Zero-width assertions fail with split

Ovid has asked for the wisdom of the Perl Monks concerning the following question:

It's late and I'm tired, but I'm banging my head against a silly little problem. It's a long story as to why I need this, but I've stumbled on the following curious problem:

#!/usr/bin/perl

use strict;
use warnings;
use Test::More qw/no_plan/;

my $token  = '-----';
my $data   = "1,2,${token}0${token},4,5";
my $split  = qr/(?<!$token),(?!$token)/;
my @fields = split $split, $data;
my @expected = ( 1,2, "${token}0${token}", 4, 5 );
is_deeply \@fields, \@expected;
[download]

I need to be able to split on a value if and only if that value is not immediately preceded and followed by the same fixed-width string. The code above actually assigns the following to @fields:

@fields = (
  '1',
  '2,-----0-----,4',
  '5'
);
[download]

It's as if it's doing a logical OR with the two zero-width assertions instead of a logical AND. I suppose it's not unreasonable that it do this, but is there some simple way I can enforce my desired behavior of splitting the resulting string to this?

@fields = (
  '1',
  '2',
  '-----0-----',
  '4',
  '5'
);
[download]

Also, where is this behavior documented? I'm sure it is somewhere. I just can't find it.

Cheers,
Ovid

New address of my CGI Course.

Comment on Zero-width assertions fail with split Select or Download Code

Replies are listed 'Best First'.
Re: Zero-width assertions fail with split by merlyn (Sage) on Sep 19, 2005 at 05:00 UTC
De Morgan is kicking your hiney. It's working as I would expect. You are asking to split on commas that have neither dashes before NOR dashes afterwards, as in "the delimiter is NOT dashes ahead AND a comma AND NOT dashes afterwards". That's the only time the split regex would match. Dashes on either side would prevent the comma from being considered. Your experimental results are borne out. And simply splitting on commas would give you the list you want, so you'll have to show an example where simply splitting on commas doesn't do it. -- Randal L. Schwartz, Perl hacker Be sure to read my standard disclaimer if this is a reply.	[reply]
Re^2: Zero-width assertions fail with split by Ovid (Cardinal) on Sep 19, 2005 at 05:06 UTC
That makes sense and, as noted, the results aren't entirely suprising, though I confess I struggled with it. Is there some simple way of getting the split behavior to do what I want? Cheers, Ovid New address of my CGI Course.	[reply]
Re^3: Zero-width assertions fail with split by hv (Prior) on Sep 19, 2005 at 12:00 UTC
You want "not both preceded and followed by TOKEN", which is equivalent to "either not preceded by TOKEN, or not followed by TOKEN", and that we can do: `my $split = qr/ (?<!$token) , \| , (?!$token) /x;` [download] Hugo	[reply] [d/l]
Re^3: Zero-width assertions fail with split by gargle (Chaplain) on Sep 19, 2005 at 06:16 UTC
Hi, It may be silly, but, doesn't this do what you want? `perl -e '$a = "1,2,---0---,3,4";@f=split ",",$a;map { print $_."\n"} @f'` [download] -- if ( 1 ) { $postman->ring() for (1..2); }	[reply] [d/l]
Re: Zero-width assertions fail with split by BrowserUk (Patriarch) on Sep 19, 2005 at 07:24 UTC
How's this? `#! perl -slw use strict; my $test = 'xxx,xyyy,zzz,zzz,ppp,qqq,rrr'; my $re = qr[ (?<!zzz),(?!zzz) # a comma neither preceeded nor followed by the +token. \| (?<=zzz),(?!zzz) # or predeeded but not followed \| (?<!zzz),(?=zzz) # or not preceeded but followed. ]x; print for split $re, $test; __END__ P:\test>junk xxx xyyy zzz,zzz ppp qqq rrr` [download] Examine what is said, not who speaks -- Silence betokens consent -- Love the truth but pardon error. Lingua non convalesco, consenesco et abolesco. -- Rule 1 has a caveat! -- Who broke the cabal? "Science is about questioning the status quo. Questioning authority". The "good enough" maybe good enough for the now, and perfection maybe unobtainable, but that should not preclude us from striving for perfection, when time, circumstance or desire allow.	[reply] [d/l]
Re: Zero-width assertions fail with split by sk (Curate) on Sep 19, 2005 at 06:50 UTC
I think your `-----0-----` led to everyone suggest just split on , but i think you had `-----,-----` in mind. Here is my take at this but does not do it in one split and not sure if it is even possible! #!/usr/bin/perl use strict; use warnings; use Data::Dumper; my $token = '-----'; my $data = "1,2,${token}0${token},4,5"; my $split = qr/(?<=$token),(?=$token)/; my @fields = split $split, $data; my @newfields = (); my $i = 0; while ($i < @fields/2) { my @f1 = split /,/,$fields[$i]; my @f2 = split /,/,$fields[$i+1]; my @dash = (pop(@f1), shift(@f2)); @newfields = (@newfields, @f1,join (',',@dash),@f2); $i += 2; } my @expected = ( 1,2, "${token}0${token}", 4, 5 ); print Dumper (\@newfields); print Dumper (\@expected); __END__ $VAR1 = [ '1', '2', '-----0-----', '4', '5,' ]; $VAR1 = [ 1, 2, '-----0-----', 4, 5 ]; [download] if you pass in `my $data = "1,2,${token},${token},4,5";` Output `$VAR1 = [ '1', '2', '-----,-----', '4', '5' ];` [download]	[reply] [d/l] [select]
Re: Zero-width assertions fail with split by demerphq (Chancellor) on Sep 19, 2005 at 06:30 UTC
This behaviour is as I expect. Split on any comma not preceded by '----' and not followed by '----'. To get the desired output simply split on a comma. I think the problem here is in your expectation and not in the code itself. --- $world=~s/war/peace/g	[reply]