use MARC::File::XML ( BinaryEncoding => 'utf-8' );
my $conf = {}; # configuration hashref
-my $count = 0;
+my $count = 0; my $scount = 0;
my $start = time;
$| = 1;
my $batch = undef; my $record = undef;
#$batch = MARC::Batch->new('XML', $records); # The other part of the hack
- $batch = MARC::Batch->new('XML', $file);
+ $batch = MARC::Batch->new($conf->{marctype}, $file);
$batch->strict_off();
$batch->warnings_off();
unless (marc_isvalid($marc))
{ dump_exception($marc); next; }
dump_fingerprints($marc);
+ $scount++; progress_ticker();
}
}
$marc{date1} = substr($my_008,7,4) if ($my_008);
$marc{date2} = substr($my_008,11,4) if ($my_008); # UNUSED
+ unless ($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;
+ print XF ">> using 260c as date1 on $id\n";
+ }
+ }
+
# item_form
if ( $marc{record_type} =~ /[gkroef]/ ) { # MAP, VIS
$marc{item_form} = substr($my_008,29,1) if ($my_008);
}
}
+
+=pod
+
+Looks like case b isn't a terribly good idea most of the time -- it
+will generate many spurious matches in series books in particular. So
+for now, simply turning it off.
+
if ($marc->{edition}) { # case b : edition
print OF join("\t", $marc->{id}, "case b",
$marc->{item_form}, $marc->{date1},
$marc->{title}, $marc->{edition}), "\n";
}
+=cut
+
if ($marc->{issn}) { # case c : issn
print OF join("\t", $marc->{id}, "case c",
$marc->{item_form}, $marc->{date1},
'incoming',
'incumbent',
'exception|x=s',
+ 'marctype|m=s',
'output|o=s',
'runtype|r=s',
'subfield|s=s',
if ($c->{incoming}) {
$c->{tag} = 903 unless defined $c->{tag};
$c->{subfield} = 'a' unless defined $c->{subfield};
+ $c->{marctype} = 'XML' unless defined $c->{marctype};
$c->{output} = 'incoming.fp' unless defined $c->{output};
$c->{exception} = 'incoming.ex' unless defined $c->{exception};
$c->{runtype} = 'full' unless defined $c->{runtype};
} elsif ($c->{incumbent}) {
$c->{tag} = 901 unless defined $c->{tag};
$c->{subfield} = 'c' unless defined $c->{subfield};
+ $c->{marctype} = 'XML' unless defined $c->{marctype};
$c->{output} = 'incumbent.fp' unless defined $c->{output};
$c->{exception} = 'incumbent.ex' unless defined $c->{exception};
$c->{runtype} = 'full' unless defined $c->{runtype};
sub progress_ticker {
return if $conf->{quiet};
- printf("\r> %d (%d/s)", $count, ($count / (time - $start + 1)))
- if ($count % 100 == 0);
+ printf("\r> %d recs seen; %d processed", $count, $scount);
+ printf(" (%d/s)", ($count / (time - $start + 1)))
+ if ($count % 500 == 0);
}
=head2 show_help