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 $last_field_name; # cache to prevent repeated fields |
76 |
|
|
77 |
sub data2xml { |
sub data2xml { |
78 |
|
|
79 |
use xmlify; |
use xmlify; |
104 |
} |
} |
105 |
|
|
106 |
my @sorted_tags; |
my @sorted_tags; |
107 |
if ($cache->{tags_by_order}->{$type}) { |
if ($cache->{tags_by_order}) { |
108 |
@sorted_tags = @{$cache->{tags_by_order}->{$type}}; |
@sorted_tags = @{$cache->{tags_by_order}}; |
109 |
} else { |
} else { |
110 |
@sorted_tags = sort by_order keys %{$config->{indexer}}; |
@sorted_tags = sort by_order keys %{$config->{indexer}}; |
111 |
$cache->{tags_by_order}->{$type} = \@sorted_tags; |
$cache->{tags_by_order} = \@sorted_tags; |
112 |
} |
} |
113 |
|
|
114 |
# lookup key |
# lookup key |
115 |
my $lookup_key; |
my $lookup_key; |
116 |
|
|
117 |
|
# cache for field in pages |
118 |
|
delete $cache->{display_data}; |
119 |
|
delete $cache->{swish_data}; |
120 |
|
delete $cache->{swish_exact_data}; |
121 |
|
delete $cache->{index_data}; |
122 |
|
delete $cache->{index_delimiter}; |
123 |
|
my @page_fields; # names of fields |
124 |
|
|
125 |
|
|
126 |
|
# subs used to produce output |
127 |
|
|
128 |
|
sub get_field_name($$$) { |
129 |
|
my ($config,$field,$field_usage) = @_; |
130 |
|
|
131 |
|
# find field name (signular, plural) |
132 |
|
my $field_name = ""; |
133 |
|
if ($config->{indexer}->{$field}->{name_singular} && $field_usage == 1) { |
134 |
|
$field_name = $config->{indexer}->{$field}->{name_singular}; |
135 |
|
} elsif ($config->{indexer}->{$field}->{name_plural}) { |
136 |
|
$field_name = $config->{indexer}->{$field}->{name_plural}; |
137 |
|
} elsif ($config->{indexer}->{$field}->{name}) { |
138 |
|
$field_name = $config->{indexer}->{$field}->{name}; |
139 |
|
} else { |
140 |
|
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
141 |
|
} |
142 |
|
if ($field_name) { |
143 |
|
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 |
|
|
153 |
|
|
154 |
|
# begin real work: go field by field |
155 |
foreach my $field (@sorted_tags) { |
foreach my $field (@sorted_tags) { |
156 |
|
|
157 |
$field=x($field); |
$field=x($field); |
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); |
167 |
|
|
168 |
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"; |
169 |
|
|
170 |
|
# is this field page-by-page? |
171 |
|
my $iterate_by_page = $config->{indexer}->{$field}->{iterate_by_page}; |
172 |
|
push @page_fields,$field if ($iterate_by_page); |
173 |
|
my %page_max = (); |
174 |
|
# default line_delimiter if using |
175 |
|
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 |
|
|
187 |
my $format = x($x->{content}); |
my $format = x($x->{content}); |
188 |
my $delimiter = x($x->{delimiter}) || ' '; |
my $delimiter = x($x->{delimiter}) || ' '; |
189 |
|
|
190 |
my $repeat_off = 0; # repeatable offset |
my $repeat_off = 0; # init repeatable offset |
191 |
|
|
192 |
# swish, swish_exact, display, index, index_lookup |
# swish, swish_exact, display, index, index_lookup |
193 |
# swish and display defaults |
# swish and display defaults |
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 chk_eval($) { |
214 |
|
my $data = shift; |
215 |
|
return if (! defined($data)); |
216 |
|
if ($data && $data =~ s/\s*eval{([^}]+)}\s*//) { |
217 |
|
if (eval "$1") { |
218 |
|
return $data; |
219 |
|
} else { |
220 |
|
return undef; |
221 |
|
} |
222 |
|
} else { |
223 |
|
return $data; |
224 |
