But the numbers presented are zero-padded integers, so the comparison is equivalent. If the numeric comparator were used with the simple split, a non-numeric comparison warning will be thrown (he does have warnings on, right?). Not all files listed have the .jpg file extension, so (split /_|\./, $a)[-2] won't work. I was also trying to keep it short/simple, preventing a long code block and making it easier for the OP to build his own. The best I came up with for using the numeric comparator sorted based upon the last block of digits:
my $re = qr/\d+/;
print join "\n", sort {($a =~ /$re/g)[-1] <=> ($b =~ /$re/g)[-1]; } @n
+ames
But then I'd have to link to perlretut and Regexp Quote Like Operators as well... | [reply]
[d/l]
[select] |
... the numbers presented are zero-padded integers [with three digits each], so the comparison is equivalent.
Aye, there's the rub. The OPer only presents three examples and, in general, about 35% of the info needed to build some robust code. Numeric and string comparison are the same until someone sez "Oh, by the way..."
It's just a guess, but I'd be inclined to fortify myself on the numeric side based on the presence of the word "number" in the OP: confronted with real-world data (whatever that may be), a numeric comparison seems more likely to continue to work.
I was thinking of something along the lines of Marshall:
>perl -wMstrict -le
"my @names = qw(FLY_ABC123_01.jpg PG_ABC123_03 SPL_ABC123_02.jpg
FOO_ABC123_011 SPL_ABC123_021 SPL_ABC123_021.jpg);
sub sort_123 {
return
map $_->[0],
sort { $a->[1] <=> $b->[1] }
map [ $_, m{ .* _ (\d+) }xms ],
@_
;
}
my @sorted = sort_123(@names);
print qq{'$_'} for @sorted;
"
'FLY_ABC123_01.jpg'
'SPL_ABC123_02.jpg'
'PG_ABC123_03'
'FOO_ABC123_011'
'SPL_ABC123_021'
'SPL_ABC123_021.jpg'
| [reply]
[d/l] |