Better way to sort an array with unusual value - arrays

Question

I have this array:
#raw_stack = (
'900244~dfasdf~ddd3',
'900122~dfasdf~ddd1',
'900244~dfasdf~ddd2',
'900456~dfasdf~ddd4',
'900312~dfasdf~ddd3',
'900456~dfasdf~ddd5',
);
I'd like to sort it by the first '~' element.
Is there a more elegant way to solve this rather than
looping and splitting through each value?

Answer 1

Use Schwartzian transform:
my #raw_stack = (
'900244~dfasdf~ddd3',
'900122~dfasdf~ddd1',
'900244~dfasdf~ddd2',
'900456~dfasdf~ddd4',
'900312~dfasdf~ddd3',
'900456~dfasdf~ddd5',
);
my #sorted =
map { $_->[0] }
sort { $a->[1] <=> $b->[1] }
map { [$_, (split/~/)[0]] } #raw_stack;
dump#sorted;
Benchmark:
#!/usr/bin/perl
use 5.010;
use strict;
use warnings;
use Benchmark qw(:all);
my $s = '~dfasdf~ddd3';
my #arr = ();
for(0..20000) {
push #arr, int(rand(100000)) . $s;
}
my $count = -3;
cmpthese($count, {
'ST' => sub {
my #sorted =
map { $_->[0] }
sort { $a->[1] <=> $b->[1] }
map { [$_, (split/~/)[0]] } #arr;
},
'SORT' => sub {
my #sorted =
sort {
my ($a_0) = split /~/, $a;
my ($b_0) = split /~/, $b;
$a_0 <=> $b_0
} #arr;
},
});
result:
array of 200 elements:
Rate SORT ST
SORT 267/s -- -61%
ST 689/s 158% --
array of 2000 elements:
Rate SORT ST
SORT 18.0/s -- -71%
ST 61.5/s 242% --
array of 20000 elements:
Rate SORT ST
SORT 1.35/s -- -73%
ST 4.96/s 266% --

Answer 2

Sort and list slices?
sort { ( split( /~/, $a ) )[0] <=> ( split( /~/, $b ) )[0] } #raw_stack;

Answer 3

These might help. They show you how to extract parts of strings to use them to sort the larger strings:
Perlmonks How do I sort an array by (anything)?
perlfaq4 How do I sort an array by (anything)?
Stackoverflow Perl - Sort CSV on a certain column?

Answer 4

Is it always 6 digits? If so, the following would be the simplest and fastest:
my #sorted_stack = sort #raw_stack;
If not,
my #sorted_stack =
sort {
my ($a_0) = split /~/, $a;
my ($b_0) = split /~/, $b;
$a_0 <=> $b_0
} #raw_stack;
A Schwartzian transform might be cleaner if you're used to that, but it's actually slower in this case: [Update: Apparently, it's actually faster than my second solution for larger lists. It's never faster than the first, though ]
my #sorted_stack =
map $_->[0],
sort { $a->[1] <=> $b->[1] }
map [ $_, split /~/ ],
#raw_stack;