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 |
|
|
use index_DBI; |
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 |
|
|
|
30 |
|
|
|
31 |
dpavlin |
30 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
32 |
dpavlin |
7 |
|
33 |
dpavlin |
14 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
34 |
|
|
|
35 |
dpavlin |
53 |
|
36 |
dpavlin |
7 |
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 |
dpavlin |
9 |
# 'user' => 'show_user_detail', |
43 |
|
|
'index' => 'show_index', |
44 |
dpavlin |
7 |
); |
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 |
dpavlin |
47 |
my @path_arr = $q->param('path'); |
78 |
|
|
my $full = $q->param('full'); |
79 |
|
|
|
80 |
dpavlin |
71 |
for(my $i = 1; $i <=30; $i++) { |
81 |
dpavlin |
7 |
|
82 |
dpavlin |
9 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
83 |
dpavlin |
71 |
next if (! $q->param("v$i")); |
84 |
dpavlin |
9 |
next if (! $q->param("f$i")); |
85 |
dpavlin |
7 |
|
86 |
|
|
# re-write query from +/- to and/and not |
87 |
dpavlin |
73 |
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 |
dpavlin |
7 |
|
99 |
dpavlin |
73 |
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 |
dpavlin |
7 |
} |
110 |
dpavlin |
73 |
$s =~ s/\*+/*/g; |
111 |
|
|
push @swish_q,$s; |
112 |
dpavlin |
7 |
} |
113 |
dpavlin |
73 |
# 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 |
dpavlin |
7 |
} |
117 |
|
|
|
118 |
dpavlin |
9 |
my $tmpl = $self->load_tmpl('results.html'); |
119 |
|
|
|
120 |
dpavlin |
80 |
sub esc_html { |
121 |
|
|
my $html = shift; |
122 |
|
|
$html =~ s/</</g; |
123 |
|
|
$html =~ s/>/>/g; |
124 |
|
|
return $html; |
125 |
|
|
} |
126 |
|
|
|
127 |
dpavlin |
7 |
# call swish |
128 |
|
|
my $sh = SWISH->connect('Fork', |
129 |
|
|
prog => $SWISH, |
130 |
|
|
indexes => $INDEX, |
131 |
dpavlin |
13 |
properties => [qw/swishdocpath swishrank swishtitle headline html/], |
132 |
dpavlin |
7 |
results => sub { |
133 |
|
|
my ($sh,$hit) = @_; |
134 |
|
|
|
135 |
|
|
push @swish_results, { |
136 |
|
|
nr => ($#swish_results + 2), |
137 |
|
|
path => $hit->swishdocpath, |
138 |
dpavlin |
80 |
headline => esc_html($from_utf8->convert($hit->headline)), |
139 |
dpavlin |
41 |
html => back2html($from_utf8->convert($hit->html)), |
140 |
dpavlin |
7 |
rank => $hit->swishrank }; |
141 |
|
|
|
142 |
|
|
}, |
143 |
|
|
#startnum => 0, |
144 |
dpavlin |
47 |
maxhits => $MAX_HITS |
145 |
dpavlin |
7 |
); |
146 |
|
|
|
147 |
|
|
die $SWISH::errstr unless $sh; |
148 |
dpavlin |
47 |
# construct swish query |
149 |
|
|
my $sw_q = join(" and ",@s_arr); |
150 |
|
|
if (@path_arr) { |
151 |
|
|
$sw_q .= "and (swishdocpath=\""; |
152 |
|
|
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
153 |
|
|
$sw_q .= "\")"; |
154 |
|
|
$tmpl->param('full',1); # show full records |
155 |
|
|
} |
156 |
dpavlin |
7 |
|
157 |
dpavlin |
47 |
my $hits = $sh->query($sw_q); |
158 |
|
|
|
159 |
dpavlin |
7 |
$tmpl->param('hits',$hits); |
160 |
dpavlin |
47 |
$tmpl->param('search',$sw_q); |
161 |
dpavlin |
7 |
|
162 |
dpavlin |
76 |
$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0); |
163 |
|
|
$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0); |
164 |
dpavlin |
51 |
|
165 |
dpavlin |
7 |
# create a Pager object |
166 |
