use feature ':5.10';
use strict;

use List::MoreUtils qw( natatime  );

my $test_sentence
    = "Perl is a high-level, general-purpose, interpreted, dynamic programming language.";

sub uniq_phrases {
    my $sentence = shift;
    my ( $min, $max );
    given ( scalar @_ ) {
        when (2) { ( $min, $max ) = @_; }
        when (1) { ( $min, $max ) = ($_[0], $_[0]); }
        default { ( $min, $max ) = ( 2, 4 ); };
    }

    my @words = split( /\s+/, $sentence );
    my @pairs;
    for my $size ( $min .. $max ) {
        my %seen;
        for my $window ( 0 .. ( $#words - $size ) ) {
            my $it = natatime $size, @words[ $window .. $#words ];
            while ( my @p = $it->() ) {
                next if @p != $size;
                my $p = join( " ", @p );
                next if $seen{$p}++;
                push @pairs, $p;
            }
        }
    }

    return wantarray ? @pairs : \@pairs;
}

say join( "\n", sort { $a cmp $b } uniq_phrases( $test_sentence, 4 ) );

__END__

##</code><code>##

(use '[clojure.contrib.str-utils :only (re-split str-join)])

(def test-sentence
  "Perl is a high-level, general-purpose, interpreted, dynamic programming language.")

(defn uniq-phrases
  "Find unique phrases of the given lengths (def 2-4 words; max may be omitted)"
  ([sentence] (uniq-phrases sentence 2 4))
  ([sentence len] (uniq-phrases sentence len len))
  ([sentence min-len max-len]
    (let [words (re-split #"\s+" sentence)
          pairs (set (mapcat #(partition %1 1 words) (range min-len (inc max-len))))]
      (map #(str-join " " %) pairs))))

(defn print-phrases [p]
  (println (str-join "\n" (sort p))))

(print-phrases (uniq-phrases test-sentence 2 9))

; user=> (time (print-phrases (uniq-phrases test-sentence 4 4)))
; Perl is a high-level,
; a high-level, general-purpose, interpreted,
; general-purpose, interpreted, dynamic programming
; high-level, general-purpose, interpreted, dynamic
; interpreted, dynamic programming language.
; is a high-level, general-purpose,
; "Elapsed time: 1.227 msecs"