1 |
dpavlin |
1132 |
#!/usr/bin/perl |
2 |
|
|
|
3 |
|
|
use warnings; |
4 |
|
|
use strict; |
5 |
|
|
|
6 |
|
|
use CGI qw/:standard/; |
7 |
|
|
use CGI::Carp qw/fatalsToBrowser/; |
8 |
|
|
use File::Slurp; |
9 |
|
|
use YAML; |
10 |
dpavlin |
1138 |
use Data::Page; |
11 |
|
|
use Data::Dump qw/dump/; |
12 |
dpavlin |
1149 |
use SWISH::API; |
13 |
|
|
use JSON; |
14 |
dpavlin |
1249 |
use Text::Unaccent::PurePerl qw/unac_string/; |
15 |
dpavlin |
1250 |
use HTML::FillInForm::Lite; |
16 |
dpavlin |
1328 |
use Encode; |
17 |
dpavlin |
1132 |
|
18 |
dpavlin |
1138 |
my $range_around = 5; |
19 |
dpavlin |
1240 |
my @entries_per_page = ( 30, 50, 100, 500 ); |
20 |
dpavlin |
1141 |
my $debug = param('debug'); |
21 |
dpavlin |
1138 |
|
22 |
dpavlin |
1142 |
print header( |
23 |
|
|
-charset => 'utf-8', |
24 |
|
|
); |
25 |
dpavlin |
1132 |
|
26 |
|
|
sub dump_yaml { |
27 |
dpavlin |
1135 |
my $name = shift; |
28 |
dpavlin |
1171 |
print qq|<div class=dump><tt>$name</tt><pre>|, YAML::Dump( @_ ), qq|</pre></div>| if $debug; |
29 |
dpavlin |
1132 |
} |
30 |
|
|
|
31 |
dpavlin |
1139 |
sub show_pager { |
32 |
dpavlin |
1141 |
my ($pager) = @_; |
33 |
dpavlin |
1138 |
|
34 |
|
|
my @show_pages; |
35 |
|
|
my $after_current = 0; |
36 |
|
|
|
37 |
|
|
if ( $pager->current_page <= $range_around + 2 ) { |
38 |
|
|
@show_pages = ( $pager->first_page .. $pager->current_page ); |
39 |
|
|
$after_current = $range_around - $pager->current_page; |
40 |
|
|
} else { |
41 |
|
|
@show_pages = ( $pager->first_page, '', $pager->current_page - $range_around .. $pager->current_page ); |
42 |
|
|
} |
43 |
|
|
|
44 |
dpavlin |
1139 |
if ( $pager->current_page + $after_current + $range_around + 1 >= $pager->last_page ) { |
45 |
dpavlin |
1138 |
push @show_pages, ( $pager->current_page + 1 .. $pager->last_page ); |
46 |
|
|
} else { |
47 |
|
|
push @show_pages, ( $pager->current_page + 1 .. $pager->current_page + $after_current + $range_around, '', $pager->last_page ); |
48 |
|
|
} |
49 |
|
|
|
50 |
dpavlin |
1140 |
# dump_yaml( 'show_pages', \@show_pages ); |
51 |
dpavlin |
1138 |
|
52 |
dpavlin |
1139 |
return '' unless $#show_pages; |
53 |
|
|
|
54 |
|
|
my ( $prev, $next ) = ( '<<', '>>' ); |
55 |
|
|
|
56 |
dpavlin |
1141 |
sub li_a_href { |
57 |
|
|
my ( $page, $label, $attr ) = @_; |
58 |
|
|
param( 'current_page', $page ); |
59 |
|
|
my $url = self_url( -query => 1 ); |
60 |
|
|
$attr ||= ''; |
61 |
|
|
$label ||= $page; |
62 |
|
|
qq|<li$attr><a href="$url" title="$page">$label</a></li>|; |
63 |
|
|
} |
64 |
|
|
|
65 |
dpavlin |
1139 |
return |
66 |
dpavlin |
1141 |
$pager->previous_page ? li_a_href( $pager->previous_page, $prev ) : qq|<li class=skip>$prev</li>| |
67 |
|
|
, ( map { |
68 |
dpavlin |
1142 |
if ( $_ eq $pager->current_page ) { |
69 |
dpavlin |
1141 |
qq|<li class=current_page>$_</li>|; |
70 |
dpavlin |
1139 |
} elsif ( $_ eq '' ) { |
71 |
dpavlin |
1141 |
qq|<li class=skip>...</li>|; |
72 |
dpavlin |
1139 |
