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 |
sub data2xml { |
sub data2xml { |
76 |
|
|
77 |
use xmlify; |
use xmlify; |
101 |
return $va <=> $vb; |
return $va <=> $vb; |
102 |
} |
} |
103 |
|
|
104 |
foreach my $field (sort by_order keys %{$config->{indexer}}) { |
my @sorted_tags; |
105 |
|
if ($cache->{tags_by_order}) { |
106 |
|
@sorted_tags = @{$cache->{tags_by_order}}; |
107 |
|
} else { |
108 |
|
@sorted_tags = sort by_order keys %{$config->{indexer}}; |
109 |
|
$cache->{tags_by_order} = \@sorted_tags; |
110 |
|
} |
111 |
|
|
112 |
|
# lookup key |
113 |
|
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) { |
146 |
|
|
147 |
$field=x($field); |
$field=x($field); |
148 |
$field_usage{$field}++; |
$field_usage{$field}++; |
149 |
|
|
150 |
my $swish_data = ""; |
my $swish_data = ""; |
151 |
|
my $swish_exact_data = ""; |
152 |
my $display_data = ""; |
my $display_data = ""; |
153 |
my $line_delimiter; |
my $line_delimiter; |
154 |
|
|
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 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
# swish, swish_exact, display, index, index_lookup |
174 |
|
# swish and display defaults |
175 |
|
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 |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
($s,$se,$d,$i) = (0,0,0,1) if (lc($x->{type}) eq "index"); |
179 |
|
($s,$se,$d,$i) = (0,1,0,0) if (lc($x->{type}) eq "swish_exact"); |
180 |
|
$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") { |
191 |
|
|
192 |
# placeholder for all repeatable entries for index |
# placeholder for all repeatable entries for index |
193 |
my @index_data; |
my @index_data; |
194 |
my $index_filter; |
|
195 |
|
sub mkformat { |
196 |
|
my $x = shift || die "mkformat needs tag reference"; |
197 |
|
my $data = shift || return; |
198 |
|
my $format_name = x($x->{format_name}) || return $data; |
199 |
|
my $fmt = x($config->{format}->{$format_name}->{content}) || die "<format name=\"$format_name\"> is not defined!"; |
200 |
|
my $format_delimiter = x($x->{format_delimiter}); |
201 |
|
my @data; |
202 |
|
if ($format_delimiter) { |
203 |
|
@data = split(/$format_delimiter/,$data); |
204 |
|
} else { |
205 |
|
push @data,$data; |
206 |
|
} |
207 |
|
|
208 |
|
if ($fmt) { |
209 |
|
my $nr = scalar $fmt =~ s/%s/%s/g; |
210 |
|
if (($#data+1) == $nr) { |
211 |
|
return sprintf($fmt,@data); |
212 |
|
} else { |
213 |
|
print STDERR "mkformat: [$data] can't be split on [$format_delimiter] to $nr fields!\n"; |
214 |
|
return $data; |
215 |
|
} |
216 |
|
} else { |
217 |
|
print STDERR "usage of link '$format_name' without defined format (<link> tag)\n"; |
218 |
|
} |
219 |
|
} |
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"; |
228 |
last; |
last; |
229 |
} |
} |
230 |
|
|
231 |
|
# is this field is lookup? |
232 |
|
if ($display && $x->{lookup}) { |
233 |
|
my $null = "<!-- null -->"; |
234 |
|
if ($use_lhash_cache) { |
235 |
|
if (!defined($cache->{lhash}->{$display})) { |
236 |
|
my $new_display = $lhash{$display}; |
237 |
|
if (defined($new_display)) { |
238 |
|
#print STDERR "lookup cache store '$display' = '$new_display'\n"; |
239 |
|
$display = $new_display; |
240 |
|
$cache->{lhash}->{$display} = $new_display; |
241 |
|
} else { |
242 |
|
print STDERR "WARNING: lookup for '$display' didn't find anything.\n"; |
243 |
|
$display = ""; |
244 |
|
$cache->{lhash}->{$display} = $null; |
245 |
|
} |
246 |
|
} else { |
247 |
|
$display = $cache->{lhash}->{$display}; |
248 |
|
} |
249 |
|
} else { |
250 |
|
$display = $lhash{$display} || $null; |
251 |
|
} |
252 |
|
} |
253 |
|
|
254 |
# filter="name" ; filter this field through |
# filter="name" ; filter this field through |
255 |
# filter/[name].pm |
# filter/[name].pm |
256 |
my $filter = $x->{filter}; |
my $filter = $x->{filter}; |
257 |
if ($filter) { |
if ($filter && !$cache->{filter_loaded}->{$filter}) { |
258 |
require "filter/".$filter.".pm"; |
require "filter/".$filter.".pm"; |
259 |
|
$cache->{filter_loaded}->{$filter}++; |
260 |
} |
} |
261 |
# type="swish" ; field for swish |
# type="swish" ; field for swish |
262 |
if ($s && $swish) { |
if ($swish) { |
263 |
if ($filter) { |
if ($filter && ($s || $se)) { |
264 |
no strict 'refs'; |
no strict 'refs'; |
265 |
$swish_data .= join(" ",&$filter($swish)); |
my $tmp = join(" ",&$filter($swish)) if ($s || $se); |
266 |
|
$swish_data .= $tmp if ($s); |
267 |
|
$swish_exact_data .= $tmp if ($se); |
268 |
|
|
269 |
} else { |
} else { |
270 |
$swish_data .= $swish; |
$swish_data .= $swish if ($s); |
271 |
|
$swish_exact_data .= $swish if ($se); |
272 |
} |
} |
273 |
} |
} |
274 |
|
|
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'; |
282 |
$display_data .= join($delimiter,&$filter($display)); |
if ($display_data) { |
283 |
|
$display_data .= $delimiter.join($delimiter,mkformat($x,&$filter($display))); |
284 |
|
} else { |
285 |
|
$display_data = join($delimiter,mkformat($x,&$filter($display))); |
286 |
|
} |
287 |
} else { |
} else { |
288 |
if ($display_data) { |
if ($display_data) { |
289 |
$display_data .= $delimiter.$display; |
$display_data .= $delimiter.mkformat($x,$display); |
290 |
} else { |
} else { |
291 |
$display_data .= $display; |
$display_data = mkformat($x,$display); |
292 |
} |
} |
293 |
} |
} |
294 |
} |
} |
295 |
|
|
296 |
# type="index" ; insert into index |
# type="index" ; insert into index |
297 |
if ($i && $display) { |
if ($i && $display) { |
298 |
push @index_data, $display; |
if ($filter) { |
299 |
$index_filter = $filter if ($filter); |
no strict 'refs'; |
300 |
|
$display = &$filter($display); |
301 |
|
} |
302 |
|
if ($x->{append} && @index_data) { |
303 |
|
$index_data[$#index_data].=$display; |
304 |
|
} else { |
305 |
|
push @index_data, $display; |
306 |
|
} |
307 |
} |
} |
|
} |
|
308 |
|
|
309 |
# fill data in index |
# store fields in lookup |
310 |
if (@index_data) { |
if ($il && $display) { |
311 |
if ($index_filter) { |
if (lc($x->{type}) eq "lookup_key") { |
312 |
no strict 'refs'; |
if ($lookup_key) { |
313 |
foreach my $d (&$index_filter(@index_data)) { |
print STDERR "WARNING: try to redefine lookup_key (keys shouldn't be repeatable fields!)"; |
314 |
$index->insert($field, $d, $path); |
} else { |
315 |
|
$lookup_key = $display; |
316 |
|
} |
317 |
|
} elsif (lc($x->{type}) eq "lookup_val") { |
318 |
|
if ($lookup_key) { |
319 |
|
$lhash{$lookup_key} = $display; |
320 |
|
} else { |
321 |
|
print STDERR "WARNING: no lookup_key defined for '$display'?"; |
322 |
|
} |
323 |
} |
} |
324 |
} else { |
|
325 |
foreach my $d (@index_data) { |
} |
326 |
$index->insert($field, $d, $path); |
|
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 |
|
|
357 |
|
# fill data in index |
358 |
|
foreach my $d (@index_data) { |
359 |
|
$index->insert($field, $d, $path); |
360 |
|
} |
361 |
} |
} |
362 |
|
|
363 |
# now try to parse variables from configuration file |
# now try to parse variables from configuration file |
369 |
my ($s,$d,$i) = (1,1,0); # swish, display default |
my ($s,$d,$i) = (1,1,0); # swish, display default |
370 |
$s = 0 if (lc($x->{type}) eq "display"); |
$s = 0 if (lc($x->{type}) eq "display"); |
371 |
$d = 0 if (lc($x->{type}) eq "swish"); |
$d = 0 if (lc($x->{type}) eq "swish"); |
372 |
|
# no support for swish exact in config. |
373 |
|
# IMHO, it's useless |
374 |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
($s,$d,$i) = (0,0,1) if (lc($x->{type}) eq "index"); |
375 |
|
|
376 |
if ($val) { |
if ($val) { |
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 |
|
} else { |
447 |
|
|
448 |
$xml .= xmlify($field."_swish", unac_string($codepage,$swish_data)); |
# fallback to empty field name if needed |
449 |
} |
$html .= get_field_name($config,$field,$field_usage{$field}) || ''; |
450 |
|
$html .= "#-#".$display_data."###\n"; |
451 |
|
} |
452 |
|
} |
453 |
|
if ($swish_data) { |
454 |
|
# 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> |
501 |
|
|
502 |
my $show_progress = $cfg_global->val('global', 'show_progress'); |
my $show_progress = $cfg_global->val('global', 'show_progress'); |
503 |
|
|
504 |
|
my $unac_filter = $cfg_global->val('global', 'unac_filter'); |
505 |
|
if ($unac_filter) { |
506 |
|
require $unac_filter; |
507 |
|
} |
508 |
|
|
509 |
foreach my $database ($cfg->Sections) { |
foreach my $database ($cfg->Sections) { |
510 |
|
|
511 |
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"; |
512 |
my $add_xml = $cfg -> val($database, 'xml'); # optional |
my $add_xml = $cfg -> val($database, 'xml'); # optional |
513 |
|
|
514 |
|
# create new lookup file |
515 |
|
my $lookup_file = $cfg -> val($database, 'lookup_newfile'); # optional |
516 |
|
if ($lookup_file) { |
517 |
|
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_NEWDB, 0644; |
518 |
|
tie %lhash, 'TDB_File', $lookup_file, TDB_CLEAR_IF_FIRST, O_RDWR, 0644; |
519 |
|
print STDERR "creating lookup file '$lookup_file'\n"; |
520 |
|
} |
521 |
|
|
522 |
|
# open existing lookup file |
523 |
|
$lookup_file = $cfg -> val($database, 'lookup_open'); # optional |
524 |
|
if ($lookup_file) { |
525 |
|
#tie %lhash, 'GDBM_File', $lookup_file, &GDBM_READER, 0644; |
526 |
|
tie %lhash, 'TDB_File', $lookup_file, TDB_DEFAULT, O_RDWR, 0644; |
527 |
|
print STDERR "opening lookup file '$lookup_file'\n"; |
528 |
|
} |
529 |
|
|
530 |
print STDERR "reading ./import_xml/$type.xml\n"; |
print STDERR "reading ./import_xml/$type.xml\n"; |
531 |
|
|
532 |
# extract just type basic |
# extract just type basic |
533 |
my $type_base = $type; |
my $type_base = $type; |
534 |
$type_base =~ s/_.+$//g; |
$type_base =~ s/_.+$//g; |
535 |
|
|
536 |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config' ], forcecontent => 1); |
$config=XMLin("./import_xml/$type.xml", forcearray => [ $type2tag{$type_base}, 'config', 'format' ], forcecontent => 1); |
537 |
|
|
538 |
# output current progress indicator |
# output current progress indicator |
539 |
my $last_p = 0; |
my $last_p = 0; |
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!"; |
575 |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
$import2cp = Text::Iconv->new($config->{isis_codepage},$codepage); |
576 |
my $db = OpenIsis::open( $isis_db ); |
my $db = OpenIsis::open( $isis_db ); |
577 |
|
|
578 |
|
# check if .txt database for OpenIsis is zero length, |
579 |
|
# if so, erase it and re-open database |
580 |
|
sub check_txt_db { |
581 |
|
my $isis_db = shift || die "need isis database name"; |
582 |
|
my $reopen = 0; |
583 |
|
|
584 |
|
if (-e $isis_db.".TXT") { |
585 |
|
print STDERR "WARNING: removing $isis_db.TXT OpenIsis database...\n"; |
586 |
|
unlink $isis_db.".TXT" || warn "FATAL: unlink error on '$isis_db.TXT': $!"; |
587 |
|
$reopen++; |
588 |
|
} |
589 |
|
if (-e $isis_db.".PTR") { |
590 |
|
print STDERR "WARNING: removing $isis_db.PTR OpenIsis database...\n"; |
591 |
|
unlink $isis_db.".PTR" || warn "FATAL: unlink error on '$isis_db.PTR': $!"; |
592 |
|
$reopen++; |
593 |
|
} |
594 |
|
return OpenIsis::open( $isis_db ) if ($reopen); |
595 |
|
} |
596 |
|
|
597 |
|
# EOF error |
598 |
|
if ($db == -1) { |
599 |
|
$db = check_txt_db($isis_db); |
600 |
|
if ($db == -1) { |
601 |
|
print STDERR "FATAL: OpenIsis can't open zero size file $isis_db\n"; |
602 |
|
next; |
603 |
|
} |
604 |
|
} |
605 |
|
|
606 |
# OpenIsis::ERR_BADF |
# OpenIsis::ERR_BADF |
607 |
if ($db == -4) { |
if ($db == -4) { |
608 |
print STDERR "FATAL: OpenIsis can't find file $isis_db\n"; |
print STDERR "FATAL: OpenIsis can't find file $isis_db\n"; |
618 |
|
|
619 |
my $max_rowid = OpenIsis::maxRowid( $db ); |
my $max_rowid = OpenIsis::maxRowid( $db ); |
620 |
|
|
621 |
|
# if 0 records, try to rease isis .txt database |
622 |
|
if ($max_rowid == 0) { |
623 |
|
# force removal of database |
624 |
|
$db = check_txt_db($isis_db); |
625 |
|
$max_rowid = OpenIsis::maxRowid( $db ); |
626 |
|
} |
627 |
|
|
628 |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
print STDERR "Reading database: $isis_db [$max_rowid rows]\n"; |
629 |
|
|
630 |
my $path = $database; |
my $path = $database; |
646 |
} |
} |
647 |
} |
} |
648 |
} |
} |
649 |
|
# for this to work with current version of OpenIsis (0.9.0) |
650 |
|
# you might need my patch from |
651 |
|
# http://www.rot13.org/~dpavlin/projects/openisis-0.9.0-perl_close.diff |
652 |
|
OpenIsis::close($db); |
653 |
print STDERR "\n"; |
print STDERR "\n"; |
654 |
|
|
655 |
} elsif ($type_base eq "excel") { |
} elsif ($type_base eq "excel") { |
780 |
fakeprogress($i); |
fakeprogress($i); |
781 |
|
|
782 |
} |
} |
783 |
|
# close lookup |
784 |
|
untie %lhash if (%lhash); |
785 |
} |
} |
786 |
} |
} |
787 |
|
|