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 |
my @persist_vars = ( 'rm' ); |
81 |
|
82 |
for(my $i = 1; $i <=30; $i++) { |
83 |
|
84 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
85 |
|
86 |
next if (! $q->param("v$i")); |
87 |
next if (! $q->param("f$i")); |
88 |
|
89 |
push @persist_vars, "f$i"; |
90 |
push @persist_vars, "v$i"; |
91 |
|
92 |
# re-write query from +/- to and/and not |
93 |
my @param_vals = $q->param("v$i"); |
94 |
my @swish_q; |
95 |
while (my $search = shift @param_vals) { |
96 |
my $s; |
97 |
# remove accents |
98 |
$search = unac_string($CHARSET,$search); |
99 |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
100 |
$s .= "$1 "; |
101 |
} |
102 |
$search =~ s/^\s+//; |
103 |
$search =~ s/\s+$//; |
104 |
|
105 |
foreach (split(/\s+/,$search)) { |
106 |
if (m/^([+-])(\S+)/) { |
107 |
$s.= ($s) ? "and " : ""; |
108 |
$s.="not " if ($1 eq "-"); |
109 |
$s.="$2* "; |
110 |
} elsif (m/(and|or|not)/i) { |
111 |
$s.="$_ "; |
112 |
} else { |
113 |
$s.="$_* "; |
114 |
} |
115 |
} |
116 |
$s =~ s/\*+/*/g; |
117 |
push @swish_q,$s; |
118 |
} |
119 |
# FIXME default operator for multi-value fields is or. There is |
120 |
# no way to change it, except here for now. Is there need? |
121 |
push @s_arr, $q->param("f$i")."_swish=(".join(" or ",@swish_q).")"; |
122 |
} |
123 |
|
124 |
my $tmpl = $self->load_tmpl('results.html'); |
125 |
|
126 |
sub esc_html { |
127 |
my $html = shift; |
128 |
$html =~ s/</</g; |
129 |
$html =~ s/>/>/g; |
130 |
return $html; |
131 |
} |
132 |
|
133 |
# call swish |
134 |
my $sh = SWISH->connect('Fork', |
135 |
prog => $SWISH, |
136 |
indexes => $INDEX, |
137 |
properties => [qw/swishdocpath swishrank swishtitle headline html/], |
138 |
results => sub { |
139 |
my ($sh,$hit) = @_; |
140 |
|
141 |
push @swish_results, { |
142 |
nr => ($#swish_results + 2), |
143 |
path => $hit->swishdocpath, |
144 |
headline => esc_html($from_utf8->convert($hit->headline)), |
145 |
html => back2html($from_utf8->convert($hit->html)), |
146 |
rank => $hit->swishrank }; |
147 |
|
148 |
}, |
149 |
#startnum => 0, |
150 |
maxhits => $MAX_HITS |
151 |
); |
152 |
|
153 |
die $SWISH::errstr unless $sh; |
154 |
# construct swish query |
155 |
my $sw_q = join(" and ",@s_arr); |
156 |
if (@path_arr && $q->param('show_full')) { |
157 |
$sw_q .= "and (swishdocpath=\""; |
158 |
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
159 |
$sw_q .= "\")"; |
160 |
$tmpl->param('full',1); # show full records |
161 |
} |
162 |
|
163 |
my $hits = $sh->query($sw_q); |
164 |
|
165 |
$tmpl->param('hits',$hits); |
166 |
$tmpl->param('search',$sw_q); |
167 |
|
168 |
$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0); |
169 |
$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0); |
170 |
|
171 |
# create a Pager object |
172 |
my $pager = HTML::Pager->new( |
173 |
# required parameters |
174 |
query => $q, |
175 |
get_data_callback => sub { |
176 |
my ($offset, $rows) = @_; |
177 |
|
178 |
my @result; |
179 |
for (my $i=0; $i<$rows; $i++) { |
180 |
push @result, $swish_results[$offset+$i] if $swish_results[$offset+$i]; |
181 |
} |
182 |
return \@result; |
183 |
}, |
184 |
rows => $hits, |
185 |
page_size => $ON_PAGE, |
186 |
# some optional parameters |
187 |
persist_vars => [ @persist_vars ], |
188 |
#cell_space_color => '#000000', |
189 |
#cell_background_color => '#ffffff', |
190 |
#nav_background_color => '#dddddd', |
191 |
#javascript_presubmit => 'last_minute_javascript()', |
192 |
debug => 1, |
193 |
template => $tmpl, |
194 |
); |
195 |
|
196 |
my $html = $pager->output; |
197 |
|
198 |
return $html; |
199 |
} |
200 |
|
201 |
sub show_index { |
202 |
my $self = shift; |
203 |
my $i = shift; # field number |
204 |
|
205 |
my $q = $self->query(); |
206 |
|
207 |
my $field = $q->param("f$i"); |
208 |
my $limit = $q->param("v$i"); |
209 |
|
210 |
my $html; |
211 |
|
212 |
my $index = new index_DBI( |
213 |
$cfg_global->val('global', 'dbi_dbd'), |
214 |
$cfg_global->val('global', 'dbi_dsn'), |
215 |
$cfg_global->val('global', 'dbi_user'), |
216 |
$cfg_global->val('global', 'dbi_passwd') || '' |
217 |
); |
218 |
|
219 |
my $total = $index->check($field); |
220 |
if (! $total) { |
221 |
my $tmpl = $self->load_tmpl('no_index.html'); |
222 |
$tmpl->param('field',$field); |
223 |
$html = $tmpl->output; |
224 |
return $html; |
225 |
} |
226 |
|
227 |
my $tmpl = $self->load_tmpl('index_res.html'); |
228 |
$tmpl->param('field',$field); |
229 |
$tmpl->param('limit',$limit); |
230 |
$tmpl->param('total',$total); |
231 |
|
232 |
# FIXME I should set offset and leave out limit from fetch!! |
233 |
# if (! $q->param("PAGER_offset") { |
234 |
# $q->param("Pager_offet) |
235 |
# } |
236 |
|
237 |
my $pager = HTML::Pager->new( |
238 |
query => $q, |
239 |
get_data_callback => sub { |
240 |
my ($offset, $rows) = @_; |
241 |
|
242 |
my @result = $index->fetch($field,'item',$limit, $offset, $rows); |
243 |
return \@result; |
244 |
}, |
245 |
rows => $total, |
246 |
page_size => $ON_PAGE, |
247 |
persist_vars => [ |
248 |
'rm', |
249 |
"f$i", "v$i", "f".$i."_index", |
250 |
'offset', |
251 |
], |
252 |
debug => 1, |
253 |
template => $tmpl, |
254 |
); |
255 |
|
256 |
return $pager->output; |
257 |
} |
258 |
|
259 |
1; |