1 package WebPAC::Normalize::MARC;
4 marc marc_indicators marc_repeatable_subfield
5 marc_compose marc_leader marc_fixed
6 marc_duplicate marc_remove marc_count
15 use Storable qw/dclone/;
16 use Data::Dump qw/dump/;
19 use WebPAC::Normalize;
23 my ($marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $marc_leader);
24 my ($marc_record_offset, $marc_fetch_offset) = (0, 0);
30 WebPAC::Normalize::MARC - create MARC/ISO2709 records
39 from => 225, to => 440,
53 'a, |x ; |v. |n, |p ; |v',
58 Returns number of records produced.
62 my $created_with_marc_template;
66 warn "## marc_template(",dump($args),")",$/ if $debug;
68 foreach ( qw/subfields_rename isis_template marc_template/ ) {
69 # warn "ref($_) = ",ref($args->{$_}) if $debug;
70 die "$_ not ARRAY" if defined($args->{$_}) && ref($args->{$_}) ne 'ARRAY';
73 die "marc_template needs isis_template or marc_template"
74 if ! defined $args->{isis_template} && ! defined $args->{marc_template};
76 my $rec = WebPAC::Normalize::_get_rec() || die "_get_rec?";
77 my $r = $rec->{ $args->{from} } || return;
78 die "record field ", $args->{from}, " isn't array ",dump( $rec ) unless (ref($r) eq 'ARRAY');
80 my @subfields_rename = @{ $args->{subfields_rename} };
81 # warn "### subfields_rename [$#subfields_rename] = ",dump( @subfields_rename ) if $debug;
83 confess "need mapping in pairs for subfields_rename"
84 if $#subfields_rename % 2 != 1;
86 my ( $subfields_rename, $from_subfields );
87 our $to_subfields = {};
88 while ( my ( $from, $to ) = splice(@subfields_rename, 0, 2) ) {
90 $from_subfields->{ $from }++,
91 $to_subfields->{ $to }++
93 $subfields_rename->{ $from }->[ $f ] = [ $to => $t ];
95 warn "### subfields_rename = ",dump( $subfields_rename ),$/ if $debug;
96 warn "### from_subfields = ", dump( $from_subfields ),$/ if $debug;
97 warn "### to_subfields = ", dump( $to_subfields ),$/ if $debug;
101 $_template->{isis}->{fields_re} = join('|', keys %$from_subfields );
102 $_template->{marc}->{fields_re} = join('|', keys %$to_subfields );
106 sub _parse_template {
107 my ( $name, $templates ) = @_;
109 my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
111 foreach my $template ( @{ $templates } ) {
116 my $nr = $count->{$sf}++;
117 push @order, [ $sf, $nr ];
120 my $pos_template = $template;
121 $pos_template =~ s/($fields_re)/my_count($1)/ge;
122 my $count_key = dump( $count );
123 warn "### template: |$template| -> |$pos_template| count = $count_key order = ",dump( @order ),$/ if $debug;
124 $_template->{$name}->{pos}->{ $count_key } = $pos_template;
125 $_template->{$name}->{order}->{ $pos_template } = [ @order ];
127 warn "### from ",dump( $templates ), " using $fields_re created ", dump( $_template ),$/ if $debug;
130 _parse_template( 'marc', $args->{marc_template} );
131 _parse_template( 'isis', $args->{isis_template} );
132 warn "### _template = ",dump( $_template ),$/ if $debug;
136 our $from_rec = $rec->{ $args->{from} };
138 foreach my $r ( @$from_rec ) {
140 my $to = $args->{to};
141 my ($i1,$i2) = _get_marc_indicators( $to );
142 $m = [ $to, $i1, $i2 ];
144 $created_with_marc_template->{ $to }++;
146 warn "### r = ",dump( $r ),$/ if $debug;
148 my ( $from_mapping, $from_count, $to_count );
150 foreach my $from_sf ( keys %{$r} ) {
151 # skip everything which isn't one char subfield (e.g. 'subfields')
152 next unless $from_sf =~ m/^\w$/;
153 my $from_nr = $from_count->{$from_sf}++;
154 my $rename_to = $subfields_rename->{ $from_sf } ||
155 die "can't find subfield rename for $from_sf/$from_nr in ", dump( $subfields_rename );
156 my ( $to_sf, $to_nr ) = @{ $rename_to->[$from_nr] };
157 $to_mapping->{ $to_sf }->[ $to_nr ] = [ $from_sf => $from_nr ];
