jsFind/trunk/jsFind.pm

package jsFind;

use 5.008004;
use strict;
use warnings;

our $VERSION = '0.01';

=head1 NAME

jsFind - generate index for jsFind using B-Tree

=head1 SYNOPSIS

  use jsFind;


=head1 DESCRIPTION

This module can be used to create index files for jsFind, powerful tool for
adding a search engine to a CDROM archive or catalog without requiring the
user to install anything. 

Main difference between this module and scripts delivered with jsFind are:

=over 5

=item *

You don't need to use swish-e to create index

=item *

You can programatically (and incrementaly) create index for jsFind

=back

=head1 METHODS

This module contains two packages C<jsFind> and C<jsFind::Node>.

=head2 jsFind methods

=cut

use Exporter 'import';
use Carp;

our @ISA = qw(Exporter);

BEGIN {
        import 'jsFind::Node';
}

=head3 new

Create new tree. Arguments are C<B> which is maximum numbers of keys in
each node and optional C<Root> node. Each root node may have child nodes.

All nodes are objects from C<jsFind::Node>.

 my $t = new jsFind(B => 4);

=cut

my $DEBUG = 1;

sub new {
  my $package = shift;
  my %ARGV = @_;
  croak "Usage: {$package}::new(B => number [, Root => root node ])"
      unless exists $ARGV{B};
  if ($ARGV{B} % 2) {
    my $B = $ARGV{B} + 1;
    carp "B must be an even number.  Using $B instead.";
    $ARGV{B} = $B;
  }
    
  my $B = $ARGV{B};
  my $Root = exists($ARGV{Root}) ? $ARGV{Root} : jsFind::Node->emptynode;
  bless { B => $B, Root => $Root } => $package;
}

=head3 B_search

Search, insert, append or replace data in B-Tree

 $t->B_search(
        Key => 'key value',
        Data => { "path" => {
                        "t" => "title of document",
                        "f" => 99,
                        },
                },
        Insert => 1,
        Append => 1,
 );

Semantics:

If key not found, insert it iff C<Insert> argument is present.

If key B<is> found, replace existing data iff C<Replace> argument
is present or add new datum to existing iff C<Append> argument is present.

=cut

sub B_search {
  my $self = shift;
  my %args = @_;
  my $cur_node = $self->root;
  my $k = $args{Key};
  my $d = $args{Data};
  my @path;

  if ($cur_node->is_empty) {    # Special case for empty root
    if ($args{Insert}) {
      $cur_node->kdp_insert($k => $d);
      return $d;
    } else {
      return undef;
    }
  }

  # Descend tree to leaf
  for (;;) {

    # Didn't hit bottom yet.

    my($there, $where) = $cur_node->locate_key($k);
    if ($there) {               # Found it!
      if ($args{Replace}) {
        $cur_node->kdp_replace($where, $k => $d);
      } elsif ($args{Append}) {
        $cur_node->kdp_append($where, $k => $d);
      }
      return $cur_node->data($where);
    }
    
    # Not here---must be in a subtree.
    
    if ($cur_node->is_leaf) {   # But there are no subtrees
      return undef unless $args{Insert}; # Search failed
      # Stuff it in
      $cur_node->kdp_insert($k => $d);
      if ($self->node_overfull($cur_node)) { # Oops--there was no room.
        $self->split_and_promote($cur_node, @path);
      } 
      return $d;
    }

    # There are subtrees, and the key is in one of them.

    push @path, [$cur_node, $where];    # Record path from root.

    # Move down to search the subtree
    $cur_node = $cur_node->subnode($where);

    # and start over.
  }                             # for (;;) ...

  croak ("How did I get here?");
}


sub split_and_promote_old {
  my $self = shift;
  my ($cur_node, @path) = @_;
  
  for (;;) {
    my ($newleft, $newright, $kdp) = $cur_node->halves($self->B / 2);
    my ($up, $where) = @{pop @path};
    if ($up) {
      $up->kdp_insert(@$kdp);
      my ($tthere, $twhere) = $up->locate_key($kdp->[0]);
      croak "Couldn't find key `$kdp->[0]' in node after just inserting it!"
          unless $tthere;
      croak "`$kdp->[0]' went into node at `$twhere' instead of expected `$where'!"
          unless $twhere == $where;
      $up->subnode($where,   $newleft);
      $up->subnode($where+1, $newright);
      return unless $self->node_overfull($up);
      $cur_node = $up;
    } else { # We're at the top; make a new root.
      my $newroot = new jsFind::Node ([$kdp->[0]], 
                                     [$kdp->[1]], 
                                     [$newleft, $newright]);
      $self->root($newroot);
      return;
    }
  }
  