} else { |
73 |
dpavlin |
1141 |
li_a_href( $_ ); |
74 |
dpavlin |
1139 |
} |
75 |
|
|
} @show_pages ) |
76 |
dpavlin |
1141 |
, $pager->next_page ? li_a_href( $pager->next_page, $next ) : qq|<li class=skip>$next</li>| |
77 |
dpavlin |
1139 |
; |
78 |
|
|
|
79 |
dpavlin |
1138 |
} |
80 |
|
|
|
81 |
dpavlin |
1137 |
my $path = $ENV{PATH_INFO} || 'ecas'; |
82 |
dpavlin |
1173 |
$path =~ s{^/+}{}; |
83 |
|
|
$path =~ s{/+$}{}; |
84 |
dpavlin |
1132 |
my $dir = $0; |
85 |
|
|
$dir =~ s{/[^/]+.cgi}{}; |
86 |
|
|
|
87 |
dpavlin |
1173 |
dump_yaml( 'dir', $dir ); |
88 |
|
|
|
89 |
dpavlin |
1132 |
my $config = YAML::LoadFile( "$dir/$path/config.yml" ); |
90 |
|
|
|
91 |
|
|
my $database = (keys %{ $config->{databases} })[0]; |
92 |
|
|
die "$database not in $path" unless $path =~ m{\Q$database\E}; |
93 |
|
|
|
94 |
dpavlin |
1137 |
my $html_markup = "$dir/$path/html.pm"; |
95 |
|
|
my $html_markup_skip; |
96 |
|
|
if ( -e $html_markup ) { |
97 |
|
|
require $html_markup; |
98 |
|
|
$html_markup = $database . '::html'; |
99 |
|
|
} else { |
100 |
|
|
undef $html_markup; |
101 |
|
|
} |
102 |
|
|
|
103 |
dpavlin |
1165 |
my $stats; |
104 |
|
|
{ |
105 |
|
|
my $path = "$dir/../var/swish/$database.yaml"; |
106 |
|
|
$stats = YAML::LoadFile( $path ); |
107 |
|
|
dump_yaml( "stats $path", $stats ); |
108 |
|
|
} |
109 |
dpavlin |
1132 |
|
110 |
|
|
my $db = $config->{databases}->{$database}; |
111 |
|
|
|
112 |
dpavlin |
1174 |
sub read_config_txt { |
113 |
|
|
my ( $file ) = @_; |
114 |
|
|
my $input; |
115 |
dpavlin |
1219 |
my $path ="$dir/$path/$path-$file.txt"; |
116 |
|
|
if ( ! -e $path ) { |
117 |
|
|
warn "missing $path"; |
118 |
|
|
return; |
119 |
|
|
} |
120 |
|
|
foreach ( split(/[\n\r]+/, read_file( $path ) ) ) { |
121 |
dpavlin |
1174 |
my ( $val,$label ) = split(/\s*\t\s*/,$_,2); |
122 |
|
|
push @{ $input->{ '-values' } }, $val; |
123 |
|
|
$input->{ '-labels' }->{$val} = $label; |
124 |
|
|
} |
125 |
|
|
return $input; |
126 |
|
|
} |
127 |
dpavlin |
1132 |
|
128 |
dpavlin |
1174 |
my $attr_labels = read_config_txt 'labels'; |
129 |
|
|
my $attr_operators = read_config_txt 'operators'; |
130 |
dpavlin |
1155 |
|
131 |
dpavlin |
1237 |
my @attr = @{ $attr_labels->{'-values'} } if $attr_labels; |
132 |
dpavlin |
1174 |
@attr = keys %{ $stats->{attr} } unless @attr; |
133 |
dpavlin |
1155 |
|
134 |
|
|
|
135 |
dpavlin |
1174 |
warn dump( $attr_labels, $attr_operators ); |
136 |
dpavlin |
1155 |
|
137 |
dpavlin |
1157 |
my $only_input; |
138 |
dpavlin |
1228 |
my $inputs_available = 0; |
139 |
dpavlin |
1157 |
|
140 |
|
|
foreach ( @{ $db->{input} } ) { |
141 |
|
|
my $input = $_->{name} || die "no name in ",dump( $_ ); |
142 |
dpavlin |
1242 |
next unless defined $stats->{input}->{$input}; # skip inputs without data |
143 |
dpavlin |
1157 |
if ( ! $only_input->{'-labels'}->{$input} ) { |
144 |
|
|
push @{ $only_input->{'-values'} }, $input; |
145 |
|
|
$only_input->{'-labels'}->{$input} = $_->{description} || $input; |