|
} |
225 |
|
} |
226 |
|
|
227 |
|
sub mkformat($$) { |
228 |
my $x = shift || die "mkformat needs tag reference"; |
my $x = shift || die "mkformat needs tag reference"; |
229 |
my $data = shift || return; |
my $data = shift || return; |
230 |
my $format_name = x($x->{format_name}) || return $data; |
my $format_name = x($x->{format_name}) || return chk_eval($data); |
231 |
my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!"; |
my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!"; |
232 |
my $format_delimiter = x($x->{format_delimiter}); |
my $format_delimiter = x($x->{format_delimiter}); |
233 |
my @data; |
my @data; |
240 |
if ($fmt) { |
if ($fmt) { |
241 |
my $nr = scalar $fmt =~ s/%s/%s/g; |
my $nr = scalar $fmt =~ s/%s/%s/g; |
242 |
if (($#data+1) == $nr) { |
if (($#data+1) == $nr) { |
243 |
return sprintf($fmt,@data); |
return chk_eval(sprintf($fmt,@data)); |
244 |
} else { |
} else { |
245 |
print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; |
#print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; |
246 |
return $data; |
return chk_eval($data); |
247 |
} |
} |
248 |
} else { |
} else { |
249 |
print STDERR "usage of link '$format_name' without defined format (<link> tag)\n"; |
print STDERR "usage of link '$format_name' without defined format (<link> tag)\n"; |
252 |
|
|
253 |
# while because of repeatable fields |
# while because of repeatable fields |
254 |
while ($swish || $display) { |
while ($swish || $display) { |
255 |
|
my $page = $repeat_off; |
256 |
|
$page_max{$field} = $page if ($iterate_by_page && $page > ($page_max{$field} || 0)); |
257 |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
258 |
if ($repeat_off > 1000) { |
if ($repeat_off > 1000) { |
259 |
print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n"; |
print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n"; |
262 |
|
|
263 |
# is this field is lookup? |
# is this field is lookup? |
264 |
if ($display && $x->{lookup}) { |
if ($display && $x->{lookup}) { |
265 |
|
my $null = "<!-- null -->"; |
266 |
if ($use_lhash_cache) { |
if ($use_lhash_cache) { |
267 |
if (!defined($cache->{lhash}->{$display})) { |
if (!defined($cache->{lhash}->{$display})) { |
268 |
my $new_display = $lhash{$display}; |
my $new_display = $lhash{$display}; |
269 |
if ($new_display) { |
if (defined($new_display)) { |
270 |
#print STDERR "lookup cache store '$display' = '$new_display'\n"; |
#print STDERR "lookup cache store '$display' = '$new_display'\n"; |
271 |
$display = $new_display; |
$display = $new_display; |
272 |
$cache->{lhash}->{$display} = $new_display; |
$cache->{lhash}->{$display} = $new_display; |
273 |
} else { |
} else { |
274 |
print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
# print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
275 |
$display = ""; |
$display = ""; |
276 |
$cache->{lhash}->{$display} = ""; |
$cache->{lhash}->{$display} = $null; |
277 |
} |
} |
278 |
} else { |
} else { |
279 |
$display = $cache->{lhash}->{$display}; |
$display = $cache->{lhash}->{$display}; |
280 |
} |
} |
281 |
} else { |
} else { |
282 |
$display = $lhash{$display} || ""; |
$display = $lhash{$display} || $null; |
283 |
} |
} |
284 |
} |
} |
285 |
|
|
292 |
} |
} |
293 |
# type="swish" ; field for swish |
# type="swish" ; field for swish |
294 |
if ($swish) { |
if ($swish) { |
295 |
|
my $tmp; |
296 |
if ($filter && ($s || $se)) { |
if ($filter && ($s || $se)) { |
