/[webpac]/trunk/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 /trunk/WebPac.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 600 - (show annotations)
Wed Dec 8 18:24:24 2004 UTC (19 years, 3 months ago) by dpavlin
File size: 11692 byte(s)
improvement: path limit in URL works without full result display

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
14 use lib '..';
15 use index_DBI_cache;
16 use back2html;
17
18
19 # 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 my $MIN_WILDCARD =$cfg_global->val('webpac', 'min_wildcard') || 1;
30 my $TEMPLATE =$cfg_global->val('webpac', 'template');
31 my $UNAC_FILTER =$cfg_global->val('global', 'my_unac_filter');
32 my $BASE_PATH =$cfg_global->val('webpac', 'base_path');
33 # for pager
34 my $pages_per_set = $cfg_global->val('webpac', 'pages_per_set') || 10;
35
36 Text::Iconv->raise_error(0); # Conversion errors raise exceptions
37
38 my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
39
40 if ($UNAC_FILTER) {
41 require $UNAC_FILTER;
42 } else {
43 sub WebPac::my_unac_string {
44 my ($charset, $string) = (@_);
45 return $string;
46 }
47 }
48
49 # 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 $tpl = shift || die "url_ex needs template name!";
53 return suff2file($BASE_PATH, $q->url(-absolute => 1,-path => 1),$TEMPLATE_PATH,$tpl);
54 }
55
56 sub suff2file($$$$) {
57 my ($base_path, $p, $path, $tpl) = @_;
58
59 return $tpl if (! $base_path);
60
61 # strip everything to and including base path, leaving only
62 # additional (virtual) path
63 if ($base_path eq "/") {
64 $p =~ s,/*,,g;
65 my ($name,$ext) = split(/\./,$tpl);
66 $p = $name . "-" . $p . "." . $ext;
67 } elsif ($p =~ s,^.*?$base_path,,) {
68 $p =~ s,/*,,g;
69 my ($name,$ext) = split(/\./,$tpl);
70 $p = $name . $p . "." . $ext;
71 } else {
72 # if unable reset it!
73 $p = $tpl;
74 }
75
76 if ( -e "$path/$p") {
77 return $p;
78 } else {
79 return $tpl;
80 }
81
82 }
83
84 sub setup {
85 my $self = shift;
86 $self->tmpl_path($TEMPLATE_PATH);
87 $self->run_modes(
88 'search' => 'show_search_form',
89 'results' => 'show_results_list',
90 # 'user' => 'show_user_detail',
91 'index' => 'show_index',
92 );
93 $self->start_mode('search');
94 $self->mode_param('rm');
95
96 $self->header_props(-charset=>$CHARSET);
97 }
98
99 sub in_template {
100 my $q = shift || die "need CGI object!";
101 my $html = shift || die "This page is left unintentionally blank";
102 return $html if (! defined($TEMPLATE));
103
104 my ($dir,$tpl);
105 if ($TEMPLATE =~ m,^(.*?/*)([^/]+)$,) {
106 ($dir,$tpl) = ($1,$2);
107 } else {
108 die "can't parse TEMPLATE path";
109 }
110
111 my $master_tpl = suff2file($BASE_PATH, $q->url(-absolute => 1, -path => 1),$dir,$tpl);
112 if (open(T, $master_tpl)) {
113 my $template_html = join("\n",<T>);
114 close(T);
115 $template_html =~ s/##webpac##/$html/gsi;
116 return $template_html;
117 } else {
118 return "Can't read template '$master_tpl'";
119 }
120 }
121
122 #--------------------------------------------------------------------------
123
124 #
125 # make pager navigation and fill template variables
126 # compatibile with HTML::Pager
127 #
128
129 sub make_pager($$$) {
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 foreach my $val ($q->param($v)) {
