/[webpac2]/trunk/vhost/webpac2.cgi
This is repository of my old source code which isn't updated any more. Go to git.rot13.org for current projects!
ViewVC logotype

Annotation of /trunk/vhost/webpac2.cgi

Parent Directory Parent Directory | Revision Log Revision Log


Revision 1249 - (hide annotations)
Fri Jul 24 09:29:03 2009 UTC (14 years, 9 months ago) by dpavlin
File size: 8557 byte(s)
unaccent search query before transfering it to search engine

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

Properties

Name Value
svn:executable *

  ViewVC Help
Powered by ViewVC 1.1.26