9 |
use Text::Iconv; |
use Text::Iconv; |
10 |
use Config::IniFiles; |
use Config::IniFiles; |
11 |
use Encode; |
use Encode; |
12 |
|
#use GDBM_File; |
13 |
|
use Fcntl; # for O_RDWR |
14 |
|
use TDB_File; |
15 |
|
|
16 |
$|=1; |
$|=1; |
17 |
|
|
64 |
'feed' => 'feed' |
'feed' => 'feed' |
65 |
); |
); |
66 |
|
|
67 |
|
my $cache; # for cacheing |
68 |
|
|
69 |
|
# lookup hash (tied to file) |
70 |
|
my %lhash; |
71 |
|
# this option will cache all lookup entries in memory. |
72 |
|
# if you are tight on memory, turn this off |
73 |
|
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; |
103 |
return $va <=> $vb; |
return $va <=> $vb; |
104 |
} |
} |
105 |
|
|
106 |
foreach my $field (sort by_order keys %{$config->{indexer}}) { |
my @sorted_tags; |
107 |
|
if ($cache->{tags_by_order}) { |
108 |
|
@sorted_tags = @{$cache->{tags_by_order}}; |
109 |
|
} else { |
110 |
|
@sorted_tags = sort by_order keys %{$config->{indexer}}; |
111 |
|
$cache->{tags_by_order} = \@sorted_tags; |
112 |
|
} |
113 |
|
|
114 |
|
# lookup key |
115 |
|
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 |
|
sub init_visible_type($) { |
154 |
|
my $type = shift; |
155 |
|
|
156 |
|
# swish, swish_exact, display, index, index_lookup |
157 |
|
# swish and display defaults |
158 |
|
my ($s,$se,$d,$i,$il) = (1,0,1,0,0); |
159 |
|
if (lc($type) eq "display") { |
160 |
|
$s = 0; |
161 |
|
} elsif (lc($type) eq "swish") { |
162 |
|
$d = 0; |
163 |
|
} elsif (lc($type) eq "index") { |
164 |
|
($s,$se,$d,$i) = (0,1,0,1); |
165 |
|
} elsif (lc($type) eq "swish_exact") { |
166 |
|
($s,$se,$d,$i) = (0,1,0,0); |
167 |
|
} elsif (lc($type) =~ /^lookup/) { |
168 |
|
($s,$se,$d,$i,$il) = (0,1,0,0,1); |
169 |
|
} |
170 |
|
return ($s,$se,$d,$i,$il); |
171 |
|
} |
172 |
|
|
173 |
|
# begin real work: go field by field |
174 |
|
foreach my $field (@sorted_tags) { |
175 |
|
|
176 |
$field=x($field); |
$field=x($field); |
177 |
$field_usage{$field}++; |
$field_usage{$field}++; |
179 |
my $swish_data = ""; |
my $swish_data = ""; |
180 |
my $swish_exact_data = ""; |
my $swish_exact_data = ""; |
181 |
my $display_data = ""; |
my $display_data = ""; |
182 |
|
my @index_data; |
183 |
my $line_delimiter; |
my $line_delimiter; |
184 |
|
|
185 |
my ($swish,$display); |
my ($swish,$display); |
186 |
|
|
187 |
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"; |
188 |
|
|
189 |
|
# is this field page-by-page? |
190 |
|
my $iterate_by_page = $config->{indexer}->{$field}->{iterate_by_page}; |
191 |
|
push @page_fields,$field if ($iterate_by_page); |
192 |
|
my %page_max = (); |
193 |
|
# default line_delimiter if using |
194 |
|
my $page_line_delimiter = $config->{indexer}->{$field}->{page_line_delimiter} || '<br/>'; |
195 |
|
$cache->{index_delimiter}->{$field} = $config->{indexer}->{$field}->{index_delimiter}; |
196 |
|
|
197 |
|
my $format_name = $config->{indexer}->{$field}->{format_name}; |
198 |
|
my $format_delimiter = $config->{indexer}->{$field}->{format_delimiter}; |
199 |
|
if ($format_name && $format_delimiter) { |
200 |
|
$cache->{format}->{$field}->{format_name} = $format_name; |
201 |
|
$cache->{format}->{$field}->{format_delimiter} = $format_delimiter; |
202 |
|
} |
203 |
|
|
