# Arrays and functions: Sort numerically

By default, “sort” will sort alphabetically by value. But if we’re willing to write our own subroutine we can sort by pretty much any criteria we want. Create a “byArtistCount” subroutine:

sub byArtistCount {
return \$artists{\$b} <=> \$artists{\$a};
}

Add this to the sort line:

@artists = sort byArtistCount @artists;

And run the command again:

./show --format summary stand songs.txt

The top four artists should be Judy Garland, Bing Crosby, The Lennon Sisters, and Linda Ronstadt, at 51 songs, 23 songs, 14 songs, and 12 songs, respectively.

This subroutine is a special one for sorts. When sort calls a sort subroutine it is asking that subroutine which of two items should come first. Perl automatically puts the first item in \$a and the second item in \$b. If the subroutine returns a negative 1, sort assumes that \$a comes first. If the subroutine returns a positive 1, sort assumes that \$b comes first. If the subroutine returns a 0, sort assumes that both can be ordered either way.

The “<=>”” is a useful operator for sort subroutines, because it returns a negative 1 if the number on the left is lower, a positive 1 if the number on the right is lower, and a zero if both numbers are the same.

Which means that this subroutine ends up sorting the artist names according to their count in %artists; and because I’ve put \$b on the left and \$a on the right, it sorts in descending order.

If you’re comparing two pieces of text rather than two numbers, the “cmp” operator does the same thing for text that “<=>” does for numbers. Here’s a quick script that lets you play around with compares:

#!/usr/bin/perl

\$item1 = shift;
\$item2 = shift;

print "Text compare: ", \$item1 cmp \$item2, "\n";
print "Number compare: ", \$item1 <=> \$item2, "\n";

Call it “compare”, make sure you set it to chmod u+x, and play around with giving it two items:

./compare hello world

./compare 3 5