/[webpac]/branches/humanistika/WebPac.pm
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Contents of /branches/humanistika/WebPac.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 704 - (show annotations)
Sat Apr 30 15:48:40 2005 UTC (14 years, 9 months ago) by dpavlin
File size: 14150 byte(s)
upgrated branches to head

1 package WebPac;
2
3 use base 'CGI::Application';
4 use strict;
5
6 use HTML::FillInForm;
7 use SWISH::API;
8 use Text::Iconv;
9 use DBI;
10 use Config::IniFiles;
11 use Text::Unaccent;
12 use Data::Pageset;
13 use POSIX qw(locale_h);
14
15 use lib '..';
16 use index_DBI_filter;
17 use back2html;
18
19
20 # read global.conf configuration
21 my $cfg_global = new Config::IniFiles( -file => '../global.conf' ) || die "can't open 'global.conf'";
22
23 # configuration options from global.conf
24 my $TEMPLATE_PATH = $cfg_global->val('webpac', 'template_html') || die "need template_html in global.conf, section webpac";
25 my $CHARSET = $cfg_global->val('webpac', 'charset') || 'ISO-8859-1';
26 my $SWISH = $cfg_global->val('webpac', 'swish') || '/usr/bin/swish-e';
27 my $INDEX = $cfg_global->val('webpac', 'index') || die "need index in global.conf, section webpac";
28 my $MAX_HITS = $cfg_global->val('webpac', 'max_hits') || 0;
29 my $ON_PAGE =$cfg_global->val('webpac', 'on_page') || 10;
30 my $MIN_WILDCARD =$cfg_global->val('webpac', 'min_wildcard') || 1;
31 my $TEMPLATE =$cfg_global->val('webpac', 'template');
32 my $UNAC_FILTER =$cfg_global->val('global', 'my_unac_filter');
33 my $BASE_PATH =$cfg_global->val('webpac', 'base_path');
34 # for pager
35 my $pages_per_set = $cfg_global->val('webpac', 'pages_per_set') || 10;
36 my $locale = $cfg_global->val('locale') || 'hr_HR';
37
38 Text::Iconv->raise_error(0); # Conversion errors raise exceptions
39
40 my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
41
42 setlocale(LC_CTYPE, $locale);
43 setlocale(LC_COLLATE, $locale);
44
45 if ($UNAC_FILTER) {
46 require $UNAC_FILTER;
47 } else {
48 sub WebPac::my_unac_string {
49 my ($charset, $string) = (@_);
50 return $string;
51 }
52 }
53
54 # use path from cgi script to support templates in subdirs
55 sub url_ex {
56 my $q = shift || die "suff2file needs CGI object!";
57 my $tpl = shift || die "url_ex needs template name!";
58 return suff2file($BASE_PATH, $q->url(-absolute => 1,-path => 1),$TEMPLATE_PATH,$tpl);
59 }
60
61 sub suff2file($$$$) {
62 my ($base_path, $p, $path, $tpl) = @_;
63
64 return $tpl if (! $base_path);
65
66 # strip everything to and including base path, leaving only
67 # additional (virtual) path
68 if ($base_path eq "/") {
69 $p =~ s,/*,,g;
70 my ($name,$ext) = split(/\./,$tpl);
71 $p = $name . "-" . $p . "." . $ext;
72 } elsif ($p =~ s,^.*?$base_path,,) {
73 $p =~ s,/*,,g;
74 my ($name,$ext) = split(/\./,$tpl);
75 $p = $name . $p . "." . $ext;
76 } else {
77 # if unable reset it!
