WebPAC/Normalize/XML.pm

package WebPAC::Normalize::XML;

use warnings;
use strict;

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 or YAML normalisaton rules

=head1 VERSION

Version 0.03

=cut

our $VERSION = '0.03';

=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

C<tags> define additional tags that can be forced (and an be array).

=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;

        my @force_array = [ $self->{'tag'}, 'config', 'format' ];
        push @force_array, $self->{'tags'} if ($self->{'tags'});

        $self->{'import_xml'} = XMLin($xml_file,
                ForceArray => @force_array,
                ForceContent => 1,
        );

        $log->debug("import xml is ",sub { Dumper($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

Convert string from XML UTF-8 encoding to code page defined in C<xml_file>.

 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;
        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');

        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	dpavlin	12	package WebPAC::Normalize::XML;
2	dpavlin	8
3			use warnings;
4			use strict;
5
6	dpavlin	13	use base qw/WebPAC::Common WebPAC::Normalize/;
7	dpavlin	12	use XML::Simple;
8			use Data::Dumper;
9	dpavlin	13	use Text::Iconv;
10	dpavlin	269	use YAML qw/Dump LoadFile/;
11	dpavlin	8
12			=head1 NAME
13
14	dpavlin	269	WebPAC::Normalize::XML - apply XML or YAML normalisaton rules
15	dpavlin	8
16			=head1 VERSION
17
18	dpavlin	269	Version 0.03
19	dpavlin	8
20			=cut
21
22	dpavlin	269	our $VERSION = '0.03';
23	dpavlin	8
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	dpavlin	13	=head2 open
34	dpavlin	8
35	dpavlin	12	Read normalisation rules defined using XML from C<conf/normalize/*.xml> and
36			parse it.
37	dpavlin	8
38	dpavlin	13	my $n = new WebPAC::Normalize::XML;
39			$n->open(
40	dpavlin	12	tag => 'isis',
41			xml_file => '/path/to/conf/normalize/isis.xml',
42	dpavlin	13	);
43	dpavlin	8
44	dpavlin	12	C<tag> defines tag to use within C<xml_file>
45
46	dpavlin	269	C<xml_file> defines path to normalize XML
47	dpavlin	12
48	dpavlin	62	C<tags> define additional tags that can be forced (and an be array).
49
50	dpavlin	8	=cut
51
52	dpavlin	13	sub open {
53			my $self = shift;
54	dpavlin	8
55	dpavlin	13	my $arg = {@_};
56	dpavlin	8
57			my $log = $self->_get_logger();
58
59	dpavlin	12	foreach my $req (qw/tag xml_file/) {
60	dpavlin	13	$log->logconfess("need argument $req") unless $arg->{$req};
61	dpavlin	12	}
62	dpavlin	8
63	dpavlin	13	$self->{'tag'} = $arg->{'tag'};
64			my $xml_file = $arg->{'xml_file'};
65	dpavlin	8
66	dpavlin	12	$log->info("using $xml_file tag <",$self->{'tag'},">");
67	dpavlin	8
68	dpavlin	12	$log->logdie("normalisation xml file '$xml_file' doesn't exist!") if (! -e $xml_file);
69	dpavlin	8
70	dpavlin	12	$self->{'import_xml_file'} = $xml_file;
71	dpavlin	8
72	dpavlin	62	my @force_array = [ $self->{'tag'}, 'config', 'format' ];
73			push @force_array, $self->{'tags'} if ($self->{'tags'});
74
75	dpavlin	13	$self->{'import_xml'} = XMLin($xml_file,
76	dpavlin	62	ForceArray => @force_array,
77	dpavlin	38	ForceContent => 1,
78	dpavlin	8	);
79
80	dpavlin	252	$log->debug("import xml is ",sub { Dumper($self->{'import_xml'}) }, $self->{lookup} ? " using lookups" : "lookups disabled");
81	dpavlin	8
82	dpavlin	269	#print STDERR Dump($self->{import_xml});
83
84	dpavlin	12	return $self;
85	dpavlin	8	}
86
87	dpavlin	269	=head2 open_yaml
88	dpavlin	8
89	dpavlin	269	Read normalisation rules defined in YAML file located usually at
90			C<conf/normalize/*.yml> and parse it.
91
92			my $n = new WebPAC::Normalize::XML;
93			$n->open_yaml(
94			tag => 'isis',
95			path => '/path/to/conf/normalize/isis.yml',
96			);
97
98			=cut
99
100			sub open_yaml {
101			my $self = shift;
102
103			my $arg = {@_};
104
105			my $log = $self->_get_logger();
106
107			foreach my $req (qw/tag path/) {
108			$log->logconfess("need argument $req") unless $arg->{$req};
109			}
110
111			my $path = $arg->{path};
112			$self->{tag} = $arg->{tag};
113
114			$log->logdie("normalisation yaml file '$path' doesn't exist!") if (! -e $path);
115
116			$log->info("using $path normalization YAML");
117
118	dpavlin	314	$self->{'import_xml'} = LoadFile( $path ) \|\| $log->die("can't load $path: $!");
119	dpavlin	269
120			$log->debug("import yaml is ",sub { Dumper($self->{'import_xml'}) }, $self->{lookup} ? " using lookups" : "lookups disabled");
121
122			$self->{_skip_x} = 1;
123
124			return $self;
125			}
126
127	dpavlin	13	=head2 _x
128	dpavlin	8
129	dpavlin	14	Convert string from XML UTF-8 encoding to code page defined in C<xml_file>.
130	dpavlin	8
131	dpavlin	13	my $text = $n->_x('utf8 text');
132	dpavlin	8
133	dpavlin	13	Default application code page is C<ISO-8859-2>. You will probably want to
134			change that when creating new instance of object based on this one.
135	dpavlin	8
136			=cut
137
138	dpavlin	13	sub _x {
139	dpavlin	8	my $self = shift;
140	dpavlin	13	my $utf8 = shift \|\| return;
141	dpavlin	269	return $utf8 if ($self->{_skip_x});
142	dpavlin	8
143	dpavlin	13	# create UTF-8 convertor for import_xml files
144			$self->{'utf2cp'} \|\|= Text::Iconv->new('UTF-8' ,$self->{'code_page'} \|\| 'ISO-8859-2');
145	dpavlin	8
146	dpavlin	13	return $self->{'utf2cp'}->convert($utf8) \|\|
147			$self->_get_logger()->logwarn("can't convert '$utf8'");
148	dpavlin	8	}
149
150
151			=head1 AUTHOR
152
153			Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>
154
155			=head1 COPYRIGHT & LICENSE
156
157			Copyright 2005 Dobrica Pavlinusic, All Rights Reserved.
158
159			This program is free software; you can redistribute it and/or modify it
160			under the same terms as Perl itself.
161
162			=cut
163
164	dpavlin	12	1; # End of WebPAC::Normalize::XML