git.rot13.org Git - webpac/blob - WebPac.pm

   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
  32
  33 Text::Iconv->raise_error(0);     # Conversion errors raise exceptions
  34
  35 my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
  36
  37
  38 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 #               'user' => 'show_user_detail',
  45                 'index' => 'show_index',
  46         );
  47         $self->start_mode('search');
  48         $self->mode_param('rm');
  49
  50         $self->header_props(-charset=>$CHARSET);
  51 }
  52
  53 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 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 $fif = new HTML::FillInForm;
  76
  77         return in_template($fif->fill(scalarref => \$html, fobject => $q,
  78                 target => 'search'));
  79 }
  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         my @path_arr = $q->param('path');
  93         my $full = $q->param('full');
  94
  95         my @persist_vars = ( 'rm' );
  96         my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.$q->param('PAGER_offset') || 0 );
  97
  98         for(my $i = 1; $i <=30; $i++) {
  99
 100                 return show_index($self, $i) if ($q->param("f".$i."_index"));
 101
 102                 next if (! $q->param("v$i"));
 103                 next if (! $q->param("f$i"));
 104
 105                 push @persist_vars, "f$i";
 106                 push @persist_vars, "v$i";
 107
 108                 push @url_params,"f$i=".$q->url_param("f$i");
 109                 push @url_params,"v$i=".$q->url_param("v$i");
 110
 111                 # re-write query from +/- to and/and not
 112                 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
 124                         foreach (split(/\s+/,$search)) {
 125                                 if (m/^([+-])(\S+)/) {
 126                                         $s.= ($s) ? "and " : "";
 127                                         $s.="not " if ($1 eq "-");
 128                                         $s.="$2* ";
 129                                 } elsif (m/^\s*(and|or|not)\s*$/i) {
 130                                         $s.="$_ ";
 131                                 # don't add * to words with less than x chars
 132                                 } elsif (length($_) <= $MIN_WILDCARD) {
 133                                         $s.="$_ ";
 134                                 } else {
 135                                         $s.="$_* ";
 136                                 }
 137                         }
 138                         $s =~ s/\*+/*/g;
 139                         push @swish_q,$s;
 140                 }
 141                 # 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         }
 145
 146         my $tmpl = $self->load_tmpl('results.html', global_vars => 1);
 147
 148         sub esc_html {
 149                 my $html = shift;
 150                 $html =~ s/</&lt;/g;
 151                 $html =~ s/>/&gt;/g;
 152                 return $html;
 153         }
 154
 155         # call swish
 156         my $sh = SWISH->connect('Fork',
 157                 prog     => $SWISH,
 158                 indexes  => $INDEX,
 159                 properties  => [qw/swishdocpath swishrank swishtitle headline html/],
 160                 results  => sub {
 161                         my ($sh,$hit) = @_;
 162
 163                         push @swish_results, {
 164                                 nr => ($#swish_results + 2),
 165                                 path => $hit->swishdocpath,
 166                                 headline => esc_html($from_utf8->convert($hit->headline)),
 167                                 html => back2html($from_utf8->convert($hit->html)),
 168                                 rank => $hit->swishrank };
 169
 170                 },
 171                 #startnum => 0,
 172                 maxhits => $MAX_HITS
 173         );
 174
 175         die $SWISH::errstr unless $sh;
 176         # construct swish query
 177         my $sw_q = join(" and ",@s_arr);
 178         if (@path_arr && $q->param('show_full')) {
 179                 $sw_q .= "and (swishdocpath=\"";
 180                 $sw_q .= join("\" or swishdocpath=\"",@path_arr);
 181                 $sw_q .= "\")";
 182                 $tmpl->param('full',1); # show full records
 183         } elsif ($q->param('show_full')) {
 184                 # just show full path, no path defined
 185                 $tmpl->param('full',1);
 186         } else {
 187                 $tmpl->param('full',0);
 188         }
 189
 190         my $hits = $sh->query($sw_q);
 191
 192         $tmpl->param('hits',$hits);
 193         $tmpl->param('search',$sw_q);
 194
 195         $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
 196         $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
 197
 198         $tmpl->param('url_params',"?".join("&",@url_params));
 199
 200         # create a Pager object
 201         my $pager = HTML::Pager->new(
 202                 # required parameters
 203                 query => $q,
 204                 get_data_callback => sub {
 205                         my ($offset, $rows) = @_;
 206
 207                         my @result;
 208                         for (my $i=0; $i<$rows; $i++) {
 209                                 my $r = $swish_results[$offset+$i];
 210                                 if ($r && $tmpl->param('full')) {
 211                                         push @result, $r;
 212                                 } elsif ($r) {
 213                                         # if not full output, skip html
 214                                         delete $r->{html};
 215                                         push @result, $r;
 216                                 }
 217                         }
 218                         return \@result;
 219                 },
 220                 rows => $hits,
 221                 page_size => $ON_PAGE,
 222                 # some optional parameters
 223                 persist_vars => [ @persist_vars ],
 224                 #cell_space_color => '#000000',
 225                 #cell_background_color => '#ffffff',
 226                 #nav_background_color => '#dddddd',
 227                 #javascript_presubmit => 'last_minute_javascript()',
 228                 debug => 1,
 229                 template => $tmpl,
 230         );
 231
 232         my $html = $pager->output;
 233
 234         return in_template($html);
 235 }
 236
 237 sub show_index {
 238         my $self = shift;
 239         my $i = shift;          # field number
 240
 241         my $q = $self->query();
 242
 243         my $field = $q->param("f$i");
 244         my $limit = $q->param("v$i");
 245
 246         my $html;
 247
 248         my $index = new index_DBI(
 249                 $cfg_global->val('global', 'dbi_dbd'),
 250                 $cfg_global->val('global', 'dbi_dsn'),
 251                 $cfg_global->val('global', 'dbi_user'),
 252                 $cfg_global->val('global', 'dbi_passwd') || ''
 253         );
 254
 255         my $total = $index->count($field,$limit);
 256         if (! $total) {
 257                 my $tmpl = $self->load_tmpl('no_index.html');
 258                 $tmpl->param('field',$field);
 259                 $html = $tmpl->output;
 260                 return $html;
 261         }
 262
 263         my $tmpl = $self->load_tmpl('index_res.html', global_vars => 1);
 264         $tmpl->param('field',$field);
 265         $tmpl->param('limit',$limit);
 266         $tmpl->param('total',$total);
 267
 268 # FIXME I should set offset and leave out limit from fetch!!
 269 #       if (! $q->param("PAGER_offset") {
 270 #               $q->param("Pager_offet)
 271 #       }
 272
 273         my $pager = HTML::Pager->new(
 274                 query => $q,
 275                 get_data_callback => sub {
 276                         my ($offset, $rows) = @_;
 277
 278                         my @result = $index->fetch($field,$limit, $offset, $rows);
 279                         return \@result;
 280                 },
 281                 rows => $total,
 282                 page_size => $ON_PAGE,
 283                 persist_vars => [
 284                         'rm',
 285                         "f$i", "v$i", "f".$i."_index",
 286                         'offset',
 287                         ],
 288                 debug => 1,
 289                 template => $tmpl,
 290         );
 291
 292         return in_template($pager->output);
 293 }
 294
 295 1;