78 $p = $tpl;
79 }
80
81 if ( -e "$path/$p") {
82 return $p;
83 } else {
84 return $tpl;
85 }
86
87 }
88
89 sub setup {
90 my $self = shift;
91 $self->tmpl_path($TEMPLATE_PATH);
92 $self->run_modes(
93 'search' => 'show_search_form',
94 'results' => 'show_results_list',
95 # 'user' => 'show_user_detail',
96 'index' => 'show_index',
97 );
98 $self->start_mode('search');
99 $self->mode_param('rm');
100
101 $self->header_props(-charset=>$CHARSET);
102 }
103
104 sub in_template {
105 my $q = shift || die "need CGI object!";
106 my $html = shift || die "This page is left unintentionally blank";
107 return $html if (! defined($TEMPLATE));
108
109 my ($dir,$tpl);
110 if ($TEMPLATE =~ m,^(.*?/*)([^/]+)$,) {
111 ($dir,$tpl) = ($1,$2);
112 } else {
113 die "can't parse TEMPLATE path";
114 }
115
116 my $master_tpl = suff2file($BASE_PATH, $q->url(-absolute => 1, -path => 1),$dir,$tpl);
117 if (open(T, $master_tpl)) {
118 my $template_html = join("\n",<T>);
119 close(T);
120 $template_html =~ s/##webpac##/$html/gsi;
121 return $template_html;
122 } else {
123 return "Can't read template '$master_tpl'";
124 }
125 }
126
127 #--------------------------------------------------------------------------
128
129 #
130 # make pager navigation and fill template variables
131 # compatibile with HTML::Pager
132 #
133
134 sub make_pager($$$) {
135 my ($q,$tmpl,$pager) = @_;
136
137 #
138 # pager navigation
139 #
140 my ($pager_prev,$pager_next, $pager_jump) = ('','','');
141
142 my $nav_fmt=qq{ <a href="%s">%s</a> };
143
144 if ($pager->current_page() > $pager->first_page) {
145 $q->param('PAGER_offset', $pager->current_page - 1);
146 $pager_prev .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'&lt;&lt;');
147 }
148
149 if ($pager->previous_set) {
150 $q->param('PAGER_offset', $pager->previous_set);
151 $pager_prev .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'..');
152 }
153
154
155 foreach my $p (@{$pager->pages_in_set()}) {
156 next if ($p <= 0);
157 if($p == $pager->current_page()) {
158 $pager_jump .= "<b>$p</b> ";
159 } else {
160 $q->param('PAGER_offset', $p);
161 $pager_jump .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),$p);
162 }
163 }
164
165 if ($pager->next_set) {
166 $q->param('PAGER_offset', $pager->next_set);
167 $pager_next .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'..');
168 }
169
170 if ($pager->current_page() < $pager->last_page) {
171 $q->param('PAGER_offset', $pager->current_page + 1);
172 $pager_next .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'&gt;&gt;');
173 }
174
175 $tmpl->param('PAGER_PREV', $pager_prev);
176 $tmpl->param('PAGER_JUMP', $pager_jump);
177 $tmpl->param('PAGER_NEXT', $pager_next);
178
179 }
180
181 #
182 # put persisten variables in template
183 #
184
185 sub make_pager_vars {
186 my $q = shift @_;
187 my $tmpl = shift @_;
188 my @persist_vars = @_;
189 my $hidden_vars = '';
190 my $hidden_search = '';
191 foreach my $v (@persist_vars) {
192 foreach my $val ($q->param($v)) {
193 next if (! $val || $val eq '');
194 $val =~ s/"/&quot;/g;
195 $hidden_vars .= '<input type="hidden" name="'.$v.'" value="'.$val.'"/>'."\n";
196 $hidden_search .= '<input type="hidden" name="'.$v.'" value="'.$val.'"/>'."\n" if ($v ne "rm");
197 }
198 }
199
200 $tmpl->param('PAGER_HIDDEN', $hidden_vars);
201 $tmpl->param('SEARCH_HIDDEN', $hidden_search);
202 $tmpl->param('PAGER_JAVASCRIPT', qq#
203 <SCRIPT LANGUAGE="Javascript">
204 <!-- Begin
205 // dummy emulator for HTML::Pager templates
206 function PAGER_set_offset_and_submit() {
207 return true;
208 }
209 // End -->
210 </script>
211 #);
212 }
213
214 #--------------------------------------------------------------------------
215
216 sub show_search_form {
217 my $self = shift;
218
219 # Get the CGI.pm query object
220 my $q = $self->query();
221
222 my $tmpl = $self->load_tmpl(url_ex($q,'search.html'));
223 my $html = $tmpl->output;
224
225 my $fif = new HTML::FillInForm;
226
227 return in_template($q,$fif->fill(scalarref => \$html, fobject => $q,
228 target => 'search'));
229 }
230
231 sub show_results_list {
232 my $self = shift;
233
234 my $q = $self->query();
235
236 # submit was reset?