}

sub split_and_promote {
  my $self = shift;
  my ($cur_node, @path) = @_;
  
  for (;;) {
    my ($newleft, $newright, $kdp) = $cur_node->halves($self->B / 2);
    my ($up, $where) = @{pop @path} if (@path);
    if ($up) {
      $up->kdp_insert(@$kdp);
      if ($DEBUG) {
        my ($tthere, $twhere) = $up->locate_key($kdp->[0]);
        croak "Couldn't find key `$kdp->[0]' in node after just inserting it!"
          unless $tthere;
        croak "`$kdp->[0]' went into node at `$twhere' instead of expected `$where'!"
          unless $twhere == $where;
      }
      $up->subnode($where,   $newleft);
      $up->subnode($where+1, $newright);
      return unless $self->node_overfull($up);
      $cur_node = $up;
    } else { # We're at the top; make a new root.
      my $newroot = new jsFind::Node([$kdp->[0]], 
                                     [$kdp->[1]], 
                                     [$newleft, $newright]);
      $self->root($newroot);
      return;
    }
  }
}

=head3 B

Return B (maximum number of keys)

 my $max_size = $t->B;

=cut

sub B {
  $_[0]{B};
}

=head3 root

Returns root node

 my $root = $t->root;

=cut

sub root {
  my ($self, $newroot) = @_;
  $self->{Root} = $newroot if defined $newroot;
  $self->{Root};
}

=head3 node_overfull

Returns if node is overfull

 if ($node->node_overfull) { something }

=cut

sub node_overfull {
  my $self = shift;
  my $node = shift;
  $node->size > $self->B;
}

=head3 to_string

Returns your tree as formatted string.

 my $text = $root->to_string;

Mostly usefull for debugging as output leaves much to be desired.

=cut

sub to_string {
  $_[0]->root->to_string;
}

=head3 to_dot

Create Graphviz graph of your tree

 my $dot_graph = $root->to_dot;

=cut

sub to_dot {
        my $self = shift;

        my $dot = qq/digraph dns {\nrankdir=LR;\n/;
        $dot .= $self->root->to_dot;
        $dot .= qq/\n}\n/;

        return $dot;
}

=head3 to_jsfind

Create xml index files for jsFind. This should be called after
your B-Tree has been filled with data.

 $root->to_jsfind('/full/path/to/index/dir/');

Returns number of nodes in created tree.

=cut

sub to_jsfind {
        my $self = shift;

        my $path = shift || confess "to_jsfind need path to your index!";

        $path .= "/" if ($path =~ /\/$/);
        carp "create directory for index '$path': $!" if (! -w $path);

        return $self->root->to_jsfind($path,"0");
}


# private, default cmd function
sub default_cmp {
  $_[0] cmp $_[1];
}

#####################################################################

=head2 jsFind::Node methods

Each node has C<k> key-data pairs, with C<B> <= C<k> <= C<2B>, and 
each has C<k+1> subnodes, which might be null.

The node is a blessed reference to a list with three elements:

  ($keylist, $datalist, $subnodelist)

each is a reference to a list list.

The null node is represented by a blessed reference to an empty list.

=cut

package jsFind::Node;

use warnings;
use strict;

use Carp;
use File::Path;

my $KEYS = 0;
my $DATA = 1;
my $SUBNODES = 2;

=head3 new

Create New node

 my $node = new jsFind::Node ($keylist, $datalist, $subnodelist);

You can also mit argument list to create empty node.

 my $empty_node = new jsFind::Node;

=cut

sub new {
  my $self = shift;
  my $package = ref $self || $self;
  croak "Internal error:  jsFind::Node::new called with wrong number of arguments."
      unless @_ == 3 || @_ == 0;
  bless [@_] => $package;
}

=head3 locate_key

Locate key in node using linear search. This should probably be replaced
by binary search for better performance.

 my ($found, $index) = $node->locate_key($key, $cmp_coderef);

Argument C<$cmp_coderef> is optional reference to custom comparison
operator.

Returns (1, $index) if $key[$index] eq $key.

Returns (0, $index) if key could be found in $subnode[$index].

In scalar context, just returns 1 or 0.

=cut

sub locate_key {
  # Use linear search for testing, replace with binary search.
  my $self = shift;
  my $key = shift;
  my $cmp = shift || \&jsFind::default_cmp;
  my $i;
  my $cmp_result;
  my $N = $self->size;
  for ($i = 0; $i < $N; $i++) {
    $cmp_result = &$cmp($key, $self->key($i));
    last if $cmp_result <= 0;
  }
  
  # $i is now the index of the first node-key greater than $key
  # or $N if there is no such.  $cmp_result is 0 iff the key was found.
  (!$cmp_result, $i);
}


=head3 emptynode

Creates new empty node

 $node = $root->emptynode;
 $new_node = $node->emptynode;

=cut

sub emptynode {
  new($_[0]);                   # Pass package name, but not anything else.
}

=head3 is_empty

Test if node is empty

 if ($node->is_empty) { something }

=cut

# undef is empty; so is a blessed empty list.
sub is_empty {
  my $self = shift;
  !defined($self) || $#$self < 0;
}

=head3 key

Return C<$i>th key from node

 my $key = $node->key($i);

=cut

sub key {
#  my ($self, $n) = @_;
#  $self->[$KEYS][$n];

#       speedup
   $_[0]->[$KEYS][$_[1]];
}

=head3 data

Return C<$i>th data from node

 my $data = $node->data($i);

