A small command-line python script that creates a local backup of your Flickr data. It mirrors images, titles, description, tags, albums and collections.
Switch branches/tags
Nothing to show
Clone or download
markdoliner Fix tests.
I neglected to update a function name in the tests when I changed it in the script.
Latest commit c65ade8 Jun 3, 2018
Failed to load latest commit information.



A small command-line python script that creates a local backup of your Flickr data. It mirrors images, video metadata, titles, description, tags, albums and collections.

Available at https://github.com/markdoliner/flickrmirrorer


  • python 2.something or python 3.anything
  • python dateutil
    • Ubuntu: apt-get install python-dateutil
  • python flickrapi library 2.0 or newer.
  • python requests


The script was developed on Linux. It should work on other Unixy operating systems such as OS X, hopefully without changes. It could probably be made to work on Microsoft Windows with minor changes.

e.x. ./flickrmirrorer.py /mnt/backup/flickr/

See --help for options.


The script allows you to mirror only photos, only videos, or both. See the --ignore-videos and --ignore-photos command line options.

Your local backup can be cleaned automatically, so that files that were deleted in Flickr are deleted locally. Deletion is disabled by default. See the --delete-unknown command line option.

The script displays a summary of its actions if --statistics is passed on the command line.

Running via Cron

Running this script regularly via cron is a good way to keep your backup up to date. For example, create the file /etc/cron.d/flickr_backup containing the following:

# Run Flickr photo mirroring script.
# Sleep between 0 and 4 hours to distribute load on Flickr's API servers.
0 3 * * 2  root  sleep $((`bash -c 'echo $RANDOM'` \% 14400)) && /usr/local/bin/flickrmirrorer.py -q /mnt/backup/flickr/

If you run the cronjob as a user other than yourself you may need to take additional steps to make sure the cron user is able to authenticate. The steps are something like this:

  1. Run the script as yourself the first time around. It should pop open your web browser and request permission.
  2. After granting permission an authorization token is stored in ~/.flickr/9c5c431017e712bde232a2f142703bb2/auth.token
  3. Copy this file to the home directory of the cron user:
sudo mkdir -p /root/.flickr/9c5c431017e712bde232a2f142703bb2/
sudo cp ~/.flickr/9c5c431017e712bde232a2f142703bb2/auth.token \


The script creates this directory hierarchy:

dest_dir/Not in any album/
dest_dir/Not in any album/12345.jpg -> ../photostream/12345.jpg
dest_dir/Albums/Waterfalls - 6789/
dest_dir/Albums/Waterfalls - 6789/1_12346.jpg -> ../../photostream/12346.jpg
dest_dir/Albums/Waterfalls - 6789/2_12347.jpg -> ../../photostream/12347.jpg
dest_dir/Collections/Nature - 2634-98761234/Waterfalls - 6789 -> ../../Albums/Waterfalls - 6789
dest_dir/Collections/Nature - 2634-98761234/Mountains - 6790  -> ../../Albums/Mountains - 6790

The metadata files contain JSON data dumped from the Flickr API. It's not the prettiest thing in the world... but it does contain all the necessary data in case you want to recover from it.

The album and collection directories contain symlinks to the files in the photostream. The symlink names in albums are numbered so as to preserve the order.

Routine status is printed to stdout by default. Errors are printed to stderr. To see more options run with the --help flag.

A note about videos

The Flickr API does not support downloading original video files. If this script encounters videos in your photostream, it asks you download them (you must be logged in to your Flickr account).

Running unit tests

Run py.test.


  • Handle download errors better:
    • Add retry logic.
    • Continue trying to download other photos.
    • Stop running only if there are many download errors.
  • Mirror comments
  • Store order of photos in photostream
  • Store order of albums in collections



  • Support for nested collections and empty collections.


  • Don't warn about downloading videos if they've already been downloaded.
  • Unknown files are no longer deleted by default.
  • Added new command line option --delete-unknown
  • Added new command line option --ignore-photos
  • Added new command line option --ignore-videos
  • Print statistics even if script is killed by CTRL+C.