102 |
} |
} |
103 |
|
|
104 |
my @sorted_tags; |
my @sorted_tags; |
105 |
if ($cache->{tags_by_order}->{$type}) { |
if ($cache->{tags_by_order}) { |
106 |
@sorted_tags = @{$cache->{tags_by_order}->{$type}}; |
@sorted_tags = @{$cache->{tags_by_order}}; |
107 |
} else { |
} else { |
108 |
@sorted_tags = sort by_order keys %{$config->{indexer}}; |
@sorted_tags = sort by_order keys %{$config->{indexer}}; |
109 |
$cache->{tags_by_order}->{$type} = \@sorted_tags; |
$cache->{tags_by_order} = \@sorted_tags; |
110 |
} |
} |
111 |
|
|
112 |
# lookup key |
# lookup key |
113 |
my $lookup_key; |
my $lookup_key; |
114 |
|
|
115 |
|
# cache for field in pages |
116 |
|
delete $cache->{display_data}; |
117 |
|
delete $cache->{swish_data}; |
118 |
|
delete $cache->{swish_exact_data}; |
119 |
|
my @page_fields; # names of fields |
120 |
|
|
121 |
|
|
122 |
|
# subs used to produce output |
123 |
|
|
124 |
|
sub get_field_name($$$) { |
125 |
|
my ($config,$field,$field_usage) = @_; |
126 |
|
|
127 |
|
# find field name (signular, plural) |
128 |
|
my $field_name = ""; |
129 |
|
if ($config->{indexer}->{$field}->{name_singular} && $field_usage == 1) { |
130 |
|
$field_name = $config->{indexer}->{$field}->{name_singular}; |
131 |
|
} elsif ($config->{indexer}->{$field}->{name_plural}) { |
132 |
|
$field_name = $config->{indexer}->{$field}->{name_plural}; |
133 |
|
} elsif ($config->{indexer}->{$field}->{name}) { |
134 |
|
$field_name = $config->{indexer}->{$field}->{name}; |
135 |
|
} else { |
136 |
|
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
137 |
|
} |
138 |
|
if ($field_name) { |
139 |
|
return x($field_name); |
140 |
|
} |
141 |
|
} |
142 |
|
|
143 |
|
|
144 |
|
# begin real work: go field by field |
145 |
foreach my $field (@sorted_tags) { |
foreach my $field (@sorted_tags) { |
146 |
|
|
147 |
$field=x($field); |
$field=x($field); |
155 |
my ($swish,$display); |
my ($swish,$display); |
156 |
|
|
157 |
my $tag = $type2tag{$type} || die "can't find which tag to use for type $type"; |
my $tag = $type2tag{$type} || die "can't find which tag to use for type $type"; |
158 |
|
|
159 |
|
# is this field page-by-page? |
160 |
|
my $iterate_by_page = $config->{indexer}->{$field}->{iterate_by_page}; |
161 |
|
push @page_fields,$field if ($iterate_by_page); |
162 |
|
my %page_max = (); |
163 |
|
# default line_delimiter if using |
164 |
|
my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>'; |
165 |
|
|
166 |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
167 |
|
|
168 |
my $format = x($x->{content}); |
my $format = x($x->{content}); |
169 |
my $delimiter = x($x->{delimiter}) || ' '; |
my $delimiter = x($x->{delimiter}) || ' '; |
170 |
|
|
171 |
my $repeat_off = 0; # repeatable offset |
my $repeat_off = 0; # init repeatable offset |
172 |
|
|
173 |
# swish, swish_exact, display, index, index_lookup |
# swish, swish_exact, display, index, index_lookup |
174 |
# swish and display defaults |
# swish and display defaults |
175 |
my ($s,$se,$d,$i,$il) = (1,0,1,0,0); |
my ($s,$se,$d,$i,$il) = (1,0,1,0,0); |
176 |
$s = 0 if (lc($x->{type}) eq "display"); |
$s = 0 if (lc($x->{type}) eq "display"); |
177 |
$d = 0 if (lc($x->{type}) eq "swish"); |
$d = 0 if (lc($x->{type}) eq "swish"); |
178 |
$se = 1 if (lc($x->{type}) eq "swish_exact"); |
($s,$se,$d,$i) = (0,0,0,1) if (lc($x->{type}) eq "index"); |
179 |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
($s,$se,$d,$i) = (0,1,0,0) if (lc($x->{type}) eq "swish_exact"); |
180 |
$il = 1 if (lc($x->{type}) =~ /^lookup/); |
$il = 1 if (lc($x->{type}) =~ /^lookup/); |
181 |
|
|
|
|
|
182 |
# what will separate last line from this one? |
# what will separate last line from this one? |
183 |
if ($display_data && $x->{append} && $x->{append} eq "1") { |
if ($display_data && $x->{append} && $x->{append} eq "1") { |
184 |
$line_delimiter = ' '; |
$line_delimiter = ' '; |
220 |
|
|
221 |
# while because of repeatable fields |
# while because of repeatable fields |
222 |
while ($swish || $display) { |
while ($swish || $display) { |
223 |
|
my $page = $repeat_off; |