297 |
no strict 'refs'; |
no strict 'refs'; |
298 |
my $tmp = join(" ",&$filter($swish)) if ($s || $se); |
$tmp = join(" ",&$filter($swish)) if ($s || $se); |
|
$swish_data .= $tmp if ($s); |
|
|
$swish_exact_data .= $tmp if ($se); |
|
|
|
|
|
} else { |
|
|
$swish_data .= $swish if ($s); |
|
|
$swish_exact_data .= $swish if ($se); |
|
299 |
} |
} |
300 |
|
|
301 |
|
$tmp = chk_eval($tmp); |
302 |
|
$swish_data .= $tmp if ($s && $tmp); |
303 |
|
$swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se); |
304 |
} |
} |
305 |
|
|
306 |
# type="display" ; field for display |
# type="display" ; field for display |
307 |
if ($d && $display) { |
if ($d && $display) { |
308 |
|
my $ldel = $delimiter; |
309 |
if ($line_delimiter && $display_data) { |
if ($line_delimiter && $display_data) { |
310 |
$display_data .= $line_delimiter; |
$ldel = $line_delimiter; |
|
undef $line_delimiter; |
|
311 |
} |
} |
312 |
if ($filter) { |
if ($filter) { |
313 |
no strict 'refs'; |
no strict 'refs'; |
314 |
if ($display_data) { |
my @arr; |
315 |
$display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display))); |
foreach my $tmp (&$filter($display)) { |
316 |
} else { |
my $tmp2 = mkformat($x,$tmp); |
317 |
$display_data = join($delimiter,mkformat($x,&$filter($display))); |
push @arr,$tmp2 if ($tmp2); |
318 |
} |
} |
319 |
|
$display_data .= $ldel if ($display_data && @arr); |
320 |
|
$display_data .= join($delimiter,@arr); |
321 |
} else { |
} else { |
322 |
if ($display_data) { |
$display_data .= $ldel if ($display_data); |
323 |
$display_data .= $delimiter.mkformat($x,$display); |
my $tmp = mkformat($x,$display); |
324 |
} else { |
$display_data .= $tmp if ($tmp); |
|
$display_data = mkformat($x,$display); |
|
|
} |
|
325 |
} |
} |
326 |
} |
} |
327 |
|
|
328 |
# type="index" ; insert into index |
# type="index" ; insert into index |
329 |
|
my $idisplay; |
330 |
if ($i && $display) { |
if ($i && $display) { |
331 |
|
$idisplay = $display; |
332 |
if ($filter) { |
if ($filter) { |
333 |
no strict 'refs'; |
no strict 'refs'; |
334 |
$display = &$filter($display); |
$idisplay = &$filter($idisplay); |
|
} |
|
|
if ($x->{append} && @index_data) { |
|
|
$index_data[$#index_data].=$display; |
|
|
} else { |
|
|
push @index_data, $display; |
|
335 |
} |
} |
336 |
|
$idisplay = chk_eval($idisplay); |
337 |
|
push @index_data, $idisplay if ($idisplay && !$iterate_by_page); |
338 |
} |
} |
339 |
|
|
340 |
# store fields in lookup |
# store fields in lookup |
343 |
if ($lookup_key) { |
if ($lookup_key) { |
344 |
print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)"; |
print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)"; |
345 |
} else { |
} else { |
346 |
$lookup_key = $display; |
if ($filter) { |
347 |
|
no strict 'refs'; |
348 |
|
$lookup_key = &$filter($display); |
349 |
|
} else { |
350 |
|
$lookup_key = $display; |
351 |
|
} |
352 |
} |
} |
353 |
} elsif (lc($x->{type}) eq "lookup_val") { |
} elsif (lc($x->{type}) eq "lookup_val") { |
354 |
if ($lookup_key) { |
if ($lookup_key) { |
355 |
$lhash{$lookup_key} = $display; |
if ($filter) { |
356 |
|
no strict 'refs'; |
357 |
|
$lhash{$lookup_key} = &$filter($display); |
358 |
|
} else { |
359 |
|
$lhash{$lookup_key} = $display; |
360 |
|
} |
361 |
} else { |
} else { |
362 |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
363 |
} |
} |
364 |
} |
} |
365 |
|
|
366 |
|
} |
367 |
|
|
368 |
|
