use WebPAC::Normalize;
+my $debug = 1;
+
=head1 NAME
WebPAC::Normalize::MARC - create MARC/ISO2709 records
],
);
+Returns number of records produced.
+
=cut
sub marc_template {
my $args = {@_};
- warn "## marc_template(",dump($args),")";
+ warn "## marc_template(",dump($args),")",$/ if $debug;
foreach ( qw/subfields_rename isis_template marc_template/ ) {
-# warn "ref($_) = ",ref($args->{$_});
+# warn "ref($_) = ",ref($args->{$_}) if $debug;
die "$_ not ARRAY" if defined($args->{$_}) && ref($args->{$_}) ne 'ARRAY';
}
die "record field ", $args->{from}, " isn't array ",dump( $rec ) unless (ref($r) eq 'ARRAY');
my @subfields_rename = @{ $args->{subfields_rename} };
-# warn "### subfields_rename [$#subfields_rename] = ",dump( @subfields_rename );
+# warn "### subfields_rename [$#subfields_rename] = ",dump( @subfields_rename ) if $debug;
confess "need mapping in pairs for subfields_rename"
if $#subfields_rename % 2 != 1;
- my ( $subfields_rename, $from_subfields, $to_subfields );
+ my ( $subfields_rename, $from_subfields );
+ our $to_subfields = {};
while ( my ( $from, $to ) = splice(@subfields_rename, 0, 2) ) {
my ( $f, $t ) = (
$from_subfields->{ $from }++,
);
$subfields_rename->{ $from }->[ $f ] = [ $to => $t ];
}
- warn "### subfields_rename = ",dump( $subfields_rename ),$/;
- warn "### from_subfields = ", dump( $from_subfields ),$/;
- warn "### to_subfields = ", dump( $to_subfields ),$/;
+ warn "### subfields_rename = ",dump( $subfields_rename ),$/ if $debug;
+ warn "### from_subfields = ", dump( $from_subfields ),$/ if $debug;
+ warn "### to_subfields = ", dump( $to_subfields ),$/ if $debug;
our $_template;
- $_template->{fields_re} = {
- isis => join('|', keys %$from_subfields ),
- marc => join('|', keys %$to_subfields ),
- };
+ $_template->{isis}->{fields_re} = join('|', keys %$from_subfields );
+ $_template->{marc}->{fields_re} = join('|', keys %$to_subfields );
my @marc_out;
sub _parse_template {
my ( $name, $templates ) = @_;
- my $fields_re = $_template->{fields_re}->{ $name } || die "can't find $name in ",dump( $_template->{fields_re} );
+ my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
foreach my $template ( @{ $templates } ) {
our $count = {};
my $pos_template = $template;
$pos_template =~ s/($fields_re)/my_count($1)/ge;
my $count_key = dump( $count );
- warn "### template: |$template| -> |$pos_template| count = $count_key order = ",dump( @order ),$/;
+ warn "### template: |$template| -> |$pos_template| count = $count_key order = ",dump( @order ),$/ if $debug;
$_template->{$name}->{pos}->{ $count_key } = $pos_template;
$_template->{$name}->{order}->{ $pos_template } = [ @order ];
}
- warn "### from ",dump( $templates ), " using $fields_re created ", dump( $_template );
+ warn "### from ",dump( $templates ), " using $fields_re created ", dump( $_template ),$/ if $debug;
}
_parse_template( 'marc', $args->{marc_template} );
_parse_template( 'isis', $args->{isis_template} );
- warn "### _template = ",dump( $_template );
+ warn "### _template = ",dump( $_template ),$/ if $debug;
my $m;
my $i2 = $r->{i2} || ' ';
$m = [ $args->{to}, $i1, $i2 ];
- warn "### r = ",dump( $r );
+ warn "### r = ",dump( $r ),$/ if $debug;
- my ( $new_r, $from_count, $to_count );
- foreach my $sf ( keys %{$r} ) {
+ my ( $from_mapping, $to_mapping, $from_count, $to_count );
+ foreach my $from_sf ( keys %{$r} ) {
# skip everything which isn't one char subfield (e.g. 'subfields')
- next unless $sf =~ m/^\w$/;
- my $nr = $from_count->{$sf}++;
- my $rename_to = $subfields_rename->{ $sf } ||
- die "can't find subfield rename for $sf/$nr in ", dump( $subfields_rename );
- warn "### rename $sf/$nr to ", dump( $rename_to->[$nr] ), $/;
- my ( $to_sf, $to_nr ) = @{ $rename_to->[$nr] };
- $new_r->{ $to_sf }->[ $to_nr ] = [ $sf => $nr ];
-
- $to_count->{ $to_sf }++;
+ next unless $from_sf =~ m/^\w$/;
+ my $from_nr = $from_count->{$from_sf}++;
+ my $rename_to = $subfields_rename->{ $from_sf } ||
