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 |
7 |
|
12 |
dpavlin |
11 |
use lib '..'; |
13 |
|
|
use index_DBI; |
14 |
dpavlin |
13 |
use back2html; |
15 |
dpavlin |
11 |
|
16 |
dpavlin |
7 |
# configuration options |
17 |
dpavlin |
51 |
# FIXME they really should go in configuration file! |
18 |
dpavlin |
7 |
my $TEMPLATE_PATH = '/data/webpac/template_html'; |
19 |
|
|
my $CHARSET = 'ISO-8859-2'; |
20 |
dpavlin |
13 |
my $SWISH = '/usr/bin/swish-e'; |
21 |
dpavlin |
7 |
my $INDEX = '/data/webpac/index/isis.index'; |
22 |
|
|
my $MAX_HITS = 500; |
23 |
|
|
my $ON_PAGE = 10; |
24 |
|
|
|
25 |
dpavlin |
30 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
26 |
dpavlin |
7 |
|
27 |
dpavlin |
14 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
28 |
|
|
|
29 |
dpavlin |
7 |
sub setup { |
30 |
|
|
my $self = shift; |
31 |
|
|
$self->tmpl_path($TEMPLATE_PATH); |
32 |
|
|
$self->run_modes( |
33 |
|
|
'search' => 'show_search_form', |
34 |
|
|
'results' => 'show_results_list', |
35 |
dpavlin |
9 |
# 'user' => 'show_user_detail', |
36 |
|
|
'index' => 'show_index', |
37 |
dpavlin |
7 |
); |
38 |
|
|
$self->start_mode('search'); |
39 |
|
|
$self->mode_param('rm'); |
40 |
|
|
|
41 |
|
|
$self->header_props(-charset=>$CHARSET); |
42 |
|
|
} |
43 |
|
|
|
44 |
|
|
sub show_search_form { |
45 |
|
|
my $self = shift; |
46 |
|
|
|
47 |
|
|
# Get the CGI.pm query object |
48 |
|
|
my $q = $self->query(); |
49 |
|
|
|
50 |
|
|
my $tmpl = $self->load_tmpl('search.html'); |
51 |
|
|
my $html = $tmpl->output; |
52 |
|
|
|
53 |
|
|
my $fif = new HTML::FillInForm; |
54 |
|
|
|
55 |
|
|
return $fif->fill(scalarref => \$html, fobject => $q, |
56 |
|
|
target => 'search'); |
57 |
|
|
} |
58 |
|
|
|
59 |
|
|
sub show_results_list { |
60 |
|
|
my $self = shift; |
61 |
|
|
|
62 |
|
|
my $q = $self->query(); |
63 |
|
|
|
64 |
|
|
my @swish_results; # results from swish |
65 |
|
|
|
66 |
|
|
# load template for this page |
67 |
|
|
|
68 |
|
|
my @s_arr; # all queries are located here |
69 |
|
|
|
70 |
dpavlin |
47 |
my @path_arr = $q->param('path'); |
71 |
|
|
my $full = $q->param('full'); |
72 |
|
|
|
73 |
dpavlin |
7 |
for(my $i = 1; $i <=10; $i++) { |
74 |
|
|
|
75 |
dpavlin |
9 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
76 |
|
|
next if (! $q->param("f$i")); |
77 |
dpavlin |
7 |
next if (! $q->param("v$i")); |
78 |
|
|
|
79 |
|
|
# re-write query from +/- to and/and not |
80 |
|
|
my $s; |
81 |
|
|
my $search = $q->param("v$i"); |
82 |
|
|
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
83 |
|
|
$s .= "$1 "; |
84 |
|
|
} |
85 |
|
|
$search =~ s/^\s+//; |
86 |
|
|
$search =~ s/\s+$//; |
87 |
|
|
|
88 |
|
|
foreach (split(/\s+/,$search)) { |
89 |
|
|
if (m/^([+-])(\S+)/) { |
90 |
|
|
$s.= ($s) ? "and " : ""; |
91 |
|
|
$s.="not " if ($1 eq "-"); |
92 |
|
|
$s.="$2* "; |
93 |
|
|
} else { |
94 |
|
|
$s.="$_* "; |
95 |
|
|
} |
96 |
|
|
} |
97 |
dpavlin |
16 |
$s =~ s/\*+/*/g; |
98 |
dpavlin |
7 |
|
99 |
dpavlin |
9 |
push @s_arr,$q->param("f$i")."_swish=($s)"; |
100 |
dpavlin |
7 |
} |
101 |
|
|
|
102 |
dpavlin |
9 |
my $tmpl = $self->load_tmpl('results.html'); |
103 |
|
|
|
104 |
dpavlin |
7 |
# call swish |
105 |
|
|
my $sh = SWISH->connect('Fork', |
106 |
|
|
prog => $SWISH, |
107 |
|
|
indexes => $INDEX, |
108 |
dpavlin |
13 |
properties => [qw/swishdocpath swishrank swishtitle headline html/], |
109 |
dpavlin |
7 |
results => sub { |
110 |
|
|
my ($sh,$hit) = @_; |
111 |
|
|
|
112 |
|
|
push @swish_results, { |
113 |
|
|
nr => ($#swish_results + 2), |
114 |
|
|
path => $hit->swishdocpath, |
115 |
dpavlin |
41 |
headline => $from_utf8->convert($hit->headline), |
116 |
|
|
html => back2html($from_utf8->convert($hit->html)), |
117 |
dpavlin |
7 |
rank => $hit->swishrank }; |
118 |
|
|
|
119 |
|
|
}, |
120 |
|
|
#startnum => 0, |
121 |
dpavlin |
47 |
maxhits => $MAX_HITS |
