r11743@llin: dpavlin | 2005-12-17 02:09:53 +0100

[webpac2] / lib / WebPAC / Normalize / XML.pm

diff --git a/lib/WebPAC/Normalize/XML.pm b/lib/WebPAC/Normalize/XML.pm
index a0727d0..1de4cde 100644 (file)
--- a/lib/WebPAC/Normalize/XML.pm
+++ b/lib/WebPAC/Normalize/XML.pm
@@ -7,18 +7,19 @@ use base qw/WebPAC::Common WebPAC::Normalize/;
 use XML::Simple;
 use Data::Dumper;
 use Text::Iconv;
+use YAML qw/Dump LoadFile/;
 
 =head1 NAME
 
-WebPAC::Normalize::XML - apply XML normalisaton rules
+WebPAC::Normalize::XML - apply XML or YAML normalisaton rules
 
 =head1 VERSION
 
-Version 0.01
+Version 0.03
 
 =cut
 
-our $VERSION = '0.01';
+our $VERSION = '0.03';
 
 =head1 SYNOPSIS
 
@@ -42,7 +43,9 @@ parse it.
 
 C<tag> defines tag to use within C<xml_file>
 
-C<xml_file> defines path to normalize XML.
+C<xml_file> defines path to normalize XML
+
+C<tags> define additional tags that can be forced (and an be array).
 
 =cut
 
@@ -66,15 +69,60 @@ sub open {
 
        $self->{'import_xml_file'} = $xml_file;
 
+       my @force_array = [ $self->{'tag'}, 'config', 'format' ];
+       push @force_array, $self->{'tags'} if ($self->{'tags'});
+
        $self->{'import_xml'} = XMLin($xml_file,
-               ForceArray => [ $self->{'tag'}, $self->{'tags'}, 'config', 'format' ],
+               ForceArray => @force_array,
+               ForceContent => 1,
        );
 
-       $log->debug("import xml is ",sub { Dumper($self->{'import_xml'}) });
+       $log->debug("import xml is ",sub { Dumper($self->{'import_xml'}) }, $self->{lookup} ? " using lookups" : "lookups disabled");
+
+       #print STDERR Dump($self->{import_xml});
 
        return $self;
 }
 
+=head2 open_yaml
+
+Read normalisation rules defined in YAML file located usually at
+C<conf/normalize/*.yml> and parse it.
+
+ my $n = new WebPAC::Normalize::XML;
+ $n->open_yaml(
+       tag => 'isis',
+       path => '/path/to/conf/normalize/isis.yml',
+ );
+
+=cut
+
+sub open_yaml {
+       my $self = shift;
+
+       my $arg = {@_};
+
+       my $log = $self->_get_logger();
+
+       foreach my $req (qw/tag path/) {
+               $log->logconfess("need argument $req") unless $arg->{$req};
+       }
+
+    my $path = $arg->{path};
+       $self->{tag} = $arg->{tag};
+
+       $log->logdie("normalisation yaml file '$path' doesn't exist!") if (! -e $path);
+
+       $log->info("using $path normalization YAML");
+
+    $self->{'import_xml'} = LoadFile( $path ) || $log->die("can't load $path: $!");
+
+       $log->debug("import yaml is ",sub { Dumper($self->{'import_xml'}) }, $self->{lookup} ? " using lookups" : "lookups disabled");
+
+       $self->{_skip_x} = 1;
+
+       return $self;
+}
 
 =head2 _x
 
@@ -90,6 +138,7 @@ change that when creating new instance of object based on this one.
 sub _x {
        my $self = shift;
        my $utf8 = shift || return;
+       return $utf8 if ($self->{_skip_x});
 
        # create UTF-8 convertor for import_xml files
        $self->{'utf2cp'} ||= Text::Iconv->new('UTF-8' ,$self->{'code_page'} || 'ISO-8859-2');