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;
  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
  31
  32 Text::Iconv->raise_error(0);     # Conversion errors raise exceptions
  33
  34 my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
  35
  36
  37 sub setup {
  38         my $self = shift;
  39         $self->tmpl_path($TEMPLATE_PATH);
  40         $self->run_modes(
  41                 'search' => 'show_search_form',
  42                 'results' => 'show_results_list',
  43 #               'user' => 'show_user_detail',
  44                 'index' => 'show_index',
  45         );
  46         $self->start_mode('search');
  47         $self->mode_param('rm');
  48
  49         $self->header_props(-charset=>$CHARSET);
  50 }
  51
  52 sub show_search_form {
  53         my $self = shift;
  54
  55         # Get the CGI.pm query object
  56         my $q = $self->query();
  57
  58         my $tmpl = $self->load_tmpl('search.html');
  59         my $html = $tmpl->output;
  60
  61         my $fif = new HTML::FillInForm;
  62
  63         return $fif->fill(scalarref => \$html, fobject => $q,
  64                 target => 'search');
  65 }
  66
  67 sub show_results_list {
  68         my $self = shift;
  69
  70         my $q = $self->query();
  71
  72         my @swish_results;      # results from swish
  73
  74         # load template for this page
  75
  76         my @s_arr;      # all queries are located here
  77
  78         my @path_arr = $q->param('path');
  79         my $full = $q->param('full');
  80
  81         my @persist_vars = ( 'rm' );
  82         my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.$q->param('PAGER_offset') || 0 );
  83
  84         for(my $i = 1; $i <=30; $i++) {
  85
  86                 return show_index($self, $i) if ($q->param("f".$i."_index"));
  87
  88                 next if (! $q->param("v$i"));
  89                 next if (! $q->param("f$i"));
  90
  91                 push @persist_vars, "f$i";
  92                 push @persist_vars, "v$i";
  93
  94                 push @url_params,"f$i=".$q->url_param("f$i");
  95                 push @url_params,"v$i=".$q->url_param("v$i");
  96
  97                 # re-write query from +/- to and/and not
  98                 my @param_vals = $q->param("v$i");
  99                 my @swish_q;
 100                 while (my $search = shift @param_vals) {
 101                         my $s;
 102                         # remove accents
 103                         $search = unac_string($CHARSET,$search);
 104                         while ($search =~ s/\s*("[^"]+")\s*/ /) {
 105                                 $s .= "$1 ";
 106                         }
 107                         $search =~ s/^\s+//;
 108                         $search =~ s/\s+$//;
 109
 110                         foreach (split(/\s+/,$search)) {
 111                                 if (m/^([+-])(\S+)/) {
 112                                         $s.= ($s) ? "and " : "";
 113                                         $s.="not " if ($1 eq "-");
 114                                         $s.="$2* ";
 115                                 } elsif (m/^\s*(and|or|not)\s*$/i) {
 116                                         $s.="$_ ";
 117                                 # don't add * to words with less than x chars
 118                                 } elsif (length($_) <= $MIN_WILDCARD) {
 119                                         $s.="$_ ";
 120                                 } else {
 121                                         $s.="$_* ";
 122                                 }
 123                         }
 124                         $s =~ s/\*+/*/g;
 125                         push @swish_q,$s;
 126                 }
 127                 # FIXME default operator for multi-value fields is or. There is
 128                 # no way to change it, except here for now. Is there need?
