/[webpac]/trunk2/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

Annotation of /trunk2/WebPac.pm

Parent Directory Parent Directory | Revision Log Revision Log


Revision 186 - (hide annotations)
Sat Nov 29 18:40:19 2003 UTC (20 years, 4 months ago) by dpavlin
Original Path: trunk/WebPac.pm
File size: 8001 byte(s)
transfer swish_exact flash in form also, not only in URL

1 dpavlin 7 package WebPac;
2    
3     use base 'CGI::Application';
4     use strict;
5    
6     use HTML::Pager;
7     use HTML::FillInForm;
8     use SWISH;
9 dpavlin 14 use Text::Iconv;
10 dpavlin 9 use DBI;
11 dpavlin 53 use Config::IniFiles;
12 dpavlin 73 use Text::Unaccent;
13 dpavlin 7
14 dpavlin 11 use lib '..';
15 dpavlin 128 use index_DBI_cache;
16 dpavlin 13 use back2html;
17 dpavlin 11
18 dpavlin 7
19 dpavlin 76 # 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 dpavlin 120 my $MIN_WILDCARD =$cfg_global->val('webpac', 'min_wildcard') || 1;
30 dpavlin 147 my $TEMPLATE =$cfg_global->val('webpac', 'template');
31 dpavlin 164 my $UNAC_FILTER =$cfg_global->val('global', 'unac_filter');
32 dpavlin 76
33 dpavlin 164 if ($UNAC_FILTER) {
34     require $UNAC_FILTER;
35     }
36 dpavlin 76
37 dpavlin 30 Text::Iconv->raise_error(0); # Conversion errors raise exceptions
38 dpavlin 7
39 dpavlin 14 my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
40    
41 dpavlin 53
42 dpavlin 7 sub setup {
43     my $self = shift;
44     $self->tmpl_path($TEMPLATE_PATH);
45     $self->run_modes(
46     'search' => 'show_search_form',
47     'results' => 'show_results_list',
48 dpavlin 9 # 'user' => 'show_user_detail',
49     'index' => 'show_index',
50 dpavlin 7 );
51     $self->start_mode('search');
52     $self->mode_param('rm');
53    
54     $self->header_props(-charset=>$CHARSET);
55     }
56    
57 dpavlin 147 sub in_template {
58     my $html = shift || "This page is left unintentionally blank";
59     return $html if (! defined($TEMPLATE));
60     if (open(T, $TEMPLATE)) {
61     my $template_html = join("\n",<T>);
62     close(T);
63     $template_html =~ s/##webpac##/$html/gsi;
64     return $template_html;
65     } else {
66     return "Can't read template '$TEMPLATE'";
67     }
68     }
69    
70 dpavlin 7 sub show_search_form {
71     my $self = shift;
72    
73     # Get the CGI.pm query object
74     my $q = $self->query();
75    
76     my $tmpl = $self->load_tmpl('search.html');
77     my $html = $tmpl->output;
78    
79     my $fif = new HTML::FillInForm;
80    
81 dpavlin 147 return in_template($fif->fill(scalarref => \$html, fobject => $q,
82     target => 'search'));
83 dpavlin 7 }
84    
85     sub show_results_list {
86     my $self = shift;
87    
88     my $q = $self->query();
89    
90     my @swish_results; # results from swish
91    
92     # load template for this page
93    
94     my @s_arr; # all queries are located here
95    
96 dpavlin 47 my @path_arr = $q->param('path');
97     my $full = $q->param('full');
98    
99 dpavlin 112 my @persist_vars = ( 'rm' );
100 dpavlin 126 my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.$q->param('PAGER_offset') || 0 );
101 dpavlin 112
102 dpavlin 150 # support parametars "f" and "v" for start
103     for(my $i = ""; $i <=30; $i++) {
104 dpavlin 7
105 dpavlin 9 return show_index($self, $i) if ($q->param("f".$i."_index"));
106 dpavlin 112
107 dpavlin 71 next if (! $q->param("v$i"));
108 dpavlin 9 next if (! $q->param("f$i"));
109 dpavlin 7
110 dpavlin 112 push @persist_vars, "f$i";
111     push @persist_vars, "v$i";
