On Thu, May 6, 2010 at 5:39 PM, Jan Friedel <jan.frie...@sun.com> wrote:
>
>
> Hi all,
>
> hope, I didn't write to the wrong alias. On the most recent
> OpenSolaris build I run, say, the following set of commands in
> my ksh93 script:
>
> typeset f2cptr="arg1 [arg]2 [arg]3 [arg]4"
> print $(for i in ${f2cptr}; do print "${i/*[a-zA-Z\]]}"; done)
>
> The output is: arg1 2 3 4
> ,while the expected output is: 1 2 3 4
>
>
> Even more strange is the following case:
>
> typeset f2cptr="arg1 [arg]2 [arg]3 [arg]4"
> print $(for i in ${f2cptr}; do print "${i/*[\]a-zA-Z]}"; done)
>
> The output is: arg1 [arg]2 [arg]3 [arg]4
> ,while the expected output is the same: 1 2 3 4
>
> Any idea? It seems to be some kind of bug introduced sometime
> recently in parsing mechanism of []. Thanks,
>
> /j.
I get the same for grep -E patterns:
ksh93 -o noglob -c 'typeset f2cptr="arg1 [arg]2 [arg]3 [arg]4" ; for i
in ${f2cptr} ; do print "${i/~(Eli)[a-z\[\]]*/}"; done'
rg1
arg]2
arg]3
arg]4
Splitting the \] out of the [...] is a workaround:
ksh93 -o noglob -c 'typeset f2cptr="arg1 [arg]2 [arg]3 [arg]4" ; for i
in ${f2cptr} ; do print "${i/~(Eli)([a-z]|\[[a-z]*\])*/}"; done'
1
2
3
4
Using the shell OR operator @(a|b) works, too. a matches *[a-z] and b
matches *[a-z]\]
ksh93 -o noglob -c 'typeset f2cptr="arg1 [arg]2 [arg]3 [arg]4" ; for i
in ${f2cptr} ; do print "${i/*@([a-z]|[a-z]\])}"; done'
1
2
3
4
Glenn, a bug for you: This is the same line as the last above but uses
the ~(Si) prefix for case insensitive shell pattern. But it does not
work:
ksh93 -o noglob -c 'typeset f2cptr="arg1 [arg]2 [arg]3 [arg]4" ; for i
in ${f2cptr} ; do print "${i/~(Si)*@([a-z]|[a-z]\])}"; done'
arg1
[arg]2
[arg]3
[arg]4
Chris
--
^---^
(@)v(@) Chris Pickett
| / IT consultant
===m==m=== pkch...@users.sourceforge.net
_______________________________________________
ksh93-integration-discuss mailing list
ksh93-integration-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/ksh93-integration-discuss
