6afdba89b9b5a5a3f25fbc739f32c300b7cdc873
[webpac] / WebPac.pm
1 package WebPac;
2
3 use base 'CGI::Application';
4 use strict;
5
6 use HTML::Pager;
7 use HTML::FillInForm;
8 use SWISH;
9 use Text::Iconv;
10 use DBI;
11 use Config::IniFiles;
12 use Text::Unaccent;
13
14 use lib '..';
15 use index_DBI;
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
30
31 Text::Iconv->raise_error(0);     # Conversion errors raise exceptions
32
33 my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
34
35
36 sub setup {
37         my $self = shift;
38         $self->tmpl_path($TEMPLATE_PATH);
39         $self->run_modes(
40                 'search' => 'show_search_form',
41                 'results' => 'show_results_list',
42 #               'user' => 'show_user_detail',
43                 'index' => 'show_index',
44         );
45         $self->start_mode('search');
46         $self->mode_param('rm');
47
48         $self->header_props(-charset=>$CHARSET);
49 }
50
51 sub show_search_form {
52         my $self = shift;
53
54         # Get the CGI.pm query object
55         my $q = $self->query();
56
57         my $tmpl = $self->load_tmpl('search.html');
58         my $html = $tmpl->output;
59
60         my $fif = new HTML::FillInForm;
61
62         return $fif->fill(scalarref => \$html, fobject => $q,
63                 target => 'search');
64 }
65  
66 sub show_results_list {
67         my $self = shift;
68
69         my $q = $self->query();
70
71         my @swish_results;      # results from swish
72
73         # load template for this page
74
75         my @s_arr;      # all queries are located here
76
77         my @path_arr = $q->param('path');
78         my $full = $q->param('full');
79
80         for(my $i = 1; $i <=30; $i++) {
81
82                 return show_index($self, $i) if ($q->param("f".$i."_index"));
83                 next if (! $q->param("v$i"));
84                 next if (! $q->param("f$i"));
85
86                 # re-write query from +/- to and/and not
87                 my @param_vals = $q->param("v$i");
88                 my @swish_q;
89                 while (my $search = shift @param_vals) {
90                         my $s;
91                         # remove accents
92                         $search = unac_string($CHARSET,$search);
93                         while ($search =~ s/\s*("[^"]+")\s*/ /) {
94                                 $s .= "$1 ";
95                         }
96                         $search =~ s/^\s+//;
97                         $search =~ s/\s+$//;
98
99                         foreach (split(/\s+/,$search)) {
100                                 if (m/^([+-])(\S+)/) {
101                                         $s.= ($s) ? "and " : "";
102                                         $s.="not " if ($1 eq "-");
103                                         $s.="$2* ";
104                                 } elsif (m/(and|or|not)/i) {
105                                         $s.="$_ ";
106                                 } else {
107                                         $s.="$_* ";
108                                 }
109                         }
110                         $s =~ s/\*+/*/g;
111                         push @swish_q,$s;
112                 }
113                 # FIXME default operator for multi-value fields is or. There is
114                 # no way to change it, except here for now. Is there need?
115                 push @s_arr, $q->param("f$i")."_swish=(".join(" or ",@swish_q).")";
116         }
117
118         my $tmpl = $self->load_tmpl('results.html');
119
120         # call swish
121         my $sh = SWISH->connect('Fork',
122                 prog     => $SWISH,
123                 indexes  => $INDEX,
124                 properties  => [qw/swishdocpath swishrank swishtitle headline html/],
125                 results  => sub {
126                         my ($sh,$hit) = @_;
127
128                         push @swish_results, {
129                                 nr => ($#swish_results + 2),
130                                 path => $hit->swishdocpath,
131                                 headline => $from_utf8->convert($hit->headline),
132                                 html => back2html($from_utf8->convert($hit->html)),
133                                 rank => $hit->swishrank };
134
135                 },
136                 #startnum => 0,
137                 maxhits => $MAX_HITS
138         );
139
140         die $SWISH::errstr unless $sh;
141         # construct swish query
142         my $sw_q = join(" and ",@s_arr);
143         if (@path_arr) {
144                 $sw_q .= "and (swishdocpath=\"";
145                 $sw_q .= join("\" or swishdocpath=\"",@path_arr);
146                 $sw_q .= "\")";
147                 $tmpl->param('full',1); # show full records
148         }
149
150         my $hits = $sh->query($sw_q);
151
152         $tmpl->param('hits',$hits);
153         $tmpl->param('search',$sw_q);
154
155         $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
156         $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
157
158         # create a Pager object
159         my $pager = HTML::Pager->new(
160                 # required parameters
161                 query => $q,
162                 get_data_callback => sub {
163                         my ($offset, $rows) = @_;
164
165                         my @result;
166                         for (my $i=0; $i<$rows; $i++) {
167                                 push @result, $swish_results[$offset+$i] if $swish_results[$offset+$i];
168                         }
169                         return \@result;
170                 },
171                 rows => $hits,
172                 page_size => $ON_PAGE,
173                 # some optional parameters
174                 persist_vars => [
175                         'rm',
176                         'f1', 'v1',
177                         'f2', 'v2',
178                         'f3', 'v3',
179                         'f4', 'v4',
180                         'f5', 'v5',
181                         'f6', 'v6',
182                         'f7', 'v7',
183                         'f8', 'v8',
184                         'f9', 'v9',
185                         ],
186                 #cell_space_color => '#000000',
187                 #cell_background_color => '#ffffff',
188                 #nav_background_color => '#dddddd',
189                 #javascript_presubmit => 'last_minute_javascript()',
190                 debug => 1,
191                 template => $tmpl,
192         );
193
194         my $html = $pager->output;
195
196         return $html;
197 }
198  
199 sub show_index {
200         my $self = shift;
201         my $i = shift;          # field number
202
203         my $q = $self->query();
204
205         my $field = $q->param("f$i");
206         my $limit = $q->param("v$i");
207
208         my $html;
209
210         my $index = new index_DBI(
211                 $cfg_global->val('global', 'dbi_dbd'),
212                 $cfg_global->val('global', 'dbi_dsn'),
213                 $cfg_global->val('global', 'dbi_user'),
214                 $cfg_global->val('global', 'dbi_passwd') || ''
215         );
216
217         my $total = $index->check($field);
218         if (! $total) {
219                 my $tmpl = $self->load_tmpl('no_index.html');
220                 $tmpl->param('field',$field);
221                 $html = $tmpl->output;
222                 return $html;
223         }
224
225         my $tmpl = $self->load_tmpl('index_res.html');
226         $tmpl->param('field',$field);
227         $tmpl->param('limit',$limit);
228         $tmpl->param('total',$total);
229
230 # FIXME I should set offset and leave out limit from fetch!!
231 #       if (! $q->param("PAGER_offset") {
232 #               $q->param("Pager_offet)
233 #       }
234
235         my $pager = HTML::Pager->new(
236                 query => $q,
237                 get_data_callback => sub {
238                         my ($offset, $rows) = @_;
239
240                         my @result = $index->fetch($field,'item',$limit, $offset, $rows);
241                         return \@result;
242                 },
243                 rows => $total,
244                 page_size => $ON_PAGE,
245                 persist_vars => [
246                         'rm', 
247                         "f$i", "v$i", "f".$i."_index", 
248                         'offset',
249                         ],
250                 debug => 1,
251                 template => $tmpl,
252         );
253
254         return $pager->output;
255 }
256
257 1;