in reply to Insert Into a Sorted Array

my @ary = (1..4, 6..10);
print "@ary\n";

for my $val( 0, 5, 11 ) {
    bubble( \@ary, $val );
    print "@ary\n";
}

sub bubble {
    unshift @{$_[0]}, $_[1];
    for my $i( 0..@{$_[0]}-2 ) {
        last if $_[0]->[$i] <= $_[0]->[$i+1];
        ( $_[0]->[$i],$_[0]->[$i+1] ) = ( $_[0]->[$i+1],$_[0]->[$i] )
    }
}

__DATA__
1 2 3 4 6 7 8 9 10
0 1 2 3 4 6 7 8 9 10
0 1 2 3 4 5 6 7 8 9 10
0 1 2 3 4 5 6 7 8 9 10 11

[download]

cheers

tachyon

Replies are listed 'Best First'.
Re^2: Insert Into a Sorted Array
by ketema (Scribe) on Sep 29, 2004 at 03:20 UTC
    This works, but is it O(N)? Sort algorithms, man I think I remember this, can't I do a binary search? Would that be faster to find the correct placement, or am I getting terms mixed up? Thank you
[reply]
      Here's your bin search. Worst case time for the search (not counting the splicing): O(log N) 
      sub compare {
  $a cmp $b
}


sub binsearch {
   # Returns the index of the position before
   # the one where $s would be found, when $s
   # is not found.

   my ($f, $s, $list) = @_;

   my $i = 0;
   my $j = $#$list;
   my $k;
   my $c;

   $a = $s;

   for (;;) {
      $k = int(($j-$i)/2) + $i;

      $b = $list->[$k];
      $c = &$f();

      return $k if ($c == 0);

      if ($c < 0) {
         $j = $k-1;
         return $j if ($i > $j);
      } else {
         $i = $k+1;
         return $k if ($i > $j);
      }      
   }
}


@a = qw( z b x c y a );
@b = qw( o m n );

@a = sort compare @a;
@b = sort compare @b;

splice(@a, binsearch(\&compare, $b[0], \@a)+1, 0, @b);

print(@a);  # abcmnoxyz

      [download]

      Of course, in your case, it will be
      splice(@known, binsearch(\&byIndex, $newValue, \@known)+1, 0, $newValue);

      Update: Passing array to binsearch using a reference now. oops!

[reply]
[d/l]
[select]

        Passing @a not \@a means that perl copies the entire array to the function for no reason, and thus loses a lot of the potential efficiency.

[reply]
        This is the fastest, especially on a 10,000 record file.
        Oh I wanted to give thanks to all of you who helped me earlier on Figuring out how to embed c into PERL. Still not perfect at it, but i got it to work for what I need. Thanks Again to all you Wise monks. Ketema
[reply]

      Yes you can do a binary search. Yet is will find the correct placement faster for large lists. In fact I have a module on cpan called File::SortedSeek that implements binary searches in large files so I appreciate the algorithm :-) However in this scenario having found the index you then have a practical problem. It is not possible to insert a value into the middle of an array, which is really just a contiguous sequence of SV* What you have to do is make space for it. Splice does that by moving a large chunk (or the entire array). The bubble algorithm also does it and is (in past testing) about twice as fast as just calling sort. Here is a binary search implementation. Benchmarking the options is left to you. 

      my @ary = (1..4, 6..10);

for my $val( 0, 5, 11, 0, 5, 11 ) {
    binary( \@ary, $val );
    print "@ary\n";
}

sub binary {
    my ( $ary, $val ) = @_;
    my ( $min, $max, $last, $i ) = ( 0, scalar @{$ary}, 0, 0 );
    while ( 1 ) {
        $i = int( ($min+$max)/2 );
        # print "i=$i\n";  
      last if $last == $i;
        $last = $i;
        if ( $ary->[$i] < $val ) {
            $min = $i;
        }
        elsif ( $ary->[$i] > $val ) {
            $max = $i;
        }
        else {
            # values are equal so we have a valid index
            last;
        }
    } 
    $i++ if $i;   # for index 0 we want that, otherwise we want next p
+osition   
    splice @$ary, $i, 0, $val;
}

__DATA__
0 1 2 3 4 6 7 8 9 10
0 1 2 3 4 5 6 7 8 9 10
0 1 2 3 4 5 6 7 8 9 10 11
0 0 1 2 3 4 5 6 7 8 9 10 11
0 0 1 2 3 4 5 5 6 7 8 9 10 11
0 0 1 2 3 4 5 5 6 7 8 9 10 11 11

      [download]

      cheers

      tachyon

[reply]
[d/l]

In Section Seekers of Perl Wisdom