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 |
dpavlin |
128 |
use index_DBI_cache; |
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 |
147 |
my $TEMPLATE =$cfg_global->val('webpac', 'template'); |
31 |
dpavlin |
76 |
|
32 |
|
|
|
33 |
dpavlin |
30 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
34 |
dpavlin |
7 |
|
35 |
dpavlin |
14 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
36 |
|
|
|
37 |
dpavlin |
53 |
|
38 |
dpavlin |
7 |
sub setup { |
39 |
|
|
my $self = shift; |
40 |
|
|
$self->tmpl_path($TEMPLATE_PATH); |
41 |
|
|
$self->run_modes( |
42 |
|
|
'search' => 'show_search_form', |
43 |
|
|
'results' => 'show_results_list', |
44 |
dpavlin |
9 |
# 'user' => 'show_user_detail', |
45 |
|
|
'index' => 'show_index', |
46 |
dpavlin |
7 |
); |
47 |
|
|
$self->start_mode('search'); |
48 |
|
|
$self->mode_param('rm'); |
49 |
|
|
|
50 |
|
|
$self->header_props(-charset=>$CHARSET); |
51 |
|
|
} |
52 |
|
|
|
53 |
dpavlin |
147 |
sub in_template { |
54 |
|
|
my $html = shift || "This page is left unintentionally blank"; |
55 |
|
|
return $html if (! defined($TEMPLATE)); |
56 |
|
|
if (open(T, $TEMPLATE)) { |
57 |
|
|
my $template_html = join("\n",<T>); |
58 |
|
|
close(T); |
59 |
|
|
$template_html =~ s/##webpac##/$html/gsi; |
60 |
|
|
return $template_html; |
61 |
|
|
} else { |
62 |
|
|
return "Can't read template '$TEMPLATE'"; |
63 |
|
|
} |
64 |
|
|
} |
65 |
|
|
|
66 |
dpavlin |
7 |
sub show_search_form { |
67 |
|
|
my $self = shift; |
68 |
|
|
|
69 |
|
|
# Get the CGI.pm query object |
70 |
|
|
my $q = $self->query(); |
71 |
|
|
|
72 |
|
|
my $tmpl = $self->load_tmpl('search.html'); |
73 |
|
|
my $html = $tmpl->output; |
74 |
|
|
|
75 |
|
|
my $fif = new HTML::FillInForm; |
76 |
|
|
|
77 |
dpavlin |
147 |
return in_template($fif->fill(scalarref => \$html, fobject => $q, |
78 |
|
|
target => 'search')); |
79 |
dpavlin |
7 |
} |
80 |
|
|
|
81 |
|
|
sub show_results_list { |
82 |
|
|
my $self = shift; |
83 |
|
|
|
84 |
|
|
my $q = $self->query(); |
85 |
|
|
|
86 |
|
|
my @swish_results; # results from swish |
87 |
|
|
|
88 |
|
|
# load template for this page |
89 |
|
|
|
90 |
|
|
my @s_arr; # all queries are located here |
91 |
|
|
|
92 |
dpavlin |
47 |
my @path_arr = $q->param('path'); |
93 |
|
|
my $full = $q->param('full'); |
94 |
|
|
|
95 |
dpavlin |
112 |
my @persist_vars = ( 'rm' ); |
96 |
dpavlin |
126 |
my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.$q->param('PAGER_offset') || 0 ); |
97 |
dpavlin |
112 |
|
98 |
dpavlin |
71 |
for(my $i = 1; $i <=30; $i++) { |
99 |
dpavlin |
7 |
|
100 |
dpavlin |
9 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
101 |
dpavlin |
112 |
|
102 |
dpavlin |
71 |
next if (! $q->param("v$i")); |
103 |
dpavlin |
9 |
next if (! $q->param("f$i")); |
104 |
dpavlin |
7 |
|
105 |
dpavlin |
112 |
push @persist_vars, "f$i"; |
106 |
|
|
push @persist_vars, "v$i"; |
107 |
|
|
|
108 |
dpavlin |
126 |
push @url_params,"f$i=".$q->url_param("f$i"); |
109 |
|
|
push @url_params,"v$i=".$q->url_param("v$i"); |
110 |
|
|
|
111 |
dpavlin |
7 |
# re-write query from +/- to and/and not |
112 |
dpavlin |
73 |
my @param_vals = $q->param("v$i"); |
113 |
|
|
my @swish_q; |
114 |
|
|
while (my $search = shift @param_vals) { |
115 |
|
|
my $s; |
116 |
|
|
# remove accents |
117 |
|
|
$search = unac_string($CHARSET,$search); |
118 |
|
|
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
119 |
|
|
$s .= "$1 "; |
120 |
|
|
} |
121 |
|
|
$search =~ s/^\s+//; |
122 |
|
|
$search =~ s/\s+$//; |
123 |
dpavlin |
7 |
|
124 |
dpavlin |
73 |
foreach (split(/\s+/,$search)) { |
125 |
|
|
if (m/^([+-])(\S+)/) { |
126 |
|
|
$s.= ($s) ? "and " : ""; |
127 |
|
|
$s.="not " if ($1 eq "-"); |
128 |
|
|
$s.="$2* "; |
129 |
dpavlin |
122 |
} elsif (m/^\s*(and|or|not)\s*$/i) { |
130 |
dpavlin |
73 |
$s.="$_ "; |
131 |
dpavlin |
120 |
# don't add * to words with less than x chars |
132 |
|
|
} elsif (length($_) <= $MIN_WILDCARD) { |
133 |
|
|
$s.="$_ "; |
134 |
dpavlin |
73 |
} else { |
135 |
|
|
$s.="$_* "; |
136 |
|
|
} |
137 |
dpavlin |
7 |
} |
138 |
dpavlin |
73 |
$s =~ s/\*+/*/g; |
139 |
|
|
push @swish_q,$s; |
140 |
dpavlin |
7 |
} |
141 |
dpavlin |
73 |
# FIXME default operator for multi-value fields is or. There is |
142 |
|
|
# no way to change it, except here for now. Is there need? |
143 |
|
|
push @s_arr, $q->param("f$i")."_swish=(".join(" or ",@swish_q).")"; |
144 |
dpavlin |
7 |
} |
145 |
|
|
|
146 |
dpavlin |
126 |
my $tmpl = $self->load_tmpl('results.html', global_vars => 1); |
147 |
dpavlin |
9 |
|
148 |
dpavlin |
80 |
sub esc_html { |
149 |
|
|
my $html = shift; |
150 |
|
|
$html =~ s/</</g; |
151 |
|
|
$html =~ s/>/>/g; |
152 |
|
|
return $html; |
153 |
|
|
} |
154 |
|
|
|
155 |
dpavlin |
7 |
# call swish |
156 |
|
|
my $sh = SWISH->connect('Fork', |
157 |
|
|
prog => $SWISH, |
158 |
|
|
indexes => $INDEX, |
159 |
dpavlin |
13 |
properties => [qw/swishdocpath swishrank swishtitle headline html/], |
160 |
dpavlin |
7 |
results => sub { |
161 |
|
|
my ($sh,$hit) = @_; |
162 |
|
|
|
163 |
|
|
push @swish_results, { |
164 |
|
|
nr => ($#swish_results + 2), |
165 |
|
|
path => $hit->swishdocpath, |
166 |
dpavlin |
80 |
headline => esc_html($from_utf8->convert($hit->headline)), |
167 |
dpavlin |
41 |
html => back2html($from_utf8->convert($hit->html)), |
168 |
dpavlin |
7 |
rank => $hit->swishrank }; |
169 |
|
|
|
170 |
|
|
}, |
171 |
|
|
#startnum => 0, |
172 |
dpavlin |
47 |
maxhits => $MAX_HITS |
173 |
dpavlin |
7 |
); |
174 |
|
|
|
175 |
|
|
die $SWISH::errstr unless $sh; |
176 |
dpavlin |
47 |
# construct swish query |
177 |
|
|
my $sw_q = join(" and ",@s_arr); |
178 |
dpavlin |
111 |
if (@path_arr && $q->param('show_full')) { |
179 |
dpavlin |
47 |
$sw_q .= "and (swishdocpath=\""; |
180 |
|
|
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
181 |
|
|
$sw_q .= "\")"; |
182 |
|
|
$tmpl->param('full',1); # show full records |
183 |
dpavlin |
121 |
} else { |
184 |
|
|
$tmpl->param('full',0); |
185 |
dpavlin |
47 |
} |
186 |
dpavlin |
7 |
|
187 |
dpavlin |
47 |
my $hits = $sh->query($sw_q); |
188 |
|
|
|
189 |
dpavlin |
7 |
$tmpl->param('hits',$hits); |
190 |
dpavlin |
47 |
$tmpl->param('search',$sw_q); |
191 |
dpavlin |
7 |
|
192 |
dpavlin |
76 |
$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0); |
193 |
|
|
$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0); |
194 |
dpavlin |
51 |
|
195 |
dpavlin |
126 |
$tmpl->param('url_params',"?".join("&",@url_params)); |
196 |
|
|
|
197 |
dpavlin |
7 |
# create a Pager object |
198 |
|
|
my $pager = HTML::Pager->new( |
199 |
|
|
