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? |
140 |
my @index_data; |
my @index_data; |
141 |
my $index_filter; |
my $index_filter; |
142 |
|
|
143 |
|
sub mkformat { |
144 |
|
my $x = shift || die "mkformat needs tag reference"; |
145 |
|
my $data = shift || return; |
146 |
|
my $format_name = x($x->{format_name}) || return $data; |
147 |
|
my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!"; |
148 |
|
my $format_delimiter = x($x->{format_delimiter}); |
149 |
|
my @data; |
150 |
|
if ($format_delimiter) { |
151 |
|
@data = split(/$format_delimiter/,$data); |
152 |
|
} else { |
153 |
|
push @data,$data; |
154 |
|
} |
155 |
|
|
156 |
|
if ($fmt) { |
157 |
|
my $nr = scalar $fmt =~ s/%s/%s/g; |
158 |
|
if (($#data+1) == $nr) { |
159 |
|
return sprintf($fmt,@data); |
160 |
|
} else { |
161 |
|
print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; |
162 |
|
return $data; |
163 |
|
} |
164 |
|
} else { |
165 |
|
print STDERR "usage of link '$format_name' without defined format (<link> tag)\n"; |
166 |
|
} |
167 |
|
} |
168 |
|
|
169 |
# while because of repeatable fields |
# while because of repeatable fields |
170 |
while ($swish || $display) { |
while ($swish || $display) { |
171 |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
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 |
|
|
203 |
} |
} |
204 |
if ($filter) { |
if ($filter) { |
205 |
no strict 'refs'; |
no strict 'refs'; |
206 |
$display_data .= $delimiter.&$filter($display); |
if ($display_data) { |
207 |
|
$display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display))); |
208 |
|
} else { |
209 |
|
$display_data = join($delimiter,mkformat($x,&$filter($display))); |
210 |
|
} |
211 |
} else { |
} else { |
212 |
$display_data .= $delimiter.$display; |
if ($display_data) { |
213 |
|
$display_data .= $delimiter.mkformat($x,$display); |
214 |
|
} else { |
215 |
|
$display_data = mkformat($x,$display); |
216 |
|
} |
217 |
} |
} |
218 |
} |
} |
219 |
|
|
248 |
my ($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"; |
352 |
my $type_base = $type; |
my $type_base = $type; |
353 |
$type_base =~ s/_.+$//g; |
$type_base =~ s/_.+$//g; |
354 |
|
|
355 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1); |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config', 'format' ], forcecontent => 1); |
356 |
|
|
357 |
# output current progress indicator |
# output current progress indicator |
358 |
my $last_p = 0; |
my $last_p = 0; |