=cut

sub data {
  my ($self, $n) = @_;
  $self->[$DATA][$n];
}

=head3 kdp_replace

Set key data pair for C<$i>th element in node

 $node->kdp_replace($i, "key value" => {
        "data key 1" => "data value 1",
        "data key 2" => "data value 2",
 };

=cut

sub kdp_replace {
  my ($self, $n, $k => $d) = @_;
  if (defined $k) {
    $self->[$KEYS][$n] = $k;
    $self->[$DATA][$n] = $d;
  }
  [$self->[$KEYS][$n], 
   $self->[$DATA][$n]];
}

=head3 kdp_insert

 # No return value.

=cut

sub kdp_insert {
  my $self = shift;
  my ($k => $d) = @_;
  my ($there, $where) = $self->locate_key($k) unless $self->is_empty;

  if ($there) { croak("Tried to insert `$k => $d' into node where `$k' was already present."); }

  # undef fix
  $where ||= 0;

  splice(@{$self->[$KEYS]}, $where, 0, $k);
  splice(@{$self->[$DATA]}, $where, 0, $d);
  splice(@{$self->[$SUBNODES]}, $where, 0, undef);
}

=head3 kdp_append

Adds new data keys and values to C<$i>th element in node

 $node->kdp_append($i, "key value" => {
        "added data key" => "added data value",
 };

=cut

sub kdp_append {
  my ($self, $n, $k => $d) = @_;
  if (defined $k) {
    $self->[$KEYS][$n] = $k;
    my ($kv,$dv) = %{$d};
    $self->[$DATA][$n]->{$kv} = $dv;
  }
  [$self->[$KEYS][$n], 
   $self->[$DATA][$n]];
}

=head3 subnode

Set new or return existing subnode

 # return 4th subnode
 my $my_node = $node->subnode(4);

 # create new subnode 5 from $my_node
 $node->subnode(5, $my_node);

=cut

sub subnode {
  my ($self, $n, $newnode) = @_;
  $self->[$SUBNODES][$n] = $newnode if defined $newnode;
  $self->[$SUBNODES][$n];
}

=head3 is_leaf

Test if node is leaf

 if ($node->is_leaf) { something }

=cut

sub is_leaf {
  my $self = shift;
  ! defined $self->[$SUBNODES][0]; # undefined subnode means leaf node.
}

=head3 size

Return number of keys in the node

 my $nr = $node->size;

=cut

sub size {
  my $self = shift;
  return scalar(@{$self->[$KEYS]});
}

=head3 halves

 # Accept an index $n
 # Divide into two nodes so that keys 0 .. $n-1 are in one node
 # and keys $n+1 ... $size are in the other.

=cut

sub halves {
  my $self = shift;
  my $n = shift;
  my $s = $self->size;
  my @right;
  my @left;

  $left[$KEYS] = [@{$self->[$KEYS]}[0 .. $n-1]];
  $left[$DATA] = [@{$self->[$DATA]}[0 .. $n-1]];
  $left[$SUBNODES] = [@{$self->[$SUBNODES]}[0 .. $n]];

  $right[$KEYS] = [@{$self->[$KEYS]}[$n+1 .. $s-1]];
  $right[$DATA] = [@{$self->[$DATA]}[$n+1 .. $s-1]];
  $right[$SUBNODES] = [@{$self->[$SUBNODES]}[$n+1 .. $s]];

  my @middle = ($self->[$KEYS][$n], $self->[$DATA][$n]);

  ($self->new(@left), $self->new(@right), \@middle);
}

=head3 to_string

Dumps tree as string

 my $str = $root->to_string;

=cut

sub to_string {
  my $self = shift;
  my $indent = shift || 0;
  my $I = ' ' x $indent;
  return '' if $self->is_empty;
  my ($k, $d, $s) = @$self;
  my $result = '';
  $result .= defined($s->[0]) ? $s->[0]->to_string($indent+2) : '';
  my $N = $self->size;
  my $i;
  for ($i = 0; $i < $N; $i++) {
#    $result .= $I . "$k->[$i] => $d->[$i]\n";
    $result .= $I . "$k->[$i]\n";
    $result .= defined($s->[$i+1]) ? $s->[$i+1]->to_string($indent+2) : '';
  }
  $result;
}

=begin comment

use Data::Dumper;

sub to_string {
  my $self = shift;
  my $indent = shift || 0;
  my $path = shift || '0';
  return '' if $self->is_empty;
  my ($k, $d, $s) = @$self;
  my $result = '';
  $result .= defined($s->[0]) ? $s->[0]->to_string($indent+1,"$path/0") : '';
  my $N = $self->size;
  for (my $i = 0; $i < $N; $i++) {
        my $dump = Dumper($d->[$i]);
        $dump =~ s/[\n\r\s]+/ /gs;
        $dump =~ s/\$VAR1\s*=\s*//;
    $result .= sprintf("%-5s [%2d] %2s: %s => %s\n", $path, $i, $indent, $k->[$i], $dump);
    $result .= defined($s->[$i+1]) ? $s->[$i+1]->to_string($indent+1,"$path/$i") : '';
  }
  $result;
}

