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 |
my $MIN_WILDCARD =$cfg_global->val('webpac', 'min_wildcard') || 1; |
30 |
|
31 |
|
32 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
33 |
|
34 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
35 |
|
36 |
|
37 |
sub setup { |
38 |
my $self = shift; |
39 |
$self->tmpl_path($TEMPLATE_PATH); |
40 |
$self->run_modes( |
41 |
'search' => 'show_search_form', |
42 |
'results' => 'show_results_list', |
43 |
# 'user' => 'show_user_detail', |
44 |
'index' => 'show_index', |
45 |
); |
46 |
$self->start_mode('search'); |
47 |
$self->mode_param('rm'); |
48 |
|
49 |
$self->header_props(-charset=>$CHARSET); |
50 |
} |
51 |
|
52 |
sub show_search_form { |
53 |
my $self = shift; |
54 |
|
55 |
# Get the CGI.pm query object |
56 |
my $q = $self->query(); |
57 |
|
58 |
my $tmpl = $self->load_tmpl('search.html'); |
59 |
my $html = $tmpl->output; |
60 |
|
61 |
my $fif = new HTML::FillInForm; |
62 |
|
63 |
return $fif->fill(scalarref => \$html, fobject => $q, |
64 |
target => 'search'); |
65 |
} |
66 |
|
67 |
sub show_results_list { |
68 |
my $self = shift; |
69 |
|
70 |
my $q = $self->query(); |
71 |
|
72 |
my @swish_results; # results from swish |
73 |
|
74 |
# load template for this page |
75 |
|
76 |
my @s_arr; # all queries are located here |
77 |
|
78 |
my @path_arr = $q->param('path'); |
79 |
my $full = $q->param('full'); |
80 |
|
81 |
my @persist_vars = ( 'rm' ); |
82 |
|
83 |
for(my $i = 1; $i <=30; $i++) { |
84 |
|
85 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
86 |
|
87 |
next if (! $q->param("v$i")); |
88 |
next if (! $q->param("f$i")); |
89 |
|
90 |
push @persist_vars, "f$i"; |
91 |
push @persist_vars, "v$i"; |
92 |
|
93 |
# re-write query from +/- to and/and not |
94 |
my @param_vals = $q->param("v$i"); |
95 |
my @swish_q; |
96 |
while (my $search = shift @param_vals) { |
97 |
my $s; |
98 |
# remove accents |
99 |
$search = unac_string($CHARSET,$search); |
100 |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
101 |
$s .= "$1 "; |
102 |
} |
103 |
$search =~ s/^\s+//; |
104 |
$search =~ s/\s+$//; |
105 |
|
106 |
foreach (split(/\s+/,$search)) { |
107 |
if (m/^([+-])(\S+)/) { |
108 |
$s.= ($s) ? "and " : ""; |
109 |
$s.="not " if ($1 eq "-"); |
110 |
$s.="$2* "; |
111 |
} elsif (m/^\s*(and|or|not)\s*$/i) { |
112 |
$s.="$_ "; |
113 |
# don't add * to words with less than x chars |
114 |
} elsif (length($_) <= $MIN_WILDCARD) { |
115 |
$s.="$_ "; |
116 |
} else { |
117 |
$s.="$_* "; |
118 |
} |
119 |
} |
120 |
$s =~ s/\*+/*/g; |
121 |
push @swish_q,$s; |
122 |
} |
123 |
# FIXME default operator for multi-value fields is or. There is |
124 |
# no way to change it, except here for now. Is there need? |
125 |
push @s_arr, $q->param("f$i")."_swish=(".join(" or ",@swish_q).")"; |
126 |
} |
127 |
|
128 |
my $tmpl = $self->load_tmpl('results.html'); |
129 |
|
130 |
sub esc_html { |
131 |
my $html = shift; |
132 |
$html =~ s/</</g; |
133 |
$html =~ s/>/>/g; |
134 |
return $html; |
135 |
} |
136 |
|
137 |
# call swish |
138 |
my $sh = SWISH->connect('Fork', |
139 |
prog => $SWISH, |
140 |
indexes => $INDEX, |
141 |
properties => [qw/swishdocpath swishrank swishtitle headline html/], |
142 |
results => sub { |
143 |
my ($sh,$hit) = @_; |
144 |
|
145 |
push @swish_results, { |
146 |
nr => ($#swish_results + 2), |
147 |
path => $hit->swishdocpath, |
148 |
headline => esc_html($from_utf8->convert($hit->headline)), |
149 |
