question about glob qualifier format (#qx)

question about glob qualifier format (#qx)

[Han Pingtian]

Hello,

I'm confused about the usage of glob qualifier format "(#qx)". The man
page says:

       				......    Also, as the syntax  is  unambiguous,
       the  expression  will  be  treated  as glob qualifiers just as long any
       parentheses contained within it are balanced; appearance of `|', `(' or
       `~'  does  not  negate the effect. 

But looks like if I put them in parentheses, it will not take effect any more:

	
	$ ls
	1.c  2.h  a  b  slink
	$ echo *((#q/))
	1.c 2.h a b slink

So I'd like to know how to comprehend the manpage here? Thanks in
advance!

Re: question about glob qualifier format (#qx)

[Peter Stephenson]

On Fri, 20 Sep 2013 19:11:10 +0800
Han Pingtian <hanpt@linux.vnet.ibm.com> wrote:
> I'm confused about the usage of glob qualifier format "(#qx)". The man
> page says:
> 
>        				......    Also, as the syntax  is  unambiguous,
>        the  expression  will  be  treated  as glob qualifiers just as long any
>        parentheses contained within it are balanced; appearance of `|', `(' or
>        `~'  does  not  negate the effect. 
> 
> But looks like if I put them in parentheses, it will not take effect any more:
> 
> 	
> 	$ ls
> 	1.c  2.h  a  b  slink
> 	$ echo *((#q/))
> 	1.c 2.h a b slink
> 
> So I'd like to know how to comprehend the manpage here? Thanks in
> advance!

Because of the ambiguity between glob qualifiers without the #q and
parenthesised groups, the shell guesses which is which.  An expression
containing "|" makes it guess that it's a parenthesised group rather
than a set of glob qualifiers.  However, if the *first* open parenthesis
(there's no reason to double them) is followed by #q (and EXTENDED_GLOB
is set), it knows for sure that this is a glob qualifier, so it doesn't
need to do any guessing.

This is there as an additional help; you don't need to change your
normal use of glob qualifiers just because you've signalled them
explicitly with #q.  (#q/) works fine.

pws

Re: question about glob qualifier format (#qx)

[Han Pingtian]

On Fri, Sep 20, 2013 at 12:38:30PM +0100, Peter Stephenson wrote:
> On Fri, 20 Sep 2013 19:11:10 +0800
> Han Pingtian <hanpt@linux.vnet.ibm.com> wrote:
> > I'm confused about the usage of glob qualifier format "(#qx)". The man
> > page says:
> > 
> >        				......    Also, as the syntax  is  unambiguous,
> >        the  expression  will  be  treated  as glob qualifiers just as long any
> >        parentheses contained within it are balanced; appearance of `|', `(' or
> >        `~'  does  not  negate the effect. 
> > 
> > But looks like if I put them in parentheses, it will not take effect any more:
> > 
> > 	
> > 	$ ls
> > 	1.c  2.h  a  b  slink
> > 	$ echo *((#q/))
> > 	1.c 2.h a b slink
> > 
> > So I'd like to know how to comprehend the manpage here? Thanks in
> > advance!
> 
> Because of the ambiguity between glob qualifiers without the #q and
> parenthesised groups, the shell guesses which is which.  An expression
> containing "|" makes it guess that it's a parenthesised group rather
> than a set of glob qualifiers.  However, if the *first* open parenthesis
> (there's no reason to double them) is followed by #q (and EXTENDED_GLOB
> is set), it knows for sure that this is a glob qualifier, so it doesn't
> need to do any guessing.
> 
> This is there as an additional help; you don't need to change your
> normal use of glob qualifiers just because you've signalled them
> explicitly with #q.  (#q/) works fine.
> 
> pws

But I'm still confusing on the manpage :)

If I want to list all symbolic links and directories in current
directory, this expression doesn't work:
    
    $ echo (*(#q@)|*(#q/))

we must write it as :

    $ echo *(#q@) *(#q/)

I cannot see any difference between '(#qx)' format and 'bare glob
qualifier' format on being disabled by '|', '('. Please advise. Thanks.

Re: question about glob qualifier format (#qx)

[Bart Schaefer]

On Sep 21,  7:39am, Han Pingtian wrote:
}
} But I'm still confusing on the manpage :)
} 
} If I want to list all symbolic links and directories in current
} directory, this expression doesn't work:
}     
}     $ echo (*(#q@)|*(#q/))

A few lines above the excerpt you previously quoted from the manual, it
says:

  If the option EXTENDED_GLOB is set, a different syntax for glob
  qualifiers is available, namely `(#qx)' where x is any of the same glob
  qualifiers used in the other format.  The qualifiers must still appear
  at the end of the pattern.

} I cannot see any difference between '(#qx)' format and 'bare glob
} qualifier' format on being disabled by '|', '('. Please advise.

The difference is with e.g. (.).  With BARE_GLOB_QUAL, *(.) matches all
plain files.  With NO_BARE_GLOB_QUAL, *(.) matches files whose name ends
in a dot, and *(#q.) is needed to match all plain files.

} we must write it as :
} 
}     $ echo *(#q@) *(#q/)

  More than one of these lists can be combined, separated by commas. The
  whole list matches if at least one of the sublists matches (they are
  `or'ed, the qualifiers in the sublists are `and'ed).

So:

    % echo *(#q@,/)      

*should* do what you want, but see below about potential bugs ....

I accidentally encountered some odd behavior while confirming this.
With NO_EXTENDED_GLOB, #q is not supposed to be available to introduce
qualifiers.  However

    % setopt NO_EXTENDED_GLOB
    % echo *(#q@)
    <list of symlinks>

Whereas

    % echo *(#q/)
    zsh: unknown file attribute

This is inconsistent, that is, sometimes (#q@) will also give "unknown"
and (#q/) will work.  The more I play with it the less consistently it
behaves.  *(#q@,/) may produce any of
    zsh: unknown file attribute
    zsh: bad pattern: *(#q@,/)
or the intended list of files, depending on ... well, I can't tell what
it depends on, possibly previous globs or what the current directory is,
or how often EXTENDED_GLOB or BARE_GLOB_QUAL have been toggled on and
off.  In fact once I even got:

    % setopt CSH_NULL_GLOB
    % print -l *(#q@) *(#q/)
    zsh: bad pattern: *(#q/)

(even when */ matches several directories, and *(/) matches them with
BARE_GLOB_QUAL).  So something about the parsing for #q may be wonky.

All of this works very nicely when BARE_GLOB_QUAL is enabled, which it
is by default, which is probably why nobody noticed before that #q is
doing strange things.