search_display search display sorted
- marc marc_indicators marc_repeatable_subfield
- marc_compose marc_leader marc_fixed
- marc_duplicate marc_remove marc_count
- marc_original_order
- marc_template
-
rec1 rec2 rec
frec frec_eq frec_ne
regex prefix suffix surround
push @EXPORT, ( 'isbn_10', 'isbn_13' );
use WebPAC::Normalize::MARC;
-push @EXPORT, ( 'marc_template' );
+push @EXPORT, ( qw/
+ marc marc_indicators marc_repeatable_subfield
+ marc_compose marc_leader marc_fixed
+ marc_duplicate marc_remove marc_count
+ marc_original_order
+ marc_template
+/);
=head1 NAME
sub _set_ds {
$rec = shift or die "no record hash";
+ $WebPAC::Normalize::MARC::rec = $rec;
}
=head2
=cut
-my ($out, $marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $marc_leader);
-my ($marc_record_offset, $marc_fetch_offset) = (0, 0);
+my $out;
sub _get_ds {
#warn "## out = ",dump($out);
sub _clean_ds {
my $a = {@_};
- ($out,$marc_record, $marc_encoding, $marc_repeatable_subfield, $marc_indicators, $marc_leader) = ();
- ($marc_record_offset, $marc_fetch_offset) = (0,0);
- $marc_encoding = $a->{marc_encoding};
+ $out = undef;
+ WebPAC::Normalize::MARC::_clean();
}
=head2 _set_lookup
$load_row_coderef = $coderef;
}
-=head2 _get_marc_fields
-
-Get all fields defined by calls to C<marc>
-
- $marc->add_fields( WebPAC::Normalize:_get_marc_fields() );
-
-We are using I<magic> which detect repeatable fields only from
-sequence of field/subfield data generated by normalization.
-
-Repeatable field is created when there is second occurence of same subfield or
-if any of indicators are different.
-
-This is sane for most cases. Something like:
-
- 900a-1 900b-1 900c-1
- 900a-2 900b-2
- 900a-3
-
-will be created from any combination of:
-
- 900a-1 900a-2 900a-3 900b-1 900b-2 900c-1
-
-and following rules:
-
- marc('900','a', rec('200','a') );
- marc('900','b', rec('200','b') );
- marc('900','c', rec('200','c') );
-
-which might not be what you have in mind. If you need repeatable subfield,
-define it using C<marc_repeatable_subfield> like this:
-
- marc_repeatable_subfield('900','a');
- marc('900','a', rec('200','a') );
- marc('900','b', rec('200','b') );
- marc('900','c', rec('200','c') );
-
-will create:
-
- 900a-1 900a-2 900a-3 900b-1 900c-1
- 900b-2
-
-There is also support for returning next or specific using:
-
- while (my $mf = WebPAC::Normalize:_get_marc_fields( fetch_next => 1 ) ) {
- # do something with $mf
- }
-
-will always return fields from next MARC record or
-
- my $mf = WebPAC::Normalize::_get_marc_fields( offset => 42 );
-
-will return 42th copy record (if it exists).
