Re: match finds wrong space.

[Hermann Peifer]

On 07/07/2010 20:36, Davide Brini wrote:
> On Wed, 07 Jul 2010 21:34:25 +0300 Aharon Robbins<address@hidden>  wrote:
>
> > > regards - Chris Willis in the UK# insjk.awk
> > >
> > > BEGIN {
> > >         s = "Mary Ann jane"
> > >         n = match( s, /\040[a-z]/ )
> > >         print n, s
> > >         }
> >
> > Hi. Current gawk is correct, and 3.0.3 is wrong. You'll note that
> > following the \040 for a space you have [a-z]. This matches *lower case
> > letters*; the "A" following the first first is an upper case letter.
>
> But it's matched in his example.
>
> > So, there's no bug.
>
> He is saying that
>
> match( s, /\040[a-z]/ )
>
> on the line
>
> "Mary Ann jane"
>
> gives 5 (meaning [a-z] matches the "A"), whereas it should give 9.
>
> I explained the reason for that in my post.

Davide,

Soemone else already explained that this is expected behaviour. Unless 
your are in C locale, the character range [a-z] can expanded to just 
about anything. Simplified  examples are:

aBbCc...XxYyZz  or  aAbBcC...xXyYz

Your locale is probably similar to the latter example, this is why it 
matches an uppercase A. In non-C locales, use character classes like 
[:lower:] and [:upper:] instead of character ranges like [a-z] and [A-Z].

Hermann