139 |
} else { |
} else { |
140 |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
141 |
} |
} |
142 |
|
|
143 |
if ($field_name) { |
if ($field_name) { |
144 |
|
$field_name = x($field_name); |
145 |
if (! $last_field_name) { |
if (! $last_field_name) { |
146 |
$last_field_name = x($field_name); |
$last_field_name = $field_name; |
147 |
return $last_field_name; |
return $last_field_name; |
148 |
} elsif ($field_name ne $last_field_name) { |
} elsif ($field_name ne $last_field_name) { |
149 |
$last_field_name = x($field_name); |
$last_field_name = $field_name; |
150 |
return $last_field_name; |
return $last_field_name; |
151 |
} |
} |
152 |
} |
} |
641 |
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
642 |
if ($lookup_file) { |
if ($lookup_file) { |
643 |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
644 |
|
if (! -e $lookup_file) { |
645 |
|
open(LOOKUP, "> $lookup_file") || die "can't create $lookup_file': $!"; |
646 |
|
close(LOOKUP); |
647 |
|
} |
648 |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
649 |
print STDERR "creating lookup file '$lookup_file'\n"; |
print STDERR "creating lookup file '$lookup_file'\n"; |
650 |
# delete memory cache for lookup file |
# delete memory cache for lookup file |
667 |
|
|
668 |
$config=XMLin("./import_xml/$type.xml", ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], ForceContent => 1 ); |
$config=XMLin("./import_xml/$type.xml", ForceArray => [ $type2tag{$type_base}, 'config', 'format' ], ForceContent => 1 ); |
669 |
|
|
670 |
|
# helper for progress bar |
671 |
|
sub fmt_time { |
672 |
|
my $t = shift || 0; |
673 |
|
my $out = ""; |
674 |
|
|
675 |
|
my ($ss,$mm,$hh) = gmtime($t); |
676 |
|
$out .= "${hh}h" if ($hh); |
677 |
|
$out .= sprintf("%02d:%02d", $mm,$ss); |
678 |
|
$out .= " " if ($hh == 0); |
679 |
|
return $out; |
680 |
|
} |
681 |
|
|
682 |
# output current progress indicator |
# output current progress indicator |
683 |
my $last_p = 0; |
my $last_p = 0; |
684 |
|
my $start_t = time(); |
685 |
sub progress { |
sub progress { |
686 |
return if (! $show_progress); |
return if (! $show_progress); |
687 |
my $current = shift; |
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 != $last_p) { |
691 |
printf STDERR ("%5d / %5d [%-51s] %-2d %% \r",$current,$total,"=" x ($p/2).">", $p ); |
my $rate = ($current / (time() - $start_t || 1)); |
692 |
|
my $eta = ($total-$current) / ($rate || 1); |
693 |
|
printf STDERR ("%5d [%-38s] %-5d %0.1f/s %s\r",$current,"=" x ($p/3)."$p%>", $total, $rate, fmt_time($eta)); |
694 |
$last_p = $p; |
$last_p = $p; |
695 |
} |
} |
696 |
} |
} |
830 |
for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) { |
for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) { |
831 |
my $cell = $oWorksheet->{Cells}[$iR][$iC]; |
my $cell = $oWorksheet->{Cells}[$iR][$iC]; |
832 |
if ($cell) { |
if ($cell) { |
833 |
$row->{int2col($iC)} = $cell->Value; |
# this conversion is a cludge. |
834 |
|
# Files from Excell could have |
835 |
|
# characters which don't fit into |
836 |
|
# destination encoding. |
837 |
|
$row->{int2col($iC)} = $utf2cp->convert($cell->Value) || $cell->Value; |
838 |
} |
} |
839 |
} |
} |
840 |
|
|
860 |
} |
} |
861 |
} elsif ($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"; |