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         # support parametars "f" and "v" for start
  99         for(my $i = ""; $i <=30; $i++) {
 100
 101                 return show_index($self, $i) if ($q->param("f".$i."_index"));
 102
 103                 next if (! $q->param("v$i"));
 104                 next if (! $q->param("f$i"));
 105
 106                 push @persist_vars, "f$i";
 107                 push @persist_vars, "v$i";
 108
 109                 push @url_params,"f$i=".$q->url_param("f$i");
 110                 push @url_params,"v$i=".$q->url_param("v$i");
 111
 112                 my $wc="*";     # swish wildcard
 113                 $wc="" if ($i eq "");   # don't apply wildcard on field 0
 114
 115                 # re-write query from +/- to and/and not
 116                 my @param_vals = $q->param("v$i");
 117                 my @swish_q;
 118                 while (my $search = shift @param_vals) {
 119                         my $s;
 120                         # remove accents
 121                         $search = unac_string($CHARSET,$search);
 122                         while ($search =~ s/\s*("[^"]+")\s*/ /) {
 123                                 $s .= "$1 ";
 124                         }
 125                         $search =~ s/^\s+//;
 126                         $search =~ s/\s+$//;
 127
 128                         foreach (split(/\s+/,$search)) {
 129                                 if (m/^([+-])(\S+)/) {
 130                                         $s.= ($s) ? "and " : "";
 131                                         $s.="not " if ($1 eq "-");
 132                                         $s.="$2$wc ";
 133                                 } elsif (m/^\s*(and|or|not)\s*$/i) {
 134                                         $s.="$_ ";
 135                                 # don't add * to words with less than x chars
 136                                 } elsif (length($_) <= $MIN_WILDCARD) {
 137                                         $s.="$_ ";
 138                                 } else {
 139                                         $s.="$_$wc ";
 140                                 }
 141                         }
 142                         $s =~ s/\*+/*/g;
 143                         push @swish_q,$s;
 144                 }
 145                 # FIXME default operator for multi-value fields is or. There is
 146                 # no way to change it, except here for now. Is there need?
 147                 push @s_arr, $q->param("f$i")."_swish=(".join(" or ",@swish_q).")";
 148         }
 149
 150         my $tmpl = $self->load_tmpl('results.html', global_vars => 1);
 151
 152         sub esc_html {
 153                 my $html = shift;
 154                 $html =~ s/</&lt;/g;
 155                 $html =~ s/>/&gt;/g;
 156                 return $html;
 157         }
 158
 159         # call swish
 160         my $sh = SWISH->connect('Fork',
 161                 prog     => $SWISH,
 162                 indexes  => $INDEX,
 163                 properties  => [qw/swishdocpath swishrank swishtitle headline html/],
 164                 results  => sub {
 165                         my ($sh,$hit) = @_;
 166
 167                         push @swish_results, {
 168                                 nr => ($#swish_results + 2),
 169                                 path => $hit->swishdocpath,
 170                                 headline => esc_html($from_utf8->convert($hit->headline)),
 171                                 html => back2html($from_utf8->convert($hit->html)),
 172                                 rank => $hit->swishrank };
 173
 174                 },
 175                 #startnum => 0,
 176                 maxhits => $MAX_HITS
 177         );
 178
 179         die $SWISH::errstr unless $sh;
 180         # construct swish query
 181         my $sw_q = join(" and ",@s_arr);
 182         if (@path_arr && $q->param('show_full')) {
 183                 $sw_q .= "and (swishdocpath=\"";
 184                 $sw_q .= join("\" or swishdocpath=\"",@path_arr);
 185                 $sw_q .= "\")";
 186                 $tmpl->param('full',1); # show full records
 187         } elsif ($q->param('show_full')) {
 188                 # just show full path, no path defined
 189                 $tmpl->param('full',1);
 190         } else {
 191                 $tmpl->param('full',0);
 192         }
 193
 194         my $hits = $sh->query($sw_q);
 195
 196         $tmpl->param('hits',$hits);
 197         $tmpl->param('search',$sw_q);
 198
 199         $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
 200         $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
 201
 202         $tmpl->param('url_params',"?".join("&",@url_params));
 203
 204         # create a Pager object
 205         my $pager = HTML::Pager->new(
 206                 # required parameters
 207                 query => $q,
 208                 get_data_callback => sub {
 209                         my ($offset, $rows) = @_;
 210
 211                         my @result;
 212                         for (my $i=0; $i<$rows; $i++) {
 213                                 my $r = $swish_results[$offset+$i];
 214                                 if ($r && $tmpl->param('full')) {
 215                                         push @result, $r;
 216                                 } elsif ($r) {
 217                                         # if not full output, skip html
 218                                         delete $r->{html};
 219                                         push @result, $r;
 220                                 }
 221                         }
 222                         return \@result;
 223                 },
 224                 rows => $hits,
 225                 page_size => $ON_PAGE,
 226                 # some optional parameters
 227                 persist_vars => [ @persist_vars ],
 228                 #cell_space_color => '#000000',
 229                 #cell_background_color => '#ffffff',
 230                 #nav_background_color => '#dddddd',
 231                 #javascript_presubmit => 'last_minute_javascript()',
 232                 debug => 1,
 233                 template => $tmpl,
 234         );
 235
 236         my $html = $pager->output;
 237
 238         return in_template($html);
 239 }
 240
 241 sub show_index {
 242         my $self = shift;
 243         my $i = shift;          # field number
 244
 245         my $q = $self->query();
 246
 247         my $field = $q->param("f$i");
 248         my $limit = $q->param("v$i");
 249
 250         my $html;
 251
 252         my $index = new index_DBI(
 253                 $cfg_global->val('global', 'dbi_dbd'),
 254                 $cfg_global->val('global', 'dbi_dsn'),
 255                 $cfg_global->val('global', 'dbi_user'),
 256                 $cfg_global->val('global', 'dbi_passwd') || ''
 257         );
 258
 259         my $total = $index->count($field,$limit);
 260         if (! $total) {
 261                 my $tmpl = $self->load_tmpl('no_index.html');
 262                 $tmpl->param('field',$field);
 263                 $html = $tmpl->output;
 264                 return $html;
 265         }
 266
 267         my $tmpl = $self->load_tmpl('index_res.html', global_vars => 1);
 268         $tmpl->param('field',$field);
 269         $tmpl->param('limit',$limit);
 270         $tmpl->param('total',$total);
 271
 272 # FIXME I should set offset and leave out limit from fetch!!
 273 #       if (! $q->param("PAGER_offset") {
 274 #               $q->param("Pager_offet)
 275 #       }
 276
 277         my $pager = HTML::Pager->new(
 278                 query => $q,
 279                 get_data_callback => sub {
 280                         my ($offset, $rows) = @_;
 281
 282                         my @result = $index->fetch($field,$limit, $offset, $rows);
 283                         return \@result;
 284                 },
 285                 rows => $total,
 286                 page_size => $ON_PAGE,
 287                 persist_vars => [
 288                         'rm',
 289                         "f$i", "v$i", "f".$i."_index",
 290                         'offset',
 291                         ],
 292                 debug => 1,
 293                 template => $tmpl,
 294         );
 295
 296         return in_template($pager->output);
 297 }
 298
 299 1;