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