| 1 |
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 |
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 |
native_encoding => 'iso-8859-2', |
| 38 |
marc_encoding => 'utf-8', |
| 39 |
lint => 1, |
| 40 |
dump => 0, |
| 41 |
) |
| 42 |
|
| 43 |
=cut |
| 44 |
|
| 45 |
sub new { |
| 46 |
my $class = shift; |
| 47 |
my $self = {@_}; |
| 48 |
bless($self, $class); |
| 49 |
|
| 50 |
my $log = $self->_get_logger; |
| 51 |
|
| 52 |
if ($self->{lint}) { |
| 53 |
$self->{lint}= new MARC::Lint or |
| 54 |
$log->warn("Can't create MARC::Lint object, linting is disabled"); |
| 55 |
} |
| 56 |
|
| 57 |
if (my $path = $self->{path}) { |
| 58 |
open($self->{fh}, '>', $path) || |
| 59 |
$log->logdie("can't open MARC output $path: $!"); |
| 60 |
binmode($self->{fh}, ':utf8'); |
| 61 |
|
| 62 |
$log->info("Creating MARC export file $path", $self->{lint} ? ' (with lint)' : '', "\n"); |
| 63 |
} else { |
| 64 |
$log->logconfess("new called without path"); |
| 65 |
} |
| 66 |
|
| 67 |
$self->{native_encoding} ||= 'iso-8859-2'; |
| 68 |
$self->{marc_encoding} ||= 'utf-8'; |
| 69 |
|
| 70 |
$self ? return $self : return undef; |
| 71 |
} |
| 72 |
|
| 73 |
=head2 add |
| 74 |
|
| 75 |
$marc->add( |
| 76 |
id => $mfn, |
| 77 |
fields => WebPAC::Normalize::_get_marc_fields(), |
| 78 |
leader => WebPAC::Normalize::marc_leader(), |
| 79 |
row => $row, |
| 80 |
); |
| 81 |
|
| 82 |
C<row> is optional parametar which is used when dumping original row to |
| 83 |
error log. |
| 84 |
|
| 85 |
=cut |
| 86 |
|
| 87 |
sub add { |
| 88 |
my $self = shift; |
| 89 |
|
| 90 |
my $arg = {@_}; |
| 91 |
|
| 92 |
my $log = $self->_get_logger; |
| 93 |
|
| 94 |
$log->logconfess("add needs fields and id arguments") |
| 95 |
unless ($arg->{fields} && defined $arg->{id}); |
| 96 |
|
| 97 |
my $marc = new MARC::Record; |
| 98 |
$marc->encoding( $self->{marc_encoding} ); |
| 99 |
|
| 100 |
my $id = $arg->{id}; |
| 101 |
|
| 102 |
$log->logconfess("fields isn't array") unless (ref($arg->{fields}) eq 'ARRAY'); |
| 103 |
|
| 104 |
my $fields = $arg->{fields}; |
| 105 |
|
| 106 |
$log->debug("original fields = ", sub { dump( $fields ) }); |
| 107 |
|
| 108 |
# recode fields to marc_encoding |
| 109 |
foreach my $j ( 0 .. $#$fields ) { |
| 110 |
foreach my $i ( 0 .. ( ( $#{$fields->[$j]} - 3 ) / 2 ) ) { |
| 111 |
my $f = $fields->[$j]->[ ($i * 2) + 4 ]; |
| 112 |
$f = decode( $self->{native_encoding}, $f ); |
| 113 |
$fields->[$j]->[ ($i * 2) + 4 ] = $f; |
| 114 |
} |
| 115 |
} |
| 116 |
|
| 117 |
# sort fields |
| 118 |
@$fields = sort { $a->[0] <=> $b->[0] } @$fields; |
| 119 |
|
| 120 |
$log->debug("recode fields = ", sub { dump( $fields ) }); |
| 121 |
|
| 122 |
$marc->add_fields( @$fields ); |
| 123 |
|
| 124 |
# tweak leader |
| 125 |
if (my $new_l = $arg->{leader}) { |
| 126 |
|
| 127 |
my $leader = $marc->leader; |
| 128 |
|
| 129 |
foreach my $o ( keys %$new_l ) { |
| 130 |
my $insert = $new_l->{$o}; |
| 131 |
$leader = substr($leader, 0, $o) . |
| 132 |
$insert . substr($leader, $o+length($insert)); |
| 133 |
} |
| 134 |
$marc->leader( $leader ); |
| 135 |
} |
| 136 |
|
| 137 |
if ($self->{lint}) { |
| 138 |
$self->{lint}->check_record( $marc ); |
| 139 |
my @w = $self->{lint}->warnings; |
| 140 |
if (@w) { |
| 141 |
$log->error("MARC lint detected warning on record $id\n", |
| 142 |
"<<<<< Original input row:\n",dump($arg->{row}), "\n", |
| 143 |
">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump( $fields ), "\n", |
| 144 |
"!!!!! MARC lint warnings:\n",join("\n",@w),"\n" |
| 145 |
); |
| 146 |
map { $self->{_marc_lint_warnings}->{$_}++ } @w; |
| 147 |
} |
| 148 |
} |
| 149 |
|
| 150 |
if ($self->{dump}) { |
| 151 |
$log->info("MARC record on record $id\n", |
| 152 |
"<<<<< Original imput row:\n",dump($arg->{row}), "\n", |
| 153 |
">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump( $fields ), "\n", |
| 154 |
); |
| 155 |
} |
| 156 |
|
| 157 |
print {$self->{fh}} $marc->as_usmarc; |
| 158 |
|
| 159 |
} |
| 160 |
|
| 161 |
=head2 finish |
| 162 |
|
| 163 |
Close MARC output file |
| 164 |
|
| 165 |
$marc->finish; |
| 166 |
|
| 167 |
It will also dump MARC lint warnings summary if called with C<lint>. |
| 168 |
|
| 169 |
=cut |
| 170 |
|
| 171 |
sub finish { |
| 172 |
my $self = shift; |
| 173 |
|
| 174 |
my $log = $self->get_logger; |
| 175 |
|
| 176 |
close( $self->{fh} ) or $log->logdie("can't close ", $self->{path}, ": $!"); |
| 177 |
|
| 178 |
if (my $w = $self->{_marc_lint_warnings}) { |
| 179 |
$log->error("MARC lint warnings summary:\n", |
| 180 |
join ("\n", |
| 181 |
map { $w->{$_} . "\t" . $_ } |
| 182 |
sort { $w->{$b} <=> $w->{$a} } keys %$w |
| 183 |
) |
| 184 |
); |
| 185 |
} |
| 186 |
} |
| 187 |
|
| 188 |
=head1 AUTHOR |
| 189 |
|
| 190 |
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
| 191 |
|
| 192 |
=head1 COPYRIGHT & LICENSE |
| 193 |
|
| 194 |
Copyright 2006 Dobrica Pavlinusic, All Rights Reserved. |
| 195 |
|
| 196 |
This program is free software; you can redistribute it and/or modify it |
| 197 |
under the same terms as Perl itself. |
| 198 |
|
| 199 |
=cut |
| 200 |
|
| 201 |
1; # End of WebPAC::Output::MARC |
Parent Directory
| 
Revision Log