Line # Revision Author
1 7 dpavlin package WebPac;
2
3 use base 'CGI::Application';
4 use strict;
5
6 use HTML::FillInForm;
7 304 dpavlin use SWISH::API;
8 14 dpavlin use Text::Iconv;
9 9 dpavlin use DBI;
10 53 dpavlin use Config::IniFiles;
11 73 dpavlin use Text::Unaccent;
12 304 dpavlin use Data::Pageset;
13 696 dpavlin use POSIX qw(locale_h);
14 7 dpavlin
15 11 dpavlin use lib '..';
16 642 dpavlin use index_DBI_filter;
17 13 dpavlin use back2html;
18 11 dpavlin
19 7 dpavlin
20 76 dpavlin # 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;
30 120 dpavlin my $MIN_WILDCARD =$cfg_global->val('webpac', 'min_wildcard') || 1;
31 147 dpavlin my $TEMPLATE =$cfg_global->val('webpac', 'template');
32 320 dpavlin my $UNAC_FILTER =$cfg_global->val('global', 'my_unac_filter');
33 198 dpavlin my $BASE_PATH =$cfg_global->val('webpac', 'base_path');
34 304 dpavlin # for pager
35 my $pages_per_set = $cfg_global->val('webpac', 'pages_per_set') || 10;
36 696 dpavlin my $locale = $cfg_global->val('locale') || 'hr_HR';
37 76 dpavlin
38 320 dpavlin Text::Iconv->raise_error(0); # Conversion errors raise exceptions
39 304 dpavlin
40 320 dpavlin my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
41
42 696 dpavlin setlocale(LC_CTYPE, $locale);
43 setlocale(LC_COLLATE, $locale);
44
45 164 dpavlin if ($UNAC_FILTER) {
46 require $UNAC_FILTER;
47 320 dpavlin } else {
48 sub WebPac::my_unac_string {
49 my ($charset, $string) = (@_);
50 return $string;
51 }
52 164 dpavlin }
53 76 dpavlin
54 198 dpavlin # 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 $tpl = shift || die "url_ex needs template name!";
58 302 dpavlin return suff2file($BASE_PATH, $q->url(-absolute => 1,-path => 1),$TEMPLATE_PATH,$tpl);
59 198 dpavlin }
60 53 dpavlin
61 198 dpavlin sub suff2file($$$$) {
62 my ($base_path, $p, $path, $tpl) = @_;
63
64 return $tpl if (! $base_path);
65
66 733 dpavlin #warn "base_path: $base_path, p: $p, path: $path, tpl: $tpl\n";
67
68 $p =~ s#/[^/]*$##;
69
70 198 dpavlin # strip everything to and including base path, leaving only
71 # additional (virtual) path
72 302 dpavlin if ($base_path eq "/") {
73 198 dpavlin $p =~ s,/*,,g;
74 my ($name,$ext) = split(/\./,$tpl);
75 302 dpavlin $p = $name . "-" . $p . "." . $ext;
76 } elsif ($p =~ s,^.*?$base_path,,) {
77 $p =~ s,/*,,g;
78 my ($name,$ext) = split(/\./,$tpl);
79 198 dpavlin $p = $name . $p . "." . $ext;
80 } else {
81 # if unable reset it!
