Slack file downloader from export archive (2021)

slack-download.py

# -*- coding: utf-8 -*-
#
# Slack file downloader from export archive
#
# Updated to python 3 with 2to3, and tqdm was added.
# Usage: slack-download.py <extracted archive export dir> <full path to download dir>

import json
import sys
import urllib.request, urllib.parse, urllib.error
import urllib.parse
import os
import glob
from tqdm import tqdm

def err(message):
    print(('[-] '+message))
    os.exit(-1)


def main(export_path):
    if not os.path.isdir(export_path):
        err('Slack exported directory required.')

    os.chdir(export_path)

    for check_file in ('users.json', 'channels.json'):
        if not os.path.isfile(check_file):
            err('It seems that given directory is not an export directory.')

        for json_path in tqdm(glob.glob('*/*.json')):
            channel, filename = os.path.split(json_path)
            date = filename.split('.')[0]

            with open(json_path, 'rb') as f:
                data = json.load(f)

            for content in data:
                files = []
                if 'file' in content:
                    files.append(content['file'])
                if 'files' in content:
                    for file in content['files']:
                        files.append(file)
                
                for file in files:
                    if 'url_private_download' not in file:
                        continue

                    url = file['url_private_download']

                    try:
                        os.mkdir(os.path.join(channel, date))
                    except OSError:
                        pass

                    # Ensure that the files each have different names for each different id
                    filename = "%s - %s" % (file['id'], urllib.parse.urlparse(url).path.split('/')[-1])
                    download_path = os.path.join(sys.argv[2], channel, date, filename)
                    os.makedirs(os.path.dirname(download_path), exist_ok=True)
                    if os.path.isfile(download_path):
                        print(('[-] Already downloaded %s -> %s' % (url, download_path)))

                    else:
                        print(('[+] Downloading %s -> %s' % (url, download_path)))

                        with open(download_path, 'wb') as f:
                            f.write(urllib.request.urlopen(url).read())

if __name__ == '__main__':
    if len(sys.argv) < 2:
        print(('Usage: %s <Slack exported directory>' % sys.argv[0]))
        sys.exit(-1)

    export_path = sys.argv[1]
    main(export_path)