/[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 190 - (hide annotations)
Sat Nov 29 19:11:23 2003 UTC (20 years, 4 months ago) by dpavlin
Original Path: trunk/WebPac.pm
File size: 8040 byte(s)
exact matches can now specify to include (or not) wildcard at end. Added
also documentation about exact matching using swish-e

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 dpavlin 190 # what about wildcards?
146     $wc = '';
147     $wc = '*' if ($q->param("e$i") & 4);
148 dpavlin 163 $exact = '_exact';
149     }
150    
151 dpavlin 73 foreach (split(/\s+/,$search)) {
152     if (m/^([+-])(\S+)/) {
153     $s.= ($s) ? "and " : "";
154     $s.="not " if ($1 eq "-");
155 dpavlin 163 $s.=$2.$wc." ";
156 dpavlin 122 } elsif (m/^\s*(and|or|not)\s*$/i) {
157 dpavlin 163 $s.=$_." ";
158 dpavlin 120 # don't add * to words with less than x chars
159     } elsif (length($_) <= $MIN_WILDCARD) {
160 dpavlin 163 $s.=$_." ";
161 dpavlin 73 } else {
162 dpavlin 163 $s.=$_.$wc." ";
163 dpavlin 73 }
164 dpavlin 7 }
165 dpavlin 73 $s =~ s/\*+/*/g;
166 dpavlin 163 $s = $pre.$s.$post if ($q->param("e$i"));
167 dpavlin 73 push @swish_q,$s;
168 dpavlin 7 }
169 dpavlin 73 # FIXME default operator for multi-value fields is or. There is
170     # no way to change it, except here for now. Is there need?
171 dpavlin 163 push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")";
172 dpavlin 7 }
173    
174 dpavlin 126 my $tmpl = $self->load_tmpl('results.html', global_vars => 1);
175 dpavlin 9
176 dpavlin 80 sub esc_html {
177     my $html = shift;
178     $html =~ s/</&lt;/g;
179     $html =~ s/>/&gt;/g;
180     return $html;
181     }
182    
183 dpavlin 7 # call swish
184     my $sh = SWISH->connect('Fork',
185     prog => $SWISH,
186     indexes => $INDEX,
187 dpavlin 13 properties => [qw/swishdocpath swishrank swishtitle headline html/],
188 dpavlin 7 results => sub {
189     my ($sh,$hit) = @_;
190    
191     push @swish_results, {
192     nr => ($#swish_results + 2),
193     path => $hit->swishdocpath,
194 dpavlin 80 headline => esc_html($from_utf8->convert($hit->headline)),
195 dpavlin 41 html => back2html($from_utf8->convert($hit->html)),
196 dpavlin 7 rank => $hit->swishrank };
197    
198     },
199     #startnum => 0,
200 dpavlin 47 maxhits => $MAX_HITS
201 dpavlin 7 );
202    
203     die $SWISH::errstr unless $sh;
204 dpavlin 47 # construct swish query
205     my $sw_q = join(" and ",@s_arr);
206 dpavlin 111 if (@path_arr && $q->param('show_full')) {
207 dpavlin 47 $sw_q .= "and (swishdocpath=\"";
208     $sw_q .= join("\" or swishdocpath=\"",@path_arr);
209     $sw_q .= "\")";
210     $tmpl->param('full',1); # show full records
211 dpavlin 149 } elsif ($q->param('show_full')) {
212     # just show full path, no path defined
213     $tmpl->param('full',1);
214 dpavlin 121 } else {
215     $tmpl->param('full',0);
216 dpavlin 47 }
217 dpavlin 7
218 dpavlin 47 my $hits = $sh->query($sw_q);
219    
220 dpavlin 7 $tmpl->param('hits',$hits);
221 dpavlin 47 $tmpl->param('search',$sw_q);
222 dpavlin 7
223 dpavlin 76 $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
224     $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
225 dpavlin 51
226 dpavlin 126 $tmpl->param('url_params',"?".join("&",@url_params));
227    
228 dpavlin 7 # create a Pager object
229     my $pager = HTML::Pager->new(
230     # required parameters
231     query => $q,
232     get_data_callback => sub {
233     my ($offset, $rows) = @_;
234    
235     my @result;
236     for (my $i=0; $i<$rows; $i++) {
237 dpavlin 114 my $r = $swish_results[$offset+$i];
238 dpavlin 121 if ($r && $tmpl->param('full')) {
239 dpavlin 114 push @result, $r;
240     } elsif ($r) {
241     # if not full output, skip html
242     delete $r->{html};
243     push @result, $r;
244     }
245 dpavlin 7 }
246     return \@result;
247     },
248     rows => $hits,
249     page_size => $ON_PAGE,
250     # some optional parameters
251 dpavlin 112 persist_vars => [ @persist_vars ],
252 dpavlin 7 #cell_space_color => '#000000',
253     #cell_background_color => '#ffffff',
254     #nav_background_color => '#dddddd',
255     #javascript_presubmit => 'last_minute_javascript()',
256     debug => 1,
257     template => $tmpl,
258     );
259    
260     my $html = $pager->output;
261    
262 dpavlin 147 return in_template($html);
263 dpavlin 7 }
264    
265 dpavlin 9 sub show_index {
266     my $self = shift;
267     my $i = shift; # field number
268    
269     my $q = $self->query();
270    
271 dpavlin 11 my $field = $q->param("f$i");
272     my $limit = $q->param("v$i");
273    
274 dpavlin 9 my $html;
275    
276 dpavlin 53 my $index = new index_DBI(
277     $cfg_global->val('global', 'dbi_dbd'),
278     $cfg_global->val('global', 'dbi_dsn'),
279     $cfg_global->val('global', 'dbi_user'),
280     $cfg_global->val('global', 'dbi_passwd') || ''
281     );
282 dpavlin 9
283 dpavlin 140 my $total = $index->count($field,$limit);
284 dpavlin 12 if (! $total) {
285 dpavlin 11 my $tmpl = $self->load_tmpl('no_index.html');
286     $tmpl->param('field',$field);
287     $html = $tmpl->output;
288     return $html;
289     }
290 dpavlin 9
291 dpavlin 124 my $tmpl = $self->load_tmpl('index_res.html', global_vars => 1);
292 dpavlin 12 $tmpl->param('field',$field);
293     $tmpl->param('limit',$limit);
294     $tmpl->param('total',$total);
295 dpavlin 11
296 dpavlin 51 # FIXME I should set offset and leave out limit from fetch!!
297 dpavlin 16 # if (! $q->param("PAGER_offset") {
298     # $q->param("Pager_offet)
299     # }
300    
301 dpavlin 12 my $pager = HTML::Pager->new(
302     query => $q,
303     get_data_callback => sub {
304     my ($offset, $rows) = @_;
305 dpavlin 11
306 dpavlin 140 my @result = $index->fetch($field,$limit, $offset, $rows);
307 dpavlin 12 return \@result;
308     },
309     rows => $total,
310     page_size => $ON_PAGE,
311     persist_vars => [
312     'rm',
313     "f$i", "v$i", "f".$i."_index",
314     'offset',
315     ],
316     debug => 1,
317     template => $tmpl,
318     );
319 dpavlin 11
320 dpavlin 147 return in_template($pager->output);
321 dpavlin 9 }
322    
323 dpavlin 7 1;

Properties

Name Value
cvs2svn:cvs-rev 1.38

  ViewVC Help
Powered by ViewVC 1.1.26