204 |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
foreach my $x (@{$config->{indexer}->{$field}->{$tag}}) { |
205 |
|
|
206 |
my $format = x($x->{content}); |
my $format = x($x->{content}); |
207 |
my $delimiter = x($x->{delimiter}) || ' '; |
my $delimiter = x($x->{delimiter}) || ' '; |
208 |
|
|
209 |
my $repeat_off = 0; # repeatable offset |
my $repeat_off = 0; # init repeatable offset |
210 |
|
|
211 |
my ($s,$se,$d,$i) = (1,0,1,0); # swish, display default |
my ($s,$se,$d,$i,$il) = init_visible_type($x->{type}); |
|
$s = 0 if (lc($x->{type}) eq "display"); |
|
|
$d = 0 if (lc($x->{type}) eq "swish"); |
|
|
$se = 1 if (lc($x->{type}) eq "swish_exact"); |
|
|
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
|
212 |
|
|
213 |
# what will separate last line from this one? |
# what will separate last line from this one? |
214 |
if ($display_data && $x->{append} && $x->{append} eq "1") { |
if ($display_data && $x->{append}) { |
215 |
$line_delimiter = ' '; |
$line_delimiter = ' '; |
216 |
} elsif ($display_data) { |
} elsif ($display_data) { |
217 |
$line_delimiter = '<br/>'; |
$line_delimiter = '<br/>'; |
221 |
($swish,$display) = (1,1); |
($swish,$display) = (1,1); |
222 |
|
|
223 |
# placeholder for all repeatable entries for index |
# placeholder for all repeatable entries for index |
|
my @index_data; |
|
|
my $index_filter; |
|
224 |
|
|
225 |
sub mkformat { |
sub mkformat($$) { |
226 |
my $x = shift || die "mkformat needs tag reference"; |
my $x = shift || die "mkformat needs tag reference"; |
227 |
my $data = shift || return; |
my $data = shift || return; |
228 |
my $format_name = x($x->{format_name}) || return $data; |
my $format_name = x($x->{format_name}) || return $data; |
240 |
if (($#data+1) == $nr) { |
if (($#data+1) == $nr) { |
241 |
return sprintf($fmt,@data); |
return sprintf($fmt,@data); |
242 |
} else { |
} else { |
243 |
print STDERR "mkformat: [$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"; |
244 |
return $data; |
return $data; |
245 |
} |
} |
246 |
} else { |
} else { |
250 |
|
|
251 |
# while because of repeatable fields |
# while because of repeatable fields |
252 |
while ($swish || $display) { |
while ($swish || $display) { |
253 |
|
my $page = $repeat_off; |
254 |
|
$page_max{$field} = $page if ($iterate_by_page && $page > ($page_max{$field} || 0)); |
255 |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
($swish,$display) = parse_format($type, $format,$row,$repeat_off++,$import2cp); |
256 |
if ($repeat_off > 1000) { |
if ($repeat_off > 1000) { |
257 |
print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n"; |
print STDERR "loop (more than 1000 repeatable fields) deteced in $row, $format\n"; |
258 |
last; |
last; |
259 |
} |
} |
260 |
|
|
261 |
|
# is this field is lookup? |
262 |
|
if ($display && $x->{lookup}) { |
263 |
|
my $null = "<!-- null -->"; |
264 |
|
if ($use_lhash_cache) { |
265 |
|
if (!defined($cache->{lhash}->{$display})) { |
266 |
|
my $new_display = $lhash{$display}; |
267 |
|
if (defined($new_display)) { |
268 |
|
#print STDERR "lookup cache store '$display' = '$new_display'\n"; |
269 |
|
$display = $new_display; |
270 |
|
$cache->{lhash}->{$display} = $new_display; |
271 |
|
} else { |
272 |
