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 |
dpavlin |
120 |
my $MIN_WILDCARD =$cfg_global->val('webpac', 'min_wildcard') || 1; |
30 |
dpavlin |
76 |
|
31 |
|
|
|
32 |
dpavlin |
30 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
33 |
dpavlin |
7 |
|
34 |
dpavlin |
14 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
35 |
|
|
|
36 |
dpavlin |
53 |
|
37 |
dpavlin |
7 |
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 |
dpavlin |
9 |
# 'user' => 'show_user_detail', |
44 |
|
|
'index' => 'show_index', |
45 |
dpavlin |
7 |
); |
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 |
dpavlin |
47 |
my @path_arr = $q->param('path'); |
79 |
|
|
my $full = $q->param('full'); |
80 |
|
|
|
81 |
dpavlin |
112 |
my @persist_vars = ( 'rm' ); |
82 |
|
|
|
83 |
dpavlin |
71 |
for(my $i = 1; $i <=30; $i++) { |
84 |
dpavlin |
7 |
|
85 |
dpavlin |
9 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
86 |
dpavlin |
112 |
|
87 |
dpavlin |
71 |
next if (! $q->param("v$i")); |
88 |
dpavlin |
9 |
next if (! $q->param("f$i")); |
89 |
dpavlin |
7 |
|
90 |
dpavlin |
112 |
push @persist_vars, "f$i"; |
91 |
|
|
push @persist_vars, "v$i"; |
92 |
|
|
|
93 |
dpavlin |
7 |
# re-write query from +/- to and/and not |
94 |
dpavlin |
73 |
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 |
dpavlin |
7 |
|
106 |
dpavlin |
73 |
foreach (split(/\s+/,$search)) { |
107 |
|
|
if (m/^([+-])(\S+)/) { |
108 |
|
|
$s.= ($s) ? "and " : ""; |
109 |
|
|
$s.="not " if ($1 eq "-"); |
110 |
|
|
$s.="$2* "; |
111 |
|
|
} elsif (m/(and|or|not)/i) { |
112 |
|
|
$s.="$_ "; |
113 |
dpavlin |
120 |
# don't add * to words with less than x chars |
114 |
|
|
} elsif (length($_) <= $MIN_WILDCARD) { |
115 |
|
|
$s.="$_ "; |
116 |
dpavlin |
73 |
} else { |
117 |
|
|
$s.="$_* "; |
118 |
|
|
} |
119 |
dpavlin |
7 |
} |
120 |
dpavlin |
73 |
$s =~ s/\*+/*/g; |
121 |
|
|
push @swish_q,$s; |
122 |
dpavlin |
7 |
} |
123 |
dpavlin |
73 |
# 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 |
dpavlin |
7 |
} |
127 |
|
|
|
128 |
dpavlin |
9 |
my $tmpl = $self->load_tmpl('results.html'); |
129 |
|
|
|
130 |
dpavlin |
80 |
sub esc_html { |
131 |
|
|
my $html = shift; |
132 |
|
|
$html =~ s/</</g; |
133 |
|
|
$html =~ s/>/>/g; |
134 |
|
|
return $html; |
135 |
|
|
} |
136 |
|
|
|
137 |
dpavlin |
7 |
# call swish |
138 |
|
|
my $sh = SWISH->connect('Fork', |
139 |
|
|
prog => $SWISH, |
140 |
|
|
indexes => $INDEX, |
141 |
dpavlin |
13 |
properties => [qw/swishdocpath swishrank swishtitle headline html/], |
142 |
dpavlin |
7 |
results => sub { |
143 |
|
|
my ($sh,$hit) = @_; |
144 |
|
|
|
145 |
|
|
push @swish_results, { |
146 |
|
|
nr => ($#swish_results + 2), |
147 |
|
|
path => $hit->swishdocpath, |
148 |
dpavlin |
80 |
headline => esc_html($from_utf8->convert($hit->headline)), |
149 |
dpavlin |
41 |
html => back2html($from_utf8->convert($hit->html)), |
150 |
dpavlin |
7 |
rank => $hit->swishrank }; |
151 |
|
|
|
152 |
|
|
}, |
153 |
|
|
#startnum => 0, |
154 |
dpavlin |
47 |
maxhits => $MAX_HITS |
155 |
dpavlin |
7 |
); |
156 |
|
|
|
157 |
|
|
die $SWISH::errstr unless $sh; |
158 |
dpavlin |
47 |
# construct swish query |
159 |
|
|
my $sw_q = join(" and ",@s_arr); |
160 |
dpavlin |
111 |
if (@path_arr && $q->param('show_full')) { |
161 |
dpavlin |
47 |
$sw_q .= "and (swishdocpath=\""; |
162 |
|
|
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
163 |
|
|
$sw_q .= "\")"; |
164 |
|
|
$tmpl->param('full',1); # show full records |
165 |
|
|
} |
166 |
dpavlin |
7 |
|
167 |
dpavlin |
47 |
my $hits = $sh->query($sw_q); |
168 |
|
|
|
