Failing test case minimizer

This is a POC implementation of the delta debugging algorithm for generating 1-minimal failing test cases. The algorithm is described in:

Simplifying Failure-Inducing Input, Ralf Hildebrandt and Andreas Zeller, 2000.

PDF download.

Demo

Consider bad.py:

#!/usr/bin/env python2

print 'hello there, how are you'

import sys
import os

os.system('ls')

sys.exit(42)

Running this code will result in a non-zero exit code:

$ python bad.py
hello there, how are you
bad.py  ddmin  LICENSE
$ echo $?
42

Let's try to minimize this test case:

$ ddmin -i bad.py -o bad0.py python
...
$ cat bad0.py
v

That was not very helpful. The problem is that ddmin regards any non-zero exit code as a failure, and running the "program" consisting of a single v will make python fail. So let's only regard an exit code of 42 as an error:

$ ddmin -i bad.py -o bad1.py --status 42 python
...
$ cat bad1.py
import sys
sys.exit(42)

Much better. We can also consider the presence of the string "hello there" in the output as an error and get this minimal test case:

$ ddmin -i bad.py -o bad2.py --writes "hello there" python
...
$ cat bad2.py
print'hello there'