Parse d source file by using compiler

[ZombineDev via Digitalmars-d-learn]

On Monday, 9 November 2015 at 05:49:25 UTC, tcak wrote:
> I checked for a flag in this page 
> http://dlang.org/dmd-linux.html , but couldn't have found any 
> for this purpose.
>
> Is there a way to parse a d source file so it generates a tree 
> in JSON, XML, or something-that-can-be-processed-easily file 
> format?
>
> ---
>
> My real purpose:
>
> I need to generate hash code (e.g. MD5) for a part of source 
> code (let's say a class, struct, or a function). So whether the 
> codes are changed or not can be detected. As you will guess, 
> comments, text formatting etc. shouldn't affect the hash result.
>
>
> Use-Case:
>
> I am writing a code generator/back up system. It will check the 
> last available code file. If important changes are done in a 
> specific part of code, it will increase version number by 1.

You're use case is really interesting! AFAIK, currently, the 
lexer part of the DMD frontend is the only part that can be 
easily used standalone. Daniel Murphy (who drove a lot of the 
work towards DDMD) published the lexer on DUB: 
http://code.dlang.org/packages/ddmd. However the package is 
outdated, because it is based on DMD v2.067.
As the whole frontend is now in D, you should be able to import 
any of the D modules and work with them, but it may not be as 
easy, as their API is geared only towards the DMD driver.
I personally want to help refactor DDMD to be usable as a 
library, so you can use for all sorts of cool things (like 
runtime JIT, IDE support, REPL, and so on), but I'm quite busy at 
the moment :(