As documented, the default for the first argument of split is ' '. ("As a special case, specifying a PATTERN of space (' ') will split on white space just as split with no arguments does.")

$ perl -MData::Dumper -e'
   $_="   56  1752.eps a b";
   print Dumper split;
'
$VAR1 = '56';
$VAR2 = '1752.eps';
$VAR3 = 'a';
$VAR4 = 'b';

$ perl -MData::Dumper -e'
   $_="   56  1752.eps a b";
   print Dumper split " ";
'
$VAR1 = '56';
$VAR2 = '1752.eps';
$VAR3 = 'a';
$VAR4 = 'b';
[download]

The next problem is that "2" is wrong for the third argument. You'd want to use "3" and ignore the last value returned.

$ perl -MData::Dumper -e'
   $_="   56  1752.eps a b";
   print Dumper split " ", $_, 2;
'
$VAR1 = '56';
$VAR2 = '1752.eps a b';

$ perl -MData::Dumper -e'
   $_="   56  1752.eps a b";
   print Dumper split " ", $_, 3;
'
$VAR1 = '56';
$VAR2 = '1752.eps';
$VAR3 = 'a b';
[download]

Solutions:

for (@data) {
   my @a = (split)[0,1];
   my @a = (split " ", $_, 3)[0,1];
   my ($x, $y) = split(" ", $_, 3);
   my ($x, $y) = split;
   ...
}
[download]

split is optimised so that it doesn't do any unnecessary work for the last one. You could also avoid split entirely.

   my ($x, $y) = /^\s*(\S+)\s+(\S+)/;
[download]

Use split ' ',$_,2. The documentation on split says:

As a special case, specifying a PATTERN of space (' ') will split on white space just as "split" with no arguments does. Thus, "split(' ')" can be used to emulate awk's default behavior, whereas "split(/ /)" will give you as many null initial fields as there are leading spaces. A "split" on "/\s+/" is like a "split(' ')" except that any leading whitespace produces a null first field. A "split" with no arguments really does a "split(' ', $_)" internally.

use strict;
use warnings;

use Data::Dumper;

my @data = ( "  56 1752.eps a b", "  56 2613.eps", "  56 3469.eps",
"   8 INPUT000 a b", "  16 INPUT001", "  16 INPUT002", "  96 MTA.ps",
"  96 MTA.6.ps a b", "  80 MTA.7.ps", "  32 head.eps", "   8 labs",
"   0 lib a b", "   8 mkexe.bat", " 112 out", "   0 screenshots",
"8720 trace.exe a b", "  16 trace.pl", "   8 tracehosts",
"1160 trace.041409.exe a b", "1160 trace.orig.exe",
);

foreach (@data) {
    print STDERR Dumper( (split)[0,1] ), "\n";
}
__END__

$VAR1 = '56';
$VAR2 = '1752.eps';

$VAR1 = '56';
$VAR2 = '2613.eps';
...
[download]

Anon, ++jethro, and ++ikegami, thanks for the great responses.

The reason I asked this question is because I'm seeing this problem when I use Sort::Fields. The script below will show what I mean:

use strict;
use warnings;
use Sort::Fields;
use Data::Dumper;

my @data = ( "  56 1752.eps", "  56 2613.eps", "  56 3469.eps",
"   8 INPUT000", "  16 INPUT001", "  16 INPUT002", "  96 MTA.1.ps",
"  96 MTA.6.ps", "  80 MTA.7.ps", "  32 head.eps", "   8 labs",
"   0 lib", "   8 mkexe.bat", " 112 out", "   0 screenshots",
"8720 trace.exe", "  16 trace.pl", "   8 tracehosts",
"1160 trace.041409.exe", "1160 trace.orig.exe",
);

# Initial spaces in column 1 don't sort the same as...
my @sorted = fieldsort( ['1n'], @data);
print STDERR "First sorted DUMP:\n", Dumper(\@sorted), "\n";

my @data2 = grep s/^/1 /, @data;

# ...initial spaces in column 2!
@sorted = fieldsort( ['2n'], @data2);
print STDERR "Second sorted DUMP:\n", Dumper(\@sorted), "\n";
[download]

If you look in make_fieldsort() sub in the Sort::Fields code, you'll see the nested map commands. I'm just getting comfortable with map, but this nested one is really throwing me for a loop (heh). I just can't seem to come up with the right solution here.

Any help for a poor, confused, I-only-seem-to-be-able-to-understand-non-nested-map-commands type person?

Thanks

-Craig

# Initial spaces in column 1 don't sort the same as...

It's impossible for a column to have initial spaces when spaces is your delimiter. The first field of most of @data is "".

use strict;
use warnings;

use Sort::Fields;
use Data::Dumper;

my @data = ( "  56 1752.eps", "  56 2613.eps", "  56 3469.eps",
"   8 INPUT000", "  16 INPUT001", "  16 INPUT002", "  96 MTA.1.ps",
"  96 MTA.6.ps", "  80 MTA.7.ps", "  32 head.eps", "   8 labs",
"   0 lib", "   8 mkexe.bat", " 112 out", "   0 screenshots",
"8720 trace.exe", "  16 trace.pl", "   8 tracehosts",
"1160 trace.041409.exe", "1160 trace.orig.exe",
);

s/^\s+// for @data;

my @sorted = fieldsort( ['1n'], @data);
print(Dumper(\@sorted));
[download]

By the way, you were using grep as map, and you were clobbering @data in the process.

ikegami-

I'm sorry, but I don't believe I understand your point. It seems that all you did to fix the problem was to remove the initial spaces in the original data.

In my opinion Sort::Fields should sort the data the same way, regardless of where the data is (field 1 or field 2). If you try to numerically sort the output of an 'ls -s' command, you can see the problem clearly:

use strict;
use warnings;
use Sort::Fields;
use Data::Dumper;

my @data = `ls -s`; chomp(@data);

my @sorted = fieldsort( ['1n'], @data);
print(Dumper(\@sorted));
[download]

This doesn't do what is intended, and is why I made the report to the author. I'm sure I could remove the initial spaces for Data::Dumper, then put them back after it's done, but that doesn't seem right to me.

-Craig

Alternatively, you can change the definition of a field.

use strict;
use warnings;

use Sort::Fields;
use Data::Dumper;

my @data = ( "  56 1752.eps", "  56 2613.eps", "  56 3469.eps",
"   8 INPUT000", "  16 INPUT001", "  16 INPUT002", "  96 MTA.1.ps",
"  96 MTA.6.ps", "  80 MTA.7.ps", "  32 head.eps", "   8 labs",
"   0 lib", "   8 mkexe.bat", " 112 out", "   0 screenshots",
"8720 trace.exe", "  16 trace.pl", "   8 tracehosts",
"1160 trace.041409.exe", "1160 trace.orig.exe",
);

my @sorted = fieldsort( "".qr/(?<!^)(?<!\s)\s+/, ['1n'], @data);
print(Dumper(\@sorted));
[download]

ikegami++

Brilliant!

I believe I understand the theory here. Now I just have to go off and figure out the specifics of what ".qr/(?<!^)(?<!\s)\s+/ is actually doing. I'll get it after a while, and will learn a lot in doing so, no doubt!

Nicely done!