82 $p = $tpl;
83 }
84
85 if ( -e "$path/$p") {
86 return $p;
87 } else {
88 return $tpl;
89 }
90
91 }
92
93 7 dpavlin sub setup {
94 my $self = shift;
95 $self->tmpl_path($TEMPLATE_PATH);
96 $self->run_modes(
97 'search' => 'show_search_form',
98 'results' => 'show_results_list',
99 9 dpavlin # 'user' => 'show_user_detail',
100 'index' => 'show_index',
101 7 dpavlin );
102 $self->start_mode('search');
103 $self->mode_param('rm');
104
105 $self->header_props(-charset=>$CHARSET);
106 }
107
108 147 dpavlin sub in_template {
109 198 dpavlin my $q = shift || die "need CGI object!";
110 my $html = shift || die "This page is left unintentionally blank";
111 147 dpavlin return $html if (! defined($TEMPLATE));
112 198 dpavlin
113 my ($dir,$tpl);
114 if ($TEMPLATE =~ m,^(.*?/*)([^/]+)$,) {
115 ($dir,$tpl) = ($1,$2);
116 } else {
117 die "can't parse TEMPLATE path";
118 }
119
120 302 dpavlin my $master_tpl = suff2file($BASE_PATH, $q->url(-absolute => 1, -path => 1),$dir,$tpl);
121 198 dpavlin if (open(T, $master_tpl)) {
122 147 dpavlin my $template_html = join("\n",<T>);
123 close(T);
124 $template_html =~ s/##webpac##/$html/gsi;
125 return $template_html;
126 } else {
127 198 dpavlin return "Can't read template '$master_tpl'";
128 147 dpavlin }
129 }
130
131 304 dpavlin #--------------------------------------------------------------------------
132
133 #
134 # make pager navigation and fill template variables
135 # compatibile with HTML::Pager
136 #
137
138 sub make_pager($$$) {
139 my ($q,$tmpl,$pager) = @_;
140
141 #
142 # pager navigation
143 #
144 my ($pager_prev,$pager_next, $pager_jump) = ('','','');
145
146 my $nav_fmt=qq{ <a href="%s">%s</a> };
147
148 if ($pager->current_page() > $pager->first_page) {
149 $q->param('PAGER_offset', $pager->current_page - 1);
150 $pager_prev .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'&lt;&lt;');
151 }
152
153 if ($pager->previous_set) {
154 $q->param('PAGER_offset', $pager->previous_set);
155 $pager_prev .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'..');
156 }
157
158
159 foreach my $p (@{$pager->pages_in_set()}) {
160 656 dpavlin next if ($p <= 0);
161 304 dpavlin if($p == $pager->current_page()) {
162 $pager_jump .= "<b>$p</b> ";
163 } else {
164 $q->param('PAGER_offset', $p);
165 $pager_jump .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),$p);
166 }
167 }
168
169 if ($pager->next_set) {
170 $q->param('PAGER_offset', $pager->next_set);
171 $pager_next .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'..');
172 }
173
174 if ($pager->current_page() < $pager->last_page) {
175 $q->param('PAGER_offset', $pager->current_page + 1);
176 $pager_next .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'&gt;&gt;');
177 }
178
179 $tmpl->param('PAGER_PREV', $pager_prev);
180 $tmpl->param('PAGER_JUMP', $pager_jump);
181 $tmpl->param('PAGER_NEXT', $pager_next);
182
183 }
184
185 #
186 # put persisten variables in template
187 #
188
189 sub make_pager_vars {
190 my $q = shift @_;
191 my $tmpl = shift @_;
192 my @persist_vars = @_;
193 my $hidden_vars = '';
194 675 dpavlin my $hidden_search = '';
195 304 dpavlin foreach my $v (@persist_vars) {
196 457 dpavlin foreach my $val ($q->param($v)) {
197 636 dpavlin next if (! $val || $val eq '');
198 688 dpavlin $val =~ s/"/&quot;/g;
199 457 dpavlin $hidden_vars .= '<input type="hidden" name="'.$v.'" value="'.$val.'"/>'."\n";
200 675 dpavlin $hidden_search .= '<input type="hidden" name="'.$v.'" value="'.$val.'"/>'."\n" if ($v ne "rm");
201 457 dpavlin }
202 304 dpavlin }
203
204 $tmpl->param('PAGER_HIDDEN', $hidden_vars);
205 675 dpavlin $tmpl->param('SEARCH_HIDDEN', $hidden_search);
206 636 dpavlin $tmpl->param('PAGER_JAVASCRIPT', qq#
207 304 dpavlin <SCRIPT LANGUAGE="Javascript">
208 <!-- Begin
209 // dummy emulator for HTML::Pager templates
210 function PAGER_set_offset_and_submit() {
211 return true;
212 }
213 // End -->
214 </script>
215 636 dpavlin #);
216 304 dpavlin }
217
218 #--------------------------------------------------------------------------
219
220 7 dpavlin sub show_search_form {
221 my $self = shift;
222
223 # Get the CGI.pm query object
224 my $q = $self->query();
225
226 198 dpavlin my $tmpl = $self->load_tmpl(url_ex($q,'search.html'));
227 7 dpavlin my $html = $tmpl->output;
228
229 my $fif = new HTML::FillInForm;
230
231 198 dpavlin return in_template($q,$fif->fill(scalarref => \$html, fobject => $q,
232 147 dpavlin target => 'search'));
233 7 dpavlin }
234
235 sub show_results_list {
236 my $self = shift;
237
238 my $q = $self->query();
239
240 676 dpavlin # submit was reset?
