File Coverage

blib/lib/App/RecordStream/Aggregator/PercentileMap.pm

Criterion	Covered	Total	%
statement	15	52	28.8
branch	0	4	0.0
condition			n/a
subroutine	5	14	35.7
pod	0	8	0.0
total	20	78	25.6

package App::RecordStream::Aggregator::PercentileMap;

our $VERSION = "4.0.23";

use strict;
use warnings;

use App::RecordStream::Aggregator::InjectInto::Field;
use App::RecordStream::DomainLanguage::Registry;

use base qw(App::RecordStream::Aggregator::InjectInto::Field);

sub _make_percentiles
{
  my $percentiles = shift;

  if(ref($percentiles) eq "ARRAY")
  {
    return $percentiles;
  }

  # be careful, split(' ', ...) is extreme magic split, not split on one space
  return [split(' ', $percentiles)];
}

sub new
{
  my $class       = shift;
  my $percentiles = shift;
  my $field       = shift;

  my $this = $class->SUPER::new($field);
  $this->{'percentiles'} = _make_percentiles($percentiles);

  return $this;
}

sub new_from_valuation
{
  my $class       = shift;
  my $percentiles = shift;
  my $valuation   = shift;

  my $this = $class->SUPER::new_from_valuation($valuation);
  $this->{'percentiles'} = _make_percentiles($percentiles);

  return $this;
}

sub initial {
  return [];
}

sub combine_field
{
  my $this   = shift;
  my $cookie = shift;
  my $value  = shift;

  push @$cookie, $value;
  return $cookie;
}

sub squish
{
  my $this   = shift;
  my $cookie = shift;

  my @sorted = sort { $a <=> $b } @$cookie;

  my %ret;

  for my $percentile (@{$this->{'percentiles'}})
  {
    my $index = int((scalar @sorted) * ($percentile / 100));

    if($index == scalar(@sorted))
    {
      $index--;
    }

    $ret{$percentile} = $sorted[$index];
  }

  return \%ret;
}

sub short_usage
{
  return "map of percentile values for field";
}

sub long_usage
{
  print <
Usage: percmap,,
   Finds the field values which  percent of values are less than.

   This is computed by creating an array of all values, sorting, and indexing
   into it at the floor((percentile / 100) * length) point

    will be perl split to determine percentiles to compute.

   Output is a hash whose keys are percentiles and whose values are
   corresponding field values.
EOF
}

sub argct
{
  return 2;
}

App::RecordStream::Aggregator->register_implementation('percentilemap', __PACKAGE__);
App::RecordStream::Aggregator->register_implementation('percmap', __PACKAGE__);

App::RecordStream::DomainLanguage::Registry::register_vfn(__PACKAGE__, 'new_from_valuation', 'percentilemap', 'SCALAR', 'VALUATION');
App::RecordStream::DomainLanguage::Registry::register_vfn(__PACKAGE__, 'new_from_valuation', 'percmap', 'SCALAR', 'VALUATION');

1;