Background
I am writing a program that will do some bulk renaming of members and functions in a directory of java source code to de-obfuscate the code based on a look-up table .csv file passed in to the program.
What this is for is the source code I have was written against a obfuscated jar. I have a de-obfuscated version of the jar that was run through a customized version RetroGaurd and I would like to parse the mapping file that was passed in to RetroGaurd to de-obfuscate the function calls my source code makes in to the jar. If I just compile my code and run it through RetroGaurd too when I decompile I loose all of my nice commenting and formatting (unless there is a option on RetroGaurd that I missed).
Problem
I found the Abstract Syntax Tree parser built in to Eclipse and it looks perfect for my uses, however I am not planning on writing my program as a plugin for Eclipse, this is going to be a stand alone jar that can be run on any machine.
My main concern is as I write my code I am getting a lot of dependencies on internal jars that Eclipse uses. I know that if I conform to the EPL for the library jars I will have no issues distributing it, but I am concerned about this project getting bigger and bigger as I write it as more and more jars from Eclipse's SDK are required.
Are there any other projects out there that would give me the ability to parse Java source code to do find and replace reliably like AST will allow me to, or is there a way to use RetroGaurd (or a program like it) to run the same de-obfuscation but keep my comments and functions the same without needing to run the de-obfuscated program though a de-compiler afterwards?