WebPAC/Output/MARC.pm

package WebPAC::Output::MARC;

use warnings;
use strict;

use base qw/WebPAC::Common/;

use MARC::Record 2.0;   # need 2.0 for utf-8 encoding see marcpm.sf.net
use MARC::Lint;
use Data::Dump qw/dump/;
use Encode qw/from_to decode/;

=head1 NAME

WebPAC::Output::MARC - Create MARC records from C<marc_*> normalisation rules

=head1 VERSION

Version 0.03

=cut

our $VERSION = '0.03';

=head1 SYNOPSIS

Create MARC records from C<marc_*> normalisation rules described in
L<WebPAC::Normalize>.


=head1 FUNCTIONS

=head2 new

  my $marc = new WebPAC::Output::MARC(
        path => '/path/to/output.marc',
        native_encoding => 'iso-8859-2',
        marc_encoding => 'utf-8',
        lint => 1,
        dump => 0,
  )

=cut

sub new {
        my $class = shift;
        my $self = {@_};
        bless($self, $class);

        my $log = $self->_get_logger;

        if ($self->{lint}) {
                $self->{lint}= new MARC::Lint or
                        $log->warn("Can't create MARC::Lint object, linting is disabled");
        }

        if (my $path = $self->{path}) {
                open($self->{fh}, '>', $path) ||
                        $log->logdie("can't open MARC output $path: $!");
                binmode($self->{fh}, ':utf8');

                $log->info("Creating MARC export file $path", $self->{lint} ? ' (with lint)' : '', "\n");
        } else {
                $log->logconfess("new called without path");
        }

        $self->{native_encoding} ||= 'iso-8859-2';
        $self->{marc_encoding} ||= 'utf-8';

        $self ? return $self : return undef;
}

=head2 add

  $marc->add(
        id => $mfn,
        fields => WebPAC::Normalize::_get_marc_fields(),
        leader => WebPAC::Normalize::marc_leader(),
        row => $row,
  );

C<row> is optional parametar which is used when dumping original row to
error log.

=cut

sub add {
        my $self = shift;

        my $arg = {@_};

        my $log = $self->_get_logger;

        $log->logconfess("add needs fields and id arguments")
                unless ($arg->{fields} && defined $arg->{id});

        my $marc = new MARC::Record;
        $marc->encoding( $self->{marc_encoding} );

        my $id = $arg->{id};

        $log->logconfess("fields isn't array") unless (ref($arg->{fields}) eq 'ARRAY');

        my $fields = $arg->{fields};

        $log->debug("original fields = ", sub { dump( $fields ) });

        # recode fields to marc_encoding
        foreach my $j ( 0 .. $#$fields ) {
                foreach my $i ( 0 .. ( ( $#{$fields->[$j]} - 3 ) / 2 ) ) {
                        my $f = $fields->[$j]->[ ($i * 2) + 4 ];
                        $f = decode( $self->{native_encoding}, $f );
                        $fields->[$j]->[ ($i * 2) + 4 ] = $f;
                }
        }

        $log->debug("recode fields = ", sub { dump( $fields ) });

        $marc->add_fields( @$fields );

        # tweak leader
        if (my $new_l = $arg->{leader}) {

                my $leader = $marc->leader;

                foreach my $o ( keys %$new_l ) {
                        my $insert = $new_l->{$o};
                        $leader = substr($leader, 0, $o) .
                                $insert . substr($leader, $o+length($insert));
                }
                $marc->leader( $leader );
        }

        if ($self->{lint}) {
                $self->{lint}->check_record( $marc );
                my @w = $self->{lint}->warnings;
                if (@w) {
                        $log->error("MARC lint detected warning on record $id\n",
                                "<<<<< Original input row:\n",dump($arg->{row}), "\n",
                                ">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump( $arg->{fields} ), "\n",
                                "!!!!! MARC lint warnings:\n",join("\n",@w),"\n"
                        );
                        map { $self->{_marc_lint_warnings}->{$_}++ } @w;
                }
        }

        if ($self->{dump}) {
                $log->info("MARC record on record $id\n",
                        "<<<<< Original imput row:\n",dump($arg->{row}), "\n",
                        ">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump( $arg->{fields} ), "\n",
                );
        }

        print {$self->{fh}} $marc->as_usmarc;

}

=head2 finish

Close MARC output file

  $marc->finish;

It will also dump MARC lint warnings summary if called with C<lint>.