+ die "can't find subfield rename for $from_sf/$from_nr in ", dump( $subfields_rename );
+ my ( $to_sf, $to_nr ) = @{ $rename_to->[$from_nr] };
+ $to_mapping->{ $to_sf }->[ $to_nr ] = [ $from_sf => $from_nr ];
+
+ my $to_nr2 = $to_count->{ $to_sf }++;
+ $from_mapping->{ $from_sf }->[ $from_nr ] = [ $to_sf => $to_nr2 ];
+
+ warn "### from $from_sf/$from_nr -> $to_sf/$to_nr\tto $from_sf/$from_nr -> $to_sf/$to_nr2\n" if $debug;
}
- warn "### new_r = ",dump( $new_r );
+ warn "### from_mapping = ",dump( $from_mapping ), "\n### to_mapping = ",dump( $to_mapping ),$/ if $debug;
my $count_key = {
from => dump( $from_count ),
to => dump( $to_count),
};
- warn "### count_key = ",dump( $count_key ), $/;
+ warn "### count_key = ",dump( $count_key ),$/ if $debug;
my $processed_templates = 0;
# this defines order of traversal
foreach ( qw/isis:from marc:to/ ) {
my ($name,$count_name) = split(/:/);
- warn "## traverse $name $count_name\n";
my $ckey = $count_key->{$count_name} || die "can't find count_key $count_name in ",dump( $count_key );
my $template = $_template->{$name}->{pos}->{ $ckey } || next;
$processed_templates++;
- warn "### selected template: |$template|\n";
+ warn "### traverse $name $count_name selected template: |$template|\n",$/ if $debug;
our $fill_in = {};
foreach my $sf ( @templates ) {
sub fill_in {
my ( $name, $r, $sf, $nr ) = @_;
- my ( $from_sf, $from_nr, $v );
+ my ( $from_sf, $from_nr );
if ( $name eq 'marc' ) {
- ( $from_sf, $from_nr ) = @{ $new_r->{$sf}->[$nr] };
+ die "no $sf/$nr in to_mapping: ",dump( $to_mapping ), " form record ",dump( $r ) unless defined $to_mapping->{$sf}->[$nr];
+ ( $from_sf, $from_nr ) = @{ $to_mapping->{$sf}->[$nr] };
} else {
( $from_sf, $from_nr ) = ( $sf, $nr );
}
my $v = $r->{ $from_sf }; # || die "no $from_sf/$from_nr";
- warn "#### fill_in( $sf, $nr ) = $from_sf/$from_nr >>>> ",dump( $v ), $/;
+ warn "#### fill_in( $sf, $nr ) = $from_sf/$from_nr >>>> ",dump( $v ),$/ if $debug;
if ( ref( $v ) eq 'ARRAY' ) {
$fill_in->{$sf}->[$nr] = $v->[$from_nr];
return $v->[$from_nr];
die "requested subfield $from_sf/$from_nr but it's ",dump( $v );
}
}
- my $fields_re = $_template->{fields_re}->{ $name } || die "can't find $name in ",dump( $_template->{fields_re} );
- warn "#### $sf <<<< $fields_re\n";
+ my $fields_re = $_template->{$name}->{fields_re} || die "can't find $name in ",dump( $_template->{$name}->{fields_re} );
+ warn "#### $sf <<<< $fields_re\n",$/ if $debug;
$sf =~ s/($fields_re)(\d+)/fill_in($name,$r,$1,$2)/ge;
- warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/;
+ warn "#### >>>> $sf with fill_in = ",dump( $fill_in ),$/ if $debug;
}
- warn "## template: |$template|\n## _template->$name = ",dump( $_template->{$name} );
-
- $sf_pos = $#m;
+ warn "## template: |$template|\n## _template->$name = ",dump( $_template->{$name} ),$/ if $debug;
foreach my $sf ( @{ $_template->{$name}->{order}->{$template} } ) {
my ( $sf, $nr ) = @$sf;
my $v = $fill_in->{$sf}->[$nr] || die "can't find fill_in $sf/$nr";
- warn "++ $sf/$nr |$v|\n";
+ if ( $name eq 'isis') {
+ ( $sf, $nr ) = @{ $from_mapping->{$sf}->[$nr] };
+ }
+ warn "++ $sf/$nr |$v|\n" if $debug;
push @$m, ( $sf, $v );
}
- warn "#### >>>> created MARC record: ", dump( $m );
+ warn "#### >>>> created MARC record: ", dump( $m ),$/ if $debug;
push @marc_out, $m;
+
+ last;
}
die "I don't have template for fields ",dump( $count_key ), "\n## available templates\n", dump( $_template ) unless $processed_templates;
- warn ">>> $processed_templates templates applied to data\n";
+ warn ">>> $processed_templates templates applied to data\n",$/ if $debug;
}
- warn "### marc_template produced following MARC records: ",dump( @marc_out );
+ my $recs = 0;
foreach my $marc ( @marc_out ) {
- warn "+++ ",dump( $marc );
+ warn "+++ ",dump( $marc ),$/ if $debug;
WebPAC::Normalize::_marc_push( $marc );
+ $recs++;
}
+
+ warn "### marc_template produced $recs MARC records: ",dump( @marc_out ),$/ if $debug;
+
+ return $recs;
}
1;
projects / webpac2 / blobdiff