in reply to Noodling with natural sorting in perl6

Thanks for your writeup!

Below I'll make some suggestion of how you could take advantage of some nice Perl 6 features, and write more idiomatic Perl 6 code:

my $filename = shift @*ARGS or die "You need to pass in a filename.";

Instead I usually write

sub MAIN($filename) { # rest of the mainline code here }

The MAIN sub does the command line parsing for you, and also gives an automatic usage message.

rx/ \w+ [<punct> \w+]* /

This also be written as rx/ [\w+] ** <punct> /. Here ** is the general quantifier, a ** 4 matches 4 a's, a ** <punct> matches an arbitrary number of a's, separated by punctuation.

sub natural_cmp ($a, $b) { my ($first, $second) = ($a, $b); my ($one, $two) = ($first, $second).map({ .=subst(/(\d+)/, -> $/{ sprintf( "%s%c%s", 0, $0.chars, $0) }, :g).lc }); return ($one cmp $two); }

If you leave out the = in .=subst, no in-place modification takes place; instead the modified value is returned, and the original variable remains untouched.

So you can write

sub natural_cmp ($a, $b) { my ($one, $two) = ($a, $b).map(*.subst(/(\d+)/, -> $/{ sprintf( "%s%c%s", 0, $0.chars, $0) }, :g).lc); return $one cmp $two; }

If you happen to need a copy, you can also say sub mysub($a is copy, $b is copy), so you don't need to come up with more variable names.

Perl 6 - links to (nearly) everything that is Perl 6.

Replies are listed 'Best First'.
Re^2: Noodling with natural sorting in perl6
by thundergnat (Deacon) on Aug 20, 2010 at 18:28 UTC
    rx/ [\w+] ** <punct> /

    Cool, thanks. I am still stuck in perl5 regex think to a large extent.

    sub natural_cmp ($a, $b) { my ($one, $two) = ($a, $b).map(*.subst(/(\d+)/, -> $/{ sprintf( "%s%c%s", 0, $0.chars, $0) }, :g).lc); return $one cmp $two; }

    Ooooo shiney! Like I said, I can be dense sometimes. Even better, do away with all the intermediate variables.

    sub natural_cmp ($a is copy, $b is copy) { return $a.subst(/(\d+)/, -> $/{ sprintf( "%s%c%s", 0, $0.chars, $0) + }, :g).lc cmp $b.subst(/(\d+)/, -> $/{ sprintf( "%s%c%s", 0, $0.chars, $0) }, + :g).lc; }
      That way you duplicated the transformation code.

      Here's a version that uses neither temporaries nor duplicated code:

      sub natural_cmp ($a is copy, $b is copy) { return [cmp] ($a, ).map: *.subst(/(\d+)/, -> $/{ sprintf( "%s%c%s", + 0, $0.chars, $0) }, :g).lc }

      This uses the reduction meta operator.

      Perl 6 - links to (nearly) everything that is Perl 6.