TemplateParser currently looks for classes for license templates, and IDs for information fields, because that's how most of the information is organized on Commons.

Now that we're in the process of spreading machine-readable to as many files as possible on Commons and all other wikis with file uploads, we're running into situations where the information template isn't necessarily built with a table (example: https://commons.wikimedia.org/wiki/Template:Audio_upload ). The table+ID approach doesn't work for those.

It would be great if TemplateParser could look for regular classes, in addition to IDs, when looking for information-like data. From a user perspective, it would be simpler if the classes had the same name as the IDs, but if it's not possible / preferable from a development standpoint, it's not a big problem.

Version: master
Severity: normal