|
# print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
273 |
|
$display = ""; |
274 |
|
$cache->{lhash}->{$display} = $null; |
275 |
|
} |
276 |
|
} else { |
277 |
|
$display = $cache->{lhash}->{$display}; |
278 |
|
} |
279 |
|
} else { |
280 |
|
$display = $lhash{$display} || $null; |
281 |
|
} |
282 |
|
} |
283 |
|
|
284 |
# filter="name" ; filter this field through |
# filter="name" ; filter this field through |
285 |
# filter/[name].pm |
# filter/[name].pm |
286 |
my $filter = $x->{filter}; |
my $filter = $x->{filter}; |
287 |
if ($filter) { |
if ($filter && !$cache->{filter_loaded}->{$filter}) { |
288 |
require "filter/".$filter.".pm"; |
require "filter/".$filter.".pm"; |
289 |
|
$cache->{filter_loaded}->{$filter}++; |
290 |
} |
} |
291 |
# type="swish" ; field for swish |
# type="swish" ; field for swish |
292 |
if ($swish) { |
if ($swish) { |
293 |
|
my $tmp = $swish; |
294 |
if ($filter && ($s || $se)) { |
if ($filter && ($s || $se)) { |
295 |
no strict 'refs'; |
no strict 'refs'; |
296 |
my $tmp = join(" ",&$filter($swish)) if ($s || $se); |
$tmp = join(" ",&$filter($tmp)) if ($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); |
|
297 |
} |
} |
298 |
|
|
299 |
|
$swish_data .= $tmp if ($s && $tmp); |
300 |
|
$swish_exact_data .= "xxbxx $tmp xxexx " if ($tmp && $tmp ne "" && $se); |
301 |
} |
} |
302 |
|
|
303 |
# type="display" ; field for display |
# type="display" ; field for display |
304 |
if ($d && $display) { |
if ($d && $display) { |
305 |
|
my $ldel = $delimiter; |
306 |
if ($line_delimiter && $display_data) { |
if ($line_delimiter && $display_data) { |
307 |
$display_data .= $line_delimiter; |
$ldel = $line_delimiter; |
|
undef $line_delimiter; |
|
308 |
} |
} |
309 |
if ($filter) { |
if ($filter) { |
310 |
no strict 'refs'; |
no strict 'refs'; |
311 |
if ($display_data) { |
my @arr; |
312 |
$display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display))); |
foreach my $tmp (&$filter($display)) { |
313 |
} else { |
my $tmp2 = mkformat($x,$tmp); |
314 |
$display_data = join($delimiter,mkformat($x,&$filter($display))); |
push @arr,$tmp2 if ($tmp2); |
315 |
} |
} |
316 |
|
$display_data .= $ldel if ($display_data && @arr); |
317 |
|
$display_data .= join($delimiter,@arr); |
318 |
} else { |
} else { |
319 |
if ($display_data) { |
$display_data .= $ldel if ($display_data); |
320 |
$display_data .= $delimiter.mkformat($x,$display); |
my $tmp = mkformat($x,$display); |
321 |
} else { |
$display_data .= $tmp if ($tmp); |
|
$display_data = mkformat($x,$display); |
|
|
} |
|
322 |
} |
} |
323 |
} |
} |
324 |
|
|
325 |
# type="index" ; insert into index |
# type="index" ; insert into index |
326 |
|
my $idisplay; |
327 |
if ($i && $display) { |
if ($i && $display) { |
328 |
push @index_data, $display; |
$idisplay = $display; |
329 |
$index_filter = $filter if ($filter); |
if ($filter) { |
330 |
|
no strict 'refs'; |
331 |
|
$idisplay = &$filter($idisplay); |
332 |
|
} |
333 |
|
push @index_data, $idisplay if ($idisplay && !$iterate_by_page); |
334 |
} |
} |
|
} |
|
335 |
|
|
336 |
# fill data in index |
# store fields in lookup |
337 |
if (@index_data) { |
if ($il && $display) { |
338 |
if ($index_filter) { |
