1 |
#!/usr/bin/perl -w |
#!/usr/bin/perl -w |
2 |
|
|
3 |
use strict; |
use strict; |
4 |
use IsisDB; |
use OpenIsis; |
5 |
use Getopt::Std; |
use Getopt::Std; |
6 |
use Data::Dumper; |
use Data::Dumper; |
7 |
use XML::Simple; |
use XML::Simple; |
684 |
my $start_t = time(); |
my $start_t = time(); |
685 |
sub progress { |
sub progress { |
686 |
return if (! $show_progress); |
return if (! $show_progress); |
687 |
my $current = shift || 1; |
my $current = shift; |
688 |
my $total = shift || 1; |
my $total = shift || 1; |
689 |
my $p = int($current * 100 / $total); |
my $p = int($current * 100 / $total); |
690 |
if ($p != $last_p) { |
if ($p != $last_p) { |
720 |
my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; |
my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; |
721 |
|
|
722 |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
723 |
my $db = new IsisDB( isisdb => $isis_db ); |
my $db = OpenIsis::open( $isis_db ); |
724 |
|
|
725 |
my $max_rowid = $db->{'maxmfn'} || die "can't find maxmfn"; |
# check if .txt database for OpenIsis is zero length, |
726 |
|
# if so, erase it and re-open database |
727 |
|
sub check_txt_db { |
728 |
|
my $isis_db = shift || die "need isis database name"; |
729 |
|
my $reopen = 0; |
730 |
|
|
731 |
|
if (-e $isis_db.".TXT") { |
732 |
|
print STDERR "WARNING: removing $isis_db.TXT OpenIsis database...\n"; |
733 |
|
unlink $isis_db.".TXT" || warn "FATAL: unlink error on '$isis_db.TXT': $!"; |
734 |
|
$reopen++; |
735 |
|
} |
736 |
|
if (-e $isis_db.".PTR") { |
737 |
|
print STDERR "WARNING: removing $isis_db.PTR OpenIsis database...\n"; |
738 |
|
unlink $isis_db.".PTR" || warn "FATAL: unlink error on '$isis_db.PTR': $!"; |
739 |
|
$reopen++; |
740 |
|
} |
741 |
|
return OpenIsis::open( $isis_db ) if ($reopen); |
742 |
|
} |
743 |
|
|
744 |
|
# EOF error |
745 |
|
if ($db == -1) { |
746 |
|
$db = check_txt_db($isis_db); |
747 |
|
if ($db == -1) { |
748 |
|
print STDERR "FATAL: OpenIsis can't open zero size file $isis_db\n"; |
749 |
|
next; |
750 |
|
} |
751 |
|
} |
752 |
|
|
753 |
|
# OpenIsis::ERR_BADF |
754 |
|
if ($db == -4) { |
755 |
|
print STDERR "FATAL: OpenIsis can't find file $isis_db\n"; |
756 |
|
next; |
757 |
|
# OpenIsis::ERR_IO |
758 |
|
} elsif ($db == -5) { |
759 |
|
print STDERR "FATAL: OpenIsis can't access file $isis_db\n"; |
760 |
|
next; |
761 |
|
} elsif ($db < 0) { |
762 |
|
print STDERR "FATAL: OpenIsis unknown error $db with file $isis_db\n"; |
763 |
|
next; |
764 |
|
} |
765 |
|
|
766 |
|
my $max_rowid = OpenIsis::maxRowid( $db ); |
767 |
|
|
768 |
|
# if 0 records, try to rease isis .txt database |
769 |
|
if ($max_rowid == 0) { |
770 |
|
# force removal of database |
771 |
|
$db = check_txt_db($isis_db); |
772 |
|
$max_rowid = OpenIsis::maxRowid( $db ); |
773 |
|
} |
774 |
|
|
775 |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
776 |
|
|
777 |
my $path = $database; |
my $path = $database; |
778 |
|
|
779 |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
for (my $row_id = 1; $row_id <= $max_rowid; $row_id++ ) { |
780 |
my $row = $db->to_hash( $row_id ); |
my $row = OpenIsis::read( $db, $row_id ); |
781 |
if ($row) { |
if ($row && $row->{mfn}) { |
782 |
|
|
|
$row->{mfn} = $row_id; |
|
|
$row->{record} = $db->{record}; |
|
|
|
|
783 |
progress($row->{mfn}, $max_rowid); |
progress($row->{mfn}, $max_rowid); |
784 |
|
|
785 |
my $swishpath = $path."#".int($row->{mfn}); |
my $swishpath = $path."#".int($row->{mfn}); |
793 |
} |
} |
794 |
} |
} |
795 |
} |
} |
796 |
|
# for this to work with current version of OpenIsis (0.9.0) |
797 |
|
# you might need my patch from |
798 |
|
# http://www.rot13.org/~dpavlin/projects/openisis-0.9.0-perl_close.diff |
799 |
|
OpenIsis::close($db); |
800 |
print STDERR "\n"; |
print STDERR "\n"; |
801 |
|
|
802 |
} elsif ($type_base eq "excel") { |
} elsif ($type_base eq "excel") { |
860 |
} |
} |
861 |
} elsif ($type_base eq "marc") { |
} elsif ($type_base eq "marc") { |
862 |
|
|
863 |
require MARC; |
require MARC::File::USMARC; |
864 |
|
|
865 |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
866 |
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
867 |
|
|
868 |
# optional argument is format |
# optional argument is format |
869 |
my $format = x($config->{marc_format}) || 'usmarc'; |
warn "marc_format is no longer used!" if ($config->{marc_format}); |
|
|
|
870 |
print STDERR "Reading MARC file '$marc_file'\n"; |
print STDERR "Reading MARC file '$marc_file'\n"; |
871 |
|
|
872 |
my $marc = new MARC; |
my $marc = MARC::File::USMARC->in( $marc_file ) |
873 |
my $nr = $marc->openmarc({ |
|| die "Can't open MARC file '$marc_file': ".$MARC::File::ERROR; |
|
file=>$marc_file, format=>$format |
|
|
}) || die "Can't open MARC file '$marc_file' with format '$format'"; |
|
874 |
|
|
875 |
# read MARC file in memory |
# count records in MARC file |
876 |
$marc->nextmarc(-1); |
sub marc_count { |
877 |
|
my $filename = shift || die; |
878 |
|
my $file = MARC::File::USMARC->in($filename) || die $MARC::File::ERROR; |
879 |
|
my $count = 0; |
880 |
|
while ($file->skip()) { |
881 |
|
$count++; |
882 |
|
} |
883 |
|
return $count; |
884 |
|
} |
885 |
|
|
886 |
my $max_rec = $marc->marc_count(); |
my $count = marc_count($marc_file) || warn "no records in '$marc_file'?"; |
887 |
|
|
888 |
for(my $i=1; $i<=$max_rec; $i++) { |
my $i = 0; |
889 |
|
|
890 |
progress($i,$max_rec); |
while( my $rec = $marc->next() ) { |
891 |
|
|
892 |
# store value for marc_sf.pm |
progress($i++,$count); |
|
$main::cache->{marc_record} = $i; |
|
893 |
|
|
894 |
my $swishpath = $database."#".$i; |
my $swishpath = $database."#".$i; |
895 |
|
|
896 |
if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) { |
if (my $xml = data2xml($type_base,$rec,$add_xml,$cfg,$database)) { |
897 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
898 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
899 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |