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 |
|
delete $cache->{index_data}; |
120 |
|
delete $cache->{index_delimiter}; |
121 |
|
my @page_fields; # names of fields |
122 |
|
|
123 |
|
|
124 |
|
# subs used to produce output |
125 |
|
|
126 |
|
sub get_field_name($$$) { |
127 |
|
my ($config,$field,$field_usage) = @_; |
128 |
|
|
129 |
|
# find field name (signular, plural) |
130 |
|
my $field_name = ""; |
131 |
|
if ($config->{indexer}->{$field}->{name_singular} && $field_usage == 1) { |
132 |
|
$field_name = $config->{indexer}->{$field}->{name_singular}; |
133 |
|
} elsif ($config->{indexer}->{$field}->{name_plural}) { |
134 |
|
$field_name = $config->{indexer}->{$field}->{name_plural}; |
135 |
|
} elsif ($config->{indexer}->{$field}->{name}) { |
136 |
|
$field_name = $config->{indexer}->{$field}->{name}; |
137 |
|
} else { |
138 |
|
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
139 |
|
} |
140 |
|
if ($field_name) { |
141 |
|
return x($field_name); |
142 |
|
} |
143 |
|
} |
144 |
|
|
145 |
|
|
146 |
|
# begin real work: go field by field |
147 |
foreach my $field (@sorted_tags) { |
foreach my $field (@sorted_tags) { |
148 |
|
|
149 |
$field=x($field); |
$field=x($field); |
152 |
my $swish_data = ""; |
my $swish_data = ""; |
153 |
my $swish_exact_data = ""; |
my $swish_exact_data = ""; |
154 |
my $display_data = ""; |
my $display_data = ""; |
155 |
|
my @index_data; |
156 |
my $line_delimiter; |
my $line_delimiter; |
157 |
|
|
158 |
my ($swish,$display); |
my ($swish,$display); |
159 |
|
|
160 |
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"; |
161 |
|
|
162 |
|
# is this field page-by-page? |
163 |
|
my $iterate_by_page = $config->{indexer}->{$field}->{iterate_by_page}; |
164 |
|
push @page_fields,$field if ($iterate_by_page); |
165 |
|
my %page_max = (); |
166 |
|
# default line_delimiter if using |
167 |
|
my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>'; |
168 |
|
$cache->{index_delimiter}->{$field} = $config->{indexer}->{$field}->{index_delimiter}; |
169 |
|
|
170 |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
171 |
|
|
172 |
my $format = x($x->{content}); |
my $format = x($x->{content}); |
173 |
my $delimiter = x($x->{delimiter}) || ' '; |
my $delimiter = x($x->{delimiter}) || ' '; |
174 |
|
|
175 |
my $repeat_off = 0; # repeatable offset |
my $repeat_off = 0; # init repeatable offset |
176 |
|
|
177 |
# swish, swish_exact, display, index, index_lookup |
# swish, swish_exact, display, index, index_lookup |
178 |
# swish and display defaults |
# swish and display defaults |
179 |
my ($s,$se,$d,$i,$il) = (1,0,1,0,0); |
my ($s,$se,$d,$i,$il) = (1,0,1,0,0); |
180 |
$s = 0 if (lc($x->{type}) eq "display"); |
$s = 0 if (lc($x->{type}) eq "display"); |
181 |
$d = 0 if (lc($x->{type}) eq "swish"); |
$d = 0 if (lc($x->{type}) eq "swish"); |
182 |
$se = 1 if (lc($x->{type}) eq "swish_exact"); |
($s,$se,$d,$i) = (0,1,0,1) if (lc($x->{type}) eq "index"); |
183 |
($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"); |
184 |
$il = 1 if (lc($x->{type}) =~ /^lookup/); |
($s,$se,$d,$i,$il) = (0,1,0,0,1) if (lc($x->{type}) =~ /^lookup/); |
|
|
|
185 |
|
|
186 |
# what will separate last line from this one? |
# what will separate last line from this one? |
187 |
if ($display_data && $x->{append} && $x->{append} eq "1") { |
if ($display_data && $x->{append}) { |
188 |
$line_delimiter = ' '; |
