The Perl Toolchain Summit needs more sponsors. If your company depends on Perl, please support this very important event.

NAME

Sort::Key::Top - select and sort top n elements

SYNOPSIS

  use Sort::Key::Top (nkeytop top);

  # select 5 first numbers by absolute value:
  @top = nkeytop { abs $_ } 5 => 1, 2, 7, 5, 5, 1, 78, 0, -2, -8, 2;
         # ==> @top = (1, 2, 1, 0, -2)

  # select 5 first numbers by absolute value and sort accordingly:
  @top = nkeytopsort { abs $_ } 5 => 1, 2, 7, 5, 5, 1, 78, 0, -2, -8, 2;
         # ==> @top = (0, 1, 1, 2, -2)

  # select 5 first words by lexicographic order:
  @a = qw(cat fish bird leon penguin horse rat elephant squirrel dog);
  @top = top 5 => @a;
         # ==> @top = qw(cat fish bird elephant dog);

DESCRIPTION

The functions available from this module select the top n elements from a list using several common orderings and custom key extraction procedures.

They are all variations around

  keytopsort { CALC_KEY($_) } $n => @data;

In array context, this function calculates the ordering key for every element in @data using the expression inside the block. Then it selects and orders the $n elements with the lower keys when compared lexicographically.

It is equivalent to the pure Perl expression:

  (sort { CALC_KEY($a) cmp CALC_KEY($b) } @data)[0 .. $n-1];

If $n is negative, the last $n elements from the bottom are selected:

  topsort 3 => qw(foo doom me bar doz hello);
       # ==> ('bar', 'doz', 'doom')

  topsort -3 => qw(foo doom me bar doz hello);
       # ==> ('foo', 'hello', 'me')

  top 3 => qw(foo doom me bar doz hello);
       # ==> ('doom', 'bar', 'doz')

  top -3 => qw(foo doom me bar doz hello);
       # ==> ('foo', 'me', 'hello')

In scalar context, the value returned by the functions on this module is the cutoff value allowing to select nth element from the array. For instance:

  # n = 5;
  scalar(topsort 5 => @data) eq (sort @data)[4]    # true

  # n = -5;
  scalar(topsort -5 => @data) eq (sort @data)[-5]  # true

Note that on scalar context, the sort variations (see below) are usually the right choice:

  scalar topsort 3 => qw(me foo doz doom me bar hello); # ==> 'doz'

  scalar top 3 => qw(me foo doz doom me bar hello); # ==> 'bar'

Note also, that the index is 1-based (it starts at one instead of at zero). The atpos set of functions explained below do the same and are 0-based.

Variations allow to:

- use the own values as the ordering keys
  topsort 5 => qw(a b ab t uu g h aa aac);
     # ==> a aa aac ab b
- use an array or hash index instead of a subroutine to extract the key
  slottop 0, 2, [4], [1], [3], [2], [4];
     # ==> [1], [2]
- return the selected values in the original order
  top 5 => qw(a b ab t uu g h aa aac);
     # ==> a b ab aa aac
- use a different ordering

For instance comparing the keys as numbers, using the locale configuration or in reverse order:

  rnkeytop { length $_ } 3 => qw(a ab aa aac b t uu g h);
     # ==> ab aa aac

  rnkeytopsort { length $_ } 3 => qw(a ab aa aac b t uu g h);
     # ==> aac ab aa

A prefix is used to indicate the required ordering:

(no prefix)

lexicographical ascending order

r

lexicographical descending order

l

lexicographical ascending order obeying locale configuration

r

lexicographical descending order obeying locale configuration

n

numerical ascending order

rn

numerical descending order

i

numerical ascending order but converting the keys to integers first

ri

numerical descending order but converting the keys to integers first

u

numerical ascending order but converting the keys to unsigned integers first

ru

numerical descending order but converting the keys to unsigned integers first

- select the head element from the list sorted
  nhead 6, 7, 3, 8, 9, 9;
      # ==> 3

