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) { |
691 |
|
$start_t = time(); |
692 |
|
} elsif ($p != $last_p) { |
693 |
my $rate = ($current / (time() - $start_t || 1)); |
my $rate = ($current / (time() - $start_t || 1)); |
694 |
my $eta = ($total-$current) / ($rate || 1); |
my $eta = ($total-$current) / ($rate || 1); |
695 |
printf STDERR ("%5d [%-38s] %-5d %0.1f/s %s\r",$current,"=" x ($p/3)."$p%>", $total, $rate, fmt_time($eta)); |
printf STDERR ("%5d [%-38s] %-5d %0.1f/s %s\r",$current,"=" x ($p/3)."$p%>", $total, $rate, fmt_time($eta)); |
813 |
} |
} |
814 |
} elsif ($type_base eq "marc") { |
} elsif ($type_base eq "marc") { |
815 |
|
|
816 |
require MARC; |
require MARC::File::USMARC; |
817 |
|
|
818 |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
819 |
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!"; |
820 |
|
|
821 |
# optional argument is format |
# optional argument is format |
822 |
my $format = x($config->{marc_format}) || 'usmarc'; |
warn "marc_format is no longer used!" if ($config->{marc_format}); |
|
|
|
823 |
print STDERR "Reading MARC file '$marc_file'\n"; |
print STDERR "Reading MARC file '$marc_file'\n"; |
824 |
|
|
825 |
my $marc = new MARC; |
my $marc = MARC::File::USMARC->in( $marc_file ) |
826 |
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'"; |
|
827 |
|
|
828 |
# read MARC file in memory |
# count records in MARC file |
829 |
$marc->nextmarc(-1); |
sub marc_count { |
830 |
|
my $filename = shift || die; |
831 |
|
my $file = MARC::File::USMARC->in($filename) || die $MARC::File::ERROR; |
832 |
|
my $count = 0; |
833 |
|
while ($file->skip()) { |
834 |
|
$count++; |
835 |
|
} |
836 |
|
return $count; |
837 |
|
} |
838 |
|
|
839 |
my $max_rec = $marc->marc_count(); |
my $count = marc_count($marc_file) || warn "no records in '$marc_file'?"; |
840 |
|
|
841 |
for(my $i=1; $i<=$max_rec; $i++) { |
my $i = 0; |
842 |
|
|
843 |
progress($i,$max_rec); |
while( my $rec = $marc->next() ) { |
844 |
|
|
845 |
# store value for marc_sf.pm |
progress($i++,$count); |
|
$main::cache->{marc_record} = $i; |
|
846 |
|
|
847 |
my $swishpath = $database."#".$i; |
my $swishpath = $database."#".$i; |
848 |
|
|
849 |
if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) { |
if (my $xml = data2xml($type_base,$rec,$add_xml,$cfg,$database)) { |
850 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
851 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
852 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |
917 |
|
|
918 |
=head1 DESCRIPTION |
=head1 DESCRIPTION |
919 |
|
|
920 |
This command will read ISIS data file using OpenIsis perl module, MARC |
This command will read ISIS data file using IsisDB perl module, MARC |
921 |
records using MARC module and optionally Micro$oft Excel files to |
records using MARC module and optionally Micro$oft Excel files to |
922 |
create one XML file for usage with I<SWISH-E> indexer. Dispite it's name, |
create one XML file for usage with I<SWISH-E> indexer. Dispite it's name, |
923 |
this script B<isn't general xml generator> from isis files (isis allready |
this script B<isn't general xml generator> from isis files (isis allready |