# store data for page-by-page repeatable fields |
369 |
|
if ($iterate_by_page) { |
370 |
|
sub iterate_fld($$$$$$) { |
371 |
|
my ($cache,$what,$field,$page,$data,$append) = @_; |
372 |
|
return if (!$data); |
373 |
|
|
374 |
|
my $ldel = $page_line_delimiter; |
375 |
|
$ldel = " " if ($append); |
376 |
|
#print STDERR "line delimiter: ",Dumper($ldel) if ($ldel); |
377 |
|
if (! $cache->{$what}->{$field}->[$page]) { |
378 |
|
$cache->{$what}->{$field}->[$page] = $data; |
379 |
|
} else { |
380 |
|
$cache->{$what}->{$field}->[$page] .= $ldel.$data; |
381 |
|
} |
382 |
|
} |
383 |
|
|
384 |
|
if ($display_data) { |
385 |
|
iterate_fld($cache,'display_data',$field,$page,$display_data,$x->{append}); |
386 |
|
} |
387 |
|
$display_data = ""; |
388 |
|
if ($swish_data) { |
389 |
|
iterate_fld($cache,'swish_data',$field,$page,$swish_data,$x->{append}); |
390 |
|
$swish_data = ""; |
391 |
|
} |
392 |
|
if ($swish_exact_data) { |
393 |
|
iterate_fld($cache,'swish_exact_data',$field,$page,$swish_exact_data,$x->{append}); |
394 |
|
$swish_exact_data = ""; |
395 |
|
} |
396 |
|
|
397 |
|
if ($idisplay) { |
398 |
|
my $ldel=$page_line_delimiter; |
399 |
|
my @index_data; |
400 |
|
if ($cache->{index_data}->{$field}->[$page]) { |
401 |
|
|
402 |
|
@index_data = @{$cache->{index_data}->{$field}->[$page]}; |
403 |
|
} |
404 |
|
if ($x->{append}) { |
405 |
|
if (@index_data) { |
406 |
|
$index_data[$#index_data] .= $idisplay; |
407 |
|
} else { |
408 |
|
push @index_data, $idisplay; |
409 |
|
} |
410 |
|
} else { |
411 |
|
push @index_data, $idisplay; |
412 |
|
} |
413 |
|
$idisplay = ""; |
414 |
|
@{$cache->{index_data}->{$field}->[$page]} = @index_data; |
415 |
|
} |
416 |
} |
} |
417 |
} |
} |
418 |
|
|
419 |
# fill data in index |
if (! $iterate_by_page) { |
420 |
foreach my $d (@index_data) { |
my $idel = $x->{index_delimiter}; |
421 |
$index->insert($field, $d, $path); |
# fill data in index |
422 |
|
foreach my $tmp (@index_data) { |
423 |
|
my $i = $d = $tmp; |
424 |
|
if ($idel && $tmp =~ m/$idel/) { |
425 |
|
($i,$d) = split(/$idel/,$tmp); |
426 |
|
} |
427 |
|
$index->insert($field, $i, $d, $path); |
428 |
|
} |
429 |
|
@index_data = (); |
430 |
} |
} |
431 |
} |
} |
432 |
|
|
449 |
$index->insert($field, $val, $path) if ($i); |
$index->insert($field, $val, $path) if ($i); |
450 |
} |
} |
451 |
|
|
452 |
|
if ($iterate_by_page) { |
453 |
|
# FIXME data from config tag will appear just |
454 |
|
# on first page!!! |
455 |
|
my $page = 0; |
456 |
|
if ($display_data) { |
457 |
|
$cache->{display_data}->{$field}->[$page] = $display_data; |
458 |
|
$display_data = ""; |
459 |
|
} |
460 |
|
if ($swish_data) { |
461 |
|
$cache->{swish_data}->{$field}->[$page] = $swish_data; |
462 |
|
$swish_data = ""; |
463 |
|
} |
464 |
|
if ($swish_exact_data) { |
465 |
|
$cache->{swish_exact_data}->{$field}->[$page] = $swish_exact_data; |
466 |
|
$swish_exact_data = ""; |
467 |
|
} |
468 |
|
} |
469 |
} |
} |
470 |
|
|
471 |
|
# save data page-by-page |
472 |
if ($display_data) { |
foreach my $field (@page_fields) { |
473 |
|
my $nr_pages = $page_max{$field} || next; |
474 |
if ($field eq "headline") { |
#print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n"; |
475 |
$xml .= xmlify("headline", $display_data); |
#print STDERR Dumper($cache->{display_data}); |
476 |
} else { |
for (my $page=0; $page <= $nr_pages; $page++) { |
477 |
|
my $display_data; |
478 |