  nkeyhead { length $_ } qw(a ab aa aac b t uu uiyii)
      # ==> 'a'
- select the tail element from the list sorted
  tail qw(a ab aa aac b t uu uiyii);
      # ==> 'uu'

  nkeytail { length $_ } qw(a ab aa aac b t uu uiyii)
      # ==> 'uiyii'
- select the element at position n from the list sorted
  atpos 3, qw(a ab aa aac b t uu uiyii);
      # ==> 'ab';

  rnkeyatpos { abs $_ } 2 => -0.3, 1.1, 4, 0.1, 0.9, -2;
      # ==> 1.1

  rnkeyatpos { abs $_ } -2 => -0.3, 1.1, 4, 0.1, 0.9, -2;
      # ==> -0.3

Note that for the atpos set of functions indexes start at zero.

- return a list composed by the elements with the first n ordered keys and then the remaining ones.
  ikeypart { length $_ } 3 => qw(a bbbb cc ddddd g fd);
      # ==> a cc g bbbb ddddd fd
- return two arrays references, the first array containing the elements with the first n ordered keys and the second with the rest.
  keypartref { length $_ } 3 => qw(a bbbb cc ddddd g fd);
      # ==> [a cc g] [bbbb ddddd fd]

The full list of available functions is:

  top ltop ntop itop utop rtop rltop rntop ritop rutop

  keytop lkeytop nkeytop ikeytop ukeytop rkeytop rlkeytop rnkeytop
  rikeytop rukeytop

  slottop lslottop nslottop islottop uslottop rslottop rlslottop rnslottop
  rislottop ruslottop

  topsort ltopsort ntopsort itopsort utopsort rtopsort rltopsort
  rntopsort ritopsort rutopsort

  keytopsort lkeytopsort nkeytopsort ikeytopsort ukeytopsort
  rkeytopsort rlkeytopsort rnkeytopsort rikeytopsort rukeytopsort

  slottopsort lslottopsort nslottopsort islottopsort uslottopsort
  rslottopsort rlslottopsort rnslottopsort rislottopsort ruslottopsort

  head lhead nhead ihead uhead rhead rlhead rnhead rihead ruhead

  keyhead lkeyhead nkeyhead ikeyhead ukeyhead rkeyhead rlkeyhead
  rnkeyhead rikeyhead rukeyhead

  slothead lslothead nslothead islothead uslothead rslothead rlslothead
  rnslothead rislothead ruslothead

  tail ltail ntail itail utail rtail rltail rntail ritail rutail

  keytail lkeytail nkeytail ikeytail ukeytail rkeytail rlkeytail
  rnkeytail rikeytail rukeytail

  slottail lslottail nslottail islottail uslottail rslottail rlslottail
  rnslottail rislottail ruslottail

  atpos latpos natpos iatpos uatpos ratpos rlatpos rnatpos riatpos
  ruatpos

  keyatpos lkeyatpos nkeyatpos ikeyatpos ukeyatpos rkeyatpos
  rlkeyatpos rnkeyatpos rikeyatpos rukeyatpos

  slotatpos lslotatpos nslotatpos islotatpos uslotatpos rslotatpos
  rlslotatpos rnslotatpos rislotatpos ruslotatpos

  part lpart npart ipart upart rpart rlpart rnpart ripart
  rupart

  keypart lkeypart nkeypart ikeypart ukeypart rkeypart
  rlkeypart rnkeypart rikeypart rukeypart

  slotpart lslotpart nslotpart islotpart uslotpart rslotpart
  rlslotpart rnslotpart rislotpart ruslotpart

SEE ALSO

Sort::Key, "sort" in perlfunc.

Sort::Key::Top::PP by Toby Inkster, provides a subset of the API of Sort::Key::Top and is written in pure Perl.

The Wikipedia article about selection algorithms http://en.wikipedia.org/wiki/Selection_algorithm.

COPYRIGHT AND LICENSE

Copyright (C) 2006-2008, 2011, 2012, 2014 by Salvador Fandiño (sfandino@yahoo.com).

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.8.8 or, at your option, any later version of Perl 5 you may have available.