$line_delimiter = ' '; |
189 |
} elsif ($display_data) { |
} elsif ($display_data) { |
190 |
$line_delimiter = '<br/>'; |
$line_delimiter = '<br/>'; |
194 |
($swish,$display) = (1,1); |
($swish,$display) = (1,1); |
195 |
|
|
196 |
# placeholder for all repeatable entries for index |
# placeholder for all repeatable entries for index |
|
my @index_data; |
|
197 |
|
|
198 |
sub mkformat { |
sub mkformat { |
199 |
my $x = shift || die "mkformat needs tag reference"; |
my $x = shift || die "mkformat needs tag reference"; |
223 |
|
|
224 |
# while because of repeatable fields |
# while because of repeatable fields |
225 |
while ($swish || $display) { |
while ($swish || $display) { |
226 |
|
my $page = $repeat_off; |
227 |
|
$page_max{$field} = $page if ($iterate_by_page && $page > ($page_max{$field} || 0)); |
228 |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
229 |
if ($repeat_off > 1000) { |
if ($repeat_off > 1000) { |
230 |
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"; |
233 |
|
|
234 |
# is this field is lookup? |
# is this field is lookup? |
235 |
if ($display && $x->{lookup}) { |
if ($display && $x->{lookup}) { |
236 |
|
my $null = "<!-- null -->"; |
237 |
if ($use_lhash_cache) { |
if ($use_lhash_cache) { |
238 |
if (!defined($cache->{lhash}->{$display})) { |
if (!defined($cache->{lhash}->{$display})) { |
239 |
my $new_display = $lhash{$display}; |
my $new_display = $lhash{$display}; |
240 |
if ($new_display) { |
if (defined($new_display)) { |
241 |
#print STDERR "lookup cache store '$display' = '$new_display'\n"; |
#print STDERR "lookup cache store '$display' = '$new_display'\n"; |
242 |
$display = $new_display; |
$display = $new_display; |
243 |
$cache->{lhash}->{$display} = $new_display; |
$cache->{lhash}->{$display} = $new_display; |
244 |
} else { |
} else { |
245 |
print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
246 |
$display = ""; |
$display = ""; |
247 |
$cache->{lhash}->{$display} = ""; |
$cache->{lhash}->{$display} = $null; |
248 |
} |
} |
249 |
} else { |
} else { |
250 |
$display = $cache->{lhash}->{$display}; |
$display = $cache->{lhash}->{$display}; |
251 |
} |
} |
252 |
} else { |
} else { |
253 |
$display = $lhash{$display} || ""; |
$display = $lhash{$display} || $null; |
254 |
} |
} |
255 |
} |
} |
256 |
|
|
279 |
if ($d && $display) { |
if ($d && $display) { |
280 |
if ($line_delimiter && $display_data) { |
if ($line_delimiter && $display_data) { |
281 |
$display_data .= $line_delimiter; |
$display_data .= $line_delimiter; |
|
undef $line_delimiter; |
|
282 |
} |
} |
283 |
if ($filter) { |
if ($filter) { |
284 |
no strict 'refs'; |
no strict 'refs'; |
297 |
} |
} |
298 |
|
|
299 |
# type="index" ; insert into index |
# type="index" ; insert into index |
300 |
|
my $idisplay; |
301 |
if ($i && $display) { |
if ($i && $display) { |
302 |
|
$idisplay = $display; |
303 |
if ($filter) { |
if ($filter) { |
304 |
no strict 'refs'; |
no strict 'refs'; |
305 |
$display = &$filter($display); |
$idisplay = &$filter($idisplay); |
|
} |
|
|
if ($x->{append} && @index_data) { |
|
|
$index_data[$#index_data].=$display; |
|
|
} else { |
|
|
push @index_data, $display; |
|
306 |
} |
} |
307 |
|
push @index_data, $idisplay if (! $iterate_by_page); |
308 |
} |
} |
309 |
|
|
310 |
# store fields in lookup |
# store fields in lookup |
322 |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
323 |
} |
} |
324 |
} |
} |
