UNIX for Dummies Questions & Answers

You're missing the point. There are an infinite number of strings that match the regular expression [ab]cd?e. "ace" is one; "ace" preceded or followed by any number of characters also matches. If your RE was anchored on both ends, then your examples would make sense. Even if you say that you will only accept strings that are three or four characters long, there are still 4 + 4 * (number of distinct characters in your current locale's LC_CTYPE definition - 1) strings that will match your RE.

@Don Cragun: Ok I see what you mean.

So to be more precise again, the regex should actually be:

Code:

/^[ab]cd?e$/

PLUS the length of the output must be between 3 and 4 (depending if "d" is included or not).

I am not sure I get this point...

I don't want to match anything actually, I just need to get all the combinations of the regex to see what it might match or not.

regex -> combinations -> match

I am just interested in the second intermediate step.

I understand it may sound weird ...

There are 127 ** 4 + 127 ** 3 combinations of three or four characters for ASCII, 255 ** 4 + 255 ** 3 combinations of three or four characters for EBCDIC and ISO-8859-* code sets, and (several thousand) ** 4 + (several thousand) ** 3 combinations of three or four characters for Unicode. Of those, 4 + 4 * (number of distinct characters in your current locale's LC_CTYPE definition - 1) strings containing three or four characters will match your RE match your RE when your RE does not contain the anchors: "ace", "ace" preceded by any character except NUL, "ace" followed by any character except NUL, "bce", "bce" preceded by any character except NUL, "bce" followed by any character except NUL, "acde", and "bcde".

When your RE includes the anchors, only two out of (number of characters in your code set - 1) ** 3 three character combinations will be matched by your RE: "ace" and "bce"; and only two out of (number of characters in your code set - 1) ** 4 four character combinations will be matched by your RE: "acde" and "bcde".

In both cases the number of strings consisting of combinations of three or four characters depends on the code set underlying the locale you're using; not on the RE you use to match those strings.