159 my $to_nr2 = $to_count->{ $to_sf }++;
160 $from_mapping->{ $from_sf }->[ $from_nr ] = [ $to_sf => $to_nr2 ];
162 warn "### from $from_sf/$from_nr -> $to_sf/$to_nr\tto $from_sf/$from_nr -> $to_sf/$to_nr2\n" if $debug;
165 warn "### from_mapping = ",dump( $from_mapping ), "\n### to_mapping = ",dump( $to_mapping ),$/ if $debug;
168 from => dump( $from_count ),
169 to => dump( $to_count),
172 warn "### count_key = ",dump( $count_key ),$/ if $debug;
174 my $processed_templates = 0;
176 # this defines order of traversal
177 foreach ( qw/isis:from marc:to/ ) {
178 my ($name,$count_name) = split(/:/);
180 my $ckey = $count_key->{$count_name} || die "can't find count_key $count_name in ",dump( $count_key );
182 my $template = $_template->{$name}->{pos}->{ $ckey } || next;
183 $processed_templates++;
185 warn "### traverse $name $count_name selected template: |$template|\n" if $debug;
189 my @templates = split(/\|/, $template );
190 @templates = ( $template ) unless @templates;
192 warn "### templates = ",dump( @templates ),$/ if $debug;
194 foreach my $sf ( @templates ) {
196 my ( $name, $r, $pre, $sf, $nr, $post ) = @_;
197 warn "#### fill_in( $name, r, '$pre', $sf, $nr, '$post' )\n" if $debug;
198 my ( $from_sf, $from_nr );
199 if ( $name eq 'marc' ) {
200 die "no $sf/$nr in to_mapping: ",dump( $to_mapping ), "\n>>>> from record ",dump( $r ), "\n>>>> full record = ",dump( $from_rec ) unless defined $to_mapping->{$sf}->[$nr];
201 ( $from_sf, $from_nr ) = @{ $to_mapping->{$sf}->[$nr] };
203 ( $from_sf, $from_nr ) = ( $sf, $nr );
205 my $v = $r->{ $from_sf }; # || die "no $from_sf/$from_nr";
206 if ( ref( $v ) eq 'ARRAY' ) {
207 $v = $pre . $v->[$from_nr] . $post;
208 } elsif ( $from_nr == 0 ) {
209 $v = $pre . $v . $post;
211 die "requested subfield $from_sf/$from_nr but it's ",dump( $v );
213 warn "#### fill_in( $sf, $nr ) = $from_sf/$from_nr >>>> ",dump( $v ),$/ if $debug;
214 $fill_in->{$sf}->[$nr] = $v;
217 my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
218 warn "#### $sf <<<< $fields_re\n" if $debug;
219 $sf =~ s/^(.*?)($fields_re)(\d+)(.*?)$/fill_in($name,$r,$1,$2,$3,$4)/ge;
220 warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/ if $debug;
223 warn "## template: |$template|\n## _template->$name = ",dump( $_template->{$name} ),$/ if $debug;
225 foreach my $sf ( @{ $_template->{$name}->{order}->{$template} } ) {
226 my ( $sf, $nr ) = @$sf;
227 my $v = $fill_in->{$sf}->[$nr];
228 die "can't find fill_in $sf/$nr" unless defined $v;
229 if ( $name eq 'isis') {
230 ( $sf, $nr ) = @{ $from_mapping->{$sf}->[$nr] };
232 warn "++ $sf/$nr |$v|\n" if $debug;
233 push @$m, ( $sf, $v );
236 warn "#### >>>> created MARC record: ", dump( $m ),$/ if $debug;
243 die "I don't have template for fields ",dump( $count_key ), "\n## available templates\n", dump( $_template ) unless $processed_templates;
244 warn ">>> $processed_templates templates applied to data\n",$/ if $debug;
250 foreach my $marc ( @marc_out ) {
251 warn "+++ ",dump( $marc ),$/ if $debug;
256 warn "### marc_template produced $recs MARC records: ",dump( @marc_out ),$/ if $debug;
263 Setup fields within MARC leader or get leader
265 marc_leader('05','c');
266 my $leader = marc_leader();
271 my ($offset,$value) = @_;
274 $marc_leader->[ $marc_record_offset ]->{ $offset } = $value;
277 if (defined($marc_leader)) {
278 die "marc_leader not array = ", dump( $marc_leader ) unless (ref($marc_leader) eq 'ARRAY');
279 return $marc_leader->[ $marc_record_offset ];
288 Create control/indentifier fields with values in fixed positions
290 marc_fixed('008', 00, '070402');
291 marc_fixed('008', 39, '|');