224 |
|
$page_max{$field} = $page if ($iterate_by_page && $page > ($page_max{$field} || 0)); |
225 |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
226 |
if ($repeat_off > 1000) { |
if ($repeat_off > 1000) { |
227 |
print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n"; |
print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n"; |
230 |
|
|
231 |
# is this field is lookup? |
# is this field is lookup? |
232 |
if ($display && $x->{lookup}) { |
if ($display && $x->{lookup}) { |
233 |
|
my $null = "<!-- null -->"; |
234 |
if ($use_lhash_cache) { |
if ($use_lhash_cache) { |
235 |
if (!defined($cache->{lhash}->{$display})) { |
if (!defined($cache->{lhash}->{$display})) { |
236 |
my $new_display = $lhash{$display}; |
my $new_display = $lhash{$display}; |
237 |
if ($new_display) { |
if (defined($new_display)) { |
238 |
#print STDERR "lookup cache store '$display' = '$new_display'\n"; |
#print STDERR "lookup cache store '$display' = '$new_display'\n"; |
239 |
$display = $new_display; |
$display = $new_display; |
240 |
$cache->{lhash}->{$display} = $new_display; |
$cache->{lhash}->{$display} = $new_display; |
241 |
} else { |
} else { |
242 |
print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
243 |
$display = ""; |
$display = ""; |
244 |
$cache->{lhash}->{$display} = ""; |
$cache->{lhash}->{$display} = $null; |
245 |
} |
} |
246 |
} else { |
} else { |
247 |
$display = $cache->{lhash}->{$display}; |
$display = $cache->{lhash}->{$display}; |
248 |
} |
} |
249 |
} else { |
} else { |
250 |
$display = $lhash{$display} || ""; |
$display = $lhash{$display} || $null; |
251 |
} |
} |
252 |
} |
} |
253 |
|
|
276 |
if ($d && $display) { |
if ($d && $display) { |
277 |
if ($line_delimiter && $display_data) { |
if ($line_delimiter && $display_data) { |
278 |
$display_data .= $line_delimiter; |
$display_data .= $line_delimiter; |
|
undef $line_delimiter; |
|
279 |
} |
} |
280 |
if ($filter) { |
if ($filter) { |
281 |
no strict 'refs'; |
no strict 'refs'; |
321 |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
322 |
} |
} |
323 |
} |
} |
324 |
|
|
325 |
|
} |
326 |
|
|
327 |
|
# store data for page-by-page repeatable fields |
328 |
|
if ($iterate_by_page) { |
329 |
|
sub iterate_fld($$$$$$) { |
330 |
|
my ($cache,$what,$field,$page,$data,$append) = @_; |
331 |
|
return if (!$data); |
332 |
|
my $line_delimiter = $page_line_delimiter; |
333 |
|
$line_delimiter = " " if ($append); |
334 |
|
if (! $cache->{$what}->{$field}->[$page]) { |
335 |
|
$cache->{$what}->{$field}->[$page] = $data; |
336 |
|
} else { |
337 |
|
$cache->{$what}->{$field}->[$page] .= $line_delimiter.$data; |
338 |
|
} |
339 |
|
} |
340 |
|
|
341 |
|
if ($display_data) { |
342 |
|
print STDERR "line delimiter: ",Dumper($line_delimiter) if ($line_delimiter); |
343 |
|
iterate_fld($cache,'display_data',$field,$page,$display_data,$x->{append}); |
344 |
|
} |
345 |
|
$display_data = ""; |
346 |
|
if ($swish_data) { |
347 |
|
iterate_fld($cache,'swish_data',$field,$page,$swish_data,$x->{append}); |
348 |
|
$swish_data = ""; |
349 |
|
} |
350 |
|
if ($swish_exact_data) { |
351 |
|
iterate_fld($cache,'swish_exact_data',$field,$page,$swish_exact_data,$x->{append}); |
352 |
|
$swish_exact_data = ""; |
353 |
|
} |
354 |
} |
} |
355 |
} |
} |
356 |
|
|
379 |
$index->insert($field, $val, $path) if ($i); |
$index->insert($field, $val, $path) if ($i); |
380 |
} |
} |
381 |
|
|
382 |
|
if ($iterate_by_page) { |
383 |
|
# FIXME data from config tag will appear just |
384 |
|
# on first page!!! |
385 |
|
my $page = 0; |
386 |
|
if ($display_data) { |
387 |
|
$cache->{display_data}->{$field}->[$page] = $display_data; |
388 |
|
$display_data = ""; |
389 |
|
} |
390 |
|
if ($swish_data) { |
391 |
|
$cache->{swish_data}->{$field}->[$page] = $swish_data; |
392 |
|
$swish_data = ""; |
393 |
|
} |
394 |
|
if ($swish_exact_data) { |
395 |
|
$cache->{swish_exact_data}->{$field}->[$page] = $swish_exact_data; |
396 |
|
$swish_exact_data = ""; |
397 |
|
} |
398 |
|
} |
399 |
} |
} |
400 |
|
|
401 |
|
# save data page-by-page |
402 |
|
