61 |
'feed' => 'feed' |
'feed' => 'feed' |
62 |
); |
); |
63 |
|
|
64 |
|
my $cache; # for cacheing |
65 |
|
|
66 |
sub data2xml { |
sub data2xml { |
67 |
|
|
68 |
use xmlify; |
use xmlify; |
92 |
return $va <=> $vb; |
return $va <=> $vb; |
93 |
} |
} |
94 |
|
|
95 |
foreach my $field (sort by_order keys %{$config->{indexer}}) { |
my @sorted_tags; |
96 |
|
if ($cache->{tags_by_order}->{$type}) { |
97 |
|
@sorted_tags = @{$cache->{tags_by_order}->{$type}}; |
98 |
|
} else { |
99 |
|
@sorted_tags = sort by_order keys %{$config->{indexer}}; |
100 |
|
$cache->{tags_by_order}->{$type} = \@sorted_tags; |
101 |
|
} |
102 |
|
|
103 |
|
foreach my $field (@sorted_tags) { |
104 |
|
|
105 |
$field=x($field); |
$field=x($field); |
106 |
$field_usage{$field}++; |
$field_usage{$field}++; |
107 |
|
|
108 |
my $swish_data = ""; |
my $swish_data = ""; |
109 |
|
my $swish_exact_data = ""; |
110 |
my $display_data = ""; |
my $display_data = ""; |
111 |
my $line_delimiter; |
my $line_delimiter; |
112 |
|
|
120 |
|
|
121 |
my $repeat_off = 0; # repeatable offset |
my $repeat_off = 0; # repeatable offset |
122 |
|
|
123 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
my ($s,$se,$d,$i) = (1,0,1,0); # swish, display default |
124 |
$s = 0 if (lc($x->{type}) eq "display"); |
$s = 0 if (lc($x->{type}) eq "display"); |
125 |
$d = 0 if (lc($x->{type}) eq "swish"); |
$d = 0 if (lc($x->{type}) eq "swish"); |
126 |
|
$se = 1 if (lc($x->{type}) eq "swish_exact"); |
127 |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
128 |
|
|
129 |
# what will separate last line from this one? |
# what will separate last line from this one? |
177 |
# filter="name" ; filter this field through |
# filter="name" ; filter this field through |
178 |
# filter/[name].pm |
# filter/[name].pm |
179 |
my $filter = $x->{filter}; |
my $filter = $x->{filter}; |
180 |
if ($filter) { |
if ($filter && !$cache->{filter_loaded}->{$filter}) { |
181 |
require "filter/".$filter.".pm"; |
require "filter/".$filter.".pm"; |
182 |
|
$cache->{filter_loaded}->{$filter}++; |
183 |
} |
} |
184 |
# type="swish" ; field for swish |
# type="swish" ; field for swish |
185 |
if ($s && $swish) { |
if ($swish) { |
186 |
if ($filter) { |
if ($filter && ($s || $se)) { |
187 |
no strict 'refs'; |
no strict 'refs'; |
188 |
$swish_data .= join(" ",&$filter($swish)); |
my $tmp = join(" ",&$filter($swish)) if ($s || $se); |
189 |
|
$swish_data .= $tmp if ($s); |
190 |
|
$swish_exact_data .= $tmp if ($se); |
191 |
|
|
192 |
} else { |
} else { |
193 |
$swish_data .= $swish; |
$swish_data .= $swish if ($s); |
194 |
|
$swish_exact_data .= $swish if ($se); |
195 |
} |
} |
196 |
} |
} |
197 |
|
|
248 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
my ($s,$d,$i) = (1,1,0); # swish, display default |
249 |
$s = 0 if (lc($x->{type}) eq "display"); |
$s = 0 if (lc($x->{type}) eq "display"); |
250 |
$d = 0 if (lc($x->{type}) eq "swish"); |
$d = 0 if (lc($x->{type}) eq "swish"); |
251 |
|
# no support for swish exact in config. |
252 |
|
# IMHO, it's useless |
253 |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
254 |
|
|
255 |
if ($val) { |
if ($val) { |
292 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
293 |
} |
} |
294 |
|
|
295 |
|
if ($swish_exact_data) { |
296 |
|
$swish_exact_data =~ s/ +/ /g; |
297 |
|
$swish_exact_data =~ s/ +$//g; |
298 |
|
|
299 |
|
# add delimiters before and after word. |
300 |
|
# That is required to produce exact match |
301 |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
302 |
|
} |
303 |
|
|
304 |
|
|
305 |
} |
} |
306 |
|
|
336 |
|
|
337 |
my $show_progress = $cfg_global->val('global', 'show_progress'); |
my $show_progress = $cfg_global->val('global', 'show_progress'); |
338 |
|
|
339 |
|
my $unac_filter = $cfg_global->val('global', 'unac_filter'); |
340 |
|
if ($unac_filter) { |
341 |
|
require $unac_filter; |
342 |
|
} |
343 |
|
|
344 |
foreach my $database ($cfg->Sections) { |
foreach my $database ($cfg->Sections) { |
345 |
|
|
346 |
my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined"; |
my $type = lc($cfg -> val($database, 'type')) || die "$database doesn't have 'type' defined"; |