293 Positions not specified will be filled with spaces (C<0x20>).
295 There will be no effort to extend last specified value to full length of
301 my ($f, $pos, $val) = @_;
302 die "need marc(field, position, value)" unless defined($f) && defined($pos);
304 confess "need val" unless defined $val;
311 if (length($old) <= $pos) {
312 $_->[1] .= ' ' x ( $pos - length($old) ) . $val;
313 warn "## marc_fixed($f,$pos,'$val') append '$old' -> '$_->[1]'\n" if ($debug > 1);
315 $_->[1] = substr($old, 0, $pos) . $val . substr($old, $pos + length($val));
316 warn "## marc_fixed($f,$pos,'$val') update '$old' -> '$_->[1]'\n" if ($debug > 1);
320 } @{ $marc_record->[ $marc_record_offset ] };
323 my $v = ' ' x $pos . $val;
324 push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $v ];
325 warn "## marc_fixed($f,$pos,'val') created '$v'\n" if ($debug > 1);
331 Save value for MARC field
333 marc('900','a', rec('200','a') );
334 marc('001', rec('000') );
339 my $f = shift or die "marc needs field";
340 die "marc field must be numer" unless ($f =~ /^\d+$/);
344 $sf = shift or die "marc needs subfield";
348 my $v = $_; # make var read-write for Encode
349 next unless (defined($v) && $v !~ /^\s*$/);
350 my ($i1,$i2) = _get_marc_indicators( $f );
352 push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $i1, $i2, $sf => $v ];
354 push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $v ];
359 =head2 marc_repeatable_subfield
361 Save values for MARC repetable subfield
363 marc_repeatable_subfield('910', 'z', rec('909') );
367 sub marc_repeatable_subfield {
369 die "marc_repeatable_subfield need field and subfield!\n" unless ($f && $sf);
370 $marc_repeatable_subfield->{ $f . $sf }++;
374 =head2 marc_indicators
376 Set both indicators for MARC field
378 marc_indicators('900', ' ', 1);
380 Any indicator value other than C<0-9> will be treated as undefined.
384 sub marc_indicators {
385 my $f = shift || die "marc_indicators need field!\n";
387 die "marc_indicators($f, ...) need i1!\n" unless(defined($i1));
388 die "marc_indicators($f, $i1, ...) need i2!\n" unless(defined($i2));
390 $i1 = ' ' if ($i1 !~ /^\d$/);
391 $i2 = ' ' if ($i2 !~ /^\d$/);
392 @{ $marc_indicators->{$f} } = ($i1,$i2);
395 sub _get_marc_indicators {
396 my $f = shift || confess "need field!\n";
397 return defined($marc_indicators->{$f}) ? @{ $marc_indicators->{$f} } : (' ',' ');
402 Save values for each MARC subfield explicitly
411 If you specify C<+> for subfield, value will be appended
412 to previous defined subfield.
417 my $f = shift or die "marc_compose needs field";
418 die "marc_compose field must be numer" unless ($f =~ /^\d+$/);
420 my ($i1,$i2) = _get_marc_indicators( $f );
421 my $m = [ $f, $i1, $i2 ];
423 warn "### marc_compose input subfields = ", dump(@_),$/ if ($debug > 2);
426 die "ERROR: marc_compose",dump($f,@_)," not valid (must be even).\nDo you need to add first() or join() around some argument?\n";
433 next unless (defined($v) && $v !~ /^\s*$/);
434 warn "## ++ marc_compose($f,$sf,$v) ", dump( $m ),$/ if ($debug > 1);
436 push @$m, ( $sf, $v );
442 warn "## marc_compose current marc = ", dump( $m ),$/ if ($debug > 1);
444 push @{ $marc_record->[ $marc_record_offset ] }, $m if ($#{$m} > 2);
447 =head2 marc_duplicate
449 Generate copy of current MARC record and continue working on copy
453 Copies can be accessed using C<< _get_marc_fields( fetch_next => 1 ) >> or
454 C<< _get_marc_fields( offset => 42 ) >>.
459 my $m = $marc_record->[ -1 ];
460 die "can't duplicate record which isn't defined" unless ($m);
461 push @{ $marc_record }, dclone( $m );
462 push @{ $marc_leader }, dclone( marc_leader() );
463 warn "## marc_duplicate = ", dump(@$marc_leader, @$marc_record), $/ if ($debug > 1);
464 $marc_record_offset = $#{ $marc_record };
465 warn "## marc_record_offset = $marc_record_offset", $/ if ($debug > 1);
471 Remove some field or subfield from MARC record.
474 marc_remove('200','a');
476 This will erase field C<200> or C<200^a> from current MARC record.
480 Will remove all fields in current MARC record.
482 This is useful after calling C<marc_duplicate> or on it's own (but, you
483 should probably just remove that subfield definition if you are not
484 using C<marc_duplicate>).
486 FIXME: support fields < 10.
493 die "marc_remove needs record number" unless defined($f);
495 my $marc = $marc_record->[ $marc_record_offset ];
497 warn "### marc_remove before = ", dump( $marc ), $/ if ($debug > 2);
501 delete( $marc_record->[ $marc_record_offset ] );
502 warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1);
507 foreach ( 0 .. $#{ $marc } ) {
508 last unless (defined $marc->[$i]);
509 warn "#### working on ",dump( @{ $marc->[$i] }), $/ if ($debug > 3);
510 if ($marc->[$i]->[0] eq $f) {
513 splice @$marc, $i, 1;
514 warn "#### slice \@\$marc, $i, 1 = ",dump( @{ $marc }), $/ if ($debug > 3);
517 foreach my $j ( 0 .. (( $#{ $marc->[$i] } - 3 ) / 2) ) {
518 my $o = ($j * 2) + 3;
519 if ($marc->[$i]->[$o] eq $sf) {
521 splice @{$marc->[$i]}, $o, 2;
522 warn "#### slice \@{\$marc->[$i]}, $o, 2 = ", dump( @{ $marc }), $/ if ($debug > 3);
523 # is record now empty?
524 if ($#{ $marc->[$i] } == 2) {
525 splice @$marc, $i, 1;
526 warn "#### slice \@\$marc, $i, 1 = ", dump( @{ $marc }), $/ if ($debug > 3);
536 warn "### marc_remove($f", $sf ? ",$sf" : "", ") after = ", dump( $marc ), $/ if ($debug > 2);
538 $marc_record->[ $marc_record_offset ] = $marc;
541 warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1);
544 =head2 marc_original_order
546 Copy all subfields preserving original order to marc field.
548 marc_original_order( marc_field_number, original_input_field_number );
550 Please note that field numbers are consistent with other commands (marc
551 field number first), but somewhat counter-intuitive (destination and then
554 You might want to use this command if you are just renaming subfields or
555 using pre-processing modify_record in C<config.yml> and don't need any
556 post-processing or want to preserve order of original subfields.
561 sub marc_original_order {
563 my ($to, $from) = @_;
564 die "marc_original_order needs from and to fields\n" unless ($from && $to);
566 return unless defined($rec->{$from});
568 my $r = $rec->{$from};
569 die "record field $from isn't array ",dump( $rec ) unless (ref($r) eq 'ARRAY');
571 my ($i1,$i2) = _get_marc_indicators( $to );
572 warn "## marc_original_order($to,$from) source = ", dump( $r ),$/ if ($debug > 1);
574 foreach my $d (@$r) {
576 if (! defined($d->{subfields}) && ref($d->{subfields}) ne 'ARRAY') {
577 warn "# marc_original_order($to,$from): field $from doesn't have subfields specification\n";
581 my @sfs = @{ $d->{subfields} };
583 die "field $from doesn't have even number of subfields specifications\n" unless($#sfs % 2 == 1);
585 warn "#--> d: ",dump($d), "\n#--> sfs: ",dump(@sfs),$/ if ($debug > 2);
587 my $m = [ $to, $i1, $i2 ];
589 while (my $sf = shift @sfs) {
591 warn "#--> sf: ",dump($sf), $/ if ($debug > 2);
592 my $offset = shift @sfs;
593 die "corrupted sufields specification for field $from\n" unless defined($offset);
596 if (ref($d->{$sf}) eq 'ARRAY') {
597 $v = $d->{$sf}->[$offset] if (defined($d->{$sf}->[$offset]));
598 } elsif ($offset == 0) {
601 die "field $from subfield '$sf' need occurence $offset which doesn't exist", dump($d->{$sf});
603 push @$m, ( $sf, $v ) if (defined($v));
607 push @{ $marc_record->[ $marc_record_offset ] }, $m;
611 warn "## marc_record = ", dump( $marc_record ),$/ if ($debug > 1);
617 Return number of MARC records created using L</marc_duplicate>.
619 print "created ", marc_count(), " records";
624 return $#{ $marc_record };
629 Clone marc records from input file, whole or just some fields/indicators
631 marc_clone; # whole record
636 foreach my $f ( keys %$rec ) {
637 warn "## marc_clone $f\n" if $debug;
638 marc_original_order( $f, $f );
642 =head1 PRIVATE FUNCTIONS
651 my $marc = shift || die "no marc?";
652 push @{ $marc_record->[ $marc_record_offset ] }, $marc;
657 Clean internal structures
662 ($marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $marc_leader, $created_with_marc_template) = ();
663 ($marc_record_offset, $marc_fetch_offset) = (0,0);
667 =head2 _get_marc_fields
669 Get all fields defined by calls to C<marc>
671 $marc->add_fields( WebPAC::Normalize:_get_marc_fields() );
673 We are using I<magic> which detect repeatable fields only from
674 sequence of field/subfield data generated by normalization.
676 This magic is disabled for all records created with C<marc_template>.
678 Repeatable field is created when there is second occurence of same subfield or
679 if any of indicators are different.
681 This is sane for most cases. Something like:
687 will be created from any combination of:
689 900a-1 900a-2 900a-3 900b-1 900b-2 900c-1
693 marc('900','a', rec('200','a') );
694 marc('900','b', rec('200','b') );
695 marc('900','c', rec('200','c') );
697 which might not be what you have in mind. If you need repeatable subfield,
698 define it using C<marc_repeatable_subfield> like this:
700 marc_repeatable_subfield('900','a');
701 marc('900','a', rec('200','a') );
702 marc('900','b', rec('200','b') );
703 marc('900','c', rec('200','c') );
707 900a-1 900a-2 900a-3 900b-1 900c-1
710 There is also support for returning next or specific using:
712 while (my $mf = WebPAC::Normalize:_get_marc_fields( fetch_next => 1 ) ) {
713 # do something with $mf
716 will always return fields from next MARC record or
718 my $mf = WebPAC::Normalize::_get_marc_fields( offset => 42 );
720 will return 42th copy record (if it exists).
