I am looking for a free, open source, portable fuzzing tool for popular image file types that is written in either Java, Python, or Jython.

Ideally, it would accept specifications for the fuzzable fields using some kind of declarative constraints. Non-procedural grammar for specifying constraints are greatly preferred. Otherwise, might as well write them all in Python or whatever.

Just specifying ranges of valid values or expressions for them. Ideally, it would support some kind of generative programming to export the fuzzer into various programming languages to suit cases where more customization was required.

If it supported a direct-manipulation GUI for controlling parameter values and ranges, that would be nice too.

The file formats that should be supported are:

• GIF
• JPEG
• PNG

So basically, it should be sort of a toolkit consisting of ready-to-run utility, a framework or library, and be capable of generating the fuzzed files directly as well as from programs it generates. It needs to be simple so that test images can be created quickly. It should have a batch capability for creating a series of images. Creating just one at a time would be too painful.

I do not want a hacking tool, just a QA tool. Basically, I just want to address concerns that it is taking too long to get commonplace image rendering/parsing libraries stable and trustworthy.