if (lc($x->{type}) eq "lookup_key") { |
339 |
no strict 'refs'; |
if ($lookup_key) { |
340 |
foreach my $d (@index_data) { |
print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)"; |
341 |
$index->insert($field, &$index_filter($d), $path); |
} else { |
342 |
|
if ($filter) { |
343 |
|
no strict 'refs'; |
344 |
|
$lookup_key = &$filter($display); |
345 |
|
} else { |
346 |
|
$lookup_key = $display; |
347 |
|
} |
348 |
|
} |
349 |
|
} elsif (lc($x->{type}) eq "lookup_val") { |
350 |
|
if ($lookup_key) { |
351 |
|
if ($filter) { |
352 |
|
no strict 'refs'; |
353 |
|
$lhash{$lookup_key} = &$filter($display); |
354 |
|
} else { |
355 |
|
$lhash{$lookup_key} = $display; |
356 |
|
} |
357 |
|
} else { |
358 |
|
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
359 |
|
} |
360 |
} |
} |
361 |
} else { |
|
362 |
foreach my $d (@index_data) { |
} |
363 |
$index->insert($field, $d, $path); |
|
364 |
|
# store data for page-by-page repeatable fields |
365 |
|
if ($iterate_by_page) { |
366 |
|
sub iterate_fld($$$$$$) { |
367 |
|
my ($cache,$what,$field,$page,$data,$append) = @_; |
368 |
|
return if (!$data); |
369 |
|
|
370 |
|
my $ldel = $page_line_delimiter; |
371 |
|
$ldel = " " if ($append); |
372 |
|
#print STDERR "line delimiter: ",Dumper($ldel) if ($ldel); |
373 |
|
if (! $cache->{$what}->{$field}->[$page]) { |
374 |
|
$cache->{$what}->{$field}->[$page] = $data; |
375 |
|
} else { |
376 |
|
$cache->{$what}->{$field}->[$page] .= $ldel.$data; |
377 |
|
} |
378 |
|
} |
379 |
|
|
380 |
|
if ($display_data) { |
381 |
|
iterate_fld($cache,'display_data',$field,$page,$display_data,$x->{append}); |
382 |
|
} |
383 |
|
$display_data = ""; |
384 |
|
if ($swish_data) { |
385 |
|
iterate_fld($cache,'swish_data',$field,$page,$swish_data,$x->{append}); |
386 |
|
$swish_data = ""; |
387 |
|
} |
388 |
|
if ($swish_exact_data) { |
389 |
|
iterate_fld($cache,'swish_exact_data',$field,$page,$swish_exact_data,$x->{append}); |
390 |
|
$swish_exact_data = ""; |
391 |
|
} |
392 |
|
|
393 |
|
if ($idisplay) { |
394 |
|
my $ldel=$page_line_delimiter; |
395 |
|
my @index_data; |
396 |
|
if ($cache->{index_data}->{$field}->[$page]) { |
397 |
|
|
398 |
|
@index_data = @{$cache->{index_data}->{$field}->[$page]}; |
399 |
|
} |
400 |
|
if ($x->{append}) { |
401 |
|
if (@index_data) { |
402 |
|
$index_data[$#index_data] .= $idisplay; |
403 |
|
} else { |
404 |
|
push @index_data, $idisplay; |
405 |
|
} |
406 |
|
} else { |
407 |
|
push @index_data, $idisplay; |
408 |
|
} |
409 |
|
$idisplay = ""; |
410 |
|
@{$cache->{index_data}->{$field}->[$page]} = @index_data; |
411 |
} |
} |
412 |
} |
} |
413 |
} |
} |
414 |
|
|
415 |
|
if (! $iterate_by_page) { |
416 |
|
my $idel = $x->{index_delimiter}; |
417 |
|
# fill data in index |
418 |
|
foreach my $tmp (@index_data) { |
419 |
|
my $i = $d = $tmp; |
420 |
|
if ($idel && $tmp =~ m/$idel/) { |
421 |
|
($i,$d) = split(/$idel/,$tmp); |
422 |
|
} |
423 |
|
$index->insert($field, $i, $d, $path); |
424 |
|
} |
425 |
|
@index_data = (); |
426 |
|
} |
427 |
} |
} |
428 |
|
|
429 |
# now try to parse variables from configuration file |
# now try to parse variables from configuration file |
432 |
my $delimiter = x($x->{delimiter}) || ' '; |
my $delimiter = x($x->{delimiter}) || ' '; |
433 |
my $val = $cfg->val($database, x($x->{content})); |
my $val = $cfg->val($database, x($x->{content})); |
434 |
|
|
435 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
# FIXME index_lookup is not supported! |
436 |
$s = 0 if (lc($x->{type}) eq "display"); |
my ($s,$se,$d,$i,$il) = init_visible_type($x->{type}); |
|
$d = 0 if (lc($x->{type}) eq "swish"); |
|
|
# no support for swish exact in config. |
|
|
# IMHO, it's useless |
|
|
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
|
437 |
|
|
438 |
if ($val) { |
if ($val) { |
439 |
$display_data .= $delimiter.$val if ($d); |
$display_data .= $delimiter.$val if ($d); |
440 |
$swish_data .= $val if ($s); |
$swish_data .= " ".$val if ($s); |
441 |
$index->insert($field, $val, $path) if ($i); |
$index->insert($field, $val, $path) if ($i); |
442 |
} |
} |
443 |
|
|
444 |
|
if ($iterate_by_page) { |
445 |
|
# FIXME data from config tag will appear just |
446 |
|
# on first page!!! |
447 |
|
my $page = 0; |
448 |
|
if ($display_data) { |
449 |
|
$cache->{display_data}->{$field}->[$page] = $display_data; |
450 |
|
$display_data = ""; |
451 |
|
} |
452 |
|
if ($swish_data) { |
453 |
|
$cache->{swish_data}->{$field}->[$page] = $swish_data; |
454 |
|
$swish_data = ""; |
455 |
|
} |
456 |
|
if ($swish_exact_data) { |
457 |
|
$cache->{swish_exact_data}->{$field}->[$page] = $swish_exact_data; |
458 |
|
$swish_exact_data = ""; |
459 |
|
} |
460 |
|
} |
461 |
} |
} |
462 |
|
|
463 |
|
# save data page-by-page |
464 |
if ($display_data) { |
foreach my $field (@page_fields) { |
465 |
|
my $nr_pages = $page_max{$field} || next; |
466 |
if ($field eq "headline") { |
#print STDERR "field '$field' iterate over ",($nr_pages || 0)," pages...\n"; |
467 |
$xml .= xmlify("headline", $display_data); |
#print STDERR Dumper($cache->{display_data}); |
468 |
} else { |
for (my $page=0; $page <= $nr_pages; $page++) { |
469 |
|
my $display_data; |
470 |
# find field name (signular, plural) |
if ($cache->{format}->{$field}) { |
471 |
my $field_name = ""; |
my $tmp = mkformat($cache->{format}->{$field},$cache->{display_data}->{$field}->[$page]); |
472 |
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}."#-#"; |
|
473 |
} else { |
} else { |
474 |
print STDERR "WARNING: field '$field' doesn't have 'name' attribute!"; |
$display_data = $cache->{display_data}->{$field}->[$page]; |
475 |
|
} |
476 |
|
if ($display_data) { # default |
477 |
|
if ($field eq "headline") { |
478 |
|
$xml .= xmlify("headline", $display_data); |
479 |
|
} else { |
480 |
|
|
481 |
|
# fallback to empty field name if needed |
482 |
|
$html .= get_field_name($config,$field,$field_usage{$field}) || ''; |
483 |
|
$html .= "#-#".$display_data."###\n"; |
484 |
|
} |
485 |
|
} |
486 |
|
|
487 |
|
my $swish_data = $cache->{swish_data}->{$field}->[$page]; |
488 |
|
if ($swish_data) { |
489 |
|
# remove extra spaces |
490 |
|
$swish_data =~ s/ +/ /g; |
491 |
|
$swish_data =~ s/ +$//g; |
492 |
|
|
493 |
|
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
494 |
|
} |
495 |
|
|
496 |
|
my $swish_exact_data = $cache->{swish_exact_data}->{$field}->[$page]; |
497 |
|
if ($swish_exact_data) { |
498 |
|
$swish_exact_data =~ s/ +/ /g; |
499 |
|
