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 |
164 |
my $UNAC_FILTER =$cfg_global->val('global', 'unac_filter'); |
32 |
dpavlin |
76 |
|
33 |
dpavlin |
164 |
if ($UNAC_FILTER) { |
34 |
|
|
require $UNAC_FILTER; |
35 |
|
|
} |
36 |
dpavlin |
76 |
|
37 |
dpavlin |
30 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
38 |
dpavlin |
7 |
|
39 |
dpavlin |
14 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
40 |
|
|
|
41 |
dpavlin |
53 |
|
42 |
dpavlin |
7 |
sub setup { |
43 |
|
|
my $self = shift; |
44 |
|
|
$self->tmpl_path($TEMPLATE_PATH); |
45 |
|
|
$self->run_modes( |
46 |
|
|
'search' => 'show_search_form', |
47 |
|
|
'results' => 'show_results_list', |
48 |
dpavlin |
9 |
# 'user' => 'show_user_detail', |
49 |
|
|
'index' => 'show_index', |
50 |
dpavlin |
7 |
); |
51 |
|
|
$self->start_mode('search'); |
52 |
|
|
$self->mode_param('rm'); |
53 |
|
|
|
54 |
|
|
$self->header_props(-charset=>$CHARSET); |
55 |
|
|
} |
56 |
|
|
|
57 |
dpavlin |
147 |
sub in_template { |
58 |
|
|
my $html = shift || "This page is left unintentionally blank"; |
59 |
|
|
return $html if (! defined($TEMPLATE)); |
60 |
|
|
if (open(T, $TEMPLATE)) { |
61 |
|
|
my $template_html = join("\n",<T>); |
62 |
|
|
close(T); |
63 |
|
|
$template_html =~ s/##webpac##/$html/gsi; |
64 |
|
|
return $template_html; |
65 |
|
|
} else { |
66 |
|
|
return "Can't read template '$TEMPLATE'"; |
67 |
|
|
} |
68 |
|
|
} |
69 |
|
|
|
70 |
dpavlin |
7 |
sub show_search_form { |
71 |
|
|
my $self = shift; |
72 |
|
|
|
73 |
|
|
# Get the CGI.pm query object |
74 |
|
|
my $q = $self->query(); |
75 |
|
|
|
76 |
|
|
my $tmpl = $self->load_tmpl('search.html'); |
77 |
|
|
my $html = $tmpl->output; |
78 |
|
|
|
79 |
|
|
my $fif = new HTML::FillInForm; |
80 |
|
|
|
81 |
dpavlin |
147 |
return in_template($fif->fill(scalarref => \$html, fobject => $q, |
82 |
|
|
target => 'search')); |
83 |
dpavlin |
7 |
} |
84 |
|
|
|
85 |
|
|
sub show_results_list { |
86 |
|
|
my $self = shift; |
87 |
|
|
|
88 |
|
|
my $q = $self->query(); |
89 |
|
|
|
90 |
|
|
my @swish_results; # results from swish |
91 |
|
|
|
92 |
|
|
# load template for this page |
93 |
|
|
|
94 |
|
|
my @s_arr; # all queries are located here |
95 |
|
|
|
96 |
dpavlin |
47 |
my @path_arr = $q->param('path'); |
97 |
|
|
my $full = $q->param('full'); |
98 |
|
|
|
99 |
dpavlin |
112 |
my @persist_vars = ( 'rm' ); |
100 |
dpavlin |
126 |
my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.$q->param('PAGER_offset') || 0 ); |
101 |
dpavlin |
112 |
|
102 |
dpavlin |
150 |
# support parametars "f" and "v" for start |
103 |
|
|
for(my $i = ""; $i <=30; $i++) { |
104 |
dpavlin |
7 |
|
105 |
dpavlin |
9 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
106 |
dpavlin |
112 |
|
107 |
dpavlin |
71 |
