in reply to Reverse engineering regular expressions
Consider the following regular expression:
m/test./In a purely ASCII environment (without UTF or Unicode character sets) this will match exactly 255 different strings (it would be 256 but \n won't match). For example, it will match:
- testa
- testb
- testc
- test1
- test2
- test$
- test%
- .......and so on....
But now look at your regular expression. With no other criteria beyond m/(.*)test(.*)/ you have a hopelessly large number of possibilities. Minimally, it could match "test". But it could also match any string of any size (up to the capacity of your computer's memory and swapfile) as long as the sequence 'test' is found somewhere within the string. In fact, given a random string of random length, the likelyhood of finding a match increases as the string grows, for in an infinite sequence of random characters there will exist an infinate number of embeded 'test' sequences.
The point is, there's no point to trying to generate every (or any) string that will match a given regular expression if that RE doesn't somehow limit the number of possibilities to be a manageable quantity.
Dave