Since both the templates names and arguments can be localized this is a complex task that needs some planning before the code is written.
How do we want WARE to internally know about the differences e.g. from enwiki to svwiki, jawiki and dewiki?
My suggestion is to base the implementation on a dataset kept separate from WARE that is read when WARE is loaded. This dataset should contain json which maps all templates to a known set of possible parameters.
When a new language version of Wikipedia appear the curator of the dataset either extracts programmatically or manually the templates and parameters and map them to a single structure shared between all language versions.
Since both the templates names and arguments can be localized this is a complex task that needs some planning before the code is written.
How do we want WARE to internally know about the differences e.g. from enwiki to svwiki, jawiki and dewiki?
My suggestion is to base the implementation on a dataset kept separate from WARE that is read when WARE is loaded. This dataset should contain json which maps all templates to a known set of possible parameters.
When a new language version of Wikipedia appear the curator of the dataset either extracts programmatically or manually the templates and parameters and map them to a single structure shared between all language versions.