# find field name (signular, plural) |
if ($cache->{format}->{$field}) { |
479 |
my $field_name = ""; |
my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]); |
480 |
if ($config->{indexer}->{$field}->{name_singular} && $field_usage{$field} == 1) { |
$display_data=$tmp if ($tmp); |
|
$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}."#-#"; |
|
481 |
} else { |
} else { |
482 |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
$display_data = $cache->{display_data}->{$field}->[$page]; |
483 |
|
} |
484 |
|
if ($display_data) { # default |
485 |
|
if ($field eq "headline") { |
486 |
|
$xml .= xmlify("headline", $display_data); |
487 |
|
} else { |
488 |
|
|
489 |
|
# fallback to empty field name if needed |
490 |
|
$html .= get_field_name($config,$field,$field_usage{$field}) || ''; |
491 |
|
$html .= "#-#".$display_data."###\n"; |
492 |
|
} |
493 |
} |
} |
494 |
if ($field_name) { |
|
495 |
$html .= x($field_name); |
my $swish_data = $cache->{swish_data}->{$field}->[$page]; |
496 |
|
if ($swish_data) { |
497 |
|
# remove extra spaces |
498 |
|
$swish_data =~ s/ +/ /g; |
499 |
|
$swish_data =~ s/ +$//g; |
500 |
|
|
501 |
|
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
502 |
|
} |
503 |
|
|
504 |
|
my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page]; |
505 |
|
if ($swish_exact_data) { |
506 |
|
$swish_exact_data =~ s/ +/ /g; |
507 |
|
$swish_exact_data =~ s/ +$//g; |
508 |
|
|
509 |
|
# add delimiters before and after word. |
510 |
|
# That is required to produce exact match |
511 |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
512 |
|
} |
513 |
|
|
514 |
|
my $idel = $cache->{index_delimiter}->{$field}; |
515 |
|
foreach my $tmp (@{$cache->{index_data}->{$field}->[$page]}) { |
516 |
|
my $i = $tmp; |
517 |
|
my $d = $tmp; |
518 |
|
if ($idel && $tmp =~ m/$idel/) { |
519 |
|
($i,$d) = split(/$idel/,$tmp); |
520 |
|
} |
521 |
|
$index->insert($field, $i, $d, $path); |
522 |
|
#print STDERR "index [$idel] $field: $i --> $d [$path]\n"; |
523 |
} |
} |
|
$html .= $display_data."###\n"; |
|
524 |
} |
} |
525 |
|
|
526 |
} |
} |
527 |
if ($swish_data) { |
|
528 |
# remove extra spaces |
if (! $iterate_by_page) { |
529 |
$swish_data =~ s/ +/ /g; |
if ($display_data) { |
530 |
$swish_data =~ s/ +$//g; |
if ($field eq "headline") { |
531 |
|
$xml .= xmlify("headline", $display_data); |
532 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
} else { |
|
} |
|
533 |
|
|
534 |
if ($swish_exact_data) { |
# fallback to empty field name if needed |
535 |
$swish_exact_data =~ s/ +/ /g; |
$html .= get_field_name($config,$field,$field_usage{$field}) || ''; |
536 |
$swish_exact_data =~ s/ +$//g; |
$html .= "#-#".$display_data."###\n"; |
537 |
|
} |
538 |
|
} |
539 |
|
if ($swish_data) { |
540 |
|
# remove extra spaces |
541 |
|
$swish_data =~ s/ +/ /g; |
542 |
|
$swish_data =~ s/ +$//g; |
543 |
|
|
544 |
# add delimiters before and after word. |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
545 |
# That is required to produce exact match |
} |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
|
|
} |
|
546 |
|
|
547 |
|
if ($swish_exact_data) { |
548 |
|
$swish_exact_data =~ s/ +/ /g; |
549 |
|
$swish_exact_data =~ s/ +$//g; |
550 |
|
|
551 |
|
# add delimiters before and after word. |
552 |
|
# That is required to produce exact match |
553 |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
554 |
|
} |
555 |
|
} |
556 |
} |
} |
557 |
|
|
558 |