-
-=cut
-
-my $fetch_pos;
-
-sub _get_marc_fields {
-
- my $arg = {@_};
- warn "### _get_marc_fields arg: ", dump($arg), $/ if ($debug > 2);
- $fetch_pos = $marc_fetch_offset;
- if ($arg->{offset}) {
- $fetch_pos = $arg->{offset};
- } elsif($arg->{fetch_next}) {
- $marc_fetch_offset++;
- }
-
- return if (! $marc_record || ref($marc_record) ne 'ARRAY');
-
- warn "### full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 2);
-
- my $marc_rec = $marc_record->[ $fetch_pos ];
-
- warn "## _get_marc_fields (at offset: $fetch_pos) -- marc_record = ", dump( @$marc_rec ), $/ if ($debug > 1);
-
- return if (! $marc_rec || ref($marc_rec) ne 'ARRAY' || $#{ $marc_rec } < 0);
-
- # first, sort all existing fields
- # XXX might not be needed, but modern perl might randomize elements in hash
- my @sorted_marc_record = sort {
- $a->[0] . ( $a->[3] || '' ) cmp $b->[0] . ( $b->[3] || '')
- } @{ $marc_rec };
-
- @sorted_marc_record = @{ $marc_rec }; ### FIXME disable sorting
-
- # output marc fields
- my @m;
-
- # count unique field-subfields (used for offset when walking to next subfield)
- my $u;
- map { $u->{ $_->[0] . ( $_->[3] || '') }++ } @sorted_marc_record;
-
- if ($debug) {
- warn "## marc_repeatable_subfield = ", dump( $marc_repeatable_subfield ), $/ if ( $marc_repeatable_subfield );
- warn "## marc_record[$fetch_pos] = ", dump( $marc_rec ), $/;
- warn "## sorted_marc_record = ", dump( \@sorted_marc_record ), $/;
- warn "## subfield count = ", dump( $u ), $/;
- }
-
- my $len = $#sorted_marc_record;
- my $visited;
- my $i = 0;
- my $field;
-
- foreach ( 0 .. $len ) {
-
- # find next element which isn't visited
- while ($visited->{$i}) {
- $i = ($i + 1) % ($len + 1);
- }
-
- # mark it visited
- $visited->{$i}++;
-
- my $row = dclone( $sorted_marc_record[$i] );
-
- # field and subfield which is key for
- # marc_repeatable_subfield and u
- my $fsf = $row->[0] . ( $row->[3] || '' );
-
- if ($debug > 1) {
-
- print "### field so far [", $#$field, "] : ", dump( $field ), " ", $field ? 'T' : 'F', $/;
- print "### this [$i]: ", dump( $row ),$/;
- print "### sf: ", $row->[3], " vs ", $field->[3],
- $marc_repeatable_subfield->{ $row->[0] . $row->[3] } ? ' (repeatable)' : '', $/,
- if ($#$field >= 0);
-
- }
-
- # if field exists
- if ( $#$field >= 0 ) {
- if (
- $row->[0] ne $field->[0] || # field
- $row->[1] ne $field->[1] || # i1
- $row->[2] ne $field->[2] # i2
- ) {
- push @m, $field;
- warn "## saved/1 ", dump( $field ),$/ if ($debug);
- $field = $row;
-
- } elsif (
- ( $row->[3] lt $field->[-2] ) # subfield which is not next (e.g. a after c)
- ||
- ( $row->[3] eq $field->[-2] && # same subfield, but not repeatable
- ! $marc_repeatable_subfield->{ $fsf }
- )
- ) {
- push @m, $field;
- warn "## saved/2 ", dump( $field ),$/ if ($debug);
- $field = $row;
-
- } else {
- # append new subfields to existing field
- push @$field, ( $row->[3], $row->[4] );
- }
- } else {
- # insert first field
- $field = $row;
- }
-
- if (! $marc_repeatable_subfield->{ $fsf }) {
- # make step to next subfield
- $i = ($i + $u->{ $fsf } ) % ($len + 1);
- }
- }
-
- if ($#$field >= 0) {
- push @m, $field;
- warn "## saved/3 ", dump( $field ),$/ if ($debug);
- }
-
- return \@m;
-}
-
-=head2 _get_marc_leader
-
-Return leader from currently fetched record by L</_get_marc_fields>
-
- print WebPAC::Normalize::_get_marc_leader();
-
-=cut
-
-sub _get_marc_leader {
- die "no fetch_pos, did you called _get_marc_fields first?" unless ( defined( $fetch_pos ) );
- return $marc_leader->[ $fetch_pos ];
-}
-
=head2 _debug
Change level of debug warnings
return $debug unless defined($l);
warn "debug level $l",$/ if ($l > 0);
$debug = $l;
+ $WebPAC::Normalize::MARC::debug = $debug;
}
=head1 Functions to create C<data_structure>
sub sorted { _field( 'sorted', @_ ) }
-=head2 marc_leader
-
-Setup fields within MARC leader or get leader
-
- marc_leader('05','c');
- my $leader = marc_leader();
-
-=cut
-
-sub marc_leader {
- my ($offset,$value) = @_;
-
- if ($offset) {
- $marc_leader->[ $marc_record_offset ]->{ $offset } = $value;
- } else {
-
- if (defined($marc_leader)) {
- die "marc_leader not array = ", dump( $marc_leader ) unless (ref($marc_leader) eq 'ARRAY');
- return $marc_leader->[ $marc_record_offset ];
- } else {
- return;
- }
- }
-}
-
-=head2 marc_fixed
-
-Create control/indentifier fields with values in fixed positions
-
- marc_fixed('008', 00, '070402');
- marc_fixed('008', 39, '|');
-
-Positions not specified will be filled with spaces (C<0x20>).