325 |
|
|
326 |
|
} |
327 |
|
|
328 |
|
# store data for page-by-page repeatable fields |
329 |
|
if ($iterate_by_page) { |
330 |
|
sub iterate_fld($$$$$$) { |
331 |
|
my ($cache,$what,$field,$page,$data,$append) = @_; |
332 |
|
return if (!$data); |
333 |
|
|
334 |
|
my $ldel = $page_line_delimiter; |
335 |
|
$ldel = " " if ($append); |
336 |
|
#print STDERR "line delimiter: ",Dumper($ldel) if ($ldel); |
337 |
|
if (! $cache->{$what}->{$field}->[$page]) { |
338 |
|
$cache->{$what}->{$field}->[$page] = $data; |
339 |
|
} else { |
340 |
|
$cache->{$what}->{$field}->[$page] .= $ldel.$data; |
341 |
|
} |
342 |
|
} |
343 |
|
|
344 |
|
if ($display_data) { |
345 |
|
iterate_fld($cache,'display_data',$field,$page,$display_data,$x->{append}); |
346 |
|
} |
347 |
|
$display_data = ""; |
348 |
|
if ($swish_data) { |
349 |
|
iterate_fld($cache,'swish_data',$field,$page,$swish_data,$x->{append}); |
350 |
|
$swish_data = ""; |
351 |
|
} |
352 |
|
if ($swish_exact_data) { |
353 |
|
iterate_fld($cache,'swish_exact_data',$field,$page,$swish_exact_data,$x->{append}); |
354 |
|
$swish_exact_data = ""; |
355 |
|
} |
356 |
|
|
357 |
|
if ($idisplay) { |
358 |
|
my $ldel=$page_line_delimiter; |
359 |
|
my @index_data; |
360 |
|
if ($cache->{index_data}->{$field}->[$page]) { |
361 |
|
|
362 |
|
@index_data = @{$cache->{index_data}->{$field}->[$page]}; |
363 |
|
} |
364 |
|
if ($x->{append}) { |
365 |
|
$index_data[$#index_data] .= $idisplay; |
366 |
|
} else { |
367 |
|
push @index_data, $idisplay; |
368 |
|
} |
369 |
|
$idisplay = ""; |
370 |
|
@{$cache->{index_data}->{$field}->[$page]} = @index_data; |
371 |
|
} |
372 |
} |
} |
373 |
} |
} |
374 |
|
|
375 |
# fill data in index |
if (! $iterate_by_page) { |
376 |
foreach my $d (@index_data) { |
my $idel = $x->{index_delimiter}; |
377 |
$index->insert($field, $d, $path); |
# fill data in index |
378 |
|
foreach my $tmp (@index_data) { |
379 |
|
my $i = $d = $tmp; |
380 |
|
if ($idel && $tmp =~ m/$idel/) { |
381 |
|
($i,$d) = split(/$idel/,$tmp); |
382 |
|
} |
383 |
|
$index->insert($field, $i, $d, $path); |
384 |
|
} |
385 |
|
@index_data = (); |
386 |
} |
} |
387 |
} |
} |
388 |
|
|
405 |
$index->insert($field, $val, $path) if ($i); |
$index->insert($field, $val, $path) if ($i); |
406 |
} |
} |
407 |
|
|
408 |
|
if ($iterate_by_page) { |
409 |
|
# FIXME data from config tag will appear just |
410 |
|
# on first page!!! |
411 |
|
my $page = 0; |
412 |
|
if ($display_data) { |
413 |
|
$cache->{display_data}->{$field}->[$page] = $display_data; |
414 |
|
$display_data = ""; |
415 |
|
} |
416 |
|
if ($swish_data) { |
417 |
|
$cache->{swish_data}->{$field}->[$page] = $swish_data; |
418 |
|
$swish_data = ""; |
419 |
|
} |
420 |
|
if ($swish_exact_data) { |
421 |
|
$cache->{swish_exact_data}->{$field}->[$page] = $swish_exact_data; |
422 |
|
$swish_exact_data = ""; |
423 |
|
} |
424 |
|
} |
425 |
} |
} |
426 |
|
|
427 |
|
# save data page-by-page |
428 |
|
foreach my $field (@page_fields) { |
429 |
|
my $nr_pages = $page_max{$field} || next; |
430 |
|
#print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n"; |
431 |
|
#print STDERR Dumper($cache->{display_data}); |
432 |
|
for (my $page=0; $page <= $nr_pages; $page++) { |
433 |
|
|
434 |
|
my $display_data = $cache->{display_data}->{$field}->[$page]; |
435 |
|
if ($display_data) { # default |
436 |
|
if ($field eq "headline") { |
437 |
|
$xml .= xmlify("headline", $display_data); |
