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 |
10 |
use Storable; |
8 |
dpavlin |
12 |
use XML::Simple; |
9 |
|
|
use Data::Dumper; |
10 |
dpavlin |
13 |
use Text::Iconv; |
11 |
dpavlin |
8 |
|
12 |
|
|
=head1 NAME |
13 |
|
|
|
14 |
dpavlin |
12 |
WebPAC::Normalize::XML - apply XML normalisaton rules |
15 |
dpavlin |
8 |
|
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 |
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 |
|
|
C<xml_file> defines path to normalize XML. |
47 |
|
|
|
48 |
dpavlin |
8 |
=cut |
49 |
|
|
|
50 |
dpavlin |
13 |
sub open { |
51 |
|
|
my $self = shift; |
52 |
dpavlin |
8 |
|
53 |
dpavlin |
13 |
my $arg = {@_}; |
54 |
dpavlin |
8 |
|
55 |
|
|
my $log = $self->_get_logger(); |
56 |
|
|
|
57 |
dpavlin |
12 |
foreach my $req (qw/tag xml_file/) { |
58 |
dpavlin |
13 |
$log->logconfess("need argument $req") unless $arg->{$req}; |
59 |
dpavlin |
12 |
} |
60 |
dpavlin |
8 |
|
61 |
dpavlin |
13 |
$self->{'tag'} = $arg->{'tag'}; |
62 |
|
|
my $xml_file = $arg->{'xml_file'}; |
63 |
dpavlin |
8 |
|
64 |
dpavlin |
12 |
$log->info("using $xml_file tag <",$self->{'tag'},">"); |
65 |
dpavlin |
8 |
|
66 |
dpavlin |
12 |
$log->logdie("normalisation xml file '$xml_file' doesn't exist!") if (! -e $xml_file); |
67 |
dpavlin |
8 |
|
68 |
dpavlin |
12 |
$self->{'import_xml_file'} = $xml_file; |
69 |
dpavlin |
8 |
|
70 |
dpavlin |
13 |
$self->{'import_xml'} = XMLin($xml_file, |
71 |
dpavlin |
12 |
ForceArray => [ $self->{'tag'}, $self->{'tags'}, 'config', 'format' ], |
72 |
dpavlin |
8 |
); |
73 |
|
|
|
74 |
|
|
$log->debug("import xml is ",sub { Dumper($self->{'import_xml'}) }); |
75 |
|
|
|
76 |
dpavlin |
12 |
return $self; |
77 |
dpavlin |
8 |
} |
78 |
|
|
|
79 |
|
|
|
80 |
dpavlin |
13 |
=head2 _x |
81 |
dpavlin |
8 |
|
82 |
dpavlin |
13 |
Convert string from UTF-8 to code page defined in C<import_xml>. |
83 |
dpavlin |
8 |
|
84 |
dpavlin |
13 |
my $text = $n->_x('utf8 text'); |
85 |
dpavlin |
8 |
|
86 |
dpavlin |
13 |
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 |
dpavlin |
8 |
|
89 |
|
|
=cut |
90 |
|
|
|
91 |
dpavlin |
13 |
sub _x { |
92 |
dpavlin |
8 |
my $self = shift; |
93 |
dpavlin |
13 |
my $utf8 = shift || return; |
94 |
dpavlin |
8 |
|
95 |
dpavlin |
13 |
# create UTF-8 convertor for import_xml files |
96 |
|
|
$self->{'utf2cp'} ||= Text::Iconv->new('UTF-8' ,$self->{'code_page'} || 'ISO-8859-2'); |
97 |
dpavlin |
8 |
|
98 |
dpavlin |
13 |
return $self->{'utf2cp'}->convert($utf8) || |
99 |
|
|
$self->_get_logger()->logwarn("can't convert '$utf8'"); |
100 |
dpavlin |
8 |
} |
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 |
dpavlin |
12 |
1; # End of WebPAC::Normalize::XML |