{}'.format(
        full[:3],
        full[3:6],
        full[6:9]
    )

def check_exists(id_):
    full_path = path.join(
        DATA_DIR,
        format_name(id_)
    )

    return path.exists(full_path)


with conn.cursor() as cursor:
    cursor.execute("select id, avatar_file_name, header_file_name from accounts")
    rows = cursor.fetchall()

id_filename = {t[0]: t[1:] for t in rows}


for id_, filenames in id_filename.items():
    avatars = glob.glob(path.join(DATA_DIR, 'avatars', format_name(id_), '*/*'))
    headers = glob.glob(path.join(DATA_DIR, 'headers', format_name(id_), '*/*'))

    for avatar in avatars:
        if path.splitext(path.basename(avatar))[0] != path.splitext(filenames[0])[0]:
            # print('avatar', path.splitext(path.basename(avatar))[0], path.splitext(filenames[0])[0])
            print(avatar)
            os.unlink(avatar)

    for header in headers:
        if path.splitext(path.basename(header))[0] != path.splitext(filenames[1])[0]:
            # print('header', path.splitext(path.basename(header))[0], path.splitext(filenames[1])[0])
            print(header)
            os.unlink(header)

conn.close()

media_fix.py

from os import path

import psycopg2

DATA_DIR = '/<CHANGE_THIS>/qdon/media_attachments/files'

URL = 'postgres://dbuser:dbpass@dbhost:5432/mastodon'

conn = psycopg2.connect(URL)


def format_name(id_):
    full = '{:09d}'.format(id_)
    return '{}/{}/{}'.format(
        full[:3],
        full[3:6],
        full[6:9]
    )

def check_exists(id_):
    full_path = path.join(
        DATA_DIR,
        format_name(id_)
    )

    return path.exists(full_path)


with conn.cursor() as cursor:
    cursor.execute("select id from media_attachments where remote_url != '' and file_file_name != ''")
    rows = cursor.fetchall()

ids = list(map(lambda x: x[0], rows))

missing_files = list(filter(
    lambda x: not check_exists(x),
    ids
))


with conn.cursor() as cursor:
    for id_ in missing_files:
        cursor.execute(
            "update media_attachments set file_file_name = '' where id = %s",
            (id_,)
        )
        print(id_)


conn.commit()
conn.close()