Wikimedia CEE Meeting 2015/Programme/Tools/List management bot

From Meta, a Wikimedia project coordination wiki

Title of your proposal[edit]

Name(s) and/or username(s)
Type of submission (Please choose one)
  • Lecture (one-to-many)

When Wiki Loves Monuments (WLM) went global, the international team asked that the lists be built using row-templates, which allow for easy automated parsing and editing. They also created a set of scripts that would parse the lists and do certain operations like adding categories to images and creating lists of pages with various errors. However, I found the code to be both too complicated (requiring the use of a database) and not flexible enough, so I wrote my own. It needed to achieve the following:

  • be modular - due to the large number of pages, I had to be able to parse different namespaces and websites independently; I have one script for each task, the main ones glued together with a shell script; secondary scripts (like the one used to create articles) are independent
  • allow caching - parsing 60.000 images takes about 2 days; the results need to be reused as much as possible
  • be aware of local Infoboxes and other templates, as they can contain a great deal of information
  • allow external data imports with as little preparation as possible

As time went by, I realized that it would make sense to use the same scripts for other lists (for instance archeological sites). This meant the code needed to be even more generic and modular. I ended up with quite a large, but easy to understand configuration for each script. This means that the code can be now reused for almost all list that respects the WLM's structure with just a few minutes of configuration.

Preliminary preparation (if necessary)
Expected outcomes

Encourage other countries to develop their own code and stop depending on WLM's international team robots to do their work.

Duration (without Q&A)

20-25 min

Specific requirements
Slides or further information
Interested attendees (Please add yourself, and you may indicate your questions to the presenter).
  1. KuboF (talk) 14:58, 22 August 2015 (UTC)[reply]