$swish_exact_data =~ s/ +$//g; |
500 |
|
|
501 |
|
# add delimiters before and after word. |
502 |
|
# That is required to produce exact match |
503 |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
504 |
} |
} |
505 |
if ($field_name) { |
|
506 |
$html .= x($field_name); |
my $idel = $cache->{index_delimiter}->{$field}; |
507 |
|
foreach my $tmp (@{$cache->{index_data}->{$field}->[$page]}) { |
508 |
|
my $i = $tmp; |
509 |
|
my $d = $tmp; |
510 |
|
if ($idel && $tmp =~ m/$idel/) { |
511 |
|
($i,$d) = split(/$idel/,$tmp); |
512 |
|
} |
513 |
|
$index->insert($field, $i, $d, $path); |
514 |
|
#print STDERR "index [$idel] $field: $i --> $d [$path]\n"; |
515 |
} |
} |
|
$html .= $display_data."###\n"; |
|
516 |
} |
} |
517 |
|
|
518 |
} |
} |
519 |
if ($swish_data) { |
|
520 |
# remove extra spaces |
if (! $iterate_by_page) { |
521 |
$swish_data =~ s/ +/ /g; |
if ($display_data) { |
522 |
$swish_data =~ s/ +$//g; |
if ($field eq "headline") { |
523 |
|
$xml .= xmlify("headline", $display_data); |
524 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
} else { |
|
} |
|
525 |
|
|
526 |
if ($swish_exact_data) { |
# fallback to empty field name if needed |
527 |
$swish_exact_data =~ s/ +/ /g; |
$html .= get_field_name($config,$field,$field_usage{$field}) || ''; |
528 |
$swish_exact_data =~ s/ +$//g; |
$html .= "#-#".$display_data."###\n"; |
529 |
|
} |
530 |
|
} |
531 |
|
if ($swish_data) { |
532 |
|
# remove extra spaces |
533 |
|
$swish_data =~ s/ +/ /g; |
534 |
|
$swish_data =~ s/ +$//g; |
535 |
|
|
536 |
# add delimiters before and after word. |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
537 |
# That is required to produce exact match |
} |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,'xxbxx '.$swish_exact_data.' xxexx')); |
|
|
} |
|
538 |
|
|
539 |
|
if ($swish_exact_data) { |
540 |
|
$swish_exact_data =~ s/ +/ /g; |
541 |
|
$swish_exact_data =~ s/ +$//g; |
542 |
|
|
543 |
|
# add delimiters before and after word. |
544 |
|
# That is required to produce exact match |
545 |
|
$xml .= xmlify($field."_swish_exact", unac_string($codepage,$swish_exact_data)); |
546 |
|
} |
547 |
|
} |
548 |
} |
} |
549 |
|
|
550 |
# dump formatted output in <html> |
# dump formatted output in <html> |
589 |
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"; |
590 |
my $add_xml = $cfg -> val($database, 'xml'); # optional |
my $add_xml = $cfg -> val($database, 'xml'); # optional |
591 |
|
|
592 |
|
# create new lookup file |
593 |
|
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
594 |
|
if ($lookup_file) { |
595 |
|
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
596 |
|
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
597 |
|
print STDERR "creating lookup file '$lookup_file'\n"; |
598 |
|
# delete memory cache for lookup file |
599 |
|
delete $cache->{lhash}; |
600 |
|
} |
601 |
|
|
602 |
|
# open existing lookup file |
603 |
|
$lookup_file = $cfg -> val($database, 'lookup_open'); # optional |
604 |
|
if ($lookup_file) { |
605 |
|
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_READER, 0644; |
606 |
|
tie %lhash, 'TDB_File', $lookup_file, TDB_DEFAULT, O_RDWR, 0644; |
607 |
|
print STDERR "opening lookup file '$lookup_file'\n"; |
608 |
|
} |
609 |
|
|
610 |
print STDERR "reading ./import_xml/$type.xml\n"; |
print STDERR "reading ./import_xml/$type.xml\n"; |
611 |
|
|
612 |
# extract just type basic |
# extract just type basic |
645 |
# now read database |
# now read database |
646 |
print STDERR "using: $type...\n"; |
print STDERR "using: $type...\n"; |
647 |
|
|
648 |
|
# erase cache for tags by order in this database |
649 |
|
delete $cache->{tags_by_order}; |
650 |
|
|
651 |
if ($type_base eq "isis") { |
if ($type_base eq "isis") { |
652 |
|
|
653 |
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 "$database doesn't have 'isis_db' defined!"; |
733 |
print STDERR "\n"; |
print STDERR "\n"; |
734 |
|
|
735 |
} elsif ($type_base eq "excel") { |
} elsif ($type_base eq "excel") { |
736 |
use Spreadsheet::ParseExcel; |
require Spreadsheet::ParseExcel; |
737 |
use Spreadsheet::ParseExcel::Utility qw(int2col); |
require Spreadsheet::ParseExcel::Utility; |
738 |
|
import Spreadsheet::ParseExcel::Utility qw(int2col); |
739 |
|
|
740 |
$import2cp = Text::Iconv->new($config->{excel_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{excel_codepage},$codepage); |
741 |
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 "$database doesn't have 'excel_file' defined!"; |
789 |
} |
} |
790 |
} elsif ($type_base eq "marc") { |
} elsif ($type_base eq "marc") { |
791 |
|
|
792 |
use MARC; |
require MARC; |
793 |
|
|
794 |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{marc_codepage},$codepage); |
795 |
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!"; |
796 |
|
|
797 |
# optional argument is format |
# optional argument is format |
798 |
my $format = x($config->{format}) || 'usmarc'; |
my $format = x($config->{marc_format}) || 'usmarc'; |
799 |
|
|
800 |
print STDERR "Reading MARC file '$marc_file'\n"; |
print STDERR "Reading MARC file '$marc_file'\n"; |
801 |
|
|
802 |
my $marc = new MARC; |
my $marc = new MARC; |
803 |
my $nr = $marc->openmarc({ |
my $nr = $marc->openmarc({ |
804 |
file=>$marc_file, format=>$format |
file=>$marc_file, format=>$format |
805 |
}) || die "Can't open MARC file '$marc_file'"; |
}) || die "Can't open MARC file '$marc_file' with format '$format'"; |
806 |
|
|
807 |
|
# read MARC file in memory |
808 |
|
$marc->nextmarc(-1); |
809 |
|
|
810 |
my $i=0; # record nr. |
my $max_rec = $marc->marc_count(); |
811 |
|
|
812 |
my $rec; |
for(my $i=1; $i<=$max_rec; $i++) { |
813 |
|
|
814 |
while ($marc->nextmarc(1)) { |
progress($i,$max_rec); |
815 |
|
|
816 |
# XXX |
# store value for marc_sf.pm |
817 |
fakeprogress($i++); |
$main::cache->{marc_record} = $i; |
818 |
|
|
819 |
my $swishpath = $database."#".$i; |
my $swishpath = $database."#".$i; |
820 |
|
|
826 |
print "Document-Type: XML\n\n$xml\n"; |
print "Document-Type: XML\n\n$xml\n"; |
827 |
} |
} |
828 |
} |
} |
829 |
|
|
830 |
|
print STDERR "\n"; |
831 |
|
|
832 |
} elsif ($type_base eq "feed") { |
} elsif ($type_base eq "feed") { |
833 |
|
|
834 |
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{feed_codepage},$codepage); |
867 |
fakeprogress($i); |
fakeprogress($i); |
868 |
|
|
869 |
} |
} |
870 |
|
# close lookup |
871 |
|
untie %lhash if (%lhash); |
872 |
} |
} |
873 |
} |
} |
874 |
|
|