Re: [bug-gawk] awk V4 patsplit: many unexpected 0 length elements in output array

[Davide Brini]

On Mon, 17 Feb 2014 08:06:54 -0500, David Kra <address@hidden> wrote:

> Alert: I am not a regexpert. It could be that my regex *does* match the
> null string, though I can't see how. It is a long assembly of several
> shorter regex's, each wrapped in () and then all  ||'d together, as in:
> "(p1)||(p2)||(p3)||(p4)||(p5)"
> 
> "([ABDE][0-9][0-9A-Z]{4}[A-Z])||([ABDE][0-9A-Z]{1}[0-9][0-9A-Z]{3}[A-Z])||([ABDE][0-9A-Z]{2}[0-9][0-9A-Z]{2}[A-Z])||([ABDE][0-9A-Z]{3}[0-9][0-9A-Z]{1}[A-Z])||([ABDE][0-9A-Z]{4}[0-9][A-Z])"
> 
> 
> The goal is to match 7 character strings that start with one of a few
> letters, ends in a letter, has alphanumerics in the middle, but must have
> at least one digit in the middle.

I haven't looked much into it, but the regexp alternation operator is
definitely "|" (a single pipe), not "||".

-- 
D.