146 |
dpavlin |
1228 |
$inputs_available++; |
147 |
dpavlin |
1157 |
} |
148 |
|
|
} |
149 |
|
|
|
150 |
|
|
warn "## only_input = ", dump( $only_input ); |
151 |
|
|
|
152 |
dpavlin |
1173 |
my @style = ( '../../style.css' ); |
153 |
|
|
push @style, "../../$path/$path.css" if -e "$dir/$path/$path.css"; |
154 |
|
|
dump_yaml( 'style', \@style ); |
155 |
|
|
|
156 |
dpavlin |
1177 |
sub search_form { |
157 |
dpavlin |
1250 |
|
158 |
|
|
my $form_html = "$dir/$path/$path-search.html"; |
159 |
|
|
if ( -e $form_html ) { |
160 |
|
|
my $html = read_file( $form_html ); |
161 |
|
|
my $q = CGI->new(); |
162 |
|
|
my $h = HTML::FillInForm::Lite->new(); |
163 |
|
|
return $h->fill(\$html, $q); |
164 |
|
|
} |
165 |
|
|
|
166 |
dpavlin |
1177 |
qq|<a name="form"></a>|, |
167 |
dpavlin |
1139 |
start_form( -action => self_url( query => 0 ) ), |
168 |
dpavlin |
1226 |
checkbox_group( |
169 |
dpavlin |
1132 |
-name => 'attr', |
170 |
dpavlin |
1174 |
%$attr_labels, |
171 |
dpavlin |
1132 |
# -linebreak => 0, |
172 |
|
|
), |
173 |
|
|
textfield( -name => 'search' ), |
174 |
dpavlin |
1227 |
$attr_operators ? popup_menu( -name => 'attr_operator', %$attr_operators ) : '', |
175 |
dpavlin |
1240 |
submit( -value => 'Search' ), |
176 |
|
|
# hidden( -name => 'entries_per_page', -default => $entries_per_page ), |
177 |
|
|
popup_menu( -name => 'entries_per_page', -values => [ @entries_per_page ], -title => 'entries per page' ), |
178 |
dpavlin |
1171 |
# we need current_page fixed at 1 so that every submit through form will reset it |
179 |
|
|
qq|<input type=hidden name=current_page value=1 >|, |
180 |
dpavlin |
1142 |
checkbox( -name => 'debug', -default => 0 ), # FIXME hidden? |
181 |
dpavlin |
1143 |
qq|<div id=inputs>|, |
182 |
dpavlin |
1228 |
$inputs_available > 1 ? |
183 |
|
|
h2( 'Select input' ) . |
184 |
dpavlin |
1143 |
checkbox_group( |
185 |
|
|
-name => 'only_input', |
186 |
dpavlin |
1157 |
%$only_input, |
187 |
dpavlin |
1143 |
-linebreak=> 'true', |
188 |
dpavlin |
1228 |
) : '', |
189 |
dpavlin |
1143 |
qq|</div>|, |
190 |
dpavlin |
1177 |
end_form, |
191 |
|
|
; |
192 |
|
|
} |
193 |
dpavlin |
1143 |
|
194 |
dpavlin |
1177 |
|
195 |
|
|
print |
196 |
|
|
start_html( |
197 |
|
|
-title => $db->{name}, |
198 |
|
|
-style => [ @style ], |
199 |
|
|
), |
200 |
|
|
h1( $db->{name} ), |
201 |
|
|
qq|<div id=description>|, $db->{description}, qq|</div>|, |
202 |
dpavlin |
1132 |
; |
203 |
|
|
|
204 |
|
|
if ( my $search = param('search') ) { |
205 |
|
|
|
206 |
dpavlin |
1328 |
$search = unac_string( Encode::decode('utf-8',$search) ); |
207 |
dpavlin |
1249 |
|
208 |
dpavlin |
1177 |
print qq| |
209 |
|
|
<a href="#form" class="skip" title="skip to search form">#</a> |
210 |
|
|
<div id="results"> |
211 |
|
|
|; |
212 |
dpavlin |
1132 |
|
213 |
dpavlin |
1149 |
my $swish = SWISH::API->new( "$dir/../var/swish/$database" ); |
214 |
|
|
$swish->abort_last_error if $swish->Error; |
215 |
dpavlin |