237 if ($q->param('reset')) {
238 $q->delete_all;
239 return $self->show_search_form();
240 }
241
242 # load template for this page
243
244 my @s_arr; # all queries are located here
245
246 my @path_arr = $q->param('path');
247 my $full = $q->param('full');
248
249 my @persist_vars = ( 'rm', 'persist_search' );
250 my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.($q->param('PAGER_offset') || 0) );
251
252 my @persist_search_vars;
253 my @url_params_persist;
254 if ($q->param("persist_search")) {
255 @persist_search_vars = split(/\s*,\s*/, $q->param("persist_search"));
256 push @url_params_persist, "persist_search=".$q->url_param("persist_search");
257 push @url_params,"persist_search=".$q->url_param("persist_search");
258 }
259
260 # support parametars "f" and "v" for start
261 for(my $i = 0; $i <=30; $i++) {
262
263 $i = '' if ($i == 0);
264
265 return show_index($self, $i) if ($q->param("f".$i."_index"));
266
267 next if (! $q->param("v$i") || $q->param("v$i") eq '');
268 next if (! $q->param("f$i"));
269
270 my $persist = grep(/^$i$/,@persist_search_vars);
271
272 push @persist_vars, "f$i";
273 push @persist_vars, "v$i";
274 push @persist_vars, "e$i" if ($q->param("e$i"));
275
276 # create url parametars (and persistent ones)
277
278 push @url_params,"f$i=".$q->url_param("f$i");
279 push @url_params_persist,"f$i=".$q->url_param("f$i") if ($persist);
280
281 foreach my $v ($q->url_param("v$i")) {
282 # escape quotes so that phrase search work
283 $v =~ s/"/%22/g;
284 push @url_params,"v$i=$v";
285 push @url_params_persist,"v$i=$v" if ($persist);
286 }
287
288 if ($q->param("e$i")) {
289 push @url_params,"e$i=".$q->url_param("e$i");
290 # push @url_params_persist,"e$i=".$q->url_param("e$i");
291 }
292
293 my $wc="*"; # swish wildcard
294 $wc="" if ($i eq ""); # don't apply wildcard on field 0
295
296 # re-write query from +/- to and/and not
297 my @param_vals = $q->param("v$i");
298 my @swish_q;
299 my ($pre,$post,$exact) = ('','','');
300 while (my $search = shift @param_vals) {
301 my $s;
302 # remove accents
303 $search = my_unac_string($CHARSET,$search);
304 while ($search =~ s/\s*("[^"]+")\s*/ /) {
305 $s .= "$1 ";
306 }
307 $search =~ s/^\s+//;
308 $search =~ s/\s+$//;
309
310 # filed e[nr] is exact match bitmask
311 # 1 = beginning, 2=end, 3=both
312 my $exact_flag = $q->param("e$i") || 0;
313 $pre = '"xxbxx ' if ($exact_flag & 1);
314 $post = ' xxexx"' if ($exact_flag & 2);
315 # add qotes on other side
316 if ($q->param("e$i")) {
317 $pre = '"' if (! $pre);
318 $post = '"' if (! $post);
319 # what about wildcards?
320 $wc = '';
321 $wc = '*' if ($q->param("e$i") & 4);
322 $exact = '_exact';
323 }
324
325 foreach (split(/\s+/,$search)) {
326 if (m/^([+-])(\S+)/) {
327 $s.= ($s) ? "and " : "";
328 $s.="not " if ($1 eq "-");
329 $s.=$2.$wc." ";
330 } elsif (m/^\s*(and|or|not)\s*$/i) {
331 $s.=$_." ";
332 # don't add * to words with less than x chars
333 } elsif (length($_) <= $MIN_WILDCARD) {
334 $s.=$_." ";
335 } else {
336 $s.=$_.$wc." ";
337 }
338 }
339 $s =~ s/\*+/*/g;
340 $s =~ s/[()]//g; # () are used in query language
341 $s = $pre.$s.$post if ($q->param("e$i"));
342 push @swish_q,$s;
343 }
344 # FIXME default operator for multi-value fields is or. There is
345 # no way to change it, except here for now. Is there need?
346 push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")";
347 }
348
349 my $tmpl = $self->load_tmpl(url_ex($q,'results.html'), global_vars => 1, die_on_bad_params => 0);
350
351 sub esc_html {
352 my $html = shift;
353 $html =~ s/</&lt;/g;
354 $html =~ s/>/&gt;/g;
355 return $html;
356 }
357
358 my $sort = 'swishrank';
359 if ($q->param("sort")) {
360 $sort = 'headline';
361 push @persist_vars, "sort";
362 }
363
364 my $sortby = $q->param("sortby");
365 if ($sortby) {
366 $sort = $sortby;
367 push @persist_vars, "sortby";
368 }
369 # used to filter entries in index and swish
370 my $filter = $q->param("filter");
371
372 # construct swish query
373 my $sw_q = join(" and ",@s_arr);
374 if (@path_arr && $q->param('show_full')) {
375 $sw_q .= " and (swishdocpath=\"";
376 $sw_q .= join("\" or swishdocpath=\"",@path_arr);
377 $sw_q .= "\")";
378 $tmpl->param('full',1); # show full records
379 # } elsif (@path_arr && $#path_arr == 0) {
380 # # I will assume that it's a filter since there isn't show_full
381 # $filter = shift @path_arr;
382 } elsif ($q->param('show_full')) {