122 |
dpavlin |
7 |
); |
123 |
|
|
|
124 |
|
|
die $SWISH::errstr unless $sh; |
125 |
|
|
|
126 |
dpavlin |
47 |
# construct swish query |
127 |
|
|
my $sw_q = join(" and ",@s_arr); |
128 |
|
|
if (@path_arr) { |
129 |
|
|
$sw_q .= "and (swishdocpath=\""; |
130 |
|
|
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
131 |
|
|
$sw_q .= "\")"; |
132 |
|
|
$tmpl->param('full',1); # show full records |
133 |
|
|
} |
134 |
dpavlin |
7 |
|
135 |
dpavlin |
47 |
my $hits = $sh->query($sw_q); |
136 |
|
|
|
137 |
dpavlin |
7 |
$tmpl->param('hits',$hits); |
138 |
dpavlin |
47 |
$tmpl->param('search',$sw_q); |
139 |
dpavlin |
7 |
|
140 |
dpavlin |
51 |
$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0); |
141 |
|
|
$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0); |
142 |
|
|
|
143 |
dpavlin |
7 |
# create a Pager object |
144 |
|
|
my $pager = HTML::Pager->new( |
145 |
|
|
# required parameters |
146 |
|
|
query => $q, |
147 |
|
|
get_data_callback => sub { |
148 |
|
|
my ($offset, $rows) = @_; |
149 |
|
|
|
150 |
|
|
my @result; |
151 |
|
|
for (my $i=0; $i<$rows; $i++) { |
152 |
|
|
push @result, $swish_results[$offset+$i] if $swish_results[$offset+$i]; |
153 |
|
|
} |
154 |
|
|
return \@result; |
155 |
|
|
}, |
156 |
|
|
rows => $hits, |
157 |
|
|
page_size => $ON_PAGE, |
158 |
|
|
# some optional parameters |
159 |
|
|
persist_vars => [ |
160 |
|
|
'rm', |
161 |
|
|
'f1', 'v1', |
162 |
|
|
'f2', 'v2', |
163 |
|
|
'f3', 'v3', |
164 |
|
|
'f4', 'v4', |
165 |
|
|
'f5', 'v5', |
166 |
|
|
'f6', 'v6', |
167 |
|
|
'f7', 'v7', |
168 |
|
|
'f8', 'v8', |
169 |
|
|
'f9', 'v9', |
170 |
|
|
], |
171 |
|
|
#cell_space_color => '#000000', |
172 |
|
|
#cell_background_color => '#ffffff', |
173 |
|
|
#nav_background_color => '#dddddd', |
174 |
|
|
#javascript_presubmit => 'last_minute_javascript()', |
175 |
|
|
debug => 1, |
176 |
|
|
template => $tmpl, |
177 |
|
|
); |
178 |
|
|
|
179 |
|
|
my $html = $pager->output; |
180 |
|
|
|
181 |
|
|
return $html; |
182 |
|
|
} |
183 |
|
|
|
184 |
dpavlin |
9 |
sub show_index { |
185 |
|
|
my $self = shift; |
186 |
|
|
my $i = shift; # field number |
187 |
|
|
|
188 |
|
|
my $q = $self->query(); |
189 |
|
|
|
190 |
dpavlin |
11 |
my $field = $q->param("f$i"); |
191 |
|
|
my $limit = $q->param("v$i"); |
192 |
|
|
|
193 |
dpavlin |
9 |
my $html; |
194 |
|
|
|
195 |
dpavlin |
11 |
my $index = new index_DBI(); |
196 |
dpavlin |
9 |
|
197 |
dpavlin |
12 |
my $total = $index->check($field); |
198 |
|
|
if (! $total) { |
199 |
dpavlin |
11 |
my $tmpl = $self->load_tmpl('no_index.html'); |
200 |
|
|
$tmpl->param('field',$field); |
201 |
|
|
$html = $tmpl->output; |
202 |
|
|
return $html; |
203 |
|
|
} |
204 |
dpavlin |
9 |
|
205 |
dpavlin |
12 |
my $tmpl = $self->load_tmpl('index_res.html'); |
206 |
|
|
$tmpl->param('field',$field); |
207 |
|
|
$tmpl->param('limit',$limit); |
208 |
|
|
$tmpl->param('total',$total); |
209 |
dpavlin |
11 |
|
210 |
dpavlin |
51 |
# FIXME I should set offset and leave out limit from fetch!! |
211 |
dpavlin |
16 |
# if (! $q->param("PAGER_offset") { |
212 |
|
|
# $q->param("Pager_offet) |
213 |
|
|
# } |
214 |
|
|
|
215 |
dpavlin |
12 |
my $pager = HTML::Pager->new( |
216 |
|
|
query => $q, |
217 |
|
|
get_data_callback => sub { |
218 |
|
|
my ($offset, $rows) = @_; |
219 |
dpavlin |
11 |
|
220 |
dpavlin |
12 |
my @result = $index->fetch($field,'item',$limit, $offset, $rows); |
221 |
|
|
return \@result; |
222 |
|
|
}, |
223 |
|
|
rows => $total, |
224 |
|
|
page_size => $ON_PAGE, |
225 |
|
|
persist_vars => [ |
226 |
|
|
'rm', |
227 |
|
|
"f$i", "v$i", "f".$i."_index", |
228 |
|
|
'offset', |
229 |
|
|
], |
230 |
|
|
debug => 1, |
231 |
|
|
template => $tmpl, |
232 |
|
|
); |
233 |
dpavlin |
11 |
|
234 |
dpavlin |
12 |
return $pager->output; |
235 |
dpavlin |
9 |
} |
236 |
|
|
|
237 |
dpavlin |
7 |
1; |