-
-There will be no effort to extend last specified value to full length of
-field in standard.
-
-=cut
-
-sub marc_fixed {
- my ($f, $pos, $val) = @_;
- die "need marc(field, position, value)" unless defined($f) && defined($pos);
-
- confess "need val" unless defined $val;
-
- my $update = 0;
-
- map {
- if ($_->[0] eq $f) {
- my $old = $_->[1];
- if (length($old) <= $pos) {
- $_->[1] .= ' ' x ( $pos - length($old) ) . $val;
- warn "## marc_fixed($f,$pos,'$val') append '$old' -> '$_->[1]'\n" if ($debug > 1);
- } else {
- $_->[1] = substr($old, 0, $pos) . $val . substr($old, $pos + length($val));
- warn "## marc_fixed($f,$pos,'$val') update '$old' -> '$_->[1]'\n" if ($debug > 1);
- }
- $update++;
- }
- } @{ $marc_record->[ $marc_record_offset ] };
-
- if (! $update) {
- my $v = ' ' x $pos . $val;
- push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $v ];
- warn "## marc_fixed($f,$pos,'val') created '$v'\n" if ($debug > 1);
- }
-}
-
-=head2 marc
-
-Save value for MARC field
-
- marc('900','a', rec('200','a') );
- marc('001', rec('000') );
-
-=cut
-
-sub marc {
- my $f = shift or die "marc needs field";
- die "marc field must be numer" unless ($f =~ /^\d+$/);
-
- my $sf;
- if ($f >= 10) {
- $sf = shift or die "marc needs subfield";
- }
-
- foreach (@_) {
- my $v = $_; # make var read-write for Encode
- next unless (defined($v) && $v !~ /^\s*$/);
- my ($i1,$i2) = defined($marc_indicators->{$f}) ? @{ $marc_indicators->{$f} } : (' ',' ');
- if (defined $sf) {
- push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $i1, $i2, $sf => $v ];
- } else {
- push @{ $marc_record->[ $marc_record_offset ] }, [ $f, $v ];
- }
- }
-}
-
-=head2 marc_repeatable_subfield
-
-Save values for MARC repetable subfield
-
- marc_repeatable_subfield('910', 'z', rec('909') );
-
-=cut
-
-sub marc_repeatable_subfield {
- my ($f,$sf) = @_;
- die "marc_repeatable_subfield need field and subfield!\n" unless ($f && $sf);
- $marc_repeatable_subfield->{ $f . $sf }++;
- marc(@_);
-}
-
-=head2 marc_indicators
-
-Set both indicators for MARC field
-
- marc_indicators('900', ' ', 1);
-
-Any indicator value other than C<0-9> will be treated as undefined.
-
-=cut
-
-sub marc_indicators {
- my $f = shift || die "marc_indicators need field!\n";
- my ($i1,$i2) = @_;
- die "marc_indicators($f, ...) need i1!\n" unless(defined($i1));
- die "marc_indicators($f, $i1, ...) need i2!\n" unless(defined($i2));
-
- $i1 = ' ' if ($i1 !~ /^\d$/);
- $i2 = ' ' if ($i2 !~ /^\d$/);
- @{ $marc_indicators->{$f} } = ($i1,$i2);
-}
-
-=head2 marc_compose
-
-Save values for each MARC subfield explicitly
-
- marc_compose('900',
- 'a', rec('200','a')
- 'b', rec('201','a')
- 'a', rec('200','b')
- 'c', rec('200','c')
- );
-
-If you specify C<+> for subfield, value will be appended
-to previous defined subfield.
