| 1 |
dpavlin |
1021 |
package WebPAC::Normalize::MARC; |
| 2 |
|
|
use Exporter 'import'; |
| 3 |
|
|
@EXPORT = qw/ |
| 4 |
|
|
marc_template |
| 5 |
|
|
/; |
| 6 |
|
|
|
| 7 |
|
|
use strict; |
| 8 |
|
|
use warnings; |
| 9 |
|
|
|
| 10 |
|
|
use Data::Dump qw/dump/; |
| 11 |
|
|
use Carp qw/confess/; |
| 12 |
|
|
|
| 13 |
|
|
use WebPAC::Normalize; |
| 14 |
|
|
|
| 15 |
|
|
=head1 NAME |
| 16 |
|
|
|
| 17 |
|
|
WebPAC::Normalize::MARC - create MARC/ISO2709 records |
| 18 |
|
|
|
| 19 |
|
|
=cut |
| 20 |
|
|
|
| 21 |
|
|
=head1 FUNCTIONS |
| 22 |
|
|
|
| 23 |
|
|
=head2 marc_template |
| 24 |
|
|
|
| 25 |
|
|
marc_template( |
| 26 |
|
|
from => 225, to => 440, |
| 27 |
|
|
subfields_rename => [ |
| 28 |
|
|
'a' => 'a', |
| 29 |
|
|
'x' => 'x', |
| 30 |
|
|
'v' => 'v', |
| 31 |
|
|
'h' => 'n', |
| 32 |
|
|
'i' => 'p', |
| 33 |
|
|
'w' => 'v', |
| 34 |
|
|
], |
| 35 |
|
|
isis_template => [ |
| 36 |
|
|
'a ; |v. |i', |
| 37 |
|
|
'a. |i ; |w', |
| 38 |
|
|
], |
| 39 |
|
|
marc_template => [ |
| 40 |
|
|
'a, |x ; |v. |n, |p ; |v', |
| 41 |
|
|
'a ; |v. |p ; |v', |
| 42 |
|
|
], |
| 43 |
|
|
); |
| 44 |
|
|
|
| 45 |
|
|
=cut |
| 46 |
|
|
|
| 47 |
|
|
sub marc_template { |
| 48 |
|
|
my $args = {@_}; |
| 49 |
|
|
warn "## marc_template(",dump($args),")"; |
| 50 |
|
|
|
| 51 |
|
|
foreach ( qw/subfields_rename isis_template marc_template/ ) { |
| 52 |
|
|
# warn "ref($_) = ",ref($args->{$_}); |
| 53 |
|
|
die "$_ not ARRAY" if defined($args->{$_}) && ref($args->{$_}) ne 'ARRAY'; |
| 54 |
|
|
} |
| 55 |
|
|
|
| 56 |
|
|
die "marc_template needs isis_template or marc_template" |
| 57 |
|
|
if ! defined $args->{isis_template} && ! defined $args->{marc_template}; |
| 58 |
|
|
|
| 59 |
|
|
my $rec = WebPAC::Normalize::_get_rec() || die "_get_rec?"; |
| 60 |
|
|
my $r = $rec->{ $args->{from} } || return; |
| 61 |
|
|
die "record field ", $args->{from}, " isn't array ",dump( $rec ) unless (ref($r) eq 'ARRAY'); |
| 62 |
|
|
|
| 63 |
|
|
my @subfields_rename = @{ $args->{subfields_rename} }; |
| 64 |
|
|
# warn "### subfields_rename [$#subfields_rename] = ",dump( @subfields_rename ); |
| 65 |
|
|
|
| 66 |
|
|
confess "need mapping in pairs for subfields_rename" |
| 67 |
|
|
if $#subfields_rename % 2 != 1; |
| 68 |
|
|
|
| 69 |
|
|
my ( $subfields_rename, $from_subfields, $to_subfields ); |
| 70 |
|
|
while ( my ( $from, $to ) = splice(@subfields_rename, 0, 2) ) { |
| 71 |
|
|
my ( $f, $t ) = ( |
| 72 |
|
|
$from_subfields->{ $from }++, |
| 73 |
|
|
$to_subfields->{ $to }++ |
| 74 |
|
|
); |
| 75 |
|
|
$subfields_rename->{ $from }->[ $f ] = [ $to => $t ]; |
| 76 |
|
|
} |
| 77 |
|
|
warn "### subfields_rename = ",dump( $subfields_rename ),$/; |
| 78 |
|
|
warn "### from_subfields = ", dump( $from_subfields ),$/; |
| 79 |
|
|
warn "### to_subfields = ", dump( $to_subfields ),$/; |
| 80 |
|
|
|
| 81 |
dpavlin |
1025 |
our $_template; |
| 82 |
dpavlin |
1021 |
|
| 83 |
dpavlin |
1025 |
$_template->{fields_re} = { |
| 84 |
|
|
isis => join('|', keys %$from_subfields ), |
| 85 |
|
|
marc => join('|', keys %$to_subfields ), |
| 86 |
|
|
}; |
| 87 |
|
|
|
| 88 |
dpavlin |
1021 |
my @marc_out; |
| 89 |