438 |
|
} else { |
439 |
|
|
440 |
if ($display_data) { |
# fallback to empty field name if needed |
441 |
|
$html .= get_field_name($config,$field,$field_usage{$field}) || ''; |
442 |
if ($field eq "headline") { |
$html .= "#-#".$display_data."###\n"; |
443 |
$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 '$field' doesn't have 'name' attribute!"; |
|
444 |
} |
} |
445 |
if ($field_name) { |
|
446 |
$html .= x($field_name); |
my $swish_data = $cache->{swish_data}->{$field}->[$page]; |
447 |
|
if ($swish_data) { |
448 |
|
# remove extra spaces |
449 |
|
$swish_data =~ s/ +/ /g; |
450 |
|
$swish_data =~ s/ +$//g; |
451 |
|
|
452 |
|
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
453 |
|
} |
454 |
|
|
455 |
|
my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page]; |
456 |
|
if ($swish_exact_data) { |
457 |
|
$swish_exact_data =~ s/ +/ /g; |
458 |
|
$swish_exact_data =~ s/ +$//g; |
459 |
|
|
460 |
|
# add delimiters before and after word. |
461 |
|
# That is required to produce exact match |
462 |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
463 |
|
} |
464 |
|
|
465 |
|
my $idel = $cache->{index_delimiter}->{$field}; |
466 |
|
foreach my $tmp (@{$cache->{index_data}->{$field}->[$page]}) { |
467 |
|
my $i = $tmp; |
468 |
|
my $d = $tmp; |
469 |
|
if ($idel && $tmp =~ m/$idel/) { |
470 |
|
($i,$d) = split(/$idel/,$tmp); |
471 |
|
} |
472 |
|
$index->insert($field, $i, $d, $path); |
473 |
|
#print STDERR "index [$idel] $field: $i --> $d [$path]\n"; |
474 |
} |
} |
|
$html .= $display_data."###\n"; |
|
475 |
} |
} |
476 |
|
|
477 |
} |
} |
478 |
if ($swish_data) { |
|
479 |
# remove extra spaces |
if (! $iterate_by_page) { |
480 |
$swish_data =~ s/ +/ /g; |
if ($display_data) { |
481 |
$swish_data =~ s/ +$//g; |
if ($field eq "headline") { |
482 |
|
$xml .= xmlify("headline", $display_data); |
483 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
} else { |
|
} |
|
484 |
|
|
485 |
if ($swish_exact_data) { |
# fallback to empty field name if needed |
486 |
$swish_exact_data =~ s/ +/ /g; |
$html .= get_field_name($config,$field,$field_usage{$field}) || ''; |
487 |
$swish_exact_data =~ s/ +$//g; |
$html .= "#-#".$display_data."###\n"; |
488 |
|
} |
489 |
# add delimiters before and after word. |
} |
490 |
# That is required to produce exact match |
if ($swish_data) { |
491 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
# remove extra spaces |
492 |
} |
$swish_data =~ s/ +/ /g; |
493 |
|
$swish_data =~ s/ +$//g; |
494 |
|
|
495 |
|
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
496 |
|
} |
497 |
|
|
498 |
|
if ($swish_exact_data) { |
499 |
|
$swish_exact_data =~ s/ +/ /g; |
500 |
|
$swish_exact_data =~ s/ +$//g; |
501 |
|
|
502 |
|
# add delimiters before and after word. |
503 |
|
# That is required to produce exact match |
504 |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
505 |
|
} |
506 |
|
} |
507 |
} |
} |
508 |
|
|
509 |
# dump formatted output in <html> |
# dump formatted output in <html> |
602 |
# now read database |
# now read database |
603 |
print STDERR "using: $type...\n"; |
print STDERR "using: $type...\n"; |
604 |
|
|
605 |
|
# erase cache for tags by order in this database |
606 |
|
delete $cache->{tags_by_order}; |
607 |
|
|
608 |
if ($type_base eq "isis") { |
if ($type_base eq "isis") { |
609 |
|
|
610 |
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!"; |