-
-=cut
-
-sub marc_compose {
- my $f = shift or die "marc_compose needs field";
- die "marc_compose field must be numer" unless ($f =~ /^\d+$/);
-
- my ($i1,$i2) = defined($marc_indicators->{$f}) ? @{ $marc_indicators->{$f} } : (' ',' ');
- my $m = [ $f, $i1, $i2 ];
-
- warn "### marc_compose input subfields = ", dump(@_),$/ if ($debug > 2);
-
- if ($#_ % 2 != 1) {
- die "ERROR: marc_compose",dump($f,@_)," not valid (must be even).\nDo you need to add first() or join() around some argument?\n";
- }
-
- while (@_) {
- my $sf = shift;
- my $v = shift;
-
- next unless (defined($v) && $v !~ /^\s*$/);
- warn "## ++ marc_compose($f,$sf,$v) ", dump( $m ),$/ if ($debug > 1);
- if ($sf ne '+') {
- push @$m, ( $sf, $v );
- } else {
- $m->[ $#$m ] .= $v;
- }
- }
-
- warn "## marc_compose current marc = ", dump( $m ),$/ if ($debug > 1);
-
- push @{ $marc_record->[ $marc_record_offset ] }, $m if ($#{$m} > 2);
-}
-
-=head2 marc_duplicate
-
-Generate copy of current MARC record and continue working on copy
-
- marc_duplicate();
-
-Copies can be accessed using C<< _get_marc_fields( fetch_next => 1 ) >> or
-C<< _get_marc_fields( offset => 42 ) >>.
-
-=cut
-
-sub marc_duplicate {
- my $m = $marc_record->[ -1 ];
- die "can't duplicate record which isn't defined" unless ($m);
- push @{ $marc_record }, dclone( $m );
- push @{ $marc_leader }, dclone( marc_leader() );
- warn "## marc_duplicate = ", dump(@$marc_leader, @$marc_record), $/ if ($debug > 1);
- $marc_record_offset = $#{ $marc_record };
- warn "## marc_record_offset = $marc_record_offset", $/ if ($debug > 1);
-
-}
-
-=head2 marc_remove
-
-Remove some field or subfield from MARC record.
-
- marc_remove('200');
- marc_remove('200','a');
-
-This will erase field C<200> or C<200^a> from current MARC record.
-
- marc_remove('*');
-
-Will remove all fields in current MARC record.
-
-This is useful after calling C<marc_duplicate> or on it's own (but, you
-should probably just remove that subfield definition if you are not
-using C<marc_duplicate>).
-
-FIXME: support fields < 10.
-
-=cut
-
-sub marc_remove {
- my ($f, $sf) = @_;
-
- die "marc_remove needs record number" unless defined($f);
-
- my $marc = $marc_record->[ $marc_record_offset ];
-
- warn "### marc_remove before = ", dump( $marc ), $/ if ($debug > 2);
-
- if ($f eq '*') {
-
- delete( $marc_record->[ $marc_record_offset ] );
- warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1);
-
- } else {
-
- my $i = 0;
- foreach ( 0 .. $#{ $marc } ) {
- last unless (defined $marc->[$i]);
- warn "#### working on ",dump( @{ $marc->[$i] }), $/ if ($debug > 3);
- if ($marc->[$i]->[0] eq $f) {
- if (! defined $sf) {
- # remove whole field
- splice @$marc, $i, 1;
- warn "#### slice \@\$marc, $i, 1 = ",dump( @{ $marc }), $/ if ($debug > 3);
- $i--;
- } else {
- foreach my $j ( 0 .. (( $#{ $marc->[$i] } - 3 ) / 2) ) {
- my $o = ($j * 2) + 3;
- if ($marc->[$i]->[$o] eq $sf) {
- # remove subfield
- splice @{$marc->[$i]}, $o, 2;
- warn "#### slice \@{\$marc->[$i]}, $o, 2 = ", dump( @{ $marc }), $/ if ($debug > 3);
- # is record now empty?
