> It uses the pcre-heavy Haskell library which in turn calls out to the system-wide pcre C library for actually compiling and running the regex.
That's not true regular expressions, but a backtracker with eclectic features.
The regex used in the regex solution is just:
mul\((\d+),(\d+)\)
If Haskell happens to provide bindings to the POSIX regcomp/regexec stuff, that would be something to try instead. \d has to be replaced with [0-9], needless to say.