72 |
# if you are tight on memory, turn this off |
# if you are tight on memory, turn this off |
73 |
my $use_lhash_cache = 1; |
my $use_lhash_cache = 1; |
74 |
|
|
75 |
|
my $last_field_name; # cache to prevent repeated fields |
76 |
|
|
77 |
sub data2xml { |
sub data2xml { |
78 |
|
|
79 |
use xmlify; |
use xmlify; |
118 |
delete $cache->{display_data}; |
delete $cache->{display_data}; |
119 |
delete $cache->{swish_data}; |
delete $cache->{swish_data}; |
120 |
delete $cache->{swish_exact_data}; |
delete $cache->{swish_exact_data}; |
121 |
|
delete $cache->{index_data}; |
122 |
|
delete $cache->{index_delimiter}; |
123 |
my @page_fields; # names of fields |
my @page_fields; # names of fields |
124 |
|
|
125 |
|
|
140 |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
141 |
} |
} |
142 |
if ($field_name) { |
if ($field_name) { |
143 |
return x($field_name); |
if (! $last_field_name) { |
144 |
|
$last_field_name = x($field_name); |
145 |
|
return $last_field_name; |
146 |
|
} elsif ($field_name ne $last_field_name) { |
147 |
|
$last_field_name = x($field_name); |
148 |
|
return $last_field_name; |
149 |
|
} |
150 |
} |
} |
151 |
} |
} |
152 |
|
|
160 |
my $swish_data = ""; |
my $swish_data = ""; |
161 |
my $swish_exact_data = ""; |
my $swish_exact_data = ""; |
162 |
my $display_data = ""; |
my $display_data = ""; |
163 |
|
my @index_data; |
164 |
my $line_delimiter; |
my $line_delimiter; |
165 |
|
|
166 |
my ($swish,$display); |
my ($swish,$display); |
173 |
my %page_max = (); |
my %page_max = (); |
174 |
# default line_delimiter if using |
# default line_delimiter if using |
175 |
my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>'; |
my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>'; |
176 |
|
$cache->{index_delimiter}->{$field} = $config->{indexer}->{$field}->{index_delimiter}; |
177 |
|
|
178 |
|
my $format_name = $config->{indexer}->{$field}->{format_name}; |
179 |
|
my $format_delimiter = $config->{indexer}->{$field}->{format_delimiter}; |
180 |
|
if ($format_name && $format_delimiter) { |
181 |
|
$cache->{format}->{$field}->{format_name} = $format_name; |
182 |
|
$cache->{format}->{$field}->{format_delimiter} = $format_delimiter; |
183 |
|
} |
184 |
|
|
185 |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
186 |
|
|
194 |
my ($s,$se,$d,$i,$il) = (1,0,1,0,0); |
my ($s,$se,$d,$i,$il) = (1,0,1,0,0); |
195 |
$s = 0 if (lc($x->{type}) eq "display"); |
$s = 0 if (lc($x->{type}) eq "display"); |
196 |
$d = 0 if (lc($x->{type}) eq "swish"); |
$d = 0 if (lc($x->{type}) eq "swish"); |
197 |
$se = 1 if (lc($x->{type}) eq "swish_exact"); |
($s,$se,$d,$i) = (0,1,0,1) if (lc($x->{type}) eq "index"); |
198 |
($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"); |
199 |
$il = 1 if (lc($x->{type}) =~ /^lookup/); |
($s,$se,$d,$i,$il) = (0,1,0,0,1) if (lc($x->{type}) =~ /^lookup/); |
|
|
|
200 |
|
|
201 |
# what will separate last line from this one? |
# what will separate last line from this one? |
202 |
if ($display_data && $x->{append} && $x->{append} eq "1") { |
if ($display_data && $x->{append}) { |
203 |
$line_delimiter = ' '; |
$line_delimiter = ' '; |
204 |
} elsif ($display_data) { |
} elsif ($display_data) { |
205 |
$line_delimiter = '<br/>'; |
$line_delimiter = '<br/>'; |
209 |
($swish,$display) = (1,1); |
($swish,$display) = (1,1); |
210 |
|
|
211 |
# placeholder for all repeatable entries for index |
# placeholder for all repeatable entries for index |
|
my @index_data; |
|
212 |
|
|
213 |
sub mkformat { |
sub mkformat($$) { |
214 |
my $x = shift || die "mkformat needs tag reference"; |
my $x = shift || die "mkformat needs tag reference"; |
215 |
my $data = shift || return; |
my $data = shift || return; |
216 |
my $format_name = x($x->{format_name}) || return $data; |
my $format_name = x($x->{format_name}) || return $data; |
282 |
no strict 'refs'; |
no strict 'refs'; |
283 |
my $tmp = join(" ",&$filter($swish)) if ($s || $se); |
my $tmp = join(" ",&$filter($swish)) if ($s || $se); |
284 |
$swish_data .= $tmp if ($s); |
$swish_data .= $tmp if ($s); |
285 |
$swish_exact_data .= $tmp if ($se); |
$swish_exact_data .= "xxbxx $tmp xxexx " if ($se && $tmp ne ""); |
286 |
|
|
287 |
} else { |
} else { |
288 |
$swish_data .= $swish if ($s); |
$swish_data .= $swish if ($s); |
289 |
$swish_exact_data .= $swish if ($se); |
$swish_exact_data .= "xxbxx $swish xxexx " if ($se && $swish ne ""); |
290 |
} |
} |
291 |
} |
} |
292 |
|
|
298 |
if ($filter) { |
if ($filter) { |
299 |
no strict 'refs'; |
no strict 'refs'; |
300 |
if ($display_data) { |
if ($display_data) { |
301 |
$display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display))); |
$display_data .= $delimiter.mkformat($x,join($delimiter,&$filter($display))); |
302 |
} else { |
} else { |
303 |
$display_data = join($delimiter,mkformat($x,&$filter($display))); |
$display_data = mkformat($x,join($delimiter,&$filter($display))); |
304 |
} |
} |
305 |
} else { |
} else { |
306 |
if ($display_data) { |
if ($display_data) { |
312 |
} |
} |
313 |
|
|
314 |
# type="index" ; insert into index |
# type="index" ; insert into index |
315 |
|
my $idisplay; |
316 |
if ($i && $display) { |
if ($i && $display) { |
317 |
|
$idisplay = $display; |
318 |
if ($filter) { |
if ($filter) { |
319 |
no strict 'refs'; |
no strict 'refs'; |
320 |
$display = &$filter($display); |
$idisplay = &$filter($idisplay); |
|
} |
|
|
if ($x->{append} && @index_data) { |
|
|
$index_data[$#index_data].=$display; |
|
|
} else { |
|
|
push @index_data, $display; |
|
321 |
} |
} |
322 |
|
push @index_data, $idisplay if (! $iterate_by_page); |
323 |
} |
} |
324 |
|
|
325 |
# store fields in lookup |
# store fields in lookup |
345 |
sub iterate_fld($$$$$$) { |
sub iterate_fld($$$$$$) { |
346 |
my ($cache,$what,$field,$page,$data,$append) = @_; |
my ($cache,$what,$field,$page,$data,$append) = @_; |
347 |
return if (!$data); |
return if (!$data); |
348 |
my $line_delimiter = $page_line_delimiter; |
|
349 |
$line_delimiter = " " if ($append); |
my $ldel = $page_line_delimiter; |
350 |
|
$ldel = " " if ($append); |
351 |
|
#print STDERR "line delimiter: ",Dumper($ldel) if ($ldel); |
352 |
if (! $cache->{$what}->{$field}->[$page]) { |
if (! $cache->{$what}->{$field}->[$page]) { |
353 |
$cache->{$what}->{$field}->[$page] = $data; |
$cache->{$what}->{$field}->[$page] = $data; |
354 |
} else { |
} else { |
355 |
$cache->{$what}->{$field}->[$page] .= $line_delimiter.$data; |
$cache->{$what}->{$field}->[$page] .= $ldel.$data; |
356 |
} |
} |
357 |
} |
} |
358 |
|
|
359 |
if ($display_data) { |
if ($display_data) { |
|
print STDERR "line delimiter: ",Dumper($line_delimiter) if ($line_delimiter); |
|
360 |
iterate_fld($cache,'display_data',$field,$page,$display_data,$x->{append}); |