|
|
|
| 90 |
dpavlin |
1023 |
sub _parse_template { |
| 91 |
dpavlin |
1024 |
my ( $name, $templates ) = @_; |
| 92 |
dpavlin |
1021 |
|
| 93 |
dpavlin |
1025 |
my $fields_re = $_template->{fields_re}->{ $name } || die "can't find $name in ",dump( $_template->{fields_re} ); |
| 94 |
|
|
|
| 95 |
dpavlin |
1023 |
foreach my $template ( @{ $templates } ) { |
| 96 |
|
|
our $count = {}; |
| 97 |
dpavlin |
1025 |
our @order = (); |
| 98 |
dpavlin |
1023 |
sub my_count { |
| 99 |
|
|
my $sf = shift; |
| 100 |
|
|
my $nr = $count->{$sf}++; |
| 101 |
dpavlin |
1025 |
push @order, [ $sf, $nr ]; |
| 102 |
dpavlin |
1023 |
return $sf . $nr; |
| 103 |
|
|
} |
| 104 |
|
|
my $pos_template = $template; |
| 105 |
|
|
$pos_template =~ s/($fields_re)/my_count($1)/ge; |
| 106 |
|
|
my $count_key = dump( $count ); |
| 107 |
dpavlin |
1025 |
warn "### template: |$template| -> |$pos_template| count = $count_key order = ",dump( @order ),$/; |
| 108 |
dpavlin |
1024 |
$_template->{$name}->{pos}->{ $count_key } = $pos_template; |
| 109 |
dpavlin |
1025 |
$_template->{$name}->{order}->{ $pos_template } = [ @order ]; |
| 110 |
dpavlin |
1021 |
} |
| 111 |
dpavlin |
1025 |
warn "### from ",dump( $templates ), " using $fields_re created ", dump( $_template ); |
| 112 |
dpavlin |
1021 |
} |
| 113 |
|
|
|
| 114 |
dpavlin |
1024 |
_parse_template( 'marc', $args->{marc_template} ); |
| 115 |
|
|
_parse_template( 'isis', $args->{isis_template} ); |
| 116 |
|
|
warn "### _template = ",dump( $_template ); |
| 117 |
dpavlin |
1023 |
|
| 118 |
dpavlin |
1021 |
my $m; |
| 119 |
|
|
|
| 120 |
|
|
foreach my $r ( @{ $rec->{ $args->{from} } } ) { |
| 121 |
|
|
|
| 122 |
|
|
my $i1 = $r->{i1} || ' '; |
| 123 |
|
|
my $i2 = $r->{i2} || ' '; |
| 124 |
|
|
$m = [ $args->{to}, $i1, $i2 ]; |
| 125 |
|
|
|
| 126 |
|
|
warn "### r = ",dump( $r ); |
| 127 |
|
|
|
| 128 |
|
|
my ( $new_r, $from_count, $to_count ); |
| 129 |
|
|
foreach my $sf ( keys %{$r} ) { |
| 130 |
|
|
# skip everything which isn't one char subfield (e.g. 'subfields') |
| 131 |
|
|
next unless $sf =~ m/^\w$/; |
| 132 |
|
|
my $nr = $from_count->{$sf}++; |
| 133 |
|
|
my $rename_to = $subfields_rename->{ $sf } || |
| 134 |
|
|
die "can't find subfield rename for $sf/$nr in ", dump( $subfields_rename ); |
| 135 |
|
|
warn "### rename $sf/$nr to ", dump( $rename_to->[$nr] ), $/; |
| 136 |
|
|
my ( $to_sf, $to_nr ) = @{ $rename_to->[$nr] }; |
| 137 |
|
|
$new_r->{ $to_sf }->[ $to_nr ] = [ $sf => $nr ]; |
| 138 |
|
|
|
| 139 |
|
|
$to_count->{ $to_sf }++; |
| 140 |
|
|
} |
| 141 |
|
|
|
| 142 |
|
|
warn "### new_r = ",dump( $new_r ); |
| 143 |
|
|
|
| 144 |
dpavlin |
1025 |
my $count_key = { |
| 145 |
|
|
from => dump( $from_count ), |
| 146 |
|
|
to => dump( $to_count), |
| 147 |
|
|
}; |
| 148 |
dpavlin |
1021 |
|
| 149 |
dpavlin |
1025 |
warn "### count_key = ",dump( $count_key ), $/; |
| 150 |
dpavlin |
1021 |
|
| 151 |
dpavlin |
1025 |
my $processed_templates = 0; |
| 152 |
dpavlin |
1021 |
|
| 153 |
dpavlin |
1025 |
# this defines order of traversal |
| 154 |
|
|
foreach ( qw/isis:from marc:to/ ) { |
| 155 |
|
|
my ($name,$count_name) = split(/:/); |