=end comment

=head3 to_dot

Recursivly walk nodes of tree

=cut

sub to_dot {
        my $self = shift;
        my $parent = shift;

        return '' if $self->is_empty;

        my $dot = '';

        my ($k, $d, $s) = @$self;
        my $N = $self->size;

        my @dot_keys;

        my $node_name = $parent || '_';
        $node_name =~ s/\W+//g;
        $node_name .= " [$N]";

        for (my $i = 0; $i <= $N; $i++) {
                if (my $key = $k->[$i]) {
                        push @dot_keys, qq{<$i>$key};
                }
                $dot .= $s->[$i]->to_dot(qq{"$node_name":$i}) if ($s->[$i]);
        }
        push @dot_keys, qq{<$N>...} if (! $self->is_leaf);

        my $label = join("|",@dot_keys);
        $dot .= qq{"$node_name" [ shape=record, label="$label" ];\n};

        $dot .= qq{$parent -> "$node_name";\n} if ($parent);

        $dot;
}

=head3 to_jsfind

Create jsFind xml files

 my $nr=$tree->to_dot('/path/to/index','0');

Returns number of elements created

=cut

sub to_jsfind {
        my $self = shift;
        my ($path,$file) = @_;

        return 0 if $self->is_empty;

        my $nr_keys = 0;

        my ($k, $d, $s) = @$self;
        my $N = $self->size;

        my ($key_xml, $data_xml) = ("<n>","<d>");

        for (my $i = 0; $i <= $N; $i++) {
                my $key = lc($k->[$i]);

                if ($key) {
                        $key_xml .= qq{<k>$key</k>};
                        $data_xml .= qq{<e>};
        #use Data::Dumper;
        #print Dumper($d->[$i]);
                        foreach my $path (keys %{$d->[$i]}) {
                                $data_xml .= '<l f="'.($d->[$i]->{$path}->{'f'} || 1).'" t="'.($d->[$i]->{$path}->{'t'} || 'no title').'">'.$path.'</l>';
                                $nr_keys++;
                        }
                        $data_xml .= qq{</e>};
                }

                $nr_keys += $s->[$i]->to_jsfind("$path/$file","$i") if ($s->[$i]);
        }

        $key_xml .= "</n>";
        $data_xml .= "</d>";

        if (! -e $path) {
                mkpath($path) || croak "can't create dir '$path': $!";
        }

        open(K, "> ${path}/${file}.xml") || croak "can't open '$path/$file.xml': $!";
        open(D, "> ${path}/_${file}.xml") || croak "can't open '$path/_$file.xml': $!";

        print K $key_xml;
        print D $data_xml;

        close(K);
        close(D);

        return $nr_keys;
}

1;
__END__

=head1 SEE ALSO

jsFind web site L<http://www.elucidsoft.net/projects/jsfind/>

B-Trees in perl web site L<http://perl.plover.com/BTree/>

=head1 AUTHORS

Mark-Jonson Dominus E<lt>mjd@pobox.comE<gt> wrote C<BTree.pm> which was
base for this module

Shawn P. Garbett E<lt>shawn@elucidsoft.netE<gt> wrote jsFind

Dobrica Pavlinusic E<lt>dpavlin@rot13.orgE<gt> wrote this module

=head1 COPYRIGHT AND LICENSE

Copyright (C) 2004 by Dobrica Pavlinusic

This program is free software; you can redistribute it and/or modify it
under the terms of the GNU General Public License as published by the Free
Software Foundation; either version 2 of the License, or (at your option)
any later version. This program is distributed in the hope that it will be
useful, but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
Public License for more details.

