abusing ragel, possibly need new approach / tool

Question

Hi,

I'm trying to use Ragel to implement a simple yes/no fsm. Unfortunately the language specification consists of the union of about a thousand regular expressions, with * operators appearing once or more in the majority of them. So, the number of possible states explodes and it seems it will be impossible to use Ragel to generate an fsm for my language. Is there a tool out there that can do what I need, or should I swap approaches? I need something better than checking input strings against each regular expression in turn. I could chop up the thousand regular expressions into chunks of ~50 and generate an fsm for each, and run every input string against all the machines, but if there's a tool that can handle this kind of job without such a hack I'd be pleased to hear of it.

Thanks!

Answer 1

Well, I've ended up breaking the machine into multiple machines in order to prevent Ragel from eating all available memory - in fact, I had to break up the machine into a couple of separate Ragel files because the generated java class had too many constants in it from the huge state tables generated. I'm still interested in hearing of a better solution for this, if anybody has one!