Building predictions package for electron


I have spent some time understanding how the completions.json is generated by using the donna tool and then applying further processing to give the required output in autocomplete-atom-api package. In atom everything was really structured and organized. I am facing challenges while approaching the same in electron. It is not as simple as matching just the regex as in atom.The variables can take up any variable name unlike in atom where we have a fixed structure like “atom.commands.—”. Here on the other hand it could be “globalShortcut.register” or “globalShortcut2.register” or any var name i give it . I would like to discuss how to approach this with someone in detail. Can anyone please help me out here with initial approach?