|
|
my $pager = HTML::Pager->new( |
167 |
|
|
# required parameters |
168 |
|
|
query => $q, |
169 |
|
|
get_data_callback => sub { |
170 |
|
|
my ($offset, $rows) = @_; |
171 |
|
|
|
172 |
|
|
my @result; |
173 |
|
|
for (my $i=0; $i<$rows; $i++) { |
174 |
|
|
push @result, $swish_results[$offset+$i] if $swish_results[$offset+$i]; |
175 |
|
|
} |
176 |
|
|
return \@result; |
177 |
|
|
}, |
178 |
|
|
rows => $hits, |
179 |
|
|
page_size => $ON_PAGE, |
180 |
|
|
# some optional parameters |
181 |
|
|
persist_vars => [ |
182 |
|
|
'rm', |
183 |
|
|
'f1', 'v1', |
184 |
|
|
'f2', 'v2', |
185 |
|
|
'f3', 'v3', |
186 |
|
|
'f4', 'v4', |
187 |
|
|
'f5', 'v5', |
188 |
|
|
'f6', 'v6', |
189 |
|
|
'f7', 'v7', |
190 |
|
|
'f8', 'v8', |
191 |
|
|
'f9', 'v9', |
192 |
|
|
], |
193 |
|
|
#cell_space_color => '#000000', |
194 |
|
|
#cell_background_color => '#ffffff', |
195 |
|
|
#nav_background_color => '#dddddd', |
196 |
|
|
#javascript_presubmit => 'last_minute_javascript()', |
197 |
|
|
debug => 1, |
198 |
|
|
template => $tmpl, |
199 |
|
|
); |
200 |
|
|
|
201 |
|
|
my $html = $pager->output; |
202 |
|
|
|
203 |
|
|
return $html; |
204 |
|
|
} |
205 |
|
|
|
206 |
dpavlin |
9 |
sub show_index { |
207 |
|
|
my $self = shift; |
208 |
|
|
my $i = shift; # field number |
209 |
|
|
|
210 |
|
|
my $q = $self->query(); |
211 |
|
|
|
212 |
dpavlin |
11 |
my $field = $q->param("f$i"); |
213 |
|
|
my $limit = $q->param("v$i"); |
214 |
|
|
|
215 |
dpavlin |
9 |
my $html; |
216 |
|
|
|
217 |
dpavlin |
53 |
my $index = new index_DBI( |
218 |
|
|
$cfg_global->val('global', 'dbi_dbd'), |
219 |
|
|
$cfg_global->val('global', 'dbi_dsn'), |
220 |
|
|
$cfg_global->val('global', 'dbi_user'), |
221 |
|
|
$cfg_global->val('global', 'dbi_passwd') || '' |
222 |
|
|
); |
223 |
dpavlin |
9 |
|
224 |
dpavlin |
12 |
my $total = $index->check($field); |
225 |
|
|
if (! $total) { |
226 |
dpavlin |
11 |
my $tmpl = $self->load_tmpl('no_index.html'); |
227 |
|
|
$tmpl->param('field',$field); |
228 |
|
|
$html = $tmpl->output; |
229 |
|
|
return $html; |
230 |
|
|
} |
231 |
dpavlin |
9 |
|
232 |
dpavlin |
12 |
my $tmpl = $self->load_tmpl('index_res.html'); |
233 |
|
|
$tmpl->param('field',$field); |
234 |
|
|
$tmpl->param('limit',$limit); |
235 |
|
|
$tmpl->param('total',$total); |
236 |
dpavlin |
11 |
|
237 |
dpavlin |
51 |
# FIXME I should set offset and leave out limit from fetch!! |
238 |
dpavlin |
16 |
# if (! $q->param("PAGER_offset") { |
239 |
|
|
# $q->param("Pager_offet) |
240 |
|
|
# } |
241 |
|
|
|
242 |
dpavlin |
12 |
my $pager = HTML::Pager->new( |
243 |
|
|
query => $q, |
244 |
|
|
get_data_callback => sub { |
245 |
|
|
my ($offset, $rows) = @_; |
246 |
dpavlin |
11 |
|
247 |
dpavlin |
12 |
my @result = $index->fetch($field,'item',$limit, $offset, $rows); |
248 |
|
|
return \@result; |
249 |
|
|
}, |
250 |
|
|
rows => $total, |
251 |
|
|
page_size => $ON_PAGE, |
252 |
|
|
persist_vars => [ |
253 |
|
|
'rm', |
254 |
|
|
"f$i", "v$i", "f".$i."_index", |
255 |
|
|
'offset', |
256 |
|
|
], |
257 |
|
|
debug => 1, |
258 |
|
|
template => $tmpl, |
259 |
|
|
); |
260 |
dpavlin |
11 |
|
261 |
dpavlin |
12 |
return $pager->output; |
262 |
dpavlin |
9 |
} |
263 |
|
|
|
264 |
dpavlin |
7 |
1; |