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_cache; |
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 |
my $MIN_WILDCARD =$cfg_global->val('webpac', 'min_wildcard') || 1; |
30 |
my $TEMPLATE =$cfg_global->val('webpac', 'template'); |
31 |
my $UNAC_FILTER =$cfg_global->val('global', 'unac_filter'); |
32 |
my $BASE_PATH =$cfg_global->val('webpac', 'base_path'); |
33 |
|
34 |
if ($UNAC_FILTER) { |
35 |
require $UNAC_FILTER; |
36 |
} |
37 |
|
38 |
Text::Iconv->raise_error(0); # Conversion errors raise exceptions |
39 |
|
40 |
my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET); |
41 |
|
42 |
# use path from cgi script to support templates in subdirs |
43 |
sub url_ex { |
44 |
my $q = shift || die "suff2file needs CGI object!"; |
45 |
my $tpl = shift || die "url_ex needs template name!"; |
46 |
return suff2file($BASE_PATH, $q->url(-path => 1),$TEMPLATE_PATH,$tpl); |
47 |
} |
48 |
|
49 |
sub suff2file($$$$) { |
50 |
my ($base_path, $p, $path, $tpl) = @_; |
51 |
|
52 |
return $tpl if (! $base_path); |
53 |
|
54 |
# strip everything to and including base path, leaving only |
55 |
# additional (virtual) path |
56 |
if ($p =~ s,^.*?$base_path,,) { |
57 |
$p =~ s,/*,,g; |
58 |
my ($name,$ext) = split(/\./,$tpl); |
59 |
$p = $name . $p . "." . $ext; |
60 |
} else { |
61 |
# if unable reset it! |
62 |
$p = $tpl; |
63 |
} |
64 |
|
65 |
if ( -e "$path/$p") { |
66 |
return $p; |
67 |
} else { |
68 |
return $tpl; |
69 |
} |
70 |
|
71 |
} |
72 |
|
73 |
sub setup { |
74 |
my $self = shift; |
75 |
$self->tmpl_path($TEMPLATE_PATH); |
76 |
$self->run_modes( |
77 |
'search' => 'show_search_form', |
78 |
'results' => 'show_results_list', |
79 |
# 'user' => 'show_user_detail', |
80 |
'index' => 'show_index', |
81 |
); |
82 |
$self->start_mode('search'); |
83 |
$self->mode_param('rm'); |
84 |
|
85 |
$self->header_props(-charset=>$CHARSET); |
86 |
} |
87 |
|
88 |
sub in_template { |
89 |
my $q = shift || die "need CGI object!"; |
90 |
my $html = shift || die "This page is left unintentionally blank"; |
91 |
return $html if (! defined($TEMPLATE)); |
92 |
|
93 |
my ($dir,$tpl); |
94 |
if ($TEMPLATE =~ m,^(.*?/*)([^/]+)$,) { |
95 |
($dir,$tpl) = ($1,$2); |
96 |
} else { |
97 |
die "can't parse TEMPLATE path"; |
98 |
} |
99 |
|
100 |
my $master_tpl = suff2file($BASE_PATH, $q->url(-path => 1),$dir,$tpl); |
101 |
if (open(T, $master_tpl)) { |
102 |
my $template_html = join("\n",<T>); |
103 |
close(T); |
104 |
$template_html =~ s/##webpac##/$html/gsi; |
105 |
return $template_html; |
106 |
} else { |
107 |
return "Can't read template '$master_tpl'"; |
108 |
} |
109 |
} |
110 |
|
111 |
sub show_search_form { |
112 |
my $self = shift; |
113 |
|
114 |
# Get the CGI.pm query object |
115 |
my $q = $self->query(); |
116 |
|
117 |
my $tmpl = $self->load_tmpl(url_ex($q,'search.html')); |
118 |
my $html = $tmpl->output; |
119 |
|
120 |
my $fif = new HTML::FillInForm; |
121 |
|
122 |
return in_template($q,$fif->fill(scalarref => \$html, fobject => $q, |
123 |
target => 'search')); |
124 |
} |
125 |
|
126 |
sub show_results_list { |
127 |
my $self = shift; |
128 |
|
129 |
my $q = $self->query(); |
130 |
|
131 |
my @swish_results; # results from swish |
132 |
|
133 |
# load template for this page |
134 |
|
135 |
my @s_arr; # all queries are located here |
136 |
|
137 |
my @path_arr = $q->param('path'); |
138 |
my $full = $q->param('full'); |
139 |
|
140 |
my @persist_vars = ( 'rm' ); |
141 |
my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.$q->param('PAGER_offset') || 0 ); |
