array => hash of occurrence indexes

[pldanutz]

Better / shorter ways to write this? It takes a list and returns a hash of arrays (key => list of indices where key appears in argument)

use 5.012; use warnings; use Data::Dumper;
sub occurrences {
  my %h;
  while (my ($i, $k) = each (@_)) {
    if (! $h {$k}) { $h {$k} = [] }
    push $h {$k}, $i
  }
  return %h
}

my %h = occurences (2, 1, 3, 4, 3, 4, 4, 5 ); print Dumper (\%h)'

[BrowserUk]

Not so much shorter, but avoids the 5.12 dependency:

sub occurs{ 
    my %h; 
    push @{ $h{ $_[$_] } }, $_ for 0 .. $#_; 
    return \%h 
};;

pp occurs( 2, 1, 3, 4, 3, 4, 4, 5 );;
{ 1 => [1], 2 => [0], 3 => [2, 4], 4 => [3, 5, 6], 5 => [7] }

/blockquote

[kcott]

G'day pldanutz,

I generally think it's better to pass around scalars than entire data structures, so returning a reference to that hash would be my preference.

If you wrapped $h{$k} in @{...}, you could dispense with if (! $h {$k}) { $h {$k} = [] } entirely.

Even if you're using 5.012 for other reasons elsewhere in your code, I see little, if any, value in using each here. It's easy to write it for any Perl5.

Putting all that together, you could code occurrences() like this instead:

sub occurences {                                                    
    my %h;          
    my $i = 0;
    push @{$h{$_}}, $i++ for @_;
    return \%h;
}

If you were looking for a golf answer, you could start with the following and then proceed to remove whitespace until illegibility was achieved. :-)

sub occurences { my ($i, %h) = 0; push @{$h{$_}}, $i++ for @_; \%h }

Here's my tests:

<Reveal this spoiler or all in this thread>

[choroba]

Pushing to the $i? That is a bit too much for 5.12.

[pldanutz]

Fixed. Didn't copy/paste from the right source.

[ww]

Please, mark changes to the original IN the original... and include a note that the post has been updated, in the post. Do NOT delete/replace; use strikeouts.