This is a thinly modified version of Matthew Barnett's mrab-regex regular expression library with added instrumentation for use when fuzzing python code with cpytraceafl.
The intention is for fuzzing harnesses to be able to substitute the builtin re module
library with this, highly compatible, module and thereby allow AFL to generate examples
which pass regular expressions used in the target or explore their limits in interesting
ways.
See the original README.mrab-regex.rst for more general information in this library's extended regex features.
Early in the startup of the fuzzing harness, (though after the call to install_rewriter()):
import regex
from sys import modules
modules["re"] = regexcode later importing/referencing the re module should instead be using this instrumented
regex code. Note that before evaluating any regexes, cpytraceafl.tracehook.set_map_start()
will need to have been initialized with a memory region to write its intrumentation data
into. Otherwise you'll get segfaults.
This works unless the code under test attempts to use the typing module, at which point its
trick involving the Pattern type will trip up over itself. Some further hackery will need
to be devised to get past this.