169 |
dpavlin |
7 |
$tmpl->param('hits',$hits); |
170 |
dpavlin |
47 |
$tmpl->param('search',$sw_q); |
171 |
dpavlin |
7 |
|
172 |
dpavlin |
76 |
$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0); |
173 |
|
|
$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0); |
174 |
dpavlin |
51 |
|
175 |
dpavlin |
7 |
# create a Pager object |
176 |
|
|
my $pager = HTML::Pager->new( |
177 |
|
|
# required parameters |
178 |
|
|
query => $q, |
179 |
|
|
get_data_callback => sub { |
180 |
|
|
my ($offset, $rows) = @_; |
181 |
|
|
|
182 |
|
|
my @result; |
183 |
|
|
for (my $i=0; $i<$rows; $i++) { |
184 |
dpavlin |
114 |
my $r = $swish_results[$offset+$i]; |
185 |
|
|
if ($r && $q->param('show_full')) { |
186 |
|
|
push @result, $r; |
187 |
|
|
} elsif ($r) { |
188 |
|
|
# if not full output, skip html |
189 |
|
|
delete $r->{html}; |
190 |
|
|
push @result, $r; |
191 |
|
|
} |
192 |
dpavlin |
7 |
} |
193 |
|
|
return \@result; |
194 |
|
|
}, |
195 |
|
|
rows => $hits, |
196 |
|
|
page_size => $ON_PAGE, |
197 |
|
|
# some optional parameters |
198 |
dpavlin |
112 |
persist_vars => [ @persist_vars ], |
199 |
dpavlin |
7 |
#cell_space_color => '#000000', |
200 |
|
|
#cell_background_color => '#ffffff', |
201 |
|
|
#nav_background_color => '#dddddd', |
202 |
|
|
#javascript_presubmit => 'last_minute_javascript()', |
203 |
|
|
debug => 1, |
204 |
|
|
template => $tmpl, |
205 |
|
|
); |
206 |
|
|
|
207 |
|
|
my $html = $pager->output; |
208 |
|
|
|
209 |
|
|
return $html; |
210 |
|
|
} |
211 |
|
|
|
212 |
dpavlin |
9 |
sub show_index { |
213 |
|
|
my $self = shift; |
214 |
|
|
my $i = shift; # field number |
215 |
|
|
|
216 |
|
|
my $q = $self->query(); |
217 |
|
|
|
218 |
dpavlin |
11 |
my $field = $q->param("f$i"); |
219 |
|
|
my $limit = $q->param("v$i"); |
220 |
|
|
|
221 |
dpavlin |
9 |
my $html; |
222 |
|
|
|
223 |
dpavlin |
53 |
my $index = new index_DBI( |
224 |
|
|
$cfg_global->val('global', 'dbi_dbd'), |
225 |
|
|
$cfg_global->val('global', 'dbi_dsn'), |
226 |
|
|
$cfg_global->val('global', 'dbi_user'), |
227 |
|
|
$cfg_global->val('global', 'dbi_passwd') || '' |
228 |
|
|
); |
229 |
dpavlin |
9 |
|
230 |
dpavlin |
12 |
my $total = $index->check($field); |
231 |
|
|
if (! $total) { |
232 |
dpavlin |
11 |
my $tmpl = $self->load_tmpl('no_index.html'); |
233 |
|
|
$tmpl->param('field',$field); |
234 |
|
|
$html = $tmpl->output; |
235 |
|
|
return $html; |
236 |
|
|
} |
237 |
dpavlin |
9 |
|
238 |
dpavlin |
12 |
my $tmpl = $self->load_tmpl('index_res.html'); |
239 |
|
|
$tmpl->param('field',$field); |
240 |
|
|
$tmpl->param('limit',$limit); |
241 |
|
|
$tmpl->param('total',$total); |
242 |
dpavlin |
11 |
|
243 |
dpavlin |
51 |
# FIXME I should set offset and leave out limit from fetch!! |
244 |
dpavlin |
16 |
# if (! $q->param("PAGER_offset") { |
245 |
|
|
# $q->param("Pager_offet) |
246 |
|
|
# } |
247 |
|
|
|
248 |
dpavlin |
12 |
my $pager = HTML::Pager->new( |
249 |
|
|
query => $q, |
250 |
|
|
get_data_callback => sub { |
251 |
|
|
my ($offset, $rows) = @_; |
252 |
dpavlin |
11 |
|
253 |
dpavlin |
12 |
my @result = $index->fetch($field,'item',$limit, $offset, $rows); |
254 |
|
|
return \@result; |
255 |
|
|
}, |
256 |
|
|
rows => $total, |
257 |
|
|
page_size => $ON_PAGE, |
258 |
|
|
persist_vars => [ |
259 |
|
|
'rm', |
260 |
|
|
"f$i", "v$i", "f".$i."_index", |
261 |
|
|
'offset', |
262 |
|
|
], |
263 |
|
|
debug => 1, |
264 |
|
|
template => $tmpl, |
265 |
|
|
); |
266 |
dpavlin |
11 |
|
267 |
dpavlin |
12 |
return $pager->output; |
268 |
dpavlin |
9 |
} |
269 |
|
|
|
270 |
dpavlin |
7 |
1; |