User:Multichill/Flickr ripper

A tool to upload large batches of Flickr files. Could be a local pywikipedia based bot or a web based bot.

Flow

1. Fire up the tool and specify a set of files to work on
2. See a global window and for each file a window.
3. Fill out fields for files and press "queue" to add the file to the queue or "skip" to ignore the file
4. File gets uploaded

Global and local

• The tool should have a global and a per file part.
• If a field is set in the global part and is set in the local part, the local field should be used.
• If a field is set in the global part and is left empty in the local part, the global field should be used.
• Would be nice to copy fields from previous/next file.

Fields

• Filename
  • Somehow derive from flickr title, can of course be overridden.
  • Check for name collisions
  • Would be nice to have some automatic counter to not have duplicate filenames.
• Original source (could be set automaticly)
• Author(s) (link to flickr user page)
• Date of the work (Taken on "XX" uploaded to flickr on "XX"?)
  • Take flickr upload date into account
  • Take date in exif into account
• Description (langs) (pull from flickr)
• Other versions (less important, leave it out)
• Permission
  • Tag for flickrreview
  • Flickrreview it with a certain username
  • Not free at flickr, need OTRS-permission
• Additional info: Other templates etc
• Licensing - Copy from flickr or override (needs otrs)
• Categories
  • From tags
  • From category auto completion tool

Would be nice to update one or more fields for a subset of images.

Sources

Existing tools to learn from:

• Special:Upload (main upload form)
• Commonplace
• Commonist
• Flickr upload bot - The standard upload bot. Makes uploading with all the correct info much easier.
• Flickr2Commons - This one is faster than the main Flickr upload bot (which can move really slowly occasionally and takes more clicks as it requires you to edit the image page), however a TUSC login is needed here and it adds a bunch of useless and redundant categories (but this one automatically adds geolocation information).
• FlickrLickr - A collaborative vetting process for selecting free (Creative Commons CC-BY) photos from Flickr, improving their metadata, and uploading them to Commons. No longer maintained and no new accounts will be created.
• Flinfo - For people who want to upload Flickr pictures by themselves.

Notes

• To find dupe calculate sha1 and compare it with wiki
  • Python : http://docs.python.org/library/hashlib.html
  • Wiki use api : http://commons.wikimedia.org/w/api.php?action=query&list=allimages&aisha1=e58ad30ebf578a655ce4bba335b7e5d9727e6e9a
• Flickr api is the best http://www.flickr.com/services/api/
  • Python implementation : http://stuvel.eu/projects/flickrapi
  • Useful queries:
    • flickr.people.findByUsername - Get userid for username
    • flickr.people.getPublicPhotos - to get all public images of a user http://www.flickr.com/services/api/explore/?method=flickr.groups.pools.getPhotos
    • flickr.groups.search - find id for group
    • flickr.groups.getInfo - get info about the group
    • flickr.groups.pools.getPhotos - get photo's in a pool
    • flickr.groups.pools.getContext - get previous and next image in a pool
    • flickr.photos.getSizes - get links to different sizes (need original)
    • flickr.photos.getInfo - get all info about an image
    • flickr.photos.getExif - get exif info
    • flickr.photos.geo.getLocation - get location, can be used directly in {{Location dec}}, zoom is an issue
• Geolocation
• Modes: Manual (1-10), semi-automatic (10-100), fully automated (100+)??