# dump formatted output in <html> |
# dump formatted output in <html> |
603 |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
604 |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
605 |
print STDERR "creating lookup file '$lookup_file'\n"; |
print STDERR "creating lookup file '$lookup_file'\n"; |
606 |
|
# delete memory cache for lookup file |
607 |
|
delete $cache->{lhash}; |
608 |
} |
} |
609 |
|
|
610 |
# open existing lookup file |
# open existing lookup file |
653 |
# now read database |
# now read database |
654 |
print STDERR "using: $type...\n"; |
print STDERR "using: $type...\n"; |
655 |
|
|
656 |
|
# erase cache for tags by order in this database |
657 |
|
delete $cache->{tags_by_order}; |
658 |
|
|
659 |
if ($type_base eq "isis") { |
if ($type_base eq "isis") { |
660 |
|
|
661 |
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 "$database doesn't have 'isis_db' defined!"; |
741 |
print STDERR "\n"; |
print STDERR "\n"; |
742 |
|
|
743 |
} elsif ($type_base eq "excel") { |
} elsif ($type_base eq "excel") { |
744 |
use Spreadsheet::ParseExcel; |
require Spreadsheet::ParseExcel; |
745 |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
require Spreadsheet::ParseExcel::Utility; |
746 |
|
import Spreadsheet::ParseExcel::Utility qw(int2col); |
747 |
|
|
748 |
$import2cp = Text::Iconv->new($config->{excel_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{excel_codepage},$codepage); |
749 |
my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!"; |
my $excel_file = $cfg -> val($database, 'excel_file') || die "$database doesn't have 'excel_file' defined!"; |
797 |
} |
} |
798 |
} elsif ($type_base eq "marc") { |
} elsif ($type_base eq "marc") { |
799 |
|
|
800 |
use MARC; |
require MARC; |
801 |
|
|
802 |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
803 |
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
my $marc_file = $cfg -> val($database, 'marc_file') || die "$database doesn't have 'marc_file' defined!"; |
804 |
|
|
805 |
# optional argument is format |
# optional argument is format |
806 |
my $format = x($config->{format}) || 'usmarc'; |
my $format = x($config->{marc_format}) || 'usmarc'; |
807 |
|
|
808 |
print STDERR "Reading MARC file '$marc_file'\n"; |
print STDERR "Reading MARC file '$marc_file'\n"; |
809 |
|
|
810 |
my $marc = new MARC; |
my $marc = new MARC; |
811 |
my $nr = $marc->openmarc({ |
my $nr = $marc->openmarc({ |
812 |
file=>$marc_file, format=>$format |
file=>$marc_file, format=>$format |
813 |
}) || die "Can't open MARC file '$marc_file'"; |
}) || die "Can't open MARC file '$marc_file' with format '$format'"; |
814 |
|
|
815 |
my $i=0; # record nr. |
# read MARC file in memory |
816 |
|
$marc->nextmarc(-1); |
817 |
|
|
818 |
my $rec; |
my $max_rec = $marc->marc_count(); |
819 |
|
|
820 |
while ($marc->nextmarc(1)) { |
for(my $i=1; $i<=$max_rec; $i++) { |
821 |
|
|
822 |
# XXX |
progress($i,$max_rec); |
823 |
fakeprogress($i++); |
|
824 |
|
# store value for marc_sf.pm |
825 |
|
$main::cache->{marc_record} = $i; |
826 |
|
|
827 |
my $swishpath = $database."#".$i; |
my $swishpath = $database."#".$i; |
828 |
|
|
834 |
print "Document-Type: XML\n\n$xml\n"; |
print "Document-Type: XML\n\n$xml\n"; |
835 |
} |
} |
836 |
} |
} |
837 |
|
|
838 |
|
print STDERR "\n"; |
839 |
|
|
840 |
} elsif ($type_base eq "feed") { |
} elsif ($type_base eq "feed") { |
841 |
|
|
842 |
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |