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 || $current == 1) { |
691 |
|
$start_t = time(); |
692 |
|
$last_p = 0; |
693 |
|
} elsif ($p != $last_p) { |
694 |
my $rate = ($current / (time() - $start_t || 1)); |
my $rate = ($current / (time() - $start_t || 1)); |
695 |
my $eta = ($total-$current) / ($rate || 1); |
my $eta = ($total-$current) / ($rate || 1); |
696 |
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)); |
699 |
} |
} |
700 |
|
|
701 |
my $fake_dir = 1; |
my $fake_dir = 1; |
702 |
|
my $fake_pos = 0; |
703 |
|
my $last_fake_t = time(); |
704 |
sub fakeprogress { |
sub fakeprogress { |
705 |
return if (! $show_progress); |
return if (! $show_progress); |
706 |
my $current = shift @_; |
my $current = shift @_; |
707 |
|
|
708 |
my @ind = ('-','\\','|','/','-','\\','|','/', '-'); |
my @ind = ('-','\\','|','/','-','\\','|','/'); |
709 |
|
|
710 |
$last_p += $fake_dir; |
if ($current < $fake_pos) { |
711 |
$fake_dir = -$fake_dir if ($last_p > 1000 || $last_p < 0); |
$start_t = time(); |
712 |
if ($last_p % 10 == 0) { |
$last_fake_t = 0; |
713 |
printf STDERR ("%5d / %5s [%-51s]\r",$current,"?"," " x ($last_p/20).$ind[($last_p/20) % $#ind]); |
$fake_dir = 1; |
714 |
|
$fake_pos = 0; |
715 |
|
} |
716 |
|
|
717 |
|
if (time()-$last_fake_t >= 1) { |
718 |
|
$last_fake_t = time(); |
719 |
|
$fake_pos += $fake_dir; |
720 |
|
$fake_dir = -$fake_dir if ($fake_pos > 38); |
721 |
|
} |
722 |
|
|
723 |
|
if ($current % 10 == 0) { |
724 |
|
my $rate = ($current / (time() - $start_t || 1)); |
725 |
|
printf STDERR ("%5d [%-38s] %0.1f/s\r",$current, " " x $fake_pos .$ind[($current / 10) % 8], $rate); |
726 |
} |
} |
727 |
} |
} |
728 |
|
|
828 |
} |
} |
829 |
} elsif ($type_base eq "marc") { |
} elsif ($type_base eq "marc") { |
830 |
|
|
831 |
require MARC; |
require MARC::File::USMARC; |
832 |
|
|
833 |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
834 |
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!"; |
835 |
|
|
836 |
# optional argument is format |
# optional argument is format |
837 |
my $format = x($config->{marc_format}) || 'usmarc'; |
warn "marc_format is no longer used!" if ($config->{marc_format}); |
|
|
|
838 |
print STDERR "Reading MARC file '$marc_file'\n"; |
print STDERR "Reading MARC file '$marc_file'\n"; |
839 |
|
|
840 |
my $marc = new MARC; |
my $marc = MARC::File::USMARC->in( $marc_file ) |
841 |
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'"; |
|
842 |
|
|
843 |
# read MARC file in memory |
# count records in MARC file |
844 |
$marc->nextmarc(-1); |
sub marc_count { |
845 |
|
my $filename = shift || die; |
846 |
|
my $file = MARC::File::USMARC->in($filename) || die $MARC::File::ERROR; |
847 |
|
my $count = 0; |
848 |
|
while ($file->skip()) { |
849 |
|
$count++; |
850 |
|
} |
851 |
|
return $count; |
852 |
|
} |
853 |
|
|
854 |
my $max_rec = $marc->marc_count(); |
my $count = marc_count($marc_file) || warn "no records in '$marc_file'?"; |
855 |
|
|
856 |
for(my $i=1; $i<=$max_rec; $i++) { |
my $i = 0; |
857 |
|
|
858 |
progress($i,$max_rec); |
while( my $rec = $marc->next() ) { |
859 |
|
|
860 |
# store value for marc_sf.pm |
progress($i++,$count); |
|
$main::cache->{marc_record} = $i; |
|
861 |
|
|
862 |
my $swishpath = $database."#".$i; |
my $swishpath = $database."#".$i; |
863 |
|
|
864 |
if (my $xml = data2xml($type_base,$marc,$add_xml,$cfg,$database)) { |
if (my $xml = data2xml($type_base,$rec,$add_xml,$cfg,$database)) { |
865 |
$xml = $cp2utf->convert($xml); |
$xml = $cp2utf->convert($xml); |
866 |
use bytes; # as opposed to chars |
use bytes; # as opposed to chars |
867 |
print "Path-Name: $swishpath\n"; |
print "Path-Name: $swishpath\n"; |
932 |
|
|
933 |
=head1 DESCRIPTION |
=head1 DESCRIPTION |
934 |
|
|
935 |
This command will read ISIS data file using OpenIsis perl module, MARC |
This command will read ISIS data file using IsisDB perl module, MARC |
936 |
records using MARC module and optionally Micro$oft Excel files to |
records using MARC module and optionally Micro$oft Excel files to |
937 |
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, |
938 |
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 |