383 # just show full path, no path defined
384 $tmpl->param('full',1);
385 } else {
386 $tmpl->param('full',0);
387 }
388
389 if ($filter) {
390 $sw_q .= " and (swishdocpath=\"$filter\")" unless (@path_arr);
391 push @persist_vars, "filter";
392 push @url_params, "filter=$filter";
393 push @url_params_persist, "filter=$filter";
394 }
395
396 my $swish_msg = ' ';
397
398 # create new swish instance
399 my $swish = SWISH::API->new($INDEX);
400 $swish_msg .= $swish->ErrorString." ".$swish->LastErrorMsg if $swish->Error;
401
402 # execute query and get number of results from SWISH-E
403 my $search = $swish->New_Search_Object;
404
405 $search->SetSort($sort);
406
407 my $results = $search->Execute($sw_q);
408 $swish_msg .= $swish->ErrorString." ".$swish->LastErrorMsg if $swish->Error;
409
410 my $hits = $results->Hits;
411
412 $tmpl->param('hits',$hits);
413 my $search_msg = $sw_q;
414 $search_msg .= '<em>'.$swish_msg.'</em>' if ($swish_msg);
415 $tmpl->param('search', $search_msg);
416
417 $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
418 $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
419
420 # URL parametars for search results
421 $tmpl->param('url_params',"?".join("&",@url_params));
422 $tmpl->param('url_params_paths',"?".join("&",@url_params).'&'.join("&",map { my $t = $_; $t =~ s/\#/%23/g; "path=$t"; } @path_arr));
423
424
425
426 #
427 # build pager
428 #
429
430 my $current_page = $q->param('PAGER_offset') || 1;
431
432 my $pager = Data::Pageset->new({
433 'total_entries' => $hits,
434 'entries_per_page' => $ON_PAGE,
435 'current_page' => $current_page,
436 'pages_per_set' => $pages_per_set,
437 });
438
439 $results->SeekResult( $pager->first - 1 );
440
441 # get number of entries on this page
442 my $i = $pager->entries_on_this_page;
443
444 # results from swish for template
445 my @pager_data_list;
446
447 for(my $i=$pager->first; $i<=$pager->last; $i++) {
448
449 my $result = $results->NextResult;
450 last if (! $result);
451
452 my $r = {
453 nr => $i,
454 path => $result->Property('swishdocpath'),
455 headline => esc_html($from_utf8->convert($result->Property('headline'))),
456 rank => $result->Property('swishrank')
457 };
458
459 $r->{html} = back2html($from_utf8->convert($result->Property('html')), join("&",@url_params_persist)) if ($q->param('show_full'));
460
461 push @pager_data_list, $r;
462 }
463
464
465
466 # put something in template
467 make_pager($q, $tmpl, $pager);
468 make_pager_vars($q, $tmpl, @persist_vars);
469 $tmpl->param('PAGER_DATA_LIST', \@pager_data_list);
470
471 my $html = $tmpl->output;
472
473 return in_template($q,$html);
474 }
475
476 sub show_index {
477 my $self = shift;
478 my $i = shift; # field number
479
480 my $q = $self->query();
481
482 my $field = $q->param("f$i");
483 my $limit = $q->param("v$i");
484
485 my $filter = $q->param("filter");
486
487 my $html;
488
489 my $index = new index_DBI(
490 $cfg_global->val('global', 'dbi_dbd'),
491 $cfg_global->val('global', 'dbi_dsn'),
492 $cfg_global->val('global', 'dbi_user'),
493 $cfg_global->val('global', 'dbi_passwd') || ''
494 );
495
496 my $total = $index->count($field,$limit,$filter);
497
498 if (! defined($total)) {
499 my $tmpl = $self->load_tmpl(url_ex($q,'no_index.html'));
500 $tmpl->param('field',$field);
501 $html = $tmpl->output;
502 return $html;
503 }
504
505 my $tmpl = $self->load_tmpl(url_ex($q,'index_res.html'), global_vars => 1, die_on_bad_params => 0);
506 $tmpl->param('field',$field);
507 $tmpl->param('limit',$limit);
508 $tmpl->param('total',$total);
509 $tmpl->param('filter',$filter);
510
511 # FIXME I should set offset and leave out limit from fetch!!
512 # if (! $q->param("PAGER_offset") {
513 # $q->param("Pager_offet)
514 # }
515
516
517 #
518 # build pager
519 #
520 my $pager = Data::Pageset->new({
521 'total_entries' => $total,
522 'entries_per_page' => $ON_PAGE,
523 'current_page' => $q->param('PAGER_offset') || 1,
524 'pages_per_set' => $pages_per_set
525 });
526
527 my @persist_vars = qw{rm f$i v$i f$i_index offset};
528
529 make_pager($q, $tmpl, $pager);
530 make_pager_vars($q, $tmpl, @persist_vars);
531
532 my @pager_data_list = $index->fetch($field,$limit, $pager->first - 1, $pager->entries_on_this_page, $filter);
533 $tmpl->param('PAGER_DATA_LIST', \@pager_data_list);
534
535 return in_template($q,$tmpl->output);
536 }
537
538 1;

Properties

Name Value
cvs2svn:cvs-rev 1.40

  ViewVC Help
Powered by ViewVC 1.1.26