1132 |
|
216 |
dpavlin |
1157 |
my @search = (); |
217 |
dpavlin |
1227 |
my @attrs = param('attr'); |
218 |
|
|
my $op = param('attr_operator'); |
219 |
|
|
|
220 |
dpavlin |
1329 |
if ( $search =~ m{(=|"|\bAND\b|\bOR\b)} ) { |
221 |
dpavlin |
1155 |
push @search, $search; |
222 |
dpavlin |
1227 |
} elsif ( @attrs ) { |
223 |
dpavlin |
1142 |
|
224 |
dpavlin |
1227 |
$op ||= 'Q*'; |
225 |
|
|
my @or; |
226 |
|
|
foreach my $attr ( @attrs ) { |
227 |
|
|
my $v = $search; |
228 |
|
|
$v =~ s/^\s+//; |
229 |
|
|
warn "-- v: $v\n"; |
230 |
|
|
sub rewrite { |
231 |
|
|
my ( $attr, $whitespace, $v ) = @_; |
232 |
|
|
warn "## filter $op $whitespace $v\n"; |
233 |
|
|
my $template = $op; |
234 |
|
|
$template =~ s{Q}{$v}; |
235 |
|
|
$whitespace = " AND " if $whitespace; |
236 |
|
|
|
237 |
dpavlin |
1340 |
# don't return -* &* and other non-word characters |
238 |
dpavlin |
1351 |
return '' if $template =~ m/^\W\*$/ || $template =~ m/\band\b/i; |
239 |
dpavlin |
1340 |
|
240 |
dpavlin |
1227 |
return |
241 |
|
|
$whitespace . |
242 |
|
|
$attr . '="' . $template . '"'; |
243 |
|
|
; |
244 |
|
|
}; |
245 |
|
|
if ( $op =~ m{\s} ) { |
246 |
|
|
my $template = $op; |
247 |
|
|
$template =~ s{Q}{$v}; |
248 |
|
|
$v = $attr . '="' . $template . '"'; |
249 |
|
|
} else { |
250 |
|
|
$v =~ s{(\s*)(\S+)}{rewrite($attr,$1,$2)}ge; |
251 |
|
|
} |
252 |
|
|
|
253 |
|
|
push @or, $v; |
254 |
|
|
|
255 |
dpavlin |
1191 |
} |
256 |
dpavlin |
1227 |
push @search, '(' . join(') OR (', @or) . ')'; |
257 |
dpavlin |
1155 |
|
258 |
|
|
} else { |
259 |
dpavlin |
1162 |
push @search, "all=\"$search\""; |
260 |
dpavlin |
1155 |
} |
261 |
|
|
|
262 |
dpavlin |
1157 |
my $q = '(' . join(') AND (', @search) . ')'; |
263 |
dpavlin |
1227 |
|
264 |
|
|
my @only_input = param('only_input'); |
265 |
|
|
$q .= ' AND ((' . join(') OR (', map { "input=\"$_\"" } @only_input) . '))' if @only_input; |
266 |
|
|
|
267 |
dpavlin |
1149 |
warn "# query: $q\n"; |
268 |
dpavlin |
1330 |
my $search_obj = $swish->new_search_object; |
269 |
|
|
if ( my $sort = param('sort') ) { |
270 |
|
|
$search_obj->set_sort( $sort ); |
271 |
|
|
} |
272 |
|
|
my $swish_results = $search_obj->execute( $q ); |
273 |
dpavlin |
1132 |
|
274 |
dpavlin |
1149 |
dump_yaml( 'swish_results', $swish_results ); |
275 |
dpavlin |
1138 |
|
276 |
dpavlin |
1175 |
my $pager = Data::Page->new; |
277 |
|
|
$pager->$_( param($_) ) foreach ( qw/entries_per_page current_page/ ); |
278 |
dpavlin |
1149 |
$pager->total_entries( $swish_results->hits ); |
279 |
|
|
|
280 |
dpavlin |
1175 |
dump_yaml( 'pager', $pager ); |
281 |
dpavlin |
1170 |
|
282 |
dpavlin |
1175 |
$swish_results->seek_result( $pager->first - 1 ); |
283 |
|
|
|
284 |
dpavlin |
1149 |
if ( ! $pager->total_entries ) { |
285 |
dpavlin |
1155 |
my $no_results = 'No results for search <b>%s</b>'; |
286 |
|
|
$no_results = $swish->error_string . '<br><b>%s</b>' if $swish->error; |
287 |
|
|