241 if ($q->param('reset')) {
242 $q->delete_all;
243 return $self->show_search_form();
244 }
245
246 7 dpavlin # load template for this page
247
248 my @s_arr; # all queries are located here
249
250 47 dpavlin my @path_arr = $q->param('path');
251 my $full = $q->param('full');
252
253 636 dpavlin my @persist_vars = ( 'rm', 'persist_search' );
254 709 dpavlin my $url_params = {
255 'rm' => 'results',
256 'show_full' => 1,
257 'last_PAGER_offset' => ($q->param('PAGER_offset') || 0),
258 };
259 112 dpavlin
260 636 dpavlin my @persist_search_vars;
261 709 dpavlin my $url_params_persist = {};
262 636 dpavlin if ($q->param("persist_search")) {
263 @persist_search_vars = split(/\s*,\s*/, $q->param("persist_search"));
264 709 dpavlin $url_params_persist->{'persist_search'} = $q->url_param("persist_search");
265 $url_params->{'persist_search'} = $q->url_param("persist_search");
266 636 dpavlin }
267
268 150 dpavlin # support parametars "f" and "v" for start
269 636 dpavlin for(my $i = 0; $i <=30; $i++) {
270 7 dpavlin
271 636 dpavlin $i = '' if ($i == 0);
272
273 9 dpavlin return show_index($self, $i) if ($q->param("f".$i."_index"));
274 112 dpavlin
275 636 dpavlin next if (! $q->param("v$i") || $q->param("v$i") eq '');
276 9 dpavlin next if (! $q->param("f$i"));
277 7 dpavlin
278 636 dpavlin my $persist = grep(/^$i$/,@persist_search_vars);
279
280 112 dpavlin push @persist_vars, "f$i";
281 push @persist_vars, "v$i";
282 186 dpavlin push @persist_vars, "e$i" if ($q->param("e$i"));
283 112 dpavlin
284 636 dpavlin # create url parametars (and persistent ones)
285
286 709 dpavlin $url_params->{"f$i"} = $q->url_param("f$i");
287 $url_params_persist->{"f$i"} = $q->url_param("f$i") if ($persist);
288 636 dpavlin
289 725 dpavlin my @v;
290
291 158 dpavlin foreach my $v ($q->url_param("v$i")) {
292 677 dpavlin # escape quotes so that phrase search work
293 $v =~ s/"/%22/g;
294 725 dpavlin push @v, $v;
295 158 dpavlin }
296 725 dpavlin $url_params->{"v$i"} = \@v;
297 $url_params_persist->{"v$i"} = \@v if ($persist);
298 126 dpavlin
299 636 dpavlin if ($q->param("e$i")) {
300 709 dpavlin $url_params->{"e$i"} = $q->url_param("e$i");
301 # $url_params_persist->{"e$i"} = $q->url_param("e$i");
302 636 dpavlin }
303
304 150 dpavlin my $wc="*"; # swish wildcard
305 $wc="" if ($i eq ""); # don't apply wildcard on field 0
306
307 7 dpavlin # re-write query from +/- to and/and not
308 73 dpavlin my @param_vals = $q->param("v$i");
309 my @swish_q;
310 163 dpavlin my ($pre,$post,$exact) = ('','','');
311 73 dpavlin while (my $search = shift @param_vals) {
312 my $s;
313 # remove accents
314 320 dpavlin $search = my_unac_string($CHARSET,$search);
315 73 dpavlin while ($search =~ s/\s*("[^"]+")\s*/ /) {
316 $s .= "$1 ";
317 }
318 $search =~ s/^\s+//;
319 $search =~ s/\s+$//;
320 7 dpavlin
321 163 dpavlin # filed e[nr] is exact match bitmask
322 # 1 = beginning, 2=end, 3=both
323 636 dpavlin my $exact_flag = $q->param("e$i") || 0;
324 $pre = '"xxbxx ' if ($exact_flag & 1);
325 $post = ' xxexx"' if ($exact_flag & 2);
326 163 dpavlin # add qotes on other side
327 if ($q->param("e$i")) {
328 $pre = '"' if (! $pre);
329 $post = '"' if (! $post);
330 190 dpavlin # what about wildcards?
