5 |
use Getopt::Std; |
use Getopt::Std; |
6 |
use Data::Dumper; |
use Data::Dumper; |
7 |
use XML::Simple; |
use XML::Simple; |
|
use Text::Unaccent 1.02; # 1.01 won't compile on my platform, |
|
8 |
use Text::Iconv; |
use Text::Iconv; |
9 |
use Config::IniFiles; |
use Config::IniFiles; |
10 |
use Encode; |
use Encode; |
16 |
|
|
17 |
my $config_file = $0; |
my $config_file = $0; |
18 |
$config_file =~ s/\.pl$/.conf/; |
$config_file =~ s/\.pl$/.conf/; |
19 |
|
$config_file = $ARGV[0] if ($ARGV[0] && -f $ARGV[0]); |
20 |
die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file); |
die "FATAL: can't find configuration file '$config_file'" if (! -e $config_file); |
21 |
|
|
22 |
my $config; |
my $config; |
168 |
($s,$se,$d,$i) = (0,1,0,0); |
($s,$se,$d,$i) = (0,1,0,0); |
169 |
} elsif (lc($type) =~ /^lookup/) { |
} elsif (lc($type) =~ /^lookup/) { |
170 |
($s,$se,$d,$i,$il) = (0,1,0,0,1); |
($s,$se,$d,$i,$il) = (0,1,0,0,1); |
171 |
|
} elsif ($type) { |
172 |
|
print STDERR "WARNING: unknown type: $type\n"; |
173 |
} |
} |
174 |
return ($s,$se,$d,$i,$il); |
return ($s,$se,$d,$i,$il); |
175 |
} |
} |
527 |
$swish_data =~ s/ +/ /g; |
$swish_data =~ s/ +/ /g; |
528 |
$swish_data =~ s/ +$//g; |
$swish_data =~ s/ +$//g; |
529 |
|
|
530 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
$xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data)); |
531 |
} |
} |
532 |
|
|
533 |
my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page]; |
my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page]; |
537 |
|
|
538 |
# add delimiters before and after word. |
# add delimiters before and after word. |
539 |
# That is required to produce exact match |
# That is required to produce exact match |
540 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
$xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data)); |
541 |
} |
} |
542 |
|
|
543 |
my $idel = $cache->{index_delimiter}->{$field}; |
my $idel = $cache->{index_delimiter}->{$field}; |
570 |
$swish_data =~ s/ +/ /g; |
$swish_data =~ s/ +/ /g; |
571 |
$swish_data =~ s/ +$//g; |
$swish_data =~ s/ +$//g; |
572 |
|
|
573 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
$xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data)); |
574 |
} |
} |
575 |
|
|
576 |
if ($swish_exact_data) { |
if ($swish_exact_data) { |
579 |
|
|
580 |
# add delimiters before and after word. |
# add delimiters before and after word. |
581 |
# That is required to produce exact match |
# That is required to produce exact match |
582 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
$xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data)); |
583 |
} |
} |
584 |
} |
} |
585 |
} |
} |
616 |
|
|
617 |
my $show_progress = $cfg_global->val('global', 'show_progress'); |
my $show_progress = $cfg_global->val('global', 'show_progress'); |
618 |
|
|
619 |
my $unac_filter = $cfg_global->val('global', 'unac_filter'); |
my $my_unac_filter = $cfg_global->val('global', 'my_unac_filter'); |
620 |
if ($unac_filter) { |
if ($my_unac_filter) { |
621 |
require $unac_filter; |
print STDERR "using $my_unac_filter to filter characters for search\n"; |
622 |
|
require $my_unac_filter; |
623 |
|
} else { |
624 |
|
print STDERR "### fallback to default my_unac_string!\n"; |
625 |
|
eval q{ |
626 |
|
sub main::my_unac_string($$) { |
627 |
|
my ($charset, $string) = (@_); |
628 |
|
return $string; |
629 |
|
} |
630 |
|
}; |
631 |
} |
} |
632 |
|
|
633 |
foreach my $database ($cfg->Sections) { |
foreach my $database ($cfg->Sections) { |
639 |
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
640 |
if ($lookup_file) { |
if ($lookup_file) { |
641 |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
642 |
|
if (! -e $lookup_file) { |
643 |
|
open(LOOKUP, "> $lookup_file") || die "can't create $lookup_file': $!"; |
644 |
|
close(LOOKUP); |
645 |
|
} |
646 |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
647 |
print STDERR "creating lookup file '$lookup_file'\n"; |
print STDERR "creating lookup file '$lookup_file'\n"; |
648 |
# delete memory cache for lookup file |
# delete memory cache for lookup file |
813 |
for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) { |
for(my $iC = $oWorksheet->{MinCol} ; defined $oWorksheet->{MaxCol} && $iC <= $oWorksheet->{MaxCol} ; $iC++) { |
814 |
my $cell = $oWorksheet->{Cells}[$iR][$iC]; |
my $cell = $oWorksheet->{Cells}[$iR][$iC]; |
815 |
if ($cell) { |
if ($cell) { |
816 |
$row->{int2col($iC)} = $cell->Value; |
# this conversion is a cludge. |
817 |
|
# Files from Excell could have |
818 |
|
# characters which don't fit into |
819 |
|
# destination encoding. |
820 |
|
$row->{int2col($iC)} = $utf2cp->convert($cell->Value) || $cell->Value; |
821 |
} |
} |
822 |
} |
} |
823 |
|
|
937 |
|
|
938 |
all2xml.pl - read various file formats and dump XML for SWISH-E |
all2xml.pl - read various file formats and dump XML for SWISH-E |
939 |
|
|
940 |
|
=head1 SYNOPSYS |
941 |
|
|
942 |
|
$ all2xml.pl [test.conf] |
943 |
|
|
944 |
=head1 DESCRIPTION |
=head1 DESCRIPTION |
945 |
|
|
946 |
This command will read ISIS data file using OpenIsis perl module, MARC |
This command will read ISIS data file using OpenIsis perl module, MARC |
949 |
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 |
950 |
has something like that). Output of this script is tailor-made for SWISH-E. |
has something like that). Output of this script is tailor-made for SWISH-E. |
951 |
|
|
952 |
|
If no configuration file is specified, it will use default one called |
953 |
|
C<all2xml.conf>. |
954 |
|
|
955 |
=head1 BUGS |
=head1 BUGS |
956 |
|
|
957 |
Documentation is really lacking. However, in true Open Source spirit, source |
Documentation is really lacking. However, in true Open Source spirit, source |