1 |
package WebPAC::Output::MARC; |
2 |
|
3 |
use warnings; |
4 |
use strict; |
5 |
|
6 |
use base qw/WebPAC::Common/; |
7 |
|
8 |
use MARC::Record; |
9 |
use MARC::Lint; |
10 |
use Data::Dump qw/dump/; |
11 |
use Encode qw/from_to decode/; |
12 |
|
13 |
=head1 NAME |
14 |
|
15 |
WebPAC::Output::MARC - Create MARC records from C<marc_*> normalisation rules |
16 |
|
17 |
=head1 VERSION |
18 |
|
19 |
Version 0.04 |
20 |
|
21 |
=cut |
22 |
|
23 |
our $VERSION = '0.04'; |
24 |
|
25 |
=head1 SYNOPSIS |
26 |
|
27 |
Create MARC records from C<marc_*> normalisation rules described in |
28 |
L<WebPAC::Normalize>. |
29 |
|
30 |
|
31 |
=head1 FUNCTIONS |
32 |
|
33 |
=head2 new |
34 |
|
35 |
my $marc = new WebPAC::Output::MARC( |
36 |
path => '/path/to/output.marc', |
37 |
marc_encoding => 'utf-8', |
38 |
lint => 1, |
39 |
dump => 0, |
40 |
) |
41 |
|
42 |
=cut |
43 |
|
44 |
sub new { |
45 |
my $class = shift; |
46 |
my $self = {@_}; |
47 |
bless($self, $class); |
48 |
|
49 |
my $log = $self->_get_logger; |
50 |
|
51 |
if ($self->{lint}) { |
52 |
$self->{lint}= new MARC::Lint or |
53 |
$log->warn("Can't create MARC::Lint object, linting is disabled"); |
54 |
} |
55 |
|
56 |
if (my $path = $self->{path}) { |
57 |
open($self->{fh}, '>', $path) || |
58 |
$log->logdie("can't open MARC output $path: $!"); |
59 |
binmode($self->{fh}, ':utf8'); |
60 |
|
61 |
$log->info("Creating MARC export file $path", $self->{lint} ? ' (with lint)' : '', "\n"); |
62 |
} else { |
63 |
$log->logconfess("new called without path"); |
64 |
} |
65 |
|
66 |
$self->{marc_encoding} ||= 'utf-8'; |
67 |
|
68 |
$self ? return $self : return undef; |
69 |
} |
70 |
|
71 |
=head2 add |
72 |
|
73 |
$marc->add( |
74 |
id => $mfn, |
75 |
fields => WebPAC::Normalize::_get_marc_fields(), |
76 |
leader => WebPAC::Normalize::_get_marc_leader(), |
77 |
row => $row, |
78 |
); |
79 |
|
80 |
C<row> is optional parametar which is used when dumping original row to |
81 |
error log. |
82 |
|
83 |
=cut |
84 |
|
85 |
sub add { |
86 |
my $self = shift; |
87 |
|
88 |
my $arg = {@_}; |
89 |
|
90 |
my $log = $self->_get_logger; |
91 |
|
92 |
$log->logconfess("add needs fields and id arguments") |
93 |
unless ($arg->{fields} && defined $arg->{id}); |
94 |
|
95 |
my $marc = new MARC::Record; |
96 |
$marc->encoding( $self->{marc_encoding} ); |
97 |
|
98 |
my $id = $arg->{id}; |
99 |
|
100 |
$log->logconfess("fields isn't array") unless (ref($arg->{fields}) eq 'ARRAY'); |
101 |
|
102 |
my $fields = $arg->{fields}; |
103 |
|
104 |
$log->debug("original fields = ", sub { dump( $fields ) }); |
105 |
|
106 |
# recode fields to marc_encoding |
107 |
foreach my $j ( 0 .. $#$fields ) { |
108 |
foreach my $i ( 0 .. ( ( $#{$fields->[$j]} - 3 ) / 2 ) ) { |
109 |
my $f = $fields->[$j]->[ ($i * 2) + 4 ]; |
110 |
$fields->[$j]->[ ($i * 2) + 4 ] = $f; |
111 |
} |
112 |
} |
113 |
|
114 |
# sort fields |
115 |
@$fields = sort { $a->[0] <=> $b->[0] } @$fields; |
116 |
|
117 |
$log->debug("recode fields = ", sub { dump( $fields ) }); |
118 |
|
119 |
$marc->add_fields( @$fields ); |
120 |
|
121 |
# tweak leader |
122 |
if (my $new_l = $arg->{leader}) { |
123 |
|
124 |
my $leader = $marc->leader; |
125 |
|
126 |
foreach my $o ( keys %$new_l ) { |
127 |
my $insert = $new_l->{$o}; |
128 |
$leader = substr($leader, 0, $o) . |
129 |
$insert . substr($leader, $o+length($insert)); |
130 |
} |
131 |
$marc->leader( $leader ); |
132 |
} |
133 |
|
134 |
if ($self->{lint}) { |
135 |
$self->{lint}->check_record( $marc ); |
136 |
my @w = $self->{lint}->warnings; |
137 |
if (@w) { |
138 |
$log->error("MARC lint detected warning on record $id\n", |
139 |
"<<<<< Original input row:\n",dump($arg->{row}), "\n", |
140 |
">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump( $fields ), "\n", |
141 |
"!!!!! MARC lint warnings:\n",join("\n",@w),"\n" |
142 |
); |
143 |
map { $self->{_marc_lint_warnings}->{$_}++ } @w; |
144 |
} |
145 |
} |
146 |
|
147 |
if ($self->{dump}) { |
148 |
$log->info("MARC record on record $id\n", |
149 |
"<<<<< Original imput row:\n",dump($arg->{row}), "\n", |
150 |
">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump( $fields ), "\n", |
151 |
); |
152 |
} |
153 |
|
154 |
print {$self->{fh}} $marc->as_usmarc; |
155 |
|
156 |
} |
157 |
|
158 |
=head2 finish |
159 |
|
160 |
Close MARC output file |
161 |
|
162 |
$marc->finish; |
163 |
|
164 |
It will also dump MARC lint warnings summary if called with C<lint>. |
165 |
|
166 |
=cut |
167 |
|
168 |
sub finish { |
169 |
my $self = shift; |
170 |
|
171 |
my $log = $self->get_logger; |
172 |
|
173 |
close( $self->{fh} ) or $log->logdie("can't close ", $self->{path}, ": $!"); |
174 |
|
175 |
if (my $w = $self->{_marc_lint_warnings}) { |
176 |
$log->error("MARC lint warnings summary:\n", |
177 |
join ("\n", |
178 |
map { $w->{$_} . "\t" . $_ } |
179 |
sort { $w->{$b} <=> $w->{$a} } keys %$w |
180 |
) |
181 |
); |
182 |
} |
183 |
} |
184 |
|
185 |
=head1 AUTHOR |
186 |
|
187 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
188 |
|
189 |
=head1 COPYRIGHT & LICENSE |
190 |
|
191 |
Copyright 2006 Dobrica Pavlinusic, All Rights Reserved. |
192 |
|
193 |
This program is free software; you can redistribute it and/or modify it |
194 |
under the same terms as Perl itself. |
195 |
|
196 |
=cut |
197 |
|
198 |
1; # End of WebPAC::Output::MARC |