1 |
dpavlin |
578 |
package WebPAC::Output::MARC; |
2 |
|
|
|
3 |
|
|
use warnings; |
4 |
|
|
use strict; |
5 |
|
|
|
6 |
|
|
use base qw/WebPAC::Common/; |
7 |
|
|
|
8 |
|
|
use MARC::Record 2.0; # need 2.0 for utf-8 encoding see marcpm.sf.net |
9 |
|
|
use MARC::Lint; |
10 |
|
|
use Data::Dump qw/dump/; |
11 |
|
|
|
12 |
|
|
=head1 NAME |
13 |
|
|
|
14 |
|
|
WebPAC::Output::MARC - Create MARC records from C<marc_*> normalisation 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 |
|
|
Create MARC records from C<marc_*> normalisation rules described in |
27 |
|
|
L<WebPAC::Normalize>. |
28 |
|
|
|
29 |
|
|
|
30 |
|
|
=head1 FUNCTIONS |
31 |
|
|
|
32 |
|
|
=head2 new |
33 |
|
|
|
34 |
|
|
my $marc = new WebPAC::Output::MARC( |
35 |
|
|
path => '/path/to/output.marc', |
36 |
|
|
encoding => 'utf-8', |
37 |
|
|
lint => 1, |
38 |
|
|
dump => 0, |
39 |
|
|
) |
40 |
|
|
|
41 |
|
|
=cut |
42 |
|
|
|
43 |
|
|
sub new { |
44 |
|
|
my $class = shift; |
45 |
|
|
my $self = {@_}; |
46 |
|
|
bless($self, $class); |
47 |
|
|
|
48 |
|
|
my $log = $self->_get_logger; |
49 |
|
|
|
50 |
|
|
if ($self->{lint}) { |
51 |
|
|
$self->{lint}= new MARC::Lint or |
52 |
|
|
$log->warn("Can't create MARC::Lint object, linting is disabled"); |
53 |
|
|
} |
54 |
|
|
|
55 |
|
|
if (my $path = $self->{path}) { |
56 |
|
|
open($self->{fh}, '>', $path) || |
57 |
|
|
$log->logdie("can't open MARC output $path: $!"); |
58 |
|
|
|
59 |
|
|
$log->info("Creating MARC export file $path", $self->{lint} ? ' (with lint)' : '', "\n"); |
60 |
|
|
} else { |
61 |
|
|
$log->logconfess("new called without path"); |
62 |
|
|
} |
63 |
|
|
|
64 |
|
|
$self->{encoding} ||= 'utf-8'; |
65 |
|
|
|
66 |
|
|
$self ? return $self : return undef; |
67 |
|
|
} |
68 |
|
|
|
69 |
|
|
=head2 add |
70 |
|
|
|
71 |
|
|
$marc->add( |
72 |
|
|
id => $mfn, |
73 |
|
|
fields => WebPAC::Normalize::_get_marc_fields(), |
74 |
|
|
leader => WebPAC::Normalize::marc_leader(), |
75 |
dpavlin |
582 |
row => $row, |
76 |
dpavlin |
578 |
); |
77 |
|
|
|
78 |
dpavlin |
582 |
C<row> is optional parametar which is used when dumping original row to |
79 |
|
|
error log. |
80 |
|
|
|
81 |
dpavlin |
578 |
=cut |
82 |
|
|
|
83 |
|
|
sub add { |
84 |
|
|
my $self = shift; |
85 |
|
|
|
86 |
|
|
my $arg = {@_}; |
87 |
|
|
|
88 |
|
|
my $log = $self->_get_logger; |
89 |
|
|
|
90 |
|
|
$log->logconfess("add needs fields and id arguments") |
91 |
|
|
unless ($arg->{fields} && defined $arg->{id}); |
92 |
|
|
|
93 |
|
|
my $marc = new MARC::Record; |
94 |
|
|
$marc->encoding( $self->{encoding} ); |
95 |
|
|
|
96 |
|
|
my $id = $self->{id}; |
97 |
|
|
|
98 |
|
|
$log->logconfess("fields isn't array") unless (ref($arg->{fields}) eq 'ARRAY'); |
99 |
|
|
|
