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 my $UNAC_FILTER =$cfg_global->val('global', 'unac_filter');
  32
  33 if ($UNAC_FILTER) {
  34         require $UNAC_FILTER;
  35 }
  36
  37 Text::Iconv->raise_error(0);     # Conversion errors raise exceptions
  38
  39 my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
  40
  41
  42 sub setup {
  43         my $self = shift;
  44         $self->tmpl_path($TEMPLATE_PATH);
  45         $self->run_modes(
  46                 'search' => 'show_search_form',
  47                 'results' => 'show_results_list',
  48 #               'user' => 'show_user_detail',
  49                 'index' => 'show_index',
  50         );
  51         $self->start_mode('search');
  52         $self->mode_param('rm');
  53
  54         $self->header_props(-charset=>$CHARSET);
  55 }
  56
  57 sub in_template {
  58         my $html = shift || "This page is left unintentionally blank";
  59         return $html if (! defined($TEMPLATE));
  60         if (open(T, $TEMPLATE)) {
  61                 my $template_html = join("\n",<T>);
  62                 close(T);
  63                 $template_html =~ s/##webpac##/$html/gsi;
  64                 return $template_html;
  65         } else {
  66                 return "Can't read template '$TEMPLATE'";
  67         }
  68 }
  69
  70 sub show_search_form {
  71         my $self = shift;
  72
  73         # Get the CGI.pm query object
  74         my $q = $self->query();
  75
  76         my $tmpl = $self->load_tmpl('search.html');
  77         my $html = $tmpl->output;
  78
  79         my $fif = new HTML::FillInForm;
  80
  81         return in_template($fif->fill(scalarref => \$html, fobject => $q,
  82                 target => 'search'));
  83 }
  84
  85 sub show_results_list {
  86         my $self = shift;
  87
  88         my $q = $self->query();
  89
  90         my @swish_results;      # results from swish
  91
  92         # load template for this page
  93
  94         my @s_arr;      # all queries are located here
  95
  96         my @path_arr = $q->param('path');
  97         my $full = $q->param('full');
  98
  99         my @persist_vars = ( 'rm' );
 100         my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.$q->param('PAGER_offset') || 0 );
 101
 102         # support parametars "f" and "v" for start
 103         for(my $i = ""; $i <=30; $i++) {
 104
 105                 return show_index($self, $i) if ($q->param("f".$i."_index"));
 106
 107                 next if (! $q->param("v$i"));
 108                 next if (! $q->param("f$i"));
 109
 110                 push @persist_vars, "f$i";
 111                 push @persist_vars, "v$i";
 112
 113                 push @url_params,"f$i=".$q->url_param("f$i");
 114                 foreach my $v ($q->url_param("v$i")) {
 115                         push @url_params,"v$i=$v";
 116                 }
 117
 118                 my $wc="*";     # swish wildcard
 119                 $wc="" if ($i eq "");   # don't apply wildcard on field 0
 120
 121                 # re-write query from +/- to and/and not
 122                 my @param_vals = $q->param("v$i");
 123                 my @swish_q;
 124                 my ($pre,$post,$exact) = ('','','');
 125                 while (my $search = shift @param_vals) {
 126                         my $s;
 127                         # remove accents
 128                         $search = unac_string($CHARSET,$search);
 129                         while ($search =~ s/\s*("[^"]+")\s*/ /) {
 130                                 $s .= "$1 ";
 131                         }
 132                         $search =~ s/^\s+//;
 133                         $search =~ s/\s+$//;
 134
 135                         # filed e[nr] is exact match bitmask
 136                         # 1 = beginning, 2=end, 3=both
 137                         $pre = '"xxbxx ' if ($q->param("e$i") & 1);
 138                         $post = ' xxexx"' if ($q->param("e$i") & 2);
 139                         # add qotes on other side
 140                         if ($q->param("e$i")) {
 141                                 $pre = '"' if (! $pre);
 142                                 $post = '"' if (! $post);
 143                                 $wc = '';       # don't use windcard in exact
 144                                 $exact = '_exact';
 145                         }
 146
 147                         foreach (split(/\s+/,$search)) {
 148                                 if (m/^([+-])(\S+)/) {
 149                                         $s.= ($s) ? "and " : "";
 150                                         $s.="not " if ($1 eq "-");
 151                                         $s.=$2.$wc." ";
 152                                 } elsif (m/^\s*(and|or|not)\s*$/i) {
 153                                         $s.=$_." ";
 154                                 # don't add * to words with less than x chars
 155                                 } elsif (length($_) <= $MIN_WILDCARD) {
 156                                         $s.=$_." ";
 157                                 } else {
 158                                         $s.=$_.$wc." ";
 159                                 }
 160                         }
 161                         $s =~ s/\*+/*/g;
 162                         $s = $pre.$s.$post if ($q->param("e$i"));
 163                         push @swish_q,$s;
 164                 }
 165                 # FIXME default operator for multi-value fields is or. There is
 166                 # no way to change it, except here for now. Is there need?