| 156 |
|
|
warn "## traverse $name $count_name\n"; |
| 157 |
dpavlin |
1021 |
|
| 158 |
dpavlin |
1025 |
my $ckey = $count_key->{$count_name} || die "can't find count_key $count_name in ",dump( $count_key ); |
| 159 |
dpavlin |
1021 |
|
| 160 |
dpavlin |
1025 |
my $template = $_template->{$name}->{pos}->{ $ckey } || next; |
| 161 |
|
|
$processed_templates++; |
| 162 |
dpavlin |
1021 |
|
| 163 |
dpavlin |
1025 |
warn "### selected template: |$template|\n"; |
| 164 |
|
|
|
| 165 |
|
|
our $fill_in = {}; |
| 166 |
|
|
|
| 167 |
|
|
my @templates = split(/\|/, $template ); |
| 168 |
|
|
@templates = ( $template ); |
| 169 |
|
|
|
| 170 |
|
|
foreach my $sf ( @templates ) { |
| 171 |
|
|
sub fill_in { |
| 172 |
|
|
my ( $name, $r, $sf, $nr ) = @_; |
| 173 |
|
|
my ( $from_sf, $from_nr, $v ); |
| 174 |
|
|
if ( $name eq 'marc' ) { |
| 175 |
|
|
( $from_sf, $from_nr ) = @{ $new_r->{$sf}->[$nr] }; |
| 176 |
|
|
} else { |
| 177 |
|
|
( $from_sf, $from_nr ) = ( $sf, $nr ); |
| 178 |
|
|
} |
| 179 |
|
|
my $v = $r->{ $from_sf }; # || die "no $from_sf/$from_nr"; |
| 180 |
|
|
warn "#### fill_in( $sf, $nr ) = $from_sf/$from_nr >>>> ",dump( $v ), $/; |
| 181 |
|
|
if ( ref( $v ) eq 'ARRAY' ) { |
| 182 |
|
|
$fill_in->{$sf}->[$nr] = $v->[$from_nr]; |
| 183 |
|
|
return $v->[$from_nr]; |
| 184 |
|
|
} elsif ( $from_nr == 0 ) { |
| 185 |
|
|
$fill_in->{$sf}->[$nr] = $v; |
| 186 |
|
|
return $v; |
| 187 |
|
|
} else { |
| 188 |
|
|
die "requested subfield $from_sf/$from_nr but it's ",dump( $v ); |
| 189 |
|
|
} |
| 190 |
dpavlin |
1021 |
} |
| 191 |
dpavlin |
1025 |
my $fields_re = $_template->{fields_re}->{ $name } || die "can't find $name in ",dump( $_template->{fields_re} ); |
| 192 |
|
|
warn "#### $sf <<<< $fields_re\n"; |
| 193 |
|
|
$sf =~ s/($fields_re)(\d+)/fill_in($name,$r,$1,$2)/ge; |
| 194 |
|
|
warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/; |
| 195 |
dpavlin |
1021 |
} |
| 196 |
|
|
|
| 197 |
dpavlin |
1025 |
warn "## template: |$template|\n## _template->$name = ",dump( $_template->{$name} ); |
| 198 |
dpavlin |
1021 |
|
| 199 |
dpavlin |
1025 |
$sf_pos = $#m; |
| 200 |
dpavlin |
1021 |
|
| 201 |
dpavlin |
1025 |
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 |
|
|
warn "++ $sf/$nr |$v|\n"; |
| 205 |
|
|
push @$m, ( $sf, $v ); |
| 206 |
|
|
} |
| 207 |
dpavlin |
1021 |
|
| 208 |
dpavlin |
1025 |
warn "#### >>>> created MARC record: ", dump( $m ); |
| 209 |
|
|
|
| 210 |
|
|
push @marc_out, $m; |
| 211 |
|
|
} |
| 212 |
|
|
|
| 213 |
|
|
die "I don't have template for fields ",dump( $count_key ), "\n## available templates\n", dump( $_template ) unless $processed_templates; |
| 214 |
|
|
warn ">>> $processed_templates templates applied to data\n"; |
| 215 |
dpavlin |
1021 |
} |
| 216 |
|
|
|
| 217 |
|
|
|
| 218 |
dpavlin |
1025 |
warn "### marc_template produced following MARC records: ",dump( @marc_out ); |
| 219 |
|
|
|
| 220 |
dpavlin |
1021 |
foreach my $marc ( @marc_out ) { |
| 221 |
|
|
warn "+++ ",dump( $marc ); |
| 222 |
|
|
WebPAC::Normalize::_marc_push( $marc ); |
| 223 |
|
|
} |
| 224 |
|
|
} |
| 225 |
|
|
|
| 226 |
|
|
1; |
Parent Directory
| 
Revision Log