=cut
1	dpavlin	1	package jsFind;
2
3			use 5.008004;
4			use strict;
5			use warnings;
6
7			our $VERSION = '0.01';
8
9			=head1 NAME
10
11			jsFind - generate index for jsFind using B-Tree
12
13			=head1 SYNOPSIS
14
15			use jsFind;
16
17
18
19			=head1 DESCRIPTION
20
21			This module can be used to create index files for jsFind, powerful tool for
22			adding a search engine to a CDROM archive or catalog without requiring the
23			user to install anything.
24
25			Main difference between this module and scripts delivered with jsFind are:
26
27			=over 5
28
29			=item *
30
31			You don't need to use swish-e to create index
32
33			=item *
34
35			You can programatically (and incrementaly) create index for jsFind
36
37			=back
38
39			=head1 METHODS
40
41			This module contains two packages C<jsFind> and C<jsFind::Node>.
42
43			=head2 jsFind methods
44
45			=cut
46
47			use Exporter 'import';
48			use Carp;
49
50			our @ISA = qw(Exporter);
51
52			BEGIN {
53			import 'jsFind::Node';
54			}
55
56			=head3 new
57
58			Create new tree. Arguments are C<B> which is maximum numbers of keys in
59			each node and optional C<Root> node. Each root node may have child nodes.
60
61			All nodes are objects from C<jsFind::Node>.
62
63			my $t = new jsFind(B => 4);
64
65			=cut
66
67			my $DEBUG = 1;
68
69			sub new {
70			my $package = shift;
71			my %ARGV = @_;
72			croak "Usage: {$package}::new(B => number [, Root => root node ])"
73			unless exists $ARGV{B};
74			if ($ARGV{B} % 2) {
75			my $B = $ARGV{B} + 1;
76			carp "B must be an even number. Using $B instead.";
77			$ARGV{B} = $B;
78			}
79
80			my $B = $ARGV{B};
81			my $Root = exists($ARGV{Root}) ? $ARGV{Root} : jsFind::Node->emptynode;
82			bless { B => $B, Root => $Root } => $package;
83			}
84
85			=head3 B_search
86
87			Search, insert, append or replace data in B-Tree
88
89	dpavlin	5	$t->B_search(
90			Key => 'key value',
91			Data => { "path" => {
92			"t" => "title of document",
93			"f" => 99,
94	dpavlin	7	},
95	dpavlin	5	},
96			Insert => 1,
97			Append => 1,
98			);
99	dpavlin	1
100			Semantics:
101
102			If key not found, insert it iff C<Insert> argument is present.
103
104			If key B<is> found, replace existing data iff C<Replace> argument
105			is present or add new datum to existing iff C<Append> argument is present.
106
107			=cut
108
109			sub B_search {
110			my $self = shift;
111			my %args = @_;
112			my $cur_node = $self->root;
113			my $k = $args{Key};
114			my $d = $args{Data};
115			my @path;
116
117			if ($cur_node->is_empty) { # Special case for empty root
118			if ($args{Insert}) {
119			$cur_node->kdp_insert($k => $d);
120			return $d;
121			} else {
122			return undef;
123			}
124			}
125
126			# Descend tree to leaf
127			for (;;) {
128
129			# Didn't hit bottom yet.
130
131			my($there, $where) = $cur_node->locate_key($k);
132			if ($there) { # Found it!
133			if ($args{Replace}) {
134			$cur_node->kdp_replace($where, $k => $d);
135			} elsif ($args{Append}) {
136			$cur_node->kdp_append($where, $k => $d);
137			}
138			return $cur_node->data($where);
139			}
140
141			# Not here---must be in a subtree.
142
143			if ($cur_node->is_leaf) { # But there are no subtrees
144			return undef unless $args{Insert}; # Search failed
145			# Stuff it in
146			$cur_node->kdp_insert($k => $d);
147			if ($self->node_overfull($cur_node)) { # Oops--there was no room.
148			$self->split_and_promote($cur_node, @path);
149			}
150			return $d;
151			}
152
153			# There are subtrees, and the key is in one of them.
154
155			push @path, [$cur_node, $where]; # Record path from root.
156
157			# Move down to search the subtree
158			$cur_node = $cur_node->subnode($where);
159
160			# and start over.
161			} # for (;;) ...
162
163			croak ("How did I get here?");
164			}
165
166
167
168			sub split_and_promote_old {
169			my $self = shift;
170			my ($cur_node, @path) = @_;
171
172			for (;;) {
173			my ($newleft, $newright, $kdp) = $cur_node->halves($self->B / 2);
174			my ($up, $where) = @{pop @path};
175			if ($up) {
176			$up->kdp_insert(@$kdp);
177			my ($tthere, $twhere) = $up->locate_key($kdp->[0]);
178			croak "Couldn't find key `$kdp->[0]' in node after just inserting it!"
179			unless $tthere;
180			croak "`$kdp->[0]' went into node at `$twhere' instead of expected `$where'!"
181			unless $twhere == $where;
182			$up->subnode($where, $newleft);
183			$up->subnode($where+1, $newright);
184			return unless $self->node_overfull($up);
185			$cur_node = $up;
186			} else { # We're at the top; make a new root.
187			my $newroot = new jsFind::Node ([$kdp->[0]],
188			[$kdp->[1]],
189			[$newleft, $newright]);
190			$self->root($newroot);
