Rather than uploading images to your gallery site, site owners simple enter Flickr image IDs. Tangelo then pulls in and stores the full API response. After that, everything is built from that from that stored response.
For additional performance, the raw image data is stored in redis cache, for fastest possible image display.
FWIW, the Flickr API rate limit is 3600 queries per hour, so it's unlikely most users would ever hit the limit, but caching pretty much prevents the possibility.
A working example can be viewed at shacker.net
A demo video of the core concepts can be viewed here
This is a gallery system for Python coders who also happen to be Flickr users, not for the general public. If you want a gallery system that doesn’t require writing or maintaining code, or if you are not familiar with Django, this project is not for you (try WordPress, or SquareSpace, or Lightroom Portfolios instead).
Anyone is free to help themselves to this source code, or even to make pull requests, but I will not support users who are learning Django or trying to deploy their first Django website.
This repo represents a full Django project, not a reusable app. As a result, it is not pip installable. Clone this repo and deploy it from a git checkout using your deployment system of choice.
Setup is the usual - configure your database and media/static paths, pip install -r requirements.txt, etc.
This project is a bit unusual in having no local.py - instead use local.yml for localhost, and put secrets in environment variables in your deployed server (env vars are read into settings via Goodconf).
Get a Flickr key and secret here. Then, in your project settings (env vars):
FLICKR_API_KEY: "123abc"
FLICKR_API_SECRET: "abc123"
FLICKR_USERNAME: "yourusername"
# See table on this page for image size reference:
# https://www.flickr.com/services/api/misc.urls.html
# b = 1024 on the long side
# h = 1024 on the long side
# n = 320px on the long side
FLICKR_IMAGE_SIZE: "h"
FLICKR_THUMBNAIL_SIZE: "n"
FLICKR_CROPPED_THUMB_SIZE: "q"
As always, keep secrets out of version control!
This project uses pip-tools and its pip-compile to generate hashes for dependency installation. To add or update deps:
# Full:
pip-compile --generate-hashes --output-file=requirements.txt base.in
# Single package
pip-compile --generate-hashes --output-file=requirements.txt -P django-jsoneditor base.in
Superuser will have a few links at the bottom of each image:
- Edit: Links to Django Admin edit view for this image
- Refetch: Reaches out to Flickr and overwrites our locally stored API data for this image
- Flush cache: Leave our db alone, but erases the cached redis data for this image so it's displayed with updates on next page load.
To wipe the entire site cache:
./manage.py nuclear
- Image model has an
album_orderfield that was initially meant for controlling the order of images as they appear in an album. But I had an issue with the drag/drop util I was using to make that easy, so I switched to date-based ordering. The old method is still half-there. Could be a user option in the future.