331 $wc = '';
332 $wc = '*' if ($q->param("e$i") & 4);
333 163 dpavlin $exact = '_exact';
334 }
335
336 73 dpavlin foreach (split(/\s+/,$search)) {
337 if (m/^([+-])(\S+)/) {
338 $s.= ($s) ? "and " : "";
339 $s.="not " if ($1 eq "-");
340 163 dpavlin $s.=$2.$wc." ";
341 122 dpavlin } elsif (m/^\s*(and|or|not)\s*$/i) {
342 163 dpavlin $s.=$_." ";
343 120 dpavlin # don't add * to words with less than x chars
344 } elsif (length($_) <= $MIN_WILDCARD) {
345 163 dpavlin $s.=$_." ";
346 73 dpavlin } else {
347 163 dpavlin $s.=$_.$wc." ";
348 73 dpavlin }
349 7 dpavlin }
350 73 dpavlin $s =~ s/\*+/*/g;
351 698 dpavlin $s =~ s/[()]//g; # () are used in query language
352 163 dpavlin $s = $pre.$s.$post if ($q->param("e$i"));
353 73 dpavlin push @swish_q,$s;
354 7 dpavlin }
355 73 dpavlin # FIXME default operator for multi-value fields is or. There is
356 # no way to change it, except here for now. Is there need?
357 163 dpavlin push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")";
358 7 dpavlin }
359
360 670 dpavlin my $tmpl = $self->load_tmpl(url_ex($q,'results.html'), global_vars => 1, die_on_bad_params => 0);
361 9 dpavlin
362 80 dpavlin sub esc_html {
363 my $html = shift;
364 $html =~ s/</&lt;/g;
365 $html =~ s/>/&gt;/g;
366 return $html;
367 }
368
369 202 dpavlin my $sort = 'swishrank';
370 if ($q->param("sort")) {
371 $sort = 'headline';
372 push @persist_vars, "sort";
373 }
374 660 dpavlin
375 my $sortby = $q->param("sortby");
376 if ($sortby) {
377 $sort = $sortby;
378 push @persist_vars, "sortby";
379 659 dpavlin }
380 698 dpavlin # used to filter entries in index and swish
381 my $filter = $q->param("filter");
382 202 dpavlin
383 47 dpavlin # construct swish query
384 my $sw_q = join(" and ",@s_arr);
385 636 dpavlin if (@path_arr && $q->param('show_full')) {
386 698 dpavlin $sw_q .= " and (swishdocpath=\"";
387 47 dpavlin $sw_q .= join("\" or swishdocpath=\"",@path_arr);
388 $sw_q .= "\")";
389 636 dpavlin $tmpl->param('full',1); # show full records
390 698 dpavlin # } elsif (@path_arr && $#path_arr == 0) {
391 # # I will assume that it's a filter since there isn't show_full
392 # $filter = shift @path_arr;
393 636 dpavlin } elsif ($q->param('show_full')) {
394 149 dpavlin # just show full path, no path defined
395 $tmpl->param('full',1);
396 121 dpavlin } else {
397 $tmpl->param('full',0);
398 47 dpavlin }
399 7 dpavlin
400 698 dpavlin if ($filter) {
401 $sw_q .= " and (swishdocpath=\"$filter\")" unless (@path_arr);
402 push @persist_vars, "filter";
403 709 dpavlin $url_params->{'filter'} = $filter;
404 $url_params_persist->{'filter'} = $filter;
405 698 dpavlin }
406
407 657 dpavlin my $swish_msg = ' ';
408
409 304 dpavlin # create new swish instance
410 my $swish = SWISH::API->new($INDEX);
411 657 dpavlin $swish_msg .= $swish->ErrorString." ".$swish->LastErrorMsg if $swish->Error;
412 47 dpavlin
413 304 dpavlin # execute query and get number of results from SWISH-E
414 my $search = $swish->New_Search_Object;
415
416 $search->SetSort($sort);
417
418 my $results = $search->Execute($sw_q);
419 657 dpavlin $swish_msg .= $swish->ErrorString." ".$swish->LastErrorMsg if $swish->Error;
420 304 dpavlin
421 my $hits = $results->Hits;
422
423 7 dpavlin $tmpl->param('hits',$hits);
424 657 dpavlin my $search_msg = $sw_q;
425 $search_msg .= '<em>'.$swish_msg.'</em>' if ($swish_msg);
426 $tmpl->param('search', $search_msg);
427 7 dpavlin
428 76 dpavlin $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
429 $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