iterate_fld($cache,'display_data',$field,$page,$display_data,$x->{append}); |
361 |
} |
} |
362 |
$display_data = ""; |
$display_data = ""; |
368 |
iterate_fld($cache,'swish_exact_data',$field,$page,$swish_exact_data,$x->{append}); |
iterate_fld($cache,'swish_exact_data',$field,$page,$swish_exact_data,$x->{append}); |
369 |
$swish_exact_data = ""; |
$swish_exact_data = ""; |
370 |
} |
} |
371 |
|
|
372 |
|
if ($idisplay) { |
373 |
|
my $ldel=$page_line_delimiter; |
374 |
|
my @index_data; |
375 |
|
if ($cache->{index_data}->{$field}->[$page]) { |
376 |
|
|
377 |
|
@index_data = @{$cache->{index_data}->{$field}->[$page]}; |
378 |
|
} |
379 |
|
if ($x->{append}) { |
380 |
|
if (@index_data) { |
381 |
|
$index_data[$#index_data] .= $idisplay; |
382 |
|
} else { |
383 |
|
push @index_data, $idisplay; |
384 |
|
} |
385 |
|
} else { |
386 |
|
push @index_data, $idisplay; |
387 |
|
} |
388 |
|
$idisplay = ""; |
389 |
|
@{$cache->{index_data}->{$field}->[$page]} = @index_data; |
390 |
|
} |
391 |
} |
} |
392 |
} |
} |
393 |
|
|
394 |
# fill data in index |
if (! $iterate_by_page) { |
395 |
foreach my $d (@index_data) { |
my $idel = $x->{index_delimiter}; |
396 |
$index->insert($field, $d, $path); |
# fill data in index |
397 |
|
foreach my $tmp (@index_data) { |
398 |
|
my $i = $d = $tmp; |
399 |
|
if ($idel && $tmp =~ m/$idel/) { |
400 |
|
($i,$d) = split(/$idel/,$tmp); |
401 |
|
} |
402 |
|
$index->insert($field, $i, $d, $path); |
403 |
|
} |
404 |
|
@index_data = (); |
405 |
} |
} |
406 |
} |
} |
407 |
|
|
449 |
#print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n"; |
#print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n"; |
450 |
#print STDERR Dumper($cache->{display_data}); |
#print STDERR Dumper($cache->{display_data}); |
451 |
for (my $page=0; $page <= $nr_pages; $page++) { |
for (my $page=0; $page <= $nr_pages; $page++) { |
452 |
|
my $display_data; |
453 |
my $display_data = $cache->{display_data}->{$field}->[$page]; |
if ($cache->{format}->{$field}) { |
454 |
|
$display_data=mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]); |
455 |
|
} else { |
456 |
|
$display_data = $cache->{display_data}->{$field}->[$page]; |
457 |
|
} |
458 |
if ($display_data) { # default |
if ($display_data) { # default |
459 |
if ($field eq "headline") { |
if ($field eq "headline") { |
460 |
$xml .= xmlify("headline", $display_data); |
$xml .= xmlify("headline", $display_data); |
482 |
|
|
483 |
# add delimiters before and after word. |
# add delimiters before and after word. |
484 |
# That is required to produce exact match |
# That is required to produce exact match |
485 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
486 |
|
} |
487 |
|
|
488 |
|
my $idel = $cache->{index_delimiter}->{$field}; |
489 |
|
foreach my $tmp (@{$cache->{index_data}->{$field}->[$page]}) { |
490 |
|
my $i = $tmp; |
491 |
|
my $d = $tmp; |
492 |
|
if ($idel && $tmp =~ m/$idel/) { |
493 |
|
($i,$d) = split(/$idel/,$tmp); |
494 |
|
} |
495 |
|
$index->insert($field, $i, $d, $path); |
496 |
|
#print STDERR "index [$idel] $field: $i --> $d [$path]\n"; |
497 |
} |
} |
498 |
} |
} |
499 |
|
|
524 |
|
|
525 |
# add delimiters before and after word. |
# add delimiters before and after word. |
526 |
# That is required to produce exact match |
# That is required to produce exact match |
527 |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
528 |
} |
} |
529 |
} |
} |
530 |
} |
} |