726 sub _get_marc_fields {
729 warn "### _get_marc_fields arg: ", dump($arg), $/ if ($debug > 2);
730 $fetch_pos = $marc_fetch_offset;
731 if ($arg->{offset}) {
732 $fetch_pos = $arg->{offset};
733 } elsif($arg->{fetch_next}) {
734 $marc_fetch_offset++;
737 return if (! $marc_record || ref($marc_record) ne 'ARRAY');
739 warn "### full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 2);
741 my $marc_rec = $marc_record->[ $fetch_pos ];
743 warn "## _get_marc_fields (at offset: $fetch_pos) -- marc_record = ", dump( @$marc_rec ), $/ if ($debug > 1);
745 return if (! $marc_rec || ref($marc_rec) ne 'ARRAY' || $#{ $marc_rec } < 0);
747 # first, sort all existing fields
748 # XXX might not be needed, but modern perl might randomize elements in hash
749 # my @sorted_marc_record = sort {
750 # $a->[0] . ( $a->[3] || '' ) cmp $b->[0] . ( $b->[3] || '')
753 my @sorted_marc_record = @{ $marc_rec }; ### FIXME disable sorting
758 # count unique field-subfields (used for offset when walking to next subfield)
760 map { $u->{ $_->[0] . ( $_->[3] || '') }++ } @sorted_marc_record;
763 warn "## marc_repeatable_subfield = ", dump( $marc_repeatable_subfield ), $/ if ( $marc_repeatable_subfield );
764 warn "## marc_record[$fetch_pos] = ", dump( $marc_rec ), $/;
765 warn "## sorted_marc_record = ", dump( \@sorted_marc_record ), $/;
766 warn "## subfield count = ", dump( $u ), $/;
769 my $len = $#sorted_marc_record;
774 warn "## created_with_marc_template = ",dump( $created_with_marc_template ) if $debug;
776 foreach ( 0 .. $len ) {
778 # find next element which isn't visited
779 while ($visited->{$i}) {
780 $i = ($i + 1) % ($len + 1);
786 my $row = dclone( $sorted_marc_record[$i] );
788 if ( $created_with_marc_template->{ $row->[0] } ) {
790 warn "## copied marc_template created ", dump( $row ),$/ if $debug;
794 # field and subfield which is key for
795 # marc_repeatable_subfield and u
796 my $fsf = $row->[0] . ( $row->[3] || '' );
800 print "### field so far [", $#$field, "] : ", dump( $field ), " ", $field ? 'T' : 'F', $/;
801 print "### this [$i]: ", dump( $row ),$/;
802 print "### sf: ", $row->[3], " vs ", $field->[3],
803 $marc_repeatable_subfield->{ $row->[0] . $row->[3] } ? ' (repeatable)' : '', $/,
809 if ( $#$field >= 0 ) {
811 $row->[0] ne $field->[0] || # field
812 $row->[1] ne $field->[1] || # i1
813 $row->[2] ne $field->[2] # i2
816 warn "## saved/1 ", dump( $field ),$/ if ($debug);
820 ( $row->[3] lt $field->[-2] ) # subfield which is not next (e.g. a after c)
822 ( $row->[3] eq $field->[-2] && # same subfield, but not repeatable
823 ! $marc_repeatable_subfield->{ $fsf }
827 warn "## saved/2 ", dump( $field ),$/ if ($debug);
831 # append new subfields to existing field
832 push @$field, ( $row->[3], $row->[4] );
839 if (! $marc_repeatable_subfield->{ $fsf }) {
840 # make step to next subfield
841 $i = ($i + $u->{ $fsf } ) % ($len + 1);
847 warn "## saved/3 ", dump( $field ),$/ if ($debug);
853 =head2 _get_marc_leader
855 Return leader from currently fetched record by L</_get_marc_fields>
857 print WebPAC::Normalize::MARC::_get_marc_leader();
861 sub _get_marc_leader {
862 die "no fetch_pos, did you called _get_marc_fields first?" unless ( defined( $fetch_pos ) );
863 return $marc_leader->[ $fetch_pos ];
866 =head2 _created_marc_records
868 my $nr_records = _created_marc_records;
872 sub _created_marc_records {
873 return $#{ $marc_record } + 1 if $marc_record;