next if (! $q->param("v$i")); |
108 |
dpavlin |
9 |
next if (! $q->param("f$i")); |
109 |
dpavlin |
7 |
|
110 |
dpavlin |
112 |
push @persist_vars, "f$i"; |
111 |
|
|
push @persist_vars, "v$i"; |
112 |
|
|
|
113 |
dpavlin |
126 |
push @url_params,"f$i=".$q->url_param("f$i"); |
114 |
dpavlin |
158 |
foreach my $v ($q->url_param("v$i")) { |
115 |
|
|
push @url_params,"v$i=$v"; |
116 |
|
|
} |
117 |
dpavlin |
183 |
push @url_params,"e$i=".$q->url_param("e$i"); |
118 |
dpavlin |
126 |
|
119 |
dpavlin |
150 |
my $wc="*"; # swish wildcard |
120 |
|
|
$wc="" if ($i eq ""); # don't apply wildcard on field 0 |
121 |
|
|
|
122 |
dpavlin |
7 |
# re-write query from +/- to and/and not |
123 |
dpavlin |
73 |
my @param_vals = $q->param("v$i"); |
124 |
|
|
my @swish_q; |
125 |
dpavlin |
163 |
my ($pre,$post,$exact) = ('','',''); |
126 |
dpavlin |
73 |
while (my $search = shift @param_vals) { |
127 |
|
|
my $s; |
128 |
|
|
# remove accents |
129 |
|
|
$search = unac_string($CHARSET,$search); |
130 |
|
|
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
131 |
|
|
$s .= "$1 "; |
132 |
|
|
} |
133 |
|
|
$search =~ s/^\s+//; |
134 |
|
|
$search =~ s/\s+$//; |
135 |
dpavlin |
7 |
|
136 |
dpavlin |
163 |
# filed e[nr] is exact match bitmask |
137 |
|
|
# 1 = beginning, 2=end, 3=both |
138 |
|
|
$pre = '"xxbxx ' if ($q->param("e$i") & 1); |
139 |
|
|
$post = ' xxexx"' if ($q->param("e$i") & 2); |
140 |
|
|
# add qotes on other side |
141 |
|
|
if ($q->param("e$i")) { |
142 |
|
|
$pre = '"' if (! $pre); |
143 |
|
|
$post = '"' if (! $post); |
144 |
|
|
$wc = ''; # don't use windcard in exact |
145 |
|
|
$exact = '_exact'; |
146 |
|
|
} |
147 |
|
|
|
148 |
dpavlin |
73 |
foreach (split(/\s+/,$search)) { |
149 |
|
|
if (m/^([+-])(\S+)/) { |
150 |
|
|
$s.= ($s) ? "and " : ""; |
151 |
|
|
$s.="not " if ($1 eq "-"); |
152 |
dpavlin |
163 |
$s.=$2.$wc." "; |
153 |
dpavlin |
122 |
} elsif (m/^\s*(and|or|not)\s*$/i) { |
154 |
dpavlin |
163 |
$s.=$_." "; |
155 |
dpavlin |
120 |
# don't add * to words with less than x chars |
156 |
|
|
} elsif (length($_) <= $MIN_WILDCARD) { |
157 |
dpavlin |
163 |
$s.=$_." "; |
158 |
dpavlin |
73 |
} else { |
159 |
dpavlin |
163 |
$s.=$_.$wc." "; |
160 |
dpavlin |
73 |
} |
161 |
dpavlin |
7 |
} |
162 |
dpavlin |
73 |
$s =~ s/\*+/*/g; |
163 |
dpavlin |
163 |
$s = $pre.$s.$post if ($q->param("e$i")); |
164 |
dpavlin |
73 |
push @swish_q,$s; |
165 |
dpavlin |
7 |
} |
166 |
dpavlin |
73 |
# FIXME default operator for multi-value fields is or. There is |
167 |
|
|
# no way to change it, except here for now. Is there need? |
168 |
dpavlin |
163 |
push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")"; |
169 |
dpavlin |
7 |
} |
170 |
|
|
|
171 |
dpavlin |
126 |
my $tmpl = $self->load_tmpl('results.html', global_vars => 1); |
