return correct count for marc files with additional junk bytes at end

[MARC-Fast] / Fast.pm

diff --git a/Fast.pm b/Fast.pm
index aa510ba..dea7a91 100644 (file)
--- a/Fast.pm
+++ b/Fast.pm
@@ -2,7 +2,7 @@ package MARC::Fast;
 
 use strict;
 use Carp;
-use Data::Dumper;
+use Data::Dump qw/dump/;
 
 BEGIN {
        use Exporter ();
@@ -88,6 +88,7 @@ sub new {
 
                if ($len < 24) {
                        carp "short read of leader, aborting\n";
+                       $self->{count}--;
                        last;
                }
 
@@ -124,6 +125,7 @@ sub new {
 
                # skip to next record
                my $o = substr($leader,0,5);
+               warn "# in record ", $self->{count}," record length isn't number but: ",dump($o),"\n" unless $o =~ m/^\d+$/;
                if ($o > 24) {
                        seek($self->{fh},$o-24,1) if ($o);
                } else {
@@ -336,7 +338,7 @@ sub to_hash {
 
                                        # repeatable subfiled -- convert it to array
                                        if ($val->{$f}) {
-                                               if ( $sf_usage->{$f} == 2 ) {
+                                               if ( ref($val->{$f}) ne 'ARRAY' ) {
                                                        $val->{$f} = [ $val->{$f}, $val ];
                                                } else {
                                                        push @{$val->{$f}}, $val;
@@ -383,6 +385,27 @@ sub to_ascii {
 1;
 __END__
 
+=head1 UTF-8 ENCODING
+
+This module does nothing with encoding. But, since MARC format is byte
+oriented even when using UTF-8 which has variable number of bytes for each
+character, file is opened in binary mode.
+
+As a result, all scalars recturned to perl don't have utf-8 flag. Solution is
+to use C<hash_filter> and L<Encode> to decode utf-8 encoding like this:
+
+  use Encode;
+
+  my $marc = new MARC::Fast(
+       marcdb => 'utf8.marc',
+       hash_filter => sub {
+               Encode::decode( 'utf-8', $_[0] );
+       },
+  );
+
+This will affect C<to_hash>, but C<fetch> will still return binary representation
+since it doesn't support C<hash_filter>.
+
 =head1 AUTHOR
 
        Dobrica Pavlinusic