=cut

sub finish {
        my $self = shift;

        my $log = $self->get_logger;

        close( $self->{fh} ) or $log->logdie("can't close ", $self->{path}, ": $!");

        if (my $w = $self->{_marc_lint_warnings}) {
                $log->error("MARC lint warnings summary:\n",
                        join ("\n",
                                map { $w->{$_} . "\t" . $_ }
                                sort { $w->{$b} <=> $w->{$a} } keys %$w
                        )
                );
        }
}

=head1 AUTHOR

Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >>

=head1 COPYRIGHT & LICENSE

Copyright 2006 Dobrica Pavlinusic, All Rights Reserved.

This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.

=cut

1; # End of WebPAC::Output::MARC
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	dpavlin	626	use Encode qw/from_to decode/;
12	dpavlin	578
13			=head1 NAME
14
15			WebPAC::Output::MARC - Create MARC records from C<marc_*> normalisation rules
16
17			=head1 VERSION
18
19	dpavlin	662	Version 0.03
20	dpavlin	578
21			=cut
22
23	dpavlin	662	our $VERSION = '0.03';
24	dpavlin	578
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	dpavlin	626	native_encoding => 'iso-8859-2',
38			marc_encoding => 'utf-8',
39	dpavlin	578	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: $!");
60	dpavlin	662	binmode($self->{fh}, ':utf8');
61	dpavlin	578
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	dpavlin	626	$self->{native_encoding} \|\|= 'iso-8859-2';
68			$self->{marc_encoding} \|\|= 'utf-8';
69	dpavlin	578
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	dpavlin	582	row => $row,
80	dpavlin	578	);
81
82	dpavlin	582	C<row> is optional parametar which is used when dumping original row to
83			error log.
84
85	dpavlin	578	=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	dpavlin	626	$marc->encoding( $self->{marc_encoding} );
99	dpavlin	578
100	dpavlin	590	my $id = $arg->{id};
101	dpavlin	578
102			$log->logconfess("fields isn't array") unless (ref($arg->{fields}) eq 'ARRAY');
103
104	dpavlin	626	my $fields = $arg->{fields};
105	dpavlin	578
106	dpavlin	626	$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			$log->debug("recode fields = ", sub { dump( $fields ) });
118
119			$marc->add_fields( @$fields );
120
121	dpavlin	578	# 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	dpavlin	582	my @w = $self->{lint}->warnings;
137			if (@w) {
138			$log->error("MARC lint detected warning on record $id\n",
139	dpavlin	621	"<<<<< Original input row:\n",dump($arg->{row}), "\n",
140	dpavlin	582	">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump( $arg->{fields} ), "\n",
141			"!!!!! MARC lint warnings:\n",join("\n",@w),"\n"
142			);
143			map { $self->{_marc_lint_warnings}->{$_}++ } @w;
144			}
145	dpavlin	578	}
146
147			if ($self->{dump}) {
148			$log->info("MARC record on record $id\n",
149	dpavlin	587	"<<<<< Original imput row:\n",dump($arg->{row}), "\n",
150	dpavlin	578	">>>>> Normalized MARC row: leader: [", $marc->leader(), "]\n", dump( $arg->{fields} ), "\n",
151			);
152			}
153
154	dpavlin	662	print {$self->{fh}} $marc->as_usmarc;
155	dpavlin	578
156			}
157
158			=head2 finish
159
160			Close MARC output file
161
162			$marc->finish;
163
164	dpavlin	582	It will also dump MARC lint warnings summary if called with C<lint>.
165
166	dpavlin	578	=cut
167
168			sub finish {
169			my $self = shift;
170
171	dpavlin	582	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	dpavlin	578	}
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