printf qq|<div class="error">$no_results</div>\n\n|, $q; |
288 |
dpavlin |
1132 |
} else { |
289 |
dpavlin |
1138 |
|
290 |
dpavlin |
1153 |
my $results = "<b>%d</b> results for search <b>%s</b> showing results %d - %d"; |
291 |
dpavlin |
1155 |
printf qq|<div class="message">$results</div>\n\n|, $pager->total_entries, $q, $pager->first, $pager->last; |
292 |
dpavlin |
1135 |
|
293 |
dpavlin |
1141 |
my $pager_html = join("\n", show_pager( $pager )); |
294 |
dpavlin |
1140 |
|
295 |
dpavlin |
1141 |
print qq|<ul class="pager">$pager_html</ul>\n\n| if $pager_html; |
296 |
|
|
|
297 |
dpavlin |
1175 |
my $nr = $pager->first; |
298 |
|
|
print qq|<ol start=$nr>\n|; |
299 |
dpavlin |
1138 |
|
300 |
dpavlin |
1170 |
my $limit = $pager->entries_on_this_page; |
301 |
|
|
|
302 |
dpavlin |
1176 |
my $nr = 1; |
303 |
|
|
|
304 |
dpavlin |
1149 |
while ( my $result = $swish_results->next_result ) { |
305 |
dpavlin |
1170 |
|
306 |
dpavlin |
1248 |
my $data = $result->property('data'); |
307 |
dpavlin |
1166 |
dump_yaml( 'data', $data ); |
308 |
dpavlin |
1248 |
# FIXME if we produce valid json we shouldn't need eval here! |
309 |
|
|
eval { $data = from_json( $data, {utf8 => 1} ); }; |
310 |
|
|
if ( $@ ) { |
311 |
|
|
warn "ERROR: $@ from ",dump( $data ); |
312 |
|
|
next; |
313 |
|
|
} |
314 |
dpavlin |
1166 |
|
315 |
dpavlin |
1176 |
my $li_class = ''; |
316 |
dpavlin |
1177 |
$li_class = qq| class="z"| if $nr % 2 == 0; |
317 |
dpavlin |
1176 |
print qq|<li$li_class>|; |
318 |
dpavlin |
1137 |
foreach my $attr ( @attr ) { |
319 |
dpavlin |
1149 |
next unless defined $data->{$attr}; |
320 |
|
|
my $v = $data->{$attr}; |
321 |
|
|
if ( $html_markup && ! $html_markup_skip->{$attr} ) { |
322 |
dpavlin |
1238 |
eval "\$v = $html_markup->$attr( \$v, \$data );"; |
323 |
dpavlin |
1137 |
if ( $@ ) { |
324 |
|
|
warn "disable html markup for $attr: $@"; |
325 |
|
|
$html_markup_skip->{$attr} = $@; |
326 |
|
|
} |
327 |
dpavlin |
1218 |
} else { |
328 |
|
|
$v =~ s{(http://\S+)}{<a href="$1">$1</a>}; |
329 |
dpavlin |
1137 |
} |
330 |
dpavlin |
1174 |
my $label = $attr_labels->{'-labels'}->{$attr} || $attr; |
331 |
|
|
print qq|<div><label>$label</label><span class=$attr>$v</span></div>\n|; |
332 |
dpavlin |
1137 |
} |
333 |
dpavlin |
1132 |
print qq|</li>\n|; |
334 |
dpavlin |
1175 |
|
335 |
|
|
last if $nr++ == $pager->last; |
336 |
dpavlin |
1132 |
} |
337 |
dpavlin |
1141 |
print qq|</ol>\n\n|; |
338 |
dpavlin |
1140 |
|
339 |
dpavlin |
1141 |
print qq|<ul class="pager bottom">$pager_html</ul>\n\n| if $pager_html; |
340 |
dpavlin |
1132 |
} |
341 |
|
|
print qq|</div>|; |
342 |
dpavlin |
1138 |
|
343 |
|
|
dump_yaml( 'pager', $pager ); |
344 |
|
|
|
345 |
dpavlin |
1132 |
} |
346 |
|
|
|
347 |
dpavlin |
1177 |
print search_form; |
348 |
|
|
|
349 |
dpavlin |
1165 |
dump_yaml( "config databases $database", $db ); |
350 |
dpavlin |
1137 |
dump_yaml( 'html_markup_skip', $html_markup_skip ); |
351 |
dpavlin |
1132 |
|
352 |
|
|
print end_html; |