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 (-f $ARGV[0]); |
$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; |
525 |
$swish_data =~ s/ +/ /g; |
$swish_data =~ s/ +/ /g; |
526 |
$swish_data =~ s/ +$//g; |
$swish_data =~ s/ +$//g; |
527 |
|
|
528 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
$xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data)); |
529 |
} |
} |
530 |
|
|
531 |
my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page]; |
my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page]; |
535 |
|
|
536 |
# add delimiters before and after word. |
# add delimiters before and after word. |
537 |
# That is required to produce exact match |
# That is required to produce exact match |
538 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
$xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data)); |
539 |
} |
} |
540 |
|
|
541 |
my $idel = $cache->{index_delimiter}->{$field}; |
my $idel = $cache->{index_delimiter}->{$field}; |
568 |
$swish_data =~ s/ +/ /g; |
$swish_data =~ s/ +/ /g; |
569 |
$swish_data =~ s/ +$//g; |
$swish_data =~ s/ +$//g; |
570 |
|
|
571 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
$xml .= xmlify($field."_swish", my_unac_string($codepage,$swish_data)); |
572 |
} |
} |
573 |
|
|
574 |
if ($swish_exact_data) { |
if ($swish_exact_data) { |
577 |
|
|
578 |
# add delimiters before and after word. |
# add delimiters before and after word. |
579 |
# That is required to produce exact match |
# That is required to produce exact match |
580 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
$xml .= xmlify($field."_swish_exact", my_unac_string($codepage,$swish_exact_data)); |
581 |
} |
} |
582 |
} |
} |
583 |
} |
} |
614 |
|
|
615 |
my $show_progress = $cfg_global->val('global', 'show_progress'); |
my $show_progress = $cfg_global->val('global', 'show_progress'); |
616 |
|
|
617 |
my $unac_filter = $cfg_global->val('global', 'unac_filter'); |
my $my_unac_filter = $cfg_global->val('global', 'my_unac_filter'); |
618 |
if ($unac_filter) { |
if ($my_unac_filter) { |
619 |
require $unac_filter; |
print STDERR "using $my_unac_filter to filter characters for search\n"; |
620 |
|
require $my_unac_filter; |
621 |
|
} else { |
622 |
|
print STDERR "### fallback to default my_unac_string!\n"; |
623 |
|
eval q{ |
624 |
|
sub main::my_unac_string($$) { |
625 |
|
my ($charset, $string) = (@_); |
626 |
|
return $string; |
627 |
|
} |
628 |
|
}; |
629 |
} |
} |
630 |
|
|
631 |
foreach my $database ($cfg->Sections) { |
foreach my $database ($cfg->Sections) { |
637 |
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
638 |
if ($lookup_file) { |
if ($lookup_file) { |
639 |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
640 |
|
if (! -e $lookup_file) { |
641 |
|
open(LOOKUP, "> $lookup_file") || die "can't create $lookup_file': $!"; |
642 |
|
close(LOOKUP); |
643 |
|
} |
644 |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
645 |
print STDERR "creating lookup file '$lookup_file'\n"; |
print STDERR "creating lookup file '$lookup_file'\n"; |
646 |
# delete memory cache for lookup file |
# delete memory cache for lookup file |