--- trunk/jsFind.pm	2004/07/21 15:34:03	7
+++ trunk/jsFind.pm	2004/07/26 20:30:12	11
@@ -3,8 +3,9 @@
 use 5.008004;
 use strict;
 use warnings;
+use HTML::Entities;
 
-our $VERSION = '0.01';
+our $VERSION = '0.03';
 
 =head1 NAME
 
@@ -305,15 +306,36 @@
 
 Returns number of nodes in created tree.
 
+There is also longer version if you want to recode your data charset
+into different one (probably UTF-8):
+
+ $root->to_jsfind('/full/path/to/index/dir/','ISO-8859-2','UTF-8');
+
+Destination encoding is UTF-8 by default, so you don't have to specify it.
+
+ $root->to_jsfind('/full/path/to/index/dir/','WINDOWS-1250');
+
 =cut
 
+my $iconv;
+my $iconv_l1;
+
 sub to_jsfind {
 	my $self = shift;
 
 	my $path = shift || confess "to_jsfind need path to your index!";
 
+	my ($from_cp,$to_cp) = @_;
+
+	$to_cp ||= 'UTF-8';
+
+	if ($from_cp && $to_cp) {
+		$iconv = Text::Iconv->new($from_cp,$to_cp);
+	}
+	$iconv_l1 = Text::Iconv->new('ISO-8859-1',$to_cp);
+
 	$path .= "/" if ($path =~ /\/$/);
-	carp "create directory for index '$path': $!" if (! -w $path);
+	#carp "creating directory for index '$path'" if (! -w $path);
 
 	return $self->root->to_jsfind($path,"0");
 }
@@ -324,6 +346,31 @@
   $_[0] cmp $_[1];
 }
 
+=head3 _recode
+
+This is internal function to recode charset.
+
+It will also try to decode HTML::Entities in data.
+
+=cut
+
+sub _recode {
+	my $self = shift;
+	my $text = shift || return;
+
+	sub _decode_html_entities {
+		my $data = shift || return;
+		$data = $iconv_l1->convert(decode_entities($data)) || croak "entity decode problem: $data";
+	}
+
+	if ($iconv) {
+		$text = $iconv->convert($text) || $text && carp "convert problem: $text";
+		$text =~ s/(\&\w+;)/_decode_html_entities($1)/ges;
+	}
+
+	return $text;
+}
+
 #####################################################################
 
 =head2 jsFind::Node methods
@@ -348,6 +395,9 @@
 
 use Carp;
 use File::Path;
+use Text::Iconv;
+
+use base 'jsFind';
 
 my $KEYS = 0;
 my $DATA = 1;
@@ -691,22 +741,45 @@
 	$dot;
 }
 
+=head3 to_xml
+
+Escape <, >, & and ", and to produce valid XML
+
+=cut
+
+my %escape = ('<'=>'&lt;', '>'=>'&gt;', '&'=>'&amp;', '"'=>'&quot;');
+my $escape_re  = join '|' => keys %escape;
+
+sub to_xml {
+	my $self = shift || confess "you should call to_xml as object!";
+
+	my $d = shift || return;
+	$d = $self->SUPER::_recode($d);
+	confess "escape_re undefined!" unless ($escape_re);
+	$d =~ s/($escape_re)/$escape{$1}/g;
+	return $d;
+}
+
 =head3 to_jsfind
 
 Create jsFind xml files
 
- my $nr=$tree->to_dot('/path/to/index','0');
+ my $nr=$tree->to_jsfind('/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;
 
+	confess("path is undefined.") unless ($path);
+	confess("file is undefined. Did you call \$t->root->to_jsfind(..) instead of \$t->to_jsfind(..) ?") unless (defined($file));
+
 	my $nr_keys = 0;
 
 	my ($k, $d, $s) = @$self;
@@ -718,22 +791,22 @@
 		my $key = lc($k->[$i]);
 
 		if ($key) {
-			$key_xml .= qq{<k>$key</k>};
-			$data_xml .= qq{<e>};
+			$key_xml .= '<k>'.$self->to_xml($key).'</k>';
+			$data_xml .= '<e>';
 	#use Data::Dumper;
 	#print Dumper($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>';
+				$data_xml .= '<l f="'.($d->[$i]->{$path}->{'f'} || 1).'" t="'.$self->to_xml($d->[$i]->{$path}->{'t'} || 'no title').'">'.$self->to_xml($path).'</l>';
 				$nr_keys++;
 			}
-			$data_xml .= qq{</e>};
+			$data_xml .= '</e>';
 		}
 
 		$nr_keys += $s->[$i]->to_jsfind("$path/$file","$i") if ($s->[$i]);
 	}
 
-	$key_xml .= "</n>";
-	$data_xml .= "</d>";
+	$key_xml .= '</n>';
+	$data_xml .= '</d>';
 
 	if (! -e $path) {
 		mkpath($path) || croak "can't create dir '$path': $!";