html => back2html($from_utf8->convert($hit->html)), |
150 |
rank => $hit->swishrank }; |
151 |
|
152 |
}, |
153 |
#startnum => 0, |
154 |
maxhits => $MAX_HITS |
155 |
); |
156 |
|
157 |
die $SWISH::errstr unless $sh; |
158 |
# construct swish query |
159 |
my $sw_q = join(" and ",@s_arr); |
160 |
if (@path_arr && $q->param('show_full')) { |
161 |
$sw_q .= "and (swishdocpath=\""; |
162 |
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
163 |
$sw_q .= "\")"; |
164 |
$tmpl->param('full',1); # show full records |
165 |
} else { |
166 |
$tmpl->param('full',0); |
167 |
} |
168 |
|
169 |
my $hits = $sh->query($sw_q); |
170 |
|
171 |
$tmpl->param('hits',$hits); |
172 |
$tmpl->param('search',$sw_q); |
173 |
|
174 |
$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0); |
175 |
$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0); |
176 |
|
177 |
# create a Pager object |
178 |
my $pager = HTML::Pager->new( |
179 |
# required parameters |
180 |
query => $q, |
181 |
get_data_callback => sub { |
182 |
my ($offset, $rows) = @_; |
183 |
|
184 |
my @result; |
185 |
for (my $i=0; $i<$rows; $i++) { |
186 |
my $r = $swish_results[$offset+$i]; |
187 |
if ($r && $tmpl->param('full')) { |
188 |
push @result, $r; |
189 |
} elsif ($r) { |
190 |
# if not full output, skip html |
191 |
delete $r->{html}; |
192 |
push @result, $r; |
193 |
} |
194 |
} |
195 |
return \@result; |
196 |
}, |
197 |
rows => $hits, |
198 |
page_size => $ON_PAGE, |
199 |
# some optional parameters |
200 |
persist_vars => [ @persist_vars ], |
201 |
#cell_space_color => '#000000', |
202 |
#cell_background_color => '#ffffff', |
203 |
#nav_background_color => '#dddddd', |
204 |
#javascript_presubmit => 'last_minute_javascript()', |
205 |
debug => 1, |
206 |
template => $tmpl, |
207 |
); |
208 |
|
209 |
my $html = $pager->output; |
210 |
|
211 |
return $html; |
212 |
} |
213 |
|
214 |
sub show_index { |
215 |
my $self = shift; |
216 |
my $i = shift; # field number |
217 |
|
218 |
my $q = $self->query(); |
219 |
|
220 |
my $field = $q->param("f$i"); |
221 |
my $limit = $q->param("v$i"); |
222 |
|
223 |
my $html; |
224 |
|
225 |
my $index = new index_DBI( |
226 |
$cfg_global->val('global', 'dbi_dbd'), |
227 |
$cfg_global->val('global', 'dbi_dsn'), |
228 |
$cfg_global->val('global', 'dbi_user'), |
229 |
$cfg_global->val('global', 'dbi_passwd') || '' |
230 |
); |
231 |
|
232 |
my $total = $index->check($field); |
233 |
if (! $total) { |
234 |
my $tmpl = $self->load_tmpl('no_index.html'); |
235 |
$tmpl->param('field',$field); |
236 |
$html = $tmpl->output; |
237 |
return $html; |
238 |
} |
239 |
|
240 |
my $tmpl = $self->load_tmpl('index_res.html'); |
241 |
$tmpl->param('field',$field); |
242 |
$tmpl->param('limit',$limit); |
243 |
$tmpl->param('total',$total); |
244 |
|
245 |
# FIXME I should set offset and leave out limit from fetch!! |
246 |
# if (! $q->param("PAGER_offset") { |
247 |
# $q->param("Pager_offet) |
248 |
# } |
249 |
|
250 |
my $pager = HTML::Pager->new( |
251 |
query => $q, |
252 |
get_data_callback => sub { |
253 |
my ($offset, $rows) = @_; |
254 |
|
255 |
my @result = $index->fetch($field,'item',$limit, $offset, $rows); |
256 |
return \@result; |
257 |
}, |
258 |
rows => $total, |
259 |
page_size => $ON_PAGE, |
260 |
persist_vars => [ |
261 |
'rm', |
262 |
"f$i", "v$i", "f".$i."_index", |
263 |
'offset', |
264 |
], |
265 |
debug => 1, |
266 |
template => $tmpl, |
267 |
); |
268 |
|
269 |
return $pager->output; |
270 |
} |
271 |
|
272 |
1; |