112 dpavlin 186 push @persist_vars, "e$i" if ($q->param("e$i"));
113 dpavlin 112
114 dpavlin 126 push @url_params,"f$i=".$q->url_param("f$i");
115 dpavlin 158 foreach my $v ($q->url_param("v$i")) {
116     push @url_params,"v$i=$v";
117     }
118 dpavlin 183 push @url_params,"e$i=".$q->url_param("e$i");
119 dpavlin 126
120 dpavlin 150 my $wc="*"; # swish wildcard
121     $wc="" if ($i eq ""); # don't apply wildcard on field 0
122    
123 dpavlin 7 # re-write query from +/- to and/and not
124 dpavlin 73 my @param_vals = $q->param("v$i");
125     my @swish_q;
126 dpavlin 163 my ($pre,$post,$exact) = ('','','');
127 dpavlin 73 while (my $search = shift @param_vals) {
128     my $s;
129     # remove accents
130     $search = unac_string($CHARSET,$search);
131     while ($search =~ s/\s*("[^"]+")\s*/ /) {
132     $s .= "$1 ";
133     }
134     $search =~ s/^\s+//;
135     $search =~ s/\s+$//;
136 dpavlin 7
137 dpavlin 163 # filed e[nr] is exact match bitmask
138     # 1 = beginning, 2=end, 3=both
139     $pre = '"xxbxx ' if ($q->param("e$i") & 1);
140     $post = ' xxexx"' if ($q->param("e$i") & 2);
141     # add qotes on other side
142     if ($q->param("e$i")) {
143     $pre = '"' if (! $pre);
144     $post = '"' if (! $post);
145     $wc = ''; # don't use windcard in exact
146     $exact = '_exact';
147     }
148    
149 dpavlin 73 foreach (split(/\s+/,$search)) {
150     if (m/^([+-])(\S+)/) {
151     $s.= ($s) ? "and " : "";
152     $s.="not " if ($1 eq "-");
153 dpavlin 163 $s.=$2.$wc." ";
154 dpavlin 122 } elsif (m/^\s*(and|or|not)\s*$/i) {
155 dpavlin 163 $s.=$_." ";
156 dpavlin 120 # don't add * to words with less than x chars
157     } elsif (length($_) <= $MIN_WILDCARD) {
158 dpavlin 163 $s.=$_." ";
159 dpavlin 73 } else {
160 dpavlin 163 $s.=$_.$wc." ";
161 dpavlin 73 }
162 dpavlin 7 }
163 dpavlin 73 $s =~ s/\*+/*/g;
164 dpavlin 163 $s = $pre.$s.$post if ($q->param("e$i"));
165 dpavlin 73 push @swish_q,$s;
166 dpavlin 7 }
167 dpavlin 73 # FIXME default operator for multi-value fields is or. There is
168     # no way to change it, except here for now. Is there need?
169 dpavlin 163 push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")";
170 dpavlin 7 }
171    
172 dpavlin 126 my $tmpl = $self->load_tmpl('results.html', global_vars => 1);
173 dpavlin 9
174 dpavlin 80 sub esc_html {
175     my $html = shift;
176     $html =~ s/</&lt;/g;
177     $html =~ s/>/&gt;/g;
178     return $html;
179     }
180    
181 dpavlin 7 # call swish
182     my $sh = SWISH->connect('Fork',
183     prog => $SWISH,
184     indexes => $INDEX,
185 dpavlin 13 properties => [qw/swishdocpath swishrank swishtitle headline html/],
186 dpavlin 7 results => sub {
187     my ($sh,$hit) = @_;
188    
189     push @swish_results, {
190     nr => ($#swish_results + 2),
191     path => $hit->swishdocpath,
192 dpavlin 80 headline => esc_html($from_utf8->convert($hit->headline)),
193 dpavlin 41 html => back2html($from_utf8->convert($hit->html)),
194 dpavlin 7 rank => $hit->swishrank };
195    
196     },
197     #startnum => 0,
198 dpavlin 47 maxhits => $MAX_HITS
199 dpavlin 7 );
200    
201     die $SWISH::errstr unless $sh;
202 dpavlin 47 # construct swish query
203     my $sw_q = join(" and ",@s_arr);
204 dpavlin 111 if (@path_arr && $q->param('show_full')) {
205 dpavlin 47 $sw_q .= "and (swishdocpath=\"";
206     $sw_q .= join("\" or swishdocpath=\"",@path_arr);
207     $sw_q .= "\")";
208     $tmpl->param('full',1); # show full records
209 dpavlin 149 } elsif ($q->param('show_full')) {