 167                 push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")";
 168         }
 169
 170         my $tmpl = $self->load_tmpl('results.html', global_vars => 1);
 171
 172         sub esc_html {
 173                 my $html = shift;
 174                 $html =~ s/</&lt;/g;
 175                 $html =~ s/>/&gt;/g;
 176                 return $html;
 177         }
 178
 179         # call swish
 180         my $sh = SWISH->connect('Fork',
 181                 prog     => $SWISH,
 182                 indexes  => $INDEX,
 183                 properties  => [qw/swishdocpath swishrank swishtitle headline html/],
 184                 results  => sub {
 185                         my ($sh,$hit) = @_;
 186
 187                         push @swish_results, {
 188                                 nr => ($#swish_results + 2),
 189                                 path => $hit->swishdocpath,
 190                                 headline => esc_html($from_utf8->convert($hit->headline)),
 191                                 html => back2html($from_utf8->convert($hit->html)),
 192                                 rank => $hit->swishrank };
 193
 194                 },
 195                 #startnum => 0,
 196                 maxhits => $MAX_HITS
 197         );
 198
 199         die $SWISH::errstr unless $sh;
 200         # construct swish query
 201         my $sw_q = join(" and ",@s_arr);
 202         if (@path_arr && $q->param('show_full')) {
 203                 $sw_q .= "and (swishdocpath=\"";
 204                 $sw_q .= join("\" or swishdocpath=\"",@path_arr);
 205                 $sw_q .= "\")";
 206                 $tmpl->param('full',1); # show full records
 207         } elsif ($q->param('show_full')) {
 208                 # just show full path, no path defined
 209                 $tmpl->param('full',1);
 210         } else {
 211                 $tmpl->param('full',0);
 212         }
 213
 214         my $hits = $sh->query($sw_q);
 215
 216         $tmpl->param('hits',$hits);
 217         $tmpl->param('search',$sw_q);
 218
 219         $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
 220         $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
 221
 222         $tmpl->param('url_params',"?".join("&",@url_params));
 223
 224         # create a Pager object
 225         my $pager = HTML::Pager->new(
 226                 # required parameters
 227                 query => $q,
 228                 get_data_callback => sub {
 229                         my ($offset, $rows) = @_;
 230
 231                         my @result;
 232                         for (my $i=0; $i<$rows; $i++) {
 233                                 my $r = $swish_results[$offset+$i];
 234                                 if ($r && $tmpl->param('full')) {
 235                                         push @result, $r;
 236                                 } elsif ($r) {
 237                                         # if not full output, skip html
 238                                         delete $r->{html};
 239                                         push @result, $r;
 240                                 }
 241                         }
 242                         return \@result;
 243                 },
 244                 rows => $hits,
 245                 page_size => $ON_PAGE,
 246                 # some optional parameters
 247                 persist_vars => [ @persist_vars ],
 248                 #cell_space_color => '#000000',
 249                 #cell_background_color => '#ffffff',
 250                 #nav_background_color => '#dddddd',
 251                 #javascript_presubmit => 'last_minute_javascript()',
 252                 debug => 1,
 253                 template => $tmpl,
 254         );
 255
 256         my $html = $pager->output;
 257
 258         return in_template($html);
 259 }
 260
 261 sub show_index {
 262         my $self = shift;
 263         my $i = shift;          # field number
 264
 265         my $q = $self->query();
 266
 267         my $field = $q->param("f$i");
 268         my $limit = $q->param("v$i");
 269
 270         my $html;
 271
 272         my $index = new index_DBI(
 273                 $cfg_global->val('global', 'dbi_dbd'),
 274                 $cfg_global->val('global', 'dbi_dsn'),
 275                 $cfg_global->val('global', 'dbi_user'),
 276                 $cfg_global->val('global', 'dbi_passwd') || ''
 277         );
 278
 279         my $total = $index->count($field,$limit);
 280         if (! $total) {
 281                 my $tmpl = $self->load_tmpl('no_index.html');
 282                 $tmpl->param('field',$field);
 283                 $html = $tmpl->output;
 284                 return $html;
 285         }
 286
 287         my $tmpl = $self->load_tmpl('index_res.html', global_vars => 1);
 288         $tmpl->param('field',$field);
 289         $tmpl->param('limit',$limit);
 290         $tmpl->param('total',$total);
 291
 292 # FIXME I should set offset and leave out limit from fetch!!
 293 #       if (! $q->param("PAGER_offset") {
 294 #               $q->param("Pager_offet)
 295 #       }
 296
 297         my $pager = HTML::Pager->new(
 298                 query => $q,
 299                 get_data_callback => sub {
 300                         my ($offset, $rows) = @_;
 301
 302                         my @result = $index->fetch($field,$limit, $offset, $rows);
 303                         return \@result;
 304                 },
 305                 rows => $total,
 306                 page_size => $ON_PAGE,
 307                 persist_vars => [
 308                         'rm',
 309                         "f$i", "v$i", "f".$i."_index",
 310                         'offset',
 311                         ],
 312                 debug => 1,
 313                 template => $tmpl,
 314         );
 315
 316         return in_template($pager->output);
 317 }
 318
 319 1;