191			return;
192			}
193			}
194
195			}
196
197			sub split_and_promote {
198			my $self = shift;
199			my ($cur_node, @path) = @_;
200
201			for (;;) {
202			my ($newleft, $newright, $kdp) = $cur_node->halves($self->B / 2);
203			my ($up, $where) = @{pop @path} if (@path);
204			if ($up) {
205			$up->kdp_insert(@$kdp);
206			if ($DEBUG) {
207			my ($tthere, $twhere) = $up->locate_key($kdp->[0]);
208			croak "Couldn't find key `$kdp->[0]' in node after just inserting it!"
209			unless $tthere;
210			croak "`$kdp->[0]' went into node at `$twhere' instead of expected `$where'!"
211			unless $twhere == $where;
212			}
213			$up->subnode($where, $newleft);
214			$up->subnode($where+1, $newright);
215			return unless $self->node_overfull($up);
216			$cur_node = $up;
217			} else { # We're at the top; make a new root.
218			my $newroot = new jsFind::Node([$kdp->[0]],
219			[$kdp->[1]],
220			[$newleft, $newright]);
221			$self->root($newroot);
222			return;
223			}
224			}
225			}
226
227			=head3 B
228
229			Return B (maximum number of keys)
230
231			my $max_size = $t->B;
232
233			=cut
234
235			sub B {
236			$_[0]{B};
237			}
238
239			=head3 root
240
241			Returns root node
242
243			my $root = $t->root;
244
245			=cut
246
247			sub root {
248			my ($self, $newroot) = @_;
249			$self->{Root} = $newroot if defined $newroot;
250			$self->{Root};
251			}
252
253			=head3 node_overfull
254
255			Returns if node is overfull
256
257			if ($node->node_overfull) { something }
258
259			=cut
260
261			sub node_overfull {
262			my $self = shift;
263			my $node = shift;
264			$node->size > $self->B;
265			}
266
267			=head3 to_string
268
269			Returns your tree as formatted string.
270
271			my $text = $root->to_string;
272
273			Mostly usefull for debugging as output leaves much to be desired.
274
275			=cut
276
277			sub to_string {
278			$_[0]->root->to_string;
279			}
280
281			=head3 to_dot
282
283			Create Graphviz graph of your tree
284
285			my $dot_graph = $root->to_dot;
286
287			=cut
288
289			sub to_dot {
290			my $self = shift;
291
292			my $dot = qq/digraph dns {\nrankdir=LR;\n/;
293			$dot .= $self->root->to_dot;
294			$dot .= qq/\n}\n/;
295
296			return $dot;
297			}
298
299			=head3 to_jsfind
300
301			Create xml index files for jsFind. This should be called after
302			your B-Tree has been filled with data.
303
304			$root->to_jsfind('/full/path/to/index/dir/');
305
306			Returns number of nodes in created tree.
307
308			=cut
309
310			sub to_jsfind {
311			my $self = shift;
312
313			my $path = shift \|\| confess "to_jsfind need path to your index!";
314
315			$path .= "/" if ($path =~ /\/$/);
316	dpavlin	7	carp "create directory for index '$path': $!" if (! -w $path);
317	dpavlin	1
318			return $self->root->to_jsfind($path,"0");
319			}
320
321
322			# private, default cmd function
323			sub default_cmp {
324			$_[0] cmp $_[1];
325			}
326
327			#####################################################################
328
329			=head2 jsFind::Node methods
330
331			Each node has C<k> key-data pairs, with C<B> <= C<k> <= C<2B>, and
332			each has C<k+1> subnodes, which might be null.
333
334			The node is a blessed reference to a list with three elements:
335
336			($keylist, $datalist, $subnodelist)
337
338			each is a reference to a list list.
339
340			The null node is represented by a blessed reference to an empty list.
341
342			=cut
343
344			package jsFind::Node;
345
346			use warnings;
347			use strict;
348
349			use Carp;
350			use File::Path;
351
352			my $KEYS = 0;
353			my $DATA = 1;
354			my $SUBNODES = 2;
355
356			=head3 new
357
358			Create New node
359
360			my $node = new jsFind::Node ($keylist, $datalist, $subnodelist);
361
362			You can also mit argument list to create empty node.
363
364			my $empty_node = new jsFind::Node;
365
366			=cut
367
368			sub new {
369			my $self = shift;
370			my $package = ref $self \|\| $self;
371			croak "Internal error: jsFind::Node::new called with wrong number of arguments."
372			unless @_ == 3 \|\| @_ == 0;
373			bless [@_] => $package;
374			}
375
376			=head3 locate_key
377
378			Locate key in node using linear search. This should probably be replaced
379			by binary search for better performance.
380
381			my ($found, $index) = $node->locate_key($key, $cmp_coderef);
382
383			Argument C<$cmp_coderef> is optional reference to custom comparison
384			operator.
385
386			Returns (1, $index) if $key[$index] eq $key.
387
388			Returns (0, $index) if key could be found in $subnode[$index].
389
390			In scalar context, just returns 1 or 0.
391
392			=cut
393
394			sub locate_key {
395			# Use linear search for testing, replace with binary search.