142 |
|
143 |
# support parametars "f" and "v" for start |
144 |
for(my $i = ""; $i <=30; $i++) { |
145 |
|
146 |
return show_index($self, $i) if ($q->param("f".$i."_index")); |
147 |
|
148 |
next if (! $q->param("v$i")); |
149 |
next if (! $q->param("f$i")); |
150 |
|
151 |
push @persist_vars, "f$i"; |
152 |
push @persist_vars, "v$i"; |
153 |
push @persist_vars, "e$i" if ($q->param("e$i")); |
154 |
|
155 |
push @url_params,"f$i=".$q->url_param("f$i"); |
156 |
foreach my $v ($q->url_param("v$i")) { |
157 |
push @url_params,"v$i=$v"; |
158 |
} |
159 |
push @url_params,"e$i=".$q->url_param("e$i"); |
160 |
|
161 |
my $wc="*"; # swish wildcard |
162 |
$wc="" if ($i eq ""); # don't apply wildcard on field 0 |
163 |
|
164 |
# re-write query from +/- to and/and not |
165 |
my @param_vals = $q->param("v$i"); |
166 |
my @swish_q; |
167 |
my ($pre,$post,$exact) = ('','',''); |
168 |
while (my $search = shift @param_vals) { |
169 |
my $s; |
170 |
# remove accents |
171 |
$search = unac_string($CHARSET,$search); |
172 |
while ($search =~ s/\s*("[^"]+")\s*/ /) { |
173 |
$s .= "$1 "; |
174 |
} |
175 |
$search =~ s/^\s+//; |
176 |
$search =~ s/\s+$//; |
177 |
|
178 |
# filed e[nr] is exact match bitmask |
179 |
# 1 = beginning, 2=end, 3=both |
180 |
$pre = '"xxbxx ' if ($q->param("e$i") & 1); |
181 |
$post = ' xxexx"' if ($q->param("e$i") & 2); |
182 |
# add qotes on other side |
183 |
if ($q->param("e$i")) { |
184 |
$pre = '"' if (! $pre); |
185 |
$post = '"' if (! $post); |
186 |
# what about wildcards? |
187 |
$wc = ''; |
188 |
$wc = '*' if ($q->param("e$i") & 4); |
189 |
$exact = '_exact'; |
190 |
} |
191 |
|
192 |
foreach (split(/\s+/,$search)) { |
193 |
if (m/^([+-])(\S+)/) { |
194 |
$s.= ($s) ? "and " : ""; |
195 |
$s.="not " if ($1 eq "-"); |
196 |
$s.=$2.$wc." "; |
197 |
} elsif (m/^\s*(and|or|not)\s*$/i) { |
198 |
$s.=$_." "; |
199 |
# don't add * to words with less than x chars |
200 |
} elsif (length($_) <= $MIN_WILDCARD) { |
201 |
$s.=$_." "; |
202 |
} else { |
203 |
$s.=$_.$wc." "; |
204 |
} |
205 |
} |
206 |
$s =~ s/\*+/*/g; |
207 |
$s = $pre.$s.$post if ($q->param("e$i")); |
208 |
push @swish_q,$s; |
209 |
} |
210 |
# FIXME default operator for multi-value fields is or. There is |
211 |
# no way to change it, except here for now. Is there need? |
212 |
push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")"; |
213 |
} |
214 |
|
215 |
my $tmpl = $self->load_tmpl(url_ex($q,'results.html'), global_vars => 1); |
216 |
|
217 |
sub esc_html { |
218 |
my $html = shift; |
219 |
$html =~ s/</</g; |
220 |
$html =~ s/>/>/g; |
221 |
return $html; |
222 |
} |
223 |
|
224 |
my $sort = 'swishrank'; |
225 |
if ($q->param("sort")) { |
226 |
$sort = 'headline'; |
227 |
push @persist_vars, "sort"; |
228 |
} |
229 |
|
230 |
# call swish |
231 |
my $sh = SWISH->connect('Fork', |
232 |
prog => $SWISH, |
233 |
indexes => $INDEX, |
234 |
properties => [qw/swishdocpath swishrank swishtitle headline html/], |
235 |
results => sub { |
236 |
my ($sh,$hit) = @_; |
237 |
|
238 |
push @swish_results, { |
239 |
nr => ($#swish_results + 2), |
240 |
path => $hit->swishdocpath, |
241 |
headline => esc_html($from_utf8->convert($hit->headline)), |
242 |
html => back2html($from_utf8->convert($hit->html)), |
243 |
rank => $hit->swishrank }; |
244 |
|
245 |
}, |
246 |
#startnum => 0, |
247 |
maxhits => $MAX_HITS, |
248 |
sortorder => $sort, |
249 |
); |
250 |
|
251 |
die $SWISH::errstr unless $sh; |
252 |
# construct swish query |
253 |