foreach my $field (@page_fields) { |
403 |
|
my $nr_pages = $page_max{$field} || next; |
404 |
|
#print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n"; |
405 |
|
#print STDERR Dumper($cache->{display_data}); |
406 |
|
for (my $page=0; $page <= $nr_pages; $page++) { |
407 |
|
|
408 |
|
my $display_data = $cache->{display_data}->{$field}->[$page]; |
409 |
|
if ($display_data) { # default |
410 |
|
if ($field eq "headline") { |
411 |
|
$xml .= xmlify("headline", $display_data); |
412 |
|
} else { |
413 |
|
|
414 |
if ($display_data) { |
# fallback to empty field name if needed |
415 |
|
$html .= get_field_name($config,$field,$field_usage{$field}) || ''; |
416 |
if ($field eq "headline") { |
$html .= "#-#".$display_data."###\n"; |
417 |
$xml .= xmlify("headline", $display_data); |
} |
|
} else { |
|
|
|
|
|
# find field name (signular, plural) |
|
|
my $field_name = ""; |
|
|
if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) { |
|
|
$field_name = $config->{indexer}->{$field}->{name_singular}."#-#"; |
|
|
} elsif ($config->{indexer}->{$field}->{name_plural}) { |
|
|
$field_name = $config->{indexer}->{$field}->{name_plural}."#-#"; |
|
|
} elsif ($config->{indexer}->{$field}->{name}) { |
|
|
$field_name = $config->{indexer}->{$field}->{name}."#-#"; |
|
|
} else { |
|
|
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
|
418 |
} |
} |
419 |
if ($field_name) { |
|
420 |
$html .= x($field_name); |
my $swish_data = $cache->{swish_data}->{$field}->[$page]; |
421 |
|
if ($swish_data) { |
422 |
|
# remove extra spaces |
423 |
|
$swish_data =~ s/ +/ /g; |
424 |
|
$swish_data =~ s/ +$//g; |
425 |
|
|
426 |
|
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
427 |
|
} |
428 |
|
|
429 |
|
my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page]; |
430 |
|
if ($swish_exact_data) { |
431 |
|
$swish_exact_data =~ s/ +/ /g; |
432 |
|
$swish_exact_data =~ s/ +$//g; |
433 |
|
|
434 |
|
# add delimiters before and after word. |
435 |
|
# That is required to produce exact match |
436 |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
437 |
} |
} |
|
$html .= $display_data."###\n"; |
|
438 |
} |
} |
439 |
|
|
440 |
} |
} |
441 |
if ($swish_data) { |
|
442 |
# remove extra spaces |
if (! $iterate_by_page) { |
443 |
$swish_data =~ s/ +/ /g; |
if ($display_data) { |
444 |
$swish_data =~ s/ +$//g; |
if ($field eq "headline") { |
445 |
|
$xml .= xmlify("headline", $display_data); |
446 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
} else { |
|
} |
|
447 |
|
|
448 |
if ($swish_exact_data) { |
# fallback to empty field name if needed |
449 |
$swish_exact_data =~ s/ +/ /g; |
$html .= get_field_name($config,$field,$field_usage{$field}) || ''; |
450 |
$swish_exact_data =~ s/ +$//g; |
$html .= "#-#".$display_data."###\n"; |
451 |
|
} |
452 |
# add delimiters before and after word. |
} |
453 |
# That is required to produce exact match |
if ($swish_data) { |
454 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
# remove extra spaces |
455 |
} |
$swish_data =~ s/ +/ /g; |
456 |
|
$swish_data =~ s/ +$//g; |
457 |
|
|
458 |
|
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
459 |
|
} |
460 |
|
|
461 |
|
if ($swish_exact_data) { |
462 |
|
$swish_exact_data =~ s/ +/ /g; |
463 |
|
$swish_exact_data =~ s/ +$//g; |
464 |
|
|
465 |
|
# add delimiters before and after word. |
466 |
|
# That is required to produce exact match |
467 |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
468 |
|
} |
469 |
|
} |
470 |
} |
} |
471 |
|
|
472 |
# dump formatted output in <html> |
# dump formatted output in <html> |
565 |
# now read database |
# now read database |
566 |
print STDERR "using: $type...\n"; |
print STDERR "using: $type...\n"; |
567 |
|
|
568 |
|
# erase cache for tags by order in this database |
569 |
|
delete $cache->{tags_by_order}; |
570 |
|
|
571 |
if ($type_base eq "isis") { |
if ($type_base eq "isis") { |
572 |
|
|
573 |
my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; |
my $isis_db = $cfg -> val($database, 'isis_db') || die "$database doesn't have 'isis_db' defined!"; |