396			my $self = shift;
397			my $key = shift;
398			my $cmp = shift \|\| \&jsFind::default_cmp;
399			my $i;
400			my $cmp_result;
401			my $N = $self->size;
402			for ($i = 0; $i < $N; $i++) {
403			$cmp_result = &$cmp($key, $self->key($i));
404			last if $cmp_result <= 0;
405			}
406
407			# $i is now the index of the first node-key greater than $key
408			# or $N if there is no such. $cmp_result is 0 iff the key was found.
409			(!$cmp_result, $i);
410			}
411
412
413			=head3 emptynode
414
415			Creates new empty node
416
417			$node = $root->emptynode;
418			$new_node = $node->emptynode;
419
420			=cut
421
422			sub emptynode {
423			new($_[0]); # Pass package name, but not anything else.
424			}
425
426			=head3 is_empty
427
428			Test if node is empty
429
430			if ($node->is_empty) { something }
431
432			=cut
433
434			# undef is empty; so is a blessed empty list.
435			sub is_empty {
436			my $self = shift;
437			!defined($self) \|\| $#$self < 0;
438			}
439
440			=head3 key
441
442			Return C<$i>th key from node
443
444			my $key = $node->key($i);
445
446			=cut
447
448			sub key {
449			# my ($self, $n) = @_;
450			# $self->[$KEYS][$n];
451
452			# speedup
453			$_[0]->[$KEYS][$_[1]];
454			}
455
456			=head3 data
457
458			Return C<$i>th data from node
459
460			my $data = $node->data($i);
461
462			=cut
463
464			sub data {
465			my ($self, $n) = @_;
466			$self->[$DATA][$n];
467			}
468
469			=head3 kdp_replace
470
471			Set key data pair for C<$i>th element in node
472
473			$node->kdp_replace($i, "key value" => {
474			"data key 1" => "data value 1",
475			"data key 2" => "data value 2",
476			};
477
478			=cut
479
480			sub kdp_replace {
481			my ($self, $n, $k => $d) = @_;
482			if (defined $k) {
483			$self->[$KEYS][$n] = $k;
484			$self->[$DATA][$n] = $d;
485			}
486			[$self->[$KEYS][$n],
487			$self->[$DATA][$n]];
488			}
489
490			=head3 kdp_insert
491
492			# No return value.
493
494			=cut
495
496			sub kdp_insert {
497			my $self = shift;
498			my ($k => $d) = @_;
499			my ($there, $where) = $self->locate_key($k) unless $self->is_empty;
500
501			if ($there) { croak("Tried to insert `$k => $d' into node where `$k' was already present."); }
502
503			# undef fix
504			$where \|\|= 0;
505
506			splice(@{$self->[$KEYS]}, $where, 0, $k);
507			splice(@{$self->[$DATA]}, $where, 0, $d);
508			splice(@{$self->[$SUBNODES]}, $where, 0, undef);
509			}
510
511			=head3 kdp_append
512
513			Adds new data keys and values to C<$i>th element in node
514
515			$node->kdp_append($i, "key value" => {
516			"added data key" => "added data value",
517			};
518
519			=cut
520
521			sub kdp_append {
522			my ($self, $n, $k => $d) = @_;
523			if (defined $k) {
524			$self->[$KEYS][$n] = $k;
525			my ($kv,$dv) = %{$d};
526			$self->[$DATA][$n]->{$kv} = $dv;
527			}
528			[$self->[$KEYS][$n],
529			$self->[$DATA][$n]];
530			}
531
532			=head3 subnode
533
534			Set new or return existing subnode
535
536			# return 4th subnode
537			my $my_node = $node->subnode(4);
538
539			# create new subnode 5 from $my_node
540			$node->subnode(5, $my_node);
541
542			=cut
543
544			sub subnode {
545			my ($self, $n, $newnode) = @_;
546			$self->[$SUBNODES][$n] = $newnode if defined $newnode;
547			$self->[$SUBNODES][$n];
548			}
549
550			=head3 is_leaf
551
552			Test if node is leaf
553
554			if ($node->is_leaf) { something }
555
556			=cut
557
558			sub is_leaf {
559			my $self = shift;
560			! defined $self->[$SUBNODES][0]; # undefined subnode means leaf node.
561			}
562
563			=head3 size
564
565			Return number of keys in the node
566
567			my $nr = $node->size;
568
569			=cut
570
571			sub size {
572			my $self = shift;
573			return scalar(@{$self->[$KEYS]});
574			}
575
576			=head3 halves
577
578			# Accept an index $n
579			# Divide into two nodes so that keys 0 .. $n-1 are in one node
580			# and keys $n+1 ... $size are in the other.
581
582			=cut
583
584			sub halves {
585			my $self = shift;
586			my $n = shift;
587			my $s = $self->size;
588			my @right;
589			my @left;
590
591			$left[$KEYS] = [@{$self->[$KEYS]}[0 .. $n-1]];
592			$left[$DATA] = [@{$self->[$DATA]}[0 .. $n-1]];
593			$left[$SUBNODES] = [@{$self->[$SUBNODES]}[0 .. $n]];
594
595			$right[$KEYS] = [@{$self->[$KEYS]}[$n+1 .. $s-1]];
596			$right[$DATA] = [@{$self->[$DATA]}[$n+1 .. $s-1]];
597			$right[$SUBNODES] = [@{$self->[$SUBNODES]}[$n+1 .. $s]];
598
599			my @middle = ($self->[$KEYS][$n], $self->[$DATA][$n]);
600
601			($self->new(@left), $self->new(@right), \@middle);
602			}
603
604			=head3 to_string
605
606			Dumps tree as string
607
608			my $str = $root->to_string;
609
610			=cut