172 |
dpavlin |
9 |
|
173 |
dpavlin |
80 |
sub esc_html { |
174 |
|
|
my $html = shift; |
175 |
|
|
$html =~ s/</</g; |
176 |
|
|
$html =~ s/>/>/g; |
177 |
|
|
return $html; |
178 |
|
|
} |
179 |
|
|
|
180 |
dpavlin |
7 |
# call swish |
181 |
|
|
my $sh = SWISH->connect('Fork', |
182 |
|
|
prog => $SWISH, |
183 |
|
|
indexes => $INDEX, |
184 |
dpavlin |
13 |
properties => [qw/swishdocpath swishrank swishtitle headline html/], |
185 |
dpavlin |
7 |
results => sub { |
186 |
|
|
my ($sh,$hit) = @_; |
187 |
|
|
|
188 |
|
|
push @swish_results, { |
189 |
|
|
nr => ($#swish_results + 2), |
190 |
|
|
path => $hit->swishdocpath, |
191 |
dpavlin |
80 |
headline => esc_html($from_utf8->convert($hit->headline)), |
192 |
dpavlin |
41 |
html => back2html($from_utf8->convert($hit->html)), |
193 |
dpavlin |
7 |
rank => $hit->swishrank }; |
194 |
|
|
|
195 |
|
|
}, |
196 |
|
|
#startnum => 0, |
197 |
dpavlin |
47 |
maxhits => $MAX_HITS |
198 |
dpavlin |
7 |
); |
199 |
|
|
|
200 |
|
|
die $SWISH::errstr unless $sh; |
201 |
dpavlin |
47 |
# construct swish query |
202 |
|
|
my $sw_q = join(" and ",@s_arr); |
203 |
dpavlin |
111 |
if (@path_arr && $q->param('show_full')) { |
204 |
dpavlin |
47 |
$sw_q .= "and (swishdocpath=\""; |
205 |
|
|
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
206 |
|
|
$sw_q .= "\")"; |
207 |
|
|
$tmpl->param('full',1); # show full records |
208 |
dpavlin |
149 |
} elsif ($q->param('show_full')) { |
209 |
|
|
# just show full path, no path defined |
210 |
|
|
$tmpl->param('full',1); |
211 |
dpavlin |
121 |
} else { |
212 |
|
|
$tmpl->param('full',0); |
213 |
dpavlin |
47 |
} |
214 |
dpavlin |
7 |
|
215 |
dpavlin |
47 |
my $hits = $sh->query($sw_q); |
216 |
|
|
|
217 |
dpavlin |
7 |
$tmpl->param('hits',$hits); |
218 |
dpavlin |
47 |
$tmpl->param('search',$sw_q); |
219 |
dpavlin |
7 |
|
220 |
dpavlin |
76 |
$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0); |
221 |
|
|
$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0); |
222 |
dpavlin |
51 |
|
223 |
dpavlin |
126 |
$tmpl->param('url_params',"?".join("&",@url_params)); |
224 |
|
|
|
225 |
dpavlin |
7 |
# create a Pager object |
226 |
|
|
my $pager = HTML::Pager->new( |
227 |
|
|
# required parameters |
228 |
|
|
query => $q, |
229 |
|
|
get_data_callback => sub { |
230 |
|
|
my ($offset, $rows) = @_; |
231 |
|
|
|
232 |
|
|
my @result; |
233 |
|
|
for (my $i=0; $i<$rows; $i++) { |
234 |
dpavlin |
114 |
my $r = $swish_results[$offset+$i]; |
235 |
dpavlin |
121 |
if ($r && $tmpl->param('full')) { |
236 |
dpavlin |
114 |
push @result, $r; |
237 |
|
|
} elsif ($r) { |
238 |
|
|
# if not full output, skip html |
239 |
|
|
delete $r->{html}; |
240 |
|
|
push @result, $r; |
241 |
|
|
} |
242 |
dpavlin |
7 |
} |