430 51 dpavlin
431 698 dpavlin # URL parametars for search results
432 709 dpavlin sub cook_url_params {
433 my $hash = shift || return;
434 725 dpavlin return join("&", map {
435 my $var = $_;
436 if (ref($hash->{$var}) eq 'ARRAY') {
437 join('&',
438 map { $var.'='.$_ } @{$hash->{$var}}
439 );
440 } else {
441 $var."=".$hash->{$var};
442 }
443 } keys %{$hash});
444 709 dpavlin }
445 698 dpavlin
446 709 dpavlin $tmpl->param('url_params',"?".cook_url_params($url_params));
447 $tmpl->param('url_params_paths',"?".cook_url_params($url_params).'&'.join("&",map { my $t = $_; $t =~ s/\#/%23/g; "path=$t"; } @path_arr));
448 698 dpavlin
449
450 709 dpavlin
451 304 dpavlin #
452 # build pager
453 #
454 126 dpavlin
455 304 dpavlin my $current_page = $q->param('PAGER_offset') || 1;
456 7 dpavlin
457 304 dpavlin my $pager = Data::Pageset->new({
458 'total_entries' => $hits,
459 'entries_per_page' => $ON_PAGE,
460 'current_page' => $current_page,
461 'pages_per_set' => $pages_per_set,
462 });
463 7 dpavlin
464 304 dpavlin $results->SeekResult( $pager->first - 1 );
465 7 dpavlin
466 304 dpavlin # get number of entries on this page
467 my $i = $pager->entries_on_this_page;
468
469 # results from swish for template
470 my @pager_data_list;
471
472 for(my $i=$pager->first; $i<=$pager->last; $i++) {
473
474 my $result = $results->NextResult;
475 last if (! $result);
476
477 my $r = {
478 nr => $i,
479 path => $result->Property('swishdocpath'),
480 headline => esc_html($from_utf8->convert($result->Property('headline'))),
481 rank => $result->Property('swishrank')
482 };
483
484 709 dpavlin #$r->{html} = back2html($from_utf8->convert($result->Property('html')), cook_url_params($url_params_persist)) if ($q->param('show_full'));
485 $r->{html} = back2html($from_utf8->convert($result->Property('html')), $filter ? 'filter='.$filter : '') if ($q->param('show_full'));
486 304 dpavlin
487 push @pager_data_list, $r;
488 }
489
490
491
492 # put something in template
493 make_pager($q, $tmpl, $pager);
494 make_pager_vars($q, $tmpl, @persist_vars);
495 $tmpl->param('PAGER_DATA_LIST', \@pager_data_list);
496
497 my $html = $tmpl->output;
498
499 198 dpavlin return in_template($q,$html);
500 7 dpavlin }
501
502 9 dpavlin sub show_index {
503 my $self = shift;
504 my $i = shift; # field number
505
506 my $q = $self->query();
507
508 11 dpavlin my $field = $q->param("f$i");
509 my $limit = $q->param("v$i");
510
511 643 dpavlin my $filter = $q->param("filter");
512
513 9 dpavlin my $html;
514
515 53 dpavlin my $index = new index_DBI(
516 $cfg_global->val('global', 'dbi_dbd'),
517 $cfg_global->val('global', 'dbi_dsn'),
518 $cfg_global->val('global', 'dbi_user'),
519 $cfg_global->val('global', 'dbi_passwd') || ''
520 );
521 9 dpavlin
522 643 dpavlin my $total = $index->count($field,$limit,$filter);
523 304 dpavlin
524 643 dpavlin if (! defined($total)) {
525 198 dpavlin my $tmpl = $self->load_tmpl(url_ex($q,'no_index.html'));
526 11 dpavlin $tmpl->param('field',$field);
527 $html = $tmpl->output;
528 return $html;
529 }
530 9 dpavlin
531 670 dpavlin my $tmpl = $self->load_tmpl(url_ex($q,'index_res.html'), global_vars => 1, die_on_bad_params => 0);
532 12 dpavlin $tmpl->param('field',$field);
533 $tmpl->param('limit',$limit);
534 $tmpl->param('total',$total);
535 685 dpavlin $tmpl->param('filter',$filter);
536 11 dpavlin
537 51 dpavlin # FIXME I should set offset and leave out limit from fetch!!
538 16 dpavlin # if (! $q->param("PAGER_offset") {
539 # $q->p