187 $hidden_vars .= '<input type="hidden" name="'.$v.'" value="'.$val.'"/>'."\n";
188 }
189 }
190
191 $tmpl->param('PAGER_HIDDEN', $hidden_vars);
192 $tmpl->param('PAGER_JAVASCRIPT', qq{
193 <SCRIPT LANGUAGE="Javascript">
194 <!-- Begin
195 // dummy emulator for HTML::Pager templates
196 function PAGER_set_offset_and_submit() {
197 return true;
198 }
199 // End -->
200 </script>
201 });
202 }
203
204 #--------------------------------------------------------------------------
205
206 sub show_search_form {
207 my $self = shift;
208
209 # Get the CGI.pm query object
210 my $q = $self->query();
211
212 my $tmpl = $self->load_tmpl(url_ex($q,'search.html'));
213 my $html = $tmpl->output;
214
215 my $fif = new HTML::FillInForm;
216
217 return in_template($q,$fif->fill(scalarref => \$html, fobject => $q,
218 target => 'search'));
219 }
220
221 sub show_results_list {
222 my $self = shift;
223
224 my $q = $self->query();
225
226 # load template for this page
227
228 my @s_arr; # all queries are located here
229
230 my @path_arr = $q->param('path');
231 my $full = $q->param('full');
232
233 my @persist_vars = ( 'rm' );
234 my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.($q->param('PAGER_offset') || 0) );
235
236 # support parametars "f" and "v" for start
237 for(my $i = ""; $i <=30; $i++) {
238
239 return show_index($self, $i) if ($q->param("f".$i."_index"));
240
241 next if (! $q->param("v$i"));
242 next if (! $q->param("f$i"));
243
244 push @persist_vars, "f$i";
245 push @persist_vars, "v$i";
246 push @persist_vars, "e$i" if ($q->param("e$i"));
247
248 push @url_params,"f$i=".$q->url_param("f$i");
249 foreach my $v ($q->url_param("v$i")) {
250 push @url_params,"v$i=$v";
251 }
252 push @url_params,"e$i=".$q->url_param("e$i");
253
254 my $wc="*"; # swish wildcard
255 $wc="" if ($i eq ""); # don't apply wildcard on field 0
256
257 # re-write query from +/- to and/and not
258 my @param_vals = $q->param("v$i");
259 my @swish_q;
260 my ($pre,$post,$exact) = ('','','');
261 while (my $search = shift @param_vals) {
262 my $s;
263 # remove accents
264 $search = my_unac_string($CHARSET,$search);
265 while ($search =~ s/\s*("[^"]+")\s*/ /) {
266 $s .= "$1 ";
267 }
268 $search =~ s/^\s+//;
269 $search =~ s/\s+$//;
270
271 # filed e[nr] is exact match bitmask
272 # 1 = beginning, 2=end, 3=both
273 $pre = '"xxbxx ' if ($q->param("e$i") & 1);
274 $post = ' xxexx"' if ($q->param("e$i") & 2);
275 # add qotes on other side
276 if ($q->param("e$i")) {
277 $pre = '"' if (! $pre);
278 $post = '"' if (! $post);
279 # what about wildcards?
280 $wc = '';
281 $wc = '*' if ($q->param("e$i") & 4);
282 $exact = '_exact';
283 }
284
285 foreach (split(/\s+/,$search)) {
286 if (m/^([+-])(\S+)/) {
287 $s.= ($s) ? "and " : "";
288 $s.="not " if ($1 eq "-");
289 $s.=$2.$wc." ";
290 } elsif (m/^\s*(and|or|not)\s*$/i) {
291 $s.=$_." ";
292 # don't add * to words with less than x chars
293 } elsif (length($_) <= $MIN_WILDCARD) {
294 $s.=$_." ";
295 } else {
296 $s.=$_.$wc." ";
297 }
298 }
299 $s =~ s/\*+/*/g;
300 $s = $pre.$s.$post if ($q->param("e$i"));
301 push @swish_q,$s;
302 }
303 # FIXME default operator for multi-value fields is or. There is
304 # no way to change it, except here for now. Is there need?