243 |
|
|
return \@result; |
244 |
|
|
}, |
245 |
|
|
rows => $hits, |
246 |
|
|
page_size => $ON_PAGE, |
247 |
|
|
# some optional parameters |
248 |
dpavlin |
112 |
persist_vars => [ @persist_vars ], |
249 |
dpavlin |
7 |
#cell_space_color => '#000000', |
250 |
|
|
#cell_background_color => '#ffffff', |
251 |
|
|
#nav_background_color => '#dddddd', |
252 |
|
|
#javascript_presubmit => 'last_minute_javascript()', |
253 |
|
|
debug => 1, |
254 |
|
|
template => $tmpl, |
255 |
|
|
); |
256 |
|
|
|
257 |
|
|
my $html = $pager->output; |
258 |
|
|
|
259 |
dpavlin |
147 |
return in_template($html); |
260 |
dpavlin |
7 |
} |
261 |
|
|
|
262 |
dpavlin |
9 |
sub show_index { |
263 |
|
|
my $self = shift; |
264 |
|
|
my $i = shift; # field number |
265 |
|
|
|
266 |
|
|
my $q = $self->query(); |
267 |
|
|
|
268 |
dpavlin |
11 |
my $field = $q->param("f$i"); |
269 |
|
|
my $limit = $q->param("v$i"); |
270 |
|
|
|
271 |
dpavlin |
9 |
my $html; |
272 |
|
|
|
273 |
dpavlin |
53 |
my $index = new index_DBI( |
274 |
|
|
$cfg_global->val('global', 'dbi_dbd'), |
275 |
|
|
$cfg_global->val('global', 'dbi_dsn'), |
276 |
|
|
$cfg_global->val('global', 'dbi_user'), |
277 |
|
|
$cfg_global->val('global', 'dbi_passwd') || '' |
278 |
|
|
); |
279 |
dpavlin |
9 |
|
280 |
dpavlin |
140 |
my $total = $index->count($field,$limit); |
281 |
dpavlin |
12 |
if (! $total) { |
282 |
dpavlin |
11 |
my $tmpl = $self->load_tmpl('no_index.html'); |
283 |
|
|
$tmpl->param('field',$field); |
284 |
|
|
$html = $tmpl->output; |
285 |
|
|
return $html; |
286 |
|
|
} |
287 |
dpavlin |
9 |
|
288 |
dpavlin |
124 |
my $tmpl = $self->load_tmpl('index_res.html', global_vars => 1); |
289 |
dpavlin |
12 |
$tmpl->param('field',$field); |
290 |
|
|
$tmpl->param('limit',$limit); |
291 |
|
|
$tmpl->param('total',$total); |
292 |
dpavlin |
11 |
|
293 |
dpavlin |
51 |
# FIXME I should set offset and leave out limit from fetch!! |
294 |
dpavlin |
16 |
# if (! $q->param("PAGER_offset") { |
295 |
|
|
# $q->param("Pager_offet) |
296 |
|
|
# } |
297 |
|
|
|
298 |
dpavlin |
12 |
my $pager = HTML::Pager->new( |
299 |
|
|
query => $q, |
300 |
|
|
get_data_callback => sub { |
301 |
|
|
my ($offset, $rows) = @_; |
302 |
dpavlin |
11 |
|
303 |
dpavlin |
140 |
my @result = $index->fetch($field,$limit, $offset, $rows); |
304 |
dpavlin |
12 |
return \@result; |
305 |
|
|
}, |
306 |
|
|
rows => $total, |
307 |
|
|
page_size => $ON_PAGE, |
308 |
|
|
persist_vars => [ |
309 |
|
|
'rm', |
310 |
|
|
"f$i", "v$i", "f".$i."_index", |
311 |
|
|
'offset', |
312 |
|
|
], |
313 |
|
|
debug => 1, |
314 |
|
|
template => $tmpl, |
315 |
|
|
); |
316 |
dpavlin |
11 |
|
317 |
dpavlin |
147 |
return in_template($pager->output); |
318 |
dpavlin |
9 |
} |
319 |
|
|
|
320 |
dpavlin |
7 |
1; |