# required parameters |
200 |
|
|
query => $q, |
201 |
|
|
get_data_callback => sub { |
202 |
|
|
my ($offset, $rows) = @_; |
203 |
|
|
|
204 |
|
|
my @result; |
205 |
|
|
for (my $i=0; $i<$rows; $i++) { |
206 |
dpavlin |
114 |
my $r = $swish_results[$offset+$i]; |
207 |
dpavlin |
121 |
if ($r && $tmpl->param('full')) { |
208 |
dpavlin |
114 |
push @result, $r; |
209 |
|
|
} elsif ($r) { |
210 |
|
|
# if not full output, skip html |
211 |
|
|
delete $r->{html}; |
212 |
|
|
push @result, $r; |
213 |
|
|
} |
214 |
dpavlin |
7 |
} |
215 |
|
|
return \@result; |
216 |
|
|
}, |
217 |
|
|
rows => $hits, |
218 |
|
|
page_size => $ON_PAGE, |
219 |
|
|
# some optional parameters |
220 |
dpavlin |
112 |
persist_vars => [ @persist_vars ], |
221 |
dpavlin |
7 |
#cell_space_color => '#000000', |
222 |
|
|
#cell_background_color => '#ffffff', |
223 |
|
|
#nav_background_color => '#dddddd', |
224 |
|
|
#javascript_presubmit => 'last_minute_javascript()', |
225 |
|
|
debug => 1, |
226 |
|
|
template => $tmpl, |
227 |
|
|
); |
228 |
|
|
|
229 |
|
|
my $html = $pager->output; |
230 |
|
|
|
231 |
dpavlin |
147 |
return in_template($html); |
232 |
dpavlin |
7 |
} |
233 |
|
|
|
234 |
dpavlin |
9 |
sub show_index { |
235 |
|
|
my $self = shift; |
236 |
|
|
my $i = shift; # field number |
237 |
|
|
|
238 |
|
|
my $q = $self->query(); |
239 |
|
|
|
240 |
dpavlin |
11 |
my $field = $q->param("f$i"); |
241 |
|
|
my $limit = $q->param("v$i"); |
242 |
|
|
|
243 |
dpavlin |
9 |
my $html; |
244 |
|
|
|
245 |
dpavlin |
53 |
my $index = new index_DBI( |
246 |
|
|
$cfg_global->val('global', 'dbi_dbd'), |
247 |
|
|
$cfg_global->val('global', 'dbi_dsn'), |
248 |
|
|
$cfg_global->val('global', 'dbi_user'), |
249 |
|
|
$cfg_global->val('global', 'dbi_passwd') || '' |
250 |
|
|
); |
251 |
dpavlin |
9 |
|
252 |
dpavlin |
140 |
my $total = $index->count($field,$limit); |
253 |
dpavlin |
12 |
if (! $total) { |
254 |
dpavlin |
11 |
my $tmpl = $self->load_tmpl('no_index.html'); |
255 |
|
|
$tmpl->param('field',$field); |
256 |
|
|
$html = $tmpl->output; |
257 |
|
|
return $html; |
258 |
|
|
} |
259 |
dpavlin |
9 |
|
260 |
dpavlin |
124 |
my $tmpl = $self->load_tmpl('index_res.html', global_vars => 1); |
261 |
dpavlin |
12 |
$tmpl->param('field',$field); |
262 |
|
|
$tmpl->param('limit',$limit); |
263 |
|
|
$tmpl->param('total',$total); |
264 |
dpavlin |
11 |
|
265 |
dpavlin |
51 |
# FIXME I should set offset and leave out limit from fetch!! |
266 |
dpavlin |
16 |
# if (! $q->param("PAGER_offset") { |
267 |
|
|
# $q->param("Pager_offet) |
268 |
|
|
# } |
269 |
|
|
|
270 |
dpavlin |
12 |
my $pager = HTML::Pager->new( |
271 |
|
|
query => $q, |
272 |
|
|
get_data_callback => sub { |
273 |
|
|
my ($offset, $rows) = @_; |
274 |
dpavlin |
11 |
|
275 |
dpavlin |
140 |
my @result = $index->fetch($field,$limit, $offset, $rows); |
276 |
dpavlin |
12 |
return \@result; |
277 |
|
|
}, |
278 |
|
|
rows => $total, |
279 |
|
|
page_size => $ON_PAGE, |
280 |
|
|
persist_vars => [ |
281 |
|
|
'rm', |
282 |
|
|
"f$i", "v$i", "f".$i."_index", |
283 |
|
|
'offset', |
284 |
|
|
], |
285 |
|
|
debug => 1, |
286 |
|
|
template => $tmpl, |
287 |
|
|
); |
288 |
dpavlin |
11 |
|
289 |
dpavlin |
147 |
return in_template($pager->output); |
290 |
dpavlin |
9 |
} |
291 |
|
|
|
292 |
dpavlin |
7 |
1; |