1 package WebPAC::Normalize::MARC;
10 use Data::Dump qw/dump/;
13 use WebPAC::Normalize;
19 WebPAC::Normalize::MARC - create MARC/ISO2709 records
28 from => 225, to => 440,
42 'a, |x ; |v. |n, |p ; |v',
47 Returns number of records produced.
53 warn "## marc_template(",dump($args),")",$/ if $debug;
55 foreach ( qw/subfields_rename isis_template marc_template/ ) {
56 # warn "ref($_) = ",ref($args->{$_}) if $debug;
57 die "$_ not ARRAY" if defined($args->{$_}) && ref($args->{$_}) ne 'ARRAY';
60 die "marc_template needs isis_template or marc_template"
61 if ! defined $args->{isis_template} && ! defined $args->{marc_template};
63 my $rec = WebPAC::Normalize::_get_rec() || die "_get_rec?";
64 my $r = $rec->{ $args->{from} } || return;
65 die "record field ", $args->{from}, " isn't array ",dump( $rec ) unless (ref($r) eq 'ARRAY');
67 my @subfields_rename = @{ $args->{subfields_rename} };
68 # warn "### subfields_rename [$#subfields_rename] = ",dump( @subfields_rename ) if $debug;
70 confess "need mapping in pairs for subfields_rename"
71 if $#subfields_rename % 2 != 1;
73 my ( $subfields_rename, $from_subfields, $to_subfields );
74 while ( my ( $from, $to ) = splice(@subfields_rename, 0, 2) ) {
76 $from_subfields->{ $from }++,
77 $to_subfields->{ $to }++
79 $subfields_rename->{ $from }->[ $f ] = [ $to => $t ];
81 warn "### subfields_rename = ",dump( $subfields_rename ),$/ if $debug;
82 warn "### from_subfields = ", dump( $from_subfields ),$/ if $debug;
83 warn "### to_subfields = ", dump( $to_subfields ),$/ if $debug;
87 $_template->{isis}->{fields_re} = join('|', keys %$from_subfields );
88 $_template->{marc}->{fields_re} = join('|', keys %$to_subfields );
93 my ( $name, $templates ) = @_;
95 my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
97 foreach my $template ( @{ $templates } ) {
102 my $nr = $count->{$sf}++;
103 push @order, [ $sf, $nr ];
106 my $pos_template = $template;
107 $pos_template =~ s/($fields_re)/my_count($1)/ge;
108 my $count_key = dump( $count );
109 warn "### template: |$template| -> |$pos_template| count = $count_key order = ",dump( @order ),$/ if $debug;
110 $_template->{$name}->{pos}->{ $count_key } = $pos_template;
111 $_template->{$name}->{order}->{ $pos_template } = [ @order ];
113 warn "### from ",dump( $templates ), " using $fields_re created ", dump( $_template ),$/ if $debug;
116 _parse_template( 'marc', $args->{marc_template} );
117 _parse_template( 'isis', $args->{isis_template} );
118 warn "### _template = ",dump( $_template ),$/ if $debug;
122 foreach my $r ( @{ $rec->{ $args->{from} } } ) {
124 my $i1 = $r->{i1} || ' ';
125 my $i2 = $r->{i2} || ' ';
126 $m = [ $args->{to}, $i1, $i2 ];
128 warn "### r = ",dump( $r ),$/ if $debug;
130 my ( $from_mapping, $to_mapping, $from_count, $to_count );
131 foreach my $sf ( keys %{$r} ) {
132 # skip everything which isn't one char subfield (e.g. 'subfields')
133 next unless $sf =~ m/^\w$/;
134 my $nr = $from_count->{$sf}++;
135 my $rename_to = $subfields_rename->{ $sf } ||
136 die "can't find subfield rename for $sf/$nr in ", dump( $subfields_rename );
137 warn "### rename $sf/$nr to ", dump( $rename_to->[$nr] ),$/ if $debug;
138 my ( $to_sf, $to_nr ) = @{ $rename_to->[$nr] };
139 $from_mapping->{ $sf }->[ $nr ] = [ $to_sf => $to_nr ];
140 $to_mapping->{ $to_sf }->[ $to_nr ] = [ $sf => $nr ];
142 $to_count->{ $to_sf }++;
145 warn "### to_mapping = ",dump( $to_mapping ),$/ if $debug;
148 from => dump( $from_count ),
149 to => dump( $to_count),
152 warn "### count_key = ",dump( $count_key ),$/ if $debug;
154 my $processed_templates = 0;
156 # this defines order of traversal
157 foreach ( qw/isis:from marc:to/ ) {
158 my ($name,$count_name) = split(/:/);
160 my $ckey = $count_key->{$count_name} || die "can't find count_key $count_name in ",dump( $count_key );
162 my $template = $_template->{$name}->{pos}->{ $ckey } || next;
163 $processed_templates++;
165 warn "### traverse $name $count_name selected template: |$template|\n",$/ if $debug;
169 my @templates = split(/\|/, $template );
170 @templates = ( $template );
172 foreach my $sf ( @templates ) {
174 my ( $name, $r, $sf, $nr ) = @_;
175 my ( $from_sf, $from_nr );
176 if ( $name eq 'marc' ) {
177 ( $from_sf, $from_nr ) = @{ $to_mapping->{$sf}->[$nr] };
179 ( $from_sf, $from_nr ) = ( $sf, $nr );
181 my $v = $r->{ $from_sf }; # || die "no $from_sf/$from_nr";
182 warn "#### fill_in( $sf, $nr ) = $from_sf/$from_nr >>>> ",dump( $v ),$/ if $debug;
183 if ( ref( $v ) eq 'ARRAY' ) {
184 $fill_in->{$sf}->[$nr] = $v->[$from_nr];
185 return $v->[$from_nr];
186 } elsif ( $from_nr == 0 ) {
187 $fill_in->{$sf}->[$nr] = $v;
190 die "requested subfield $from_sf/$from_nr but it's ",dump( $v );
193 my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
194 warn "#### $sf <<<< $fields_re\n",$/ if $debug;
195 $sf =~ s/($fields_re)(\d+)/fill_in($name,$r,$1,$2)/ge;
196 warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/ if $debug;
199 warn "## template: |$template|\n## _template->$name = ",dump( $_template->{$name} ),$/ if $debug;
201 foreach my $sf ( @{ $_template->{$name}->{order}->{$template} } ) {
202 my ( $sf, $nr ) = @$sf;
203 my $v = $fill_in->{$sf}->[$nr] || die "can't find fill_in $sf/$nr";
204 if ( $name eq 'isis') {
205 ( $sf, $nr ) = @{ $from_mapping->{$sf}->[$nr] };
207 warn "++ $sf/$nr |$v|\n" if $debug;
208 push @$m, ( $sf, $v );
211 warn "#### >>>> created MARC record: ", dump( $m ),$/ if $debug;
216 die "I don't have template for fields ",dump( $count_key ), "\n## available templates\n", dump( $_template ) unless $processed_templates;
217 warn ">>> $processed_templates templates applied to data\n",$/ if $debug;
223 foreach my $marc ( @marc_out ) {
224 warn "+++ ",dump( $marc ),$/ if $debug;
225 WebPAC::Normalize::_marc_push( $marc );
229 warn "### marc_template produced $recs MARC records: ",dump( @marc_out ),$/ if $debug;