611
612			sub to_string {
613			my $self = shift;
614			my $indent = shift \|\| 0;
615			my $I = ' ' x $indent;
616			return '' if $self->is_empty;
617			my ($k, $d, $s) = @$self;
618			my $result = '';
619			$result .= defined($s->[0]) ? $s->[0]->to_string($indent+2) : '';
620			my $N = $self->size;
621			my $i;
622			for ($i = 0; $i < $N; $i++) {
623			# $result .= $I . "$k->[$i] => $d->[$i]\n";
624			$result .= $I . "$k->[$i]\n";
625			$result .= defined($s->[$i+1]) ? $s->[$i+1]->to_string($indent+2) : '';
626			}
627			$result;
628			}
629
630			=begin comment
631
632			use Data::Dumper;
633
634			sub to_string {
635			my $self = shift;
636			my $indent = shift \|\| 0;
637			my $path = shift \|\| '0';
638			return '' if $self->is_empty;
639			my ($k, $d, $s) = @$self;
640			my $result = '';
641			$result .= defined($s->[0]) ? $s->[0]->to_string($indent+1,"$path/0") : '';
642			my $N = $self->size;
643			for (my $i = 0; $i < $N; $i++) {
644			my $dump = Dumper($d->[$i]);
645			$dump =~ s/[\n\r\s]+/ /gs;
646			$dump =~ s/\$VAR1\s=\s//;
647			$result .= sprintf("%-5s [%2d] %2s: %s => %s\n", $path, $i, $indent, $k->[$i], $dump);
648			$result .= defined($s->[$i+1]) ? $s->[$i+1]->to_string($indent+1,"$path/$i") : '';
649			}
650			$result;
651			}
652
653			=end comment
654
655			=head3 to_dot
656
657			Recursivly walk nodes of tree
658
659			=cut
660
661			sub to_dot {
662			my $self = shift;
663			my $parent = shift;
664
665			return '' if $self->is_empty;
666
667			my $dot = '';
668
669			my ($k, $d, $s) = @$self;
670			my $N = $self->size;
671
672			my @dot_keys;
673
674			my $node_name = $parent \|\| '_';
675			$node_name =~ s/\W+//g;
676			$node_name .= " [$N]";
677
678			for (my $i = 0; $i <= $N; $i++) {
679			if (my $key = $k->[$i]) {
680			push @dot_keys, qq{<$i>$key};
681			}
682			$dot .= $s->[$i]->to_dot(qq{"$node_name":$i}) if ($s->[$i]);
683			}
684			push @dot_keys, qq{<$N>...} if (! $self->is_leaf);
685
686			my $label = join("\|",@dot_keys);
687			$dot .= qq{"$node_name" [ shape=record, label="$label" ];\n};
688
689			$dot .= qq{$parent -> "$node_name";\n} if ($parent);
690
691			$dot;
692			}
693
694			=head3 to_jsfind
695
696			Create jsFind xml files
697
698			my $nr=$tree->to_dot('/path/to/index','0');
699
700			Returns number of elements created
701
702			=cut
703
704			sub to_jsfind {
705			my $self = shift;
706			my ($path,$file) = @_;
707
708			return 0 if $self->is_empty;
709
710			my $nr_keys = 0;
711
712			my ($k, $d, $s) = @$self;
713			my $N = $self->size;
714
715			my ($key_xml, $data_xml) = ("<n>","<d>");
716
717			for (my $i = 0; $i <= $N; $i++) {
718			my $key = lc($k->[$i]);
719
720			if ($key) {
721			$key_xml .= qq{<k>$key</k>};
722			$data_xml .= qq{<e>};
723			#use Data::Dumper;
724			#print Dumper($d->[$i]);
725			foreach my $path (keys %{$d->[$i]}) {
726			$data_xml .= '<l f="'.($d->[$i]->{$path}->{'f'} \|\| 1).'" t="'.($d->[$i]->{$path}->{'t'} \|\| 'no title').'">'.$path.'</l>';
727			$nr_keys++;
728			}
729			$data_xml .= qq{</e>};
730			}
731
732			$nr_keys += $s->[$i]->to_jsfind("$path/$file","$i") if ($s->[$i]);
733			}
734
735			$key_xml .= "</n>";
736			$data_xml .= "</d>";
737
738			if (! -e $path) {
739			mkpath($path) \|\| croak "can't create dir '$path': $!";
740			}
741
742			open(K, "> ${path}/${file}.xml") \|\| croak "can't open '$path/$file.xml': $!";
743			open(D, "> ${path}/_${file}.xml") \|\| croak "can't open '$path/_$file.xml': $!";
744
745			print K $key_xml;
746			print D $data_xml;
747
748			close(K);
749			close(D);
750
751			return $nr_keys;
752			}
753
754			1;
755			__END__
756
757			=head1 SEE ALSO
758
759			jsFind web site L<http://www.elucidsoft.net/projects/jsfind/>
760
761			B-Trees in perl web site L<http://perl.plover.com/BTree/>
762
763			=head1 AUTHORS
764
765			Mark-Jonson Dominus E<lt>mjd@pobox.comE<gt> wrote C<BTree.pm> which was
766			base for this module
767
768			Shawn P. Garbett E<lt>shawn@elucidsoft.netE<gt> wrote jsFind
769
770			Dobrica Pavlinusic E<lt>dpavlin@rot13.orgE<gt> wrote this module
771
772			=head1 COPYRIGHT AND LICENSE
773
774			Copyright (C) 2004 by Dobrica Pavlinusic
775
776			This program is free software; you can redistribute it and/or modify it
777			under the terms of the GNU General Public License as published by the Free
778			Software Foundation; either version 2 of the License, or (at your option)
779			any later version. This program is distributed in the hope that it will be
780			useful, but WITHOUT ANY WARRANTY; without even the implied warranty of
781			MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General
782			Public License for more details.
783
784			=cut