lazy pattern matching about leex HOT 2 CLOSED

Typical scanner generators like lex, flex and leex generate fully deterministic scanners. This is very important as otherwise it is very easy to create a pattern which can make perl and PCRE based scanners like just re take exponential increases in time. It is their back-tracking which kills them. There is a very good paper by Rob Pike demonstrating this.

Lazy matching is one of these properties. Leex generates fully deterministic scanners so it is actually usable. :-) Friedl in his book "Mastering Regular Expressions" in a section called "Unrolling C comments" describes how he arrived at one that works:

/\*([^*]|\*+[^/*])*\*+/

It's not easy to comprehend. :-) I have managed to modify it to other comment patterns with a similar structure. It is the * in the comments which makes it harder to read. If the comments were /x ... x/ then it would be:

/x(^x]|x+[^/x])*x+/

Robert

from leex.

Impressive. That helped.

Cheers
Jörgen

from leex.