210     # just show full path, no path defined
211     $tmpl->param('full',1);
212 dpavlin 121 } else {
213     $tmpl->param('full',0);
214 dpavlin 47 }
215 dpavlin 7
216 dpavlin 47 my $hits = $sh->query($sw_q);
217    
218 dpavlin 7 $tmpl->param('hits',$hits);
219 dpavlin 47 $tmpl->param('search',$sw_q);
220 dpavlin 7
221 dpavlin 76 $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
222     $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
223 dpavlin 51
224 dpavlin 126 $tmpl->param('url_params',"?".join("&",@url_params));
225    
226 dpavlin 7 # create a Pager object
227     my $pager = HTML::Pager->new(
228     # required parameters
229     query => $q,
230     get_data_callback => sub {
231     my ($offset, $rows) = @_;
232    
233     my @result;
234     for (my $i=0; $i<$rows; $i++) {
235 dpavlin 114 my $r = $swish_results[$offset+$i];
236 dpavlin 121 if ($r && $tmpl->param('full')) {
237 dpavlin 114 push @result, $r;
238     } elsif ($r) {
239     # if not full output, skip html
240     delete $r->{html};
241     push @result, $r;
242     }
243 dpavlin 7 }
244     return \@result;
245     },
246     rows => $hits,
247     page_size => $ON_PAGE,
248     # some optional parameters
249 dpavlin 112 persist_vars => [ @persist_vars ],
250 dpavlin 7 #cell_space_color => '#000000',
251     #cell_background_color => '#ffffff',
252     #nav_background_color => '#dddddd',
253     #javascript_presubmit => 'last_minute_javascript()',
254     debug => 1,
255     template => $tmpl,
256     );
257    
258     my $html = $pager->output;
259    
260 dpavlin 147 return in_template($html);
261 dpavlin 7 }
262    
263 dpavlin 9 sub show_index {
264     my $self = shift;
265     my $i = shift; # field number
266    
267     my $q = $self->query();
268    
269 dpavlin 11 my $field = $q->param("f$i");
270     my $limit = $q->param("v$i");
271    
272 dpavlin 9 my $html;
273    
274 dpavlin 53 my $index = new index_DBI(
275     $cfg_global->val('global', 'dbi_dbd'),
276     $cfg_global->val('global', 'dbi_dsn'),
277     $cfg_global->val('global', 'dbi_user'),
278     $cfg_global->val('global', 'dbi_passwd') || ''
279     );
280 dpavlin 9
281 dpavlin 140 my $total = $index->count($field,$limit);
282 dpavlin 12 if (! $total) {
283 dpavlin 11 my $tmpl = $self->load_tmpl('no_index.html');
284     $tmpl->param('field',$field);
285     $html = $tmpl->output;
286     return $html;
287     }
288 dpavlin 9
289 dpavlin 124 my $tmpl = $self->load_tmpl('index_res.html', global_vars => 1);
290 dpavlin 12 $tmpl->param('field',$field);
291     $tmpl->param('limit',$limit);
292     $tmpl->param('total',$total);
293 dpavlin 11
294 dpavlin 51 # FIXME I should set offset and leave out limit from fetch!!
295 dpavlin 16 # if (! $q->param("PAGER_offset") {
296     # $q->param("Pager_offet)
297     # }
298    
299 dpavlin 12 my $pager = HTML::Pager->new(
300     query => $q,
301     get_data_callback => sub {
302     my ($offset, $rows) = @_;
303 dpavlin 11
304 dpavlin 140 my @result = $index->fetch($field,$limit, $offset, $rows);
305 dpavlin 12 return \@result;
306     },
307     rows => $total,
308     page_size => $ON_PAGE,
309     persist_vars => [
310     'rm',
311     "f$i", "v$i", "f".$i."_index",
312     'offset',
313     ],
314     debug => 1,
315     template => $tmpl,
316     );
317 dpavlin 11
318 dpavlin 147 return in_template($pager->output);
319 dpavlin 9 }
320    
321 dpavlin 7 1;

Properties

Name Value
cvs2svn:cvs-rev 1.37

  ViewVC Help
Powered by ViewVC 1.1.26