| 1 |
package WebPAC::Normalize::XML; |
| 2 |
|
| 3 |
use warnings; |
| 4 |
use strict; |
| 5 |
|
| 6 |
use base qw/WebPAC::Common WebPAC::Normalize/; |
| 7 |
use XML::Simple; |
| 8 |
use Data::Dumper; |
| 9 |
use Text::Iconv; |
| 10 |
|
| 11 |
=head1 NAME |
| 12 |
|
| 13 |
WebPAC::Normalize::XML - apply XML normalisaton rules |
| 14 |
|
| 15 |
=head1 VERSION |
| 16 |
|
| 17 |
Version 0.02 |
| 18 |
|
| 19 |
=cut |
| 20 |
|
| 21 |
our $VERSION = '0.02'; |
| 22 |
|
| 23 |
=head1 SYNOPSIS |
| 24 |
|
| 25 |
This module uses C<conf/normalize/*.xml> files to perform normalisation |
| 26 |
from input records |
| 27 |
|
| 28 |
=cut |
| 29 |
|
| 30 |
=head1 FUNCTIONS |
| 31 |
|
| 32 |
=head2 open |
| 33 |
|
| 34 |
Read normalisation rules defined using XML from C<conf/normalize/*.xml> and |
| 35 |
parse it. |
| 36 |
|
| 37 |
my $n = new WebPAC::Normalize::XML; |
| 38 |
$n->open( |
| 39 |
tag => 'isis', |
| 40 |
xml_file => '/path/to/conf/normalize/isis.xml', |
| 41 |
); |
| 42 |
|
| 43 |
C<tag> defines tag to use within C<xml_file> |
| 44 |
|
| 45 |
C<xml_file> defines path to normalize XML. |
| 46 |
|
| 47 |
C<tags> define additional tags that can be forced (and an be array). |
| 48 |
|
| 49 |
=cut |
| 50 |
|
| 51 |
sub open { |
| 52 |
my $self = shift; |
| 53 |
|
| 54 |
my $arg = {@_}; |
| 55 |
|
| 56 |
my $log = $self->_get_logger(); |
| 57 |
|
| 58 |
foreach my $req (qw/tag xml_file/) { |
| 59 |
$log->logconfess("need argument $req") unless $arg->{$req}; |
| 60 |
} |
| 61 |
|
| 62 |
$self->{'tag'} = $arg->{'tag'}; |
| 63 |
my $xml_file = $arg->{'xml_file'}; |
| 64 |
|
| 65 |
$log->info("using $xml_file tag <",$self->{'tag'},">"); |
| 66 |
|
| 67 |
$log->logdie("normalisation xml file '$xml_file' doesn't exist!") if (! -e $xml_file); |
| 68 |
|
| 69 |
$self->{'import_xml_file'} = $xml_file; |
| 70 |
|
| 71 |
my @force_array = [ $self->{'tag'}, 'config', 'format' ]; |
| 72 |
push @force_array, $self->{'tags'} if ($self->{'tags'}); |
| 73 |
|
| 74 |
$self->{'import_xml'} = XMLin($xml_file, |
| 75 |
ForceArray => @force_array, |
| 76 |
ForceContent => 1, |
| 77 |
); |
| 78 |
|
| 79 |
$log->debug("import xml is ",sub { Dumper($self->{'import_xml'}) }); |
| 80 |
|
| 81 |
return $self; |
| 82 |
} |
| 83 |
|
| 84 |
|
| 85 |
=head2 _x |
| 86 |
|
| 87 |
Convert string from XML UTF-8 encoding to code page defined in C<xml_file>. |
| 88 |
|
| 89 |
my $text = $n->_x('utf8 text'); |
| 90 |
|
| 91 |
Default application code page is C<ISO-8859-2>. You will probably want to |
| 92 |
change that when creating new instance of object based on this one. |
| 93 |
|
| 94 |
=cut |
| 95 |
|
| 96 |
sub _x { |
| 97 |
my $self = shift; |
| 98 |
my $utf8 = shift || return; |
| 99 |
|
| 100 |
# create UTF-8 convertor for import_xml files |
| 101 |
$self->{'utf2cp'} ||= Text::Iconv->new('UTF-8' ,$self->{'code_page'} || 'ISO-8859-2'); |
| 102 |
|
| 103 |
return $self->{'utf2cp'}->convert($utf8) || |
| 104 |
$self->_get_logger()->logwarn("can't convert '$utf8'"); |
| 105 |
} |
| 106 |
|
| 107 |
|
| 108 |
=head1 AUTHOR |
| 109 |
|
| 110 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
| 111 |
|
| 112 |
=head1 COPYRIGHT & LICENSE |
| 113 |
|
| 114 |
Copyright 2005 Dobrica Pavlinusic, All Rights Reserved. |
| 115 |
|
| 116 |
This program is free software; you can redistribute it and/or modify it |
| 117 |
under the same terms as Perl itself. |
| 118 |
|
| 119 |
=cut |
| 120 |
|
| 121 |
1; # End of WebPAC::Normalize::XML |
Parent Directory
| 
Revision Log