WebPAC/Normalize/XML.pm

package WebPAC::Normalize::XML;

use warnings;
use strict;

use base qw/WebPAC::Common WebPAC::Normalize/;
use Storable;
use XML::Simple;
use Data::Dumper;
use Text::Iconv;

=head1 NAME

WebPAC::Normalize::XML - apply XML normalisaton rules

=head1 VERSION

Version 0.01

=cut

our $VERSION = '0.01';

=head1 SYNOPSIS

This module uses C<conf/normalize/*.xml> files to perform normalisation
from input records

=cut

=head1 FUNCTIONS

=head2 open

Read normalisation rules defined using XML from C<conf/normalize/*.xml> and
parse it.

 my $n = new WebPAC::Normalize::XML;
 $n->open(
        tag => 'isis',
        xml_file => '/path/to/conf/normalize/isis.xml',
 );

C<tag> defines tag to use within C<xml_file>

C<xml_file> defines path to normalize XML.

=cut

sub open {
        my $self = shift;

        my $arg = {@_};

        my $log = $self->_get_logger();

        foreach my $req (qw/tag xml_file/) {
                $log->logconfess("need argument $req") unless $arg->{$req};
        }

        $self->{'tag'} = $arg->{'tag'};
        my $xml_file = $arg->{'xml_file'};

        $log->info("using $xml_file tag <",$self->{'tag'},">");

        $log->logdie("normalisation xml file '$xml_file' doesn't exist!") if (! -e $xml_file);

        $self->{'import_xml_file'} = $xml_file;

        $self->{'import_xml'} = XMLin($xml_file,
                ForceArray => [ $self->{'tag'}, $self->{'tags'}, 'config', 'format' ],
        );

        $log->debug("import xml is ",sub { Dumper($self->{'import_xml'}) });

        return $self;
}


=head2 _x

Convert string from UTF-8 to code page defined in C<import_xml>.

 my $text = $n->_x('utf8 text');

Default application code page is C<ISO-8859-2>. You will probably want to
change that when creating new instance of object based on this one.

=cut

sub _x {
        my $self = shift;
        my $utf8 = shift || return;

        # create UTF-8 convertor for import_xml files
        $self->{'utf2cp'} ||= Text::Iconv->new('UTF-8' ,$self->{'code_page'} || 'ISO-8859-2');

        return $self->{'utf2cp'}->convert($utf8) ||
                $self->_get_logger()->logwarn("can't convert '$utf8'");
}


=head1 AUTHOR

Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>

=head1 COPYRIGHT & LICENSE

Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.

This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.

=cut

1; # End of WebPAC::Normalize::XML
1	package WebPAC::Normalize::XML;
2
3	use warnings;
4	use strict;
5
6	use base qw/WebPAC::Common WebPAC::Normalize/;
7	use Storable;
8	use XML::Simple;
9	use Data::Dumper;
10	use Text::Iconv;
11
12	=head1 NAME
13
14	WebPAC::Normalize::XML - apply XML normalisaton rules
15
16	=head1 VERSION
17
18	Version 0.01
19
20	=cut
21
22	our $VERSION = '0.01';
23
24	=head1 SYNOPSIS
25
26	This module uses C<conf/normalize/*.xml> files to perform normalisation
27	from input records
28
29	=cut
30
31	=head1 FUNCTIONS
32
33	=head2 open
34
35	Read normalisation rules defined using XML from C<conf/normalize/*.xml> and
36	parse it.
37
38	my $n = new WebPAC::Normalize::XML;
39	$n->open(
40	tag => 'isis',
41	xml_file => '/path/to/conf/normalize/isis.xml',
42	);
43
44	C<tag> defines tag to use within C<xml_file>
45
46	C<xml_file> defines path to normalize XML.
47
48	=cut
49
50	sub open {
51	my $self = shift;
52
53	my $arg = {@_};
54
55	my $log = $self->_get_logger();
56
57	foreach my $req (qw/tag xml_file/) {
58	$log->logconfess("need argument $req") unless $arg->{$req};
59	}
60
61	$self->{'tag'} = $arg->{'tag'};
62	my $xml_file = $arg->{'xml_file'};
63
64	$log->info("using $xml_file tag <",$self->{'tag'},">");
65
66	$log->logdie("normalisation xml file '$xml_file' doesn't exist!") if (! -e $xml_file);
67
68	$self->{'import_xml_file'} = $xml_file;
69
70	$self->{'import_xml'} = XMLin($xml_file,
71	ForceArray => [ $self->{'tag'}, $self->{'tags'}, 'config', 'format' ],
72	);
73
74	$log->debug("import xml is ",sub { Dumper($self->{'import_xml'}) });
75
76	return $self;
77	}
78
79
80	=head2 _x
81
82	Convert string from UTF-8 to code page defined in C<import_xml>.
83
84	my $text = $n->_x('utf8 text');
85
86	Default application code page is C<ISO-8859-2>. You will probably want to
87	change that when creating new instance of object based on this one.
88
89	=cut
90
91	sub _x {
92	my $self = shift;
93	my $utf8 = shift \|\| return;
94
95	# create UTF-8 convertor for import_xml files
96	$self->{'utf2cp'} \|\|= Text::Iconv->new('UTF-8' ,$self->{'code_page'} \|\| 'ISO-8859-2');
97
98	return $self->{'utf2cp'}->convert($utf8) \|\|
99	$self->_get_logger()->logwarn("can't convert '$utf8'");
100	}
101
102
103	=head1 AUTHOR
104
105	Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
106
107	=head1 COPYRIGHT & LICENSE
108
109	Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
110
111	This program is free software; you can redistribute it and/or modify it
112	under the same terms as Perl itself.
113
114	=cut
115
116	1; # End of WebPAC::Normalize::XML