use MARC::Batch;
use Unicode::Normalize;
use MARC::File::XML ( BinaryEncoding => 'utf-8' );
+use Equinox::Migration::SubfieldMapper;
my $conf = {}; # configuration hashref
my $count = 0; my $scount = 0;
# populate and normalize marc
$marc = populate_marc($record, $id);
+ # check for manual exclusion
+ next if this_record_is_excluded($record, $marc);
normalize_marc($marc);
unless (marc_isvalid($marc))
{ dump_exception($marc); next; }
}
unless ($marc{date1} and $marc{date1} =~ /\d{4}/) {
my $my_260 = $record->field('260');
- my $date1 = $my_260->subfield('c') if $my_260;
- if (defined $date1 and $date1 =~ /\d{4}/) {
- $marc{date1} = $date1;
- $marc{fudgedate} = 1;
- print XF ">> using 260c as date1 at rec $count\n";
+ if ($my_260 and $my_260->subfield('c')) {
+ my $date1 = $my_260->subfield('c');
+ $date1 =~ s/\D//g;
+ if (defined $date1 and $date1 =~ /\d{4}/) {
+ $marc{date1} = $date1;
+ $marc{fudgedate} = 1;
+ print XF ">> using 260c as date1 at rec $count\n";
+ }
}
}
}
$json .= 'misc:' . $marc->{misc_score} . '}';
- my $compact = join('-', $marc->{age_score}, $marc->{misc_score}, @score);
+ my $compact = join('', $marc->{age_score}, $marc->{misc_score}, @score);
$marc->{score} = "$compact\t$json";
}
=cut
sub dump_exception {
- my ($marc) = @_;
+ my ($marc, $msg) = @_;
unless (defined $marc) {
print XF "Undefined record at line $count; likely bad XML\n";
return;
}
+
print XF "Record ", $marc->{id}, " excluded: ";
+ if (defined $msg) {
+ print XF "$msg\n";
+ return
+ }
+
print XF "missing item_form; " unless ($marc->{item_form});
unless (defined $marc->{date1})
{ print XF "missing date1; " }
}
+=head2 this_record_is_excluded
+
+Returns 1 if the record B<is> and 0 if the record B<is not> excluded,
+according to the subfield mapping (generated via the C<--excludelist>
+option).
+
+=cut
+
+sub this_record_is_excluded {
+ my ($rec, $marc) = @_;
+ return 0 unless defined $conf->{excludelist};
+
+ for my $tag (keys %{ $conf->{excludelist}->{tags} }) {
+ for my $sub (keys %{$conf->{excludelist}->{tags}{$tag}}) {
+ my $f = $conf->{excludelist}->field($tag, $sub);
+
+ # if this record doesn't have the right tag/sub, it can't be
+ return 0 unless ($rec->field($tag) and $rec->field($tag)->subfield($sub));
+ # but it does, so if there are no filters to check...
+ unless ($conf->{excludelist}->filters($f))
+ { dump_exception($marc, "exclusion $tag$sub"); return 1 }
+
+ my $sub_contents = $rec->field($tag)->subfield($sub);
+ for my $filter (@{ $conf->{excludelist}->filters($f)}) {
+ if ($sub_contents =~ /$filter/i) {
+ # filter matches. no fp.
+ dump_exception($marc, "exclusion $tag$sub '$filter'");
+ return 1;
+ }
+ # no match, no exclude
+ return 0;
+ }
+ }
+ }
+}
+
=head2 initialize
Performs boring script initialization. Handles argument parsing,
'arbitrarily-lose-above=i',
'arbitrarily-lose-below=i',
'newwins',
+ 'excludelist=s',
'quiet|q',
'help|h',
);
$c->{exception} = join('.',$c->{prefix},'fp','ex');
}
+ # get SFM object if excludelist was specified
+ if ($c->{excludelist}) {
+ $c->{excludelist} =
+ Equinox::Migration::SubfieldMapper->new( file => $c->{excludelist} );
+ }
+
my @keys = keys %{$c};
show_help() unless (@ARGV and @keys);
for my $key ('tag', 'subfield', 'output', 'exception')
--fingerprints=LIST Fingerprints to generate, comma separated
Default: oclc,isbn,edition,issn,lccn,accomp,authpub
Others: baseline
+ --excludelist=FILE Name of fingerprints exclusions file
--scores=LIST Scores to calculate, comma separated
Default: oclc,dlc,num_650,num_tags,enc_level