Currently parsing of pronunciation URLs only works for English since it looks for File: in parseProperty.js.
Example: http://localhost:6927/ur.wikipedia.org/v1/page/mobile-sections/%D8%A2%D8%A6%DB%8C%D9%88%D8%B1%DB%8C_%DA%A9%D9%88%D8%B3%D9%B9 does not recognize the pronunciation URL.
To make this work for other languages we would need to have a list of File namespace prefixes for the various languages. I'm thinking it would be neat to bring over parts of the make-templates.py script from the Android repo but it would have to output JSON instead of the java files. The script main-page-names.py is a subset of this. Maybe we could make the main page names also be part of the JSON output. The key of this JSON object should be the language or a language code. We may have to special case language variants once we support them.