305 push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")";
306 }
307
308 my $tmpl = $self->load_tmpl(url_ex($q,'results.html'), global_vars => 1);
309
310 $tmpl->param('url_params',"?".join("&",@url_params));
311
312 sub esc_html {
313 my $html = shift;
314 $html =~ s/</&lt;/g;
315 $html =~ s/>/&gt;/g;
316 return $html;
317 }
318
319 my $sort = 'swishrank';
320 if ($q->param("sort")) {
321 $sort = 'headline';
322 push @persist_vars, "sort";
323 }
324
325 # construct swish query
326 my $sw_q = join(" and ",@s_arr);
327 if (@path_arr) {
328 $sw_q .= "and (swishdocpath=\"";
329 $sw_q .= join("\" or swishdocpath=\"",@path_arr);
330 $sw_q .= "\")";
331 }
332
333 if ($q->param('show_full')) {
334 # just show full path, no path defined
335 $tmpl->param('full',1);
336 } else {
337 $tmpl->param('full',0);
338 }
339
340 # create new swish instance
341 my $swish = SWISH::API->new($INDEX);
342 $swish->AbortLastError if $swish->Error;
343
344 # execute query and get number of results from SWISH-E
345 my $search = $swish->New_Search_Object;
346
347 $search->SetSort($sort);
348
349 my $results = $search->Execute($sw_q);
350 $swish->AbortLastError if $swish->Error;
351
352 my $hits = $results->Hits;
353
354 $tmpl->param('hits',$hits);
355 $tmpl->param('search',$sw_q);
356
357 $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
358 $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
359
360 #
361 # build pager
362 #
363
364 my $current_page = $q->param('PAGER_offset') || 1;
365
366 my $pager = Data::Pageset->new({
367 'total_entries' => $hits,
368 'entries_per_page' => $ON_PAGE,
369 'current_page' => $current_page,
370 'pages_per_set' => $pages_per_set,
371 });
372
373 $results->SeekResult( $pager->first - 1 );
374
375 # get number of entries on this page
376 my $i = $pager->entries_on_this_page;
377
378 # results from swish for template
379 my @pager_data_list;
380
381 for(my $i=$pager->first; $i<=$pager->last; $i++) {
382
383 my $result = $results->NextResult;
384 last if (! $result);
385
386 my $r = {
387 nr => $i,
388 path => $result->Property('swishdocpath'),
389 headline => esc_html($from_utf8->convert($result->Property('headline'))),
390 rank => $result->Property('swishrank')
391 };
392
393 $r->{html} = back2html($from_utf8->convert($result->Property('html'))) if ($q->param('show_full'));
394
395 push @pager_data_list, $r;
396 }
397
398
399
400 # put something in template
401 make_pager($q, $tmpl, $pager);
402 make_pager_vars($q, $tmpl, @persist_vars);
403 $tmpl->param('PAGER_DATA_LIST', \@pager_data_list);
404
405 my $html = $tmpl->output;
406
407 return in_template($q,$html);
408 }
409
410 sub show_index {
411 my $self = shift;
412 my $i = shift; # field number
413
414 my $q = $self->query();
415
416 my $field = $q->param("f$i");
417 my $limit = $q->param("v$i");
418
419 my $html;
420
421 my $index = new index_DBI(
422 $cfg_global->val('global', 'dbi_dbd'),
423 $cfg_global->val('global', 'dbi_dsn'),
424 $cfg_global->val('global', 'dbi_user'),
425 $cfg_global->val('global', 'dbi_passwd') || ''
426 );
427
428 my $total = $index->count($field,$limit);
429
430 if (! $total) {
431 my $tmpl = $self->load_tmpl(url_ex($q,'no_index.html'));
432 $tmpl->param('field',$field);
433 $html = $tmpl->output;
434 return $html;
435 }
436
437 my $tmpl = $self->load_tmpl(url_ex($q,'index_res.html'), global_vars => 1);
438 $tmpl->param('field',$field);
439 $tmpl->param('limit',$limit);
440 $tmpl->param('total',$total);
441
442 # FIXME I should set offset and leave out limit from fetch!!
443 # if (! $q->param("PAGER_offset") {
444 # $q->param("Pager_offet)
445 # }
446
447
448 #
449 # build pager
450 #
451 my $pager = Data::Pageset->new({
452 'total_entries' => $total,
453 'entries_per_page' => $ON_PAGE,
454 'current_page' => $q->param('PAGER_offset') || 1,
455 'pages_per_set' => $pages_per_set
456 });
457
458 my @persist_vars = qw{rm f$i v$i f$i_index offset};
459
460 make_pager($q, $tmpl, $pager);
461 make_pager_vars($q, $tmpl, @persist_vars);
462
463 my @pager_data_list = $index->fetch($field,$limit, $pager->first - 1, $pager->entries_on_this_page);
464 $tmpl->param('PAGER_DATA_LIST', \@pager_data_list);
465
466 return in_template($q,$tmpl->output);
467 }
468
469 1;

Properties

Name Value
cvs2svn:cvs-rev 1.40

  ViewVC Help
Powered by ViewVC 1.1.26