I think you may like this. The Digital Corpus of Sanskrit only lets you list words from a given text or one at a time, and their sole downloadable, the FrameNet XML, is pretty much useless outside of a very limited range of computational linguistic analyses. (If that's the Sanskrit equivalent of Perseus, the field is doooomed.) It's not impossible to get the information you want from them, but it would require scraping and aggregating the results from each text.