 129                 push @s_arr, $q->param("f$i")."_swish=(".join(" or ",@swish_q).")";
 130         }
 131
 132         my $tmpl = $self->load_tmpl('results.html', global_vars => 1);
 133
 134         sub esc_html {
 135                 my $html = shift;
 136                 $html =~ s/</&lt;/g;
 137                 $html =~ s/>/&gt;/g;
 138                 return $html;
 139         }
 140
 141         # call swish
 142         my $sh = SWISH->connect('Fork',
 143                 prog     => $SWISH,
 144                 indexes  => $INDEX,
 145                 properties  => [qw/swishdocpath swishrank swishtitle headline html/],
 146                 results  => sub {
 147                         my ($sh,$hit) = @_;
 148
 149                         push @swish_results, {
 150                                 nr => ($#swish_results + 2),
 151                                 path => $hit->swishdocpath,
 152                                 headline => esc_html($from_utf8->convert($hit->headline)),
 153                                 html => back2html($from_utf8->convert($hit->html)),
 154                                 rank => $hit->swishrank };
 155
 156                 },
 157                 #startnum => 0,
 158                 maxhits => $MAX_HITS
 159         );
 160
 161         die $SWISH::errstr unless $sh;
 162         # construct swish query
 163         my $sw_q = join(" and ",@s_arr);
 164         if (@path_arr && $q->param('show_full')) {
 165                 $sw_q .= "and (swishdocpath=\"";
 166                 $sw_q .= join("\" or swishdocpath=\"",@path_arr);
 167                 $sw_q .= "\")";
 168                 $tmpl->param('full',1); # show full records
 169         } else {
 170                 $tmpl->param('full',0);
 171         }
 172
 173         my $hits = $sh->query($sw_q);
 174
 175         $tmpl->param('hits',$hits);
 176         $tmpl->param('search',$sw_q);
 177
 178         $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
 179         $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
 180
 181         $tmpl->param('url_params',"?".join("&",@url_params));
 182
 183         # create a Pager object
 184         my $pager = HTML::Pager->new(
 185                 # required parameters
 186                 query => $q,
 187                 get_data_callback => sub {
 188                         my ($offset, $rows) = @_;
 189
 190                         my @result;
 191                         for (my $i=0; $i<$rows; $i++) {
 192                                 my $r = $swish_results[$offset+$i];
 193                                 if ($r && $tmpl->param('full')) {
 194                                         push @result, $r;
 195                                 } elsif ($r) {
 196                                         # if not full output, skip html
 197                                         delete $r->{html};
 198                                         push @result, $r;
 199                                 }
 200                         }
 201                         return \@result;
 202                 },
 203                 rows => $hits,
 204                 page_size => $ON_PAGE,
 205                 # some optional parameters
 206                 persist_vars => [ @persist_vars ],
 207                 #cell_space_color => '#000000',
 208                 #cell_background_color => '#ffffff',
 209                 #nav_background_color => '#dddddd',
 210                 #javascript_presubmit => 'last_minute_javascript()',
 211                 debug => 1,
 212                 template => $tmpl,
 213         );
 214
 215         my $html = $pager->output;
 216
 217         return $html;
 218 }
 219
 220 sub show_index {
 221         my $self = shift;
 222         my $i = shift;          # field number
 223
 224         my $q = $self->query();
 225
 226         my $field = $q->param("f$i");
 227         my $limit = $q->param("v$i");
 228
 229         my $html;
 230
 231         my $index = new index_DBI(
 232                 $cfg_global->val('global', 'dbi_dbd'),
 233                 $cfg_global->val('global', 'dbi_dsn'),
 234                 $cfg_global->val('global', 'dbi_user'),
 235                 $cfg_global->val('global', 'dbi_passwd') || ''
 236         );
 237
 238         my $total = $index->check($field);
 239         if (! $total) {
 240                 my $tmpl = $self->load_tmpl('no_index.html');
 241                 $tmpl->param('field',$field);
 242                 $html = $tmpl->output;
 243                 return $html;
 244         }
 245
 246         my $tmpl = $self->load_tmpl('index_res.html', global_vars => 1);
 247         $tmpl->param('field',$field);
 248         $tmpl->param('limit',$limit);
 249         $tmpl->param('total',$total);
 250
 251 # FIXME I should set offset and leave out limit from fetch!!
 252 #       if (! $q->param("PAGER_offset") {
 253 #               $q->param("Pager_offet)
 254 #       }
 255
 256         my $pager = HTML::Pager->new(
 257                 query => $q,
 258                 get_data_callback => sub {
 259                         my ($offset, $rows) = @_;
 260
 261                         my @result = $index->fetch($field,'item',$limit, $offset, $rows);
 262                         return \@result;
 263                 },
 264                 rows => $total,
 265                 page_size => $ON_PAGE,
 266                 persist_vars => [
 267                         'rm',
 268                         "f$i", "v$i", "f".$i."_index",
 269                         'offset',
 270                         ],
 271                 debug => 1,
 272                 template => $tmpl,
 273         );
 274
 275         return $pager->output;
 276 }
 277
 278 1;