my $sw_q = join(" and ",@s_arr); |
254 |
if (@path_arr && $q->param('show_full')) { |
255 |
$sw_q .= "and (swishdocpath=\""; |
256 |
$sw_q .= join("\" or swishdocpath=\"",@path_arr); |
257 |
$sw_q .= "\")"; |
258 |
$tmpl->param('full',1); # show full records |
259 |
} elsif ($q->param('show_full')) { |
260 |
# just show full path, no path defined |
261 |
$tmpl->param('full',1); |
262 |
} else { |
263 |
$tmpl->param('full',0); |
264 |
} |
265 |
|
266 |
my $hits = $sh->query($sw_q); |
267 |
|
268 |
$tmpl->param('hits',$hits); |
269 |
$tmpl->param('search',$sw_q); |
270 |
|
271 |
$tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0); |
272 |
$tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0); |
273 |
|
274 |
$tmpl->param('url_params',"?".join("&",@url_params)); |
275 |
|
276 |
# create a Pager object |
277 |
my $pager = HTML::Pager->new( |
278 |
# required parameters |
279 |
query => $q, |
280 |
get_data_callback => sub { |
281 |
my ($offset, $rows) = @_; |
282 |
|
283 |
my @result; |
284 |
for (my $i=0; $i<$rows; $i++) { |
285 |
my $r = $swish_results[$offset+$i]; |
286 |
if ($r && $tmpl->param('full')) { |
287 |
push @result, $r; |
288 |
} elsif ($r) { |
289 |
# if not full output, skip html |
290 |
delete $r->{html}; |
291 |
push @result, $r; |
292 |
} |
293 |
} |
294 |
return \@result; |
295 |
}, |
296 |
rows => $hits, |
297 |
page_size => $ON_PAGE, |
298 |
# some optional parameters |
299 |
persist_vars => [ @persist_vars ], |
300 |
#cell_space_color => '#000000', |
301 |
#cell_background_color => '#ffffff', |
302 |
#nav_background_color => '#dddddd', |
303 |
#javascript_presubmit => 'last_minute_javascript()', |
304 |
debug => 1, |
305 |
template => $tmpl, |
306 |
); |
307 |
|
308 |
my $html = $pager->output; |
309 |
|
310 |
return in_template($q,$html); |
311 |
} |
312 |
|
313 |
sub show_index { |
314 |
my $self = shift; |
315 |
my $i = shift; # field number |
316 |
|
317 |
my $q = $self->query(); |
318 |
|
319 |
my $field = $q->param("f$i"); |
320 |
my $limit = $q->param("v$i"); |
321 |
|
322 |
my $html; |
323 |
|
324 |
my $index = new index_DBI( |
325 |
$cfg_global->val('global', 'dbi_dbd'), |
326 |
$cfg_global->val('global', 'dbi_dsn'), |
327 |
$cfg_global->val('global', 'dbi_user'), |
328 |
$cfg_global->val('global', 'dbi_passwd') || '' |
329 |
); |
330 |
|
331 |
my $total = $index->count($field,$limit); |
332 |
if (! $total) { |
333 |
my $tmpl = $self->load_tmpl(url_ex($q,'no_index.html')); |
334 |
$tmpl->param('field',$field); |
335 |
$html = $tmpl->output; |
336 |
return $html; |
337 |
} |
338 |
|
339 |
my $tmpl = $self->load_tmpl(url_ex($q,'index_res.html'), global_vars => 1); |
340 |
$tmpl->param('field',$field); |
341 |
$tmpl->param('limit',$limit); |
342 |
$tmpl->param('total',$total); |
343 |
|
344 |
# FIXME I should set offset and leave out limit from fetch!! |
345 |
# if (! $q->param("PAGER_offset") { |
346 |
# $q->param("Pager_offet) |
347 |
# } |
348 |
|
349 |
my $pager = HTML::Pager->new( |
350 |
query => $q, |
351 |
get_data_callback => sub { |
352 |
my ($offset, $rows) = @_; |
353 |
|
354 |
my @result = $index->fetch($field,$limit, $offset, $rows); |
355 |
return \@result; |
356 |
}, |
357 |
rows => $total, |
358 |
page_size => $ON_PAGE, |
359 |
persist_vars => [ |
360 |
'rm', |
361 |
"f$i", "v$i", "f".$i."_index", |
362 |
'offset', |
363 |
], |
364 |
debug => 1, |
365 |
template => $tmpl, |
366 |
); |
367 |
|
368 |
return in_template($q,$pager->output); |
369 |
} |
370 |
|
371 |
1; |