- if ($#{ $marc->[$i] } == 2) {
- splice @$marc, $i, 1;
- warn "#### slice \@\$marc, $i, 1 = ", dump( @{ $marc }), $/ if ($debug > 3);
- $i--;
- };
- }
- }
- }
- }
- $i++;
- }
-
- warn "### marc_remove($f", $sf ? ",$sf" : "", ") after = ", dump( $marc ), $/ if ($debug > 2);
-
- $marc_record->[ $marc_record_offset ] = $marc;
- }
-
- warn "## full marc_record = ", dump( @{ $marc_record }), $/ if ($debug > 1);
-}
-
-=head2 marc_original_order
-
-Copy all subfields preserving original order to marc field.
-
- marc_original_order( marc_field_number, original_input_field_number );
-
-Please note that field numbers are consistent with other commands (marc
-field number first), but somewhat counter-intuitive (destination and then
-source).
-
-You might want to use this command if you are just renaming subfields or
-using pre-processing modify_record in C<config.yml> and don't need any
-post-processing or want to preserve order of original subfields.
-
-
-=cut
-
-sub marc_original_order {
-
- my ($to, $from) = @_;
- die "marc_original_order needs from and to fields\n" unless ($from && $to);
-
- return unless defined($rec->{$from});
-
- my $r = $rec->{$from};
- die "record field $from isn't array ",dump( $rec ) unless (ref($r) eq 'ARRAY');
-
- my ($i1,$i2) = defined($marc_indicators->{$to}) ? @{ $marc_indicators->{$to} } : (' ',' ');
- warn "## marc_original_order($to,$from) source = ", dump( $r ),$/ if ($debug > 1);
-
- foreach my $d (@$r) {
-
- if (! defined($d->{subfields}) && ref($d->{subfields}) ne 'ARRAY') {
- warn "# marc_original_order($to,$from): field $from doesn't have subfields specification\n";
- next;
- }
-
- my @sfs = @{ $d->{subfields} };
-
- die "field $from doesn't have even number of subfields specifications\n" unless($#sfs % 2 == 1);
-
- warn "#--> d: ",dump($d), "\n#--> sfs: ",dump(@sfs),$/ if ($debug > 2);
-
- my $m = [ $to, $i1, $i2 ];
-
- while (my $sf = shift @sfs) {
-
- warn "#--> sf: ",dump($sf), $/ if ($debug > 2);
- my $offset = shift @sfs;
- die "corrupted sufields specification for field $from\n" unless defined($offset);
-
- my $v;
- if (ref($d->{$sf}) eq 'ARRAY') {
- $v = $d->{$sf}->[$offset] if (defined($d->{$sf}->[$offset]));
- } elsif ($offset == 0) {
- $v = $d->{$sf};
- } else {
- die "field $from subfield '$sf' need occurence $offset which doesn't exist", dump($d->{$sf});
- }
- push @$m, ( $sf, $v ) if (defined($v));
- }
-
- if ($#{$m} > 2) {
- push @{ $marc_record->[ $marc_record_offset ] }, $m;
- }
- }
-
- warn "## marc_record = ", dump( $marc_record ),$/ if ($debug > 1);
-}
-
-
-=head2 marc_count
-
-Return number of MARC records created using L</marc_duplicate>.
-
- print "created ", marc_count(), " records";
-
-=cut
-
-sub marc_count {
- return $#{ $marc_record };
-}
-
-=head2 _marc_push
-
- _marc_push( $marc );
-
-=cut
-
-sub _marc_push {
- my $marc = shift || die "no marc?";
- push @{ $marc_record->[ $marc_record_offset ] }, $marc;
-}
-
=head1 Functions to extract data from input
sub id {
my $mfn = $config->{_mfn} || die "no _mfn in config data";
- return $mfn . $#{$marc_record} ? $#{$marc_record} + 1 : '';
+ return $mfn . ( WebPAC::Normalize::MARC::_created_marc_records() || '' );
}
=head2 join_with
projects / webpac2 / blobdiff