Question

Is there any way to parse or model the contents of a JavaScript file and determine what undefined references it requires, and what references it exports?

I’m looking for a way to automatically determine the correct loading order of giant collection of JavaScript files for a web app. Each file is essentially a module but many files have strange names and extra snippets of utility-type code; the code and its organization is so obtuse we’re going to be doing this manually for weeks if it can’t be automated.

Answer 1

This is kind of a broad question, which likely implies a nontrivial solution. A start would be looking at something like Jurassic to see if it lets you look at its AST, and working out what variables in which scope are resolved or not. If these are to run in browser, also consider that in Javascript the global scope is the window object, which you can “import” and “export” through, and this can be aliased to who knows what so you’ll need to do some sort of dataflow analysis.

A possible alternative would be looking at the implementation of a minifier (I’d try UglifyJS because it explicitly supports parsing into an AST then builds minification on that), since this sounds like kind of what they need to do when determining which variable names they shouldn’t abbreviate. That said minifiers are allowed to be imprecise and accept false positives, which might be a problem for you.