100 |
|
|
$marc->add_fields( @{ $arg->{fields} } ); |
101 |
|
|
|
102 |
|
|
# tweak leader |
103 |
|
|
if (my $new_l = $arg->{leader}) { |
104 |
|
|
|
105 |
|
|
my $leader = $marc->leader; |
106 |
|
|
|
107 |
|
|
foreach my $o ( keys %$new_l ) { |
108 |
|
|
my $insert = $new_l->{$o}; |
109 |
|
|
$leader = substr($leader, 0, $o) . |
110 |
|
|
$insert . substr($leader, $o+length($insert)); |
111 |
|
|
} |
112 |
|
|
$marc->leader( $leader ); |
113 |
|
|
} |
114 |
|
|
|
115 |
|
|
if ($self->{lint}) { |
116 |
|
|
$self->{lint}->check_record( $marc ); |
117 |
dpavlin |
582 |
my @w = $self->{lint}->warnings; |
118 |
|
|
if (@w) { |
119 |
|
|
$log->error("MARC lint detected warning on record $id\n", |
120 |
|
|
"<<<<< Original imput row:\n",dump($arg->{row}), "\n", |
121 |
|
|
">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump( $arg->{fields} ), "\n", |
122 |
|
|
"!!!!! MARC lint warnings:\n",join("\n",@w),"\n" |
123 |
|
|
); |
124 |
|
|
map { $self->{_marc_lint_warnings}->{$_}++ } @w; |
125 |
|
|
} |
126 |
dpavlin |
578 |
} |
127 |
|
|
|
128 |
|
|
if ($self->{dump}) { |
129 |
|
|
$log->info("MARC record on record $id\n", |
130 |
|
|
"<<<<< Original imput row:\n",dump($self->{row}), "\n", |
131 |
|
|
">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump( $arg->{fields} ), "\n", |
132 |
|
|
); |
133 |
|
|
} |
134 |
|
|
|
135 |
|
|
print {$self->{fh}} $marc->as_usmarc; |
136 |
|
|
|
137 |
|
|
} |
138 |
|
|
|
139 |
|
|
=head2 finish |
140 |
|
|
|
141 |
|
|
Close MARC output file |
142 |
|
|
|
143 |
|
|
$marc->finish; |
144 |
|
|
|
145 |
dpavlin |
582 |
It will also dump MARC lint warnings summary if called with C<lint>. |
146 |
|
|
|
147 |
dpavlin |
578 |
=cut |
148 |
|
|
|
149 |
|
|
sub finish { |
150 |
|
|
my $self = shift; |
151 |
|
|
|
152 |
dpavlin |
582 |
my $log = $self->get_logger; |
153 |
|
|
|
154 |
|
|
close( $self->{fh} ) or $log->logdie("can't close ", $self->{path}, ": $!"); |
155 |
|
|
|
156 |
|
|
if (my $w = $self->{_marc_lint_warnings}) { |
157 |
|
|
$log->error("MARC lint warnings summary:\n", |
158 |
|
|
join ("\n", |
159 |
|
|
map { $w->{$_} . "\t" . $_ } |
160 |
|
|
sort { $w->{$b} <=> $w->{$a} } keys %$w |
161 |
|
|
) |
162 |
|
|
); |
163 |
|
|
} |
164 |
dpavlin |
578 |
} |
165 |
|
|
|
166 |
|
|
=head1 AUTHOR |
167 |
|
|
|
168 |
|
|
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
169 |
|
|
|
170 |
|
|
=head1 COPYRIGHT & LICENSE |
171 |
|
|
|
172 |
|
|
Copyright 2006 Dobrica Pavlinusic, All Rights Reserved. |
173 |
|
|
|
174 |
|
|
This program is free software; you can redistribute it and/or modify it |
175 |
|
|
under the same terms as Perl itself. |
176 |
|
|
|
177 |
|
|
=cut |
178 |
|
|
|
179 |
|
|
1; # End of WebPAC::Output::MARC |