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

   1 package WebPac;
   2
   3 use base 'CGI::Application';
   4 use strict;
   5
   6 use HTML::FillInForm;
   7 use SWISH::API;
   8 use Text::Iconv;
   9 use DBI;
  10 use Config::IniFiles;
  11 use Text::Unaccent;
  12 use Data::Pageset;
  13
  14 use lib '..';
  15 use index_DBI_filter;
  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', 'my_unac_filter');
  32 my $BASE_PATH =$cfg_global->val('webpac', 'base_path');
  33 # for pager
  34 my $pages_per_set = $cfg_global->val('webpac', 'pages_per_set') || 10;
  35
  36 Text::Iconv->raise_error(0);     # Conversion errors raise exceptions
  37
  38 my $from_utf8 = Text::Iconv->new('UTF8', $CHARSET);
  39
  40 if ($UNAC_FILTER) {
  41         require $UNAC_FILTER;
  42 } else {
  43         sub WebPac::my_unac_string {
  44                 my ($charset, $string) = (@_);
  45                 return $string;
  46         }
  47 }
  48
  49 # use path from cgi script to support templates in subdirs
  50 sub url_ex {
  51         my $q = shift || die "suff2file needs CGI object!";
  52         my $tpl = shift || die "url_ex needs template name!";
  53         return suff2file($BASE_PATH, $q->url(-absolute => 1,-path => 1),$TEMPLATE_PATH,$tpl);
  54 }
  55
  56 sub suff2file($$$$) {
  57         my ($base_path, $p, $path, $tpl) = @_;
  58
  59         return $tpl if (! $base_path);
  60
  61         # strip everything to and including base path, leaving only
  62         # additional (virtual) path
  63         if ($base_path eq "/") {
  64                 $p =~ s,/*,,g;
  65                 my ($name,$ext) = split(/\./,$tpl);
  66                 $p = $name . "-" . $p . "." . $ext;
  67         } elsif ($p =~ s,^.*?$base_path,,) {
  68                 $p =~ s,/*,,g;
  69                 my ($name,$ext) = split(/\./,$tpl);
  70                 $p = $name . $p . "." . $ext;
  71         } else {
  72                 # if unable reset it!
  73                 $p = $tpl;
  74         }
  75
  76         if ( -e "$path/$p") {
  77                 return $p;
  78         } else {
  79                 return $tpl;
  80         }
  81
  82 }
  83
  84 sub setup {
  85         my $self = shift;
  86         $self->tmpl_path($TEMPLATE_PATH);
  87         $self->run_modes(
  88                 'search' => 'show_search_form',
  89                 'results' => 'show_results_list',
  90 #               'user' => 'show_user_detail',
  91                 'index' => 'show_index',
  92         );
  93         $self->start_mode('search');
  94         $self->mode_param('rm');
  95
  96         $self->header_props(-charset=>$CHARSET);
  97 }
  98
  99 sub in_template {
 100         my $q = shift || die "need CGI object!";
 101         my $html = shift || die "This page is left unintentionally blank";
 102         return $html if (! defined($TEMPLATE));
 103
 104         my ($dir,$tpl);
 105         if ($TEMPLATE =~ m,^(.*?/*)([^/]+)$,) {
 106                 ($dir,$tpl) = ($1,$2);
 107         } else {
 108                 die "can't parse TEMPLATE path";
 109         }
 110
 111         my $master_tpl = suff2file($BASE_PATH, $q->url(-absolute => 1, -path => 1),$dir,$tpl);
 112         if (open(T, $master_tpl)) {
 113                 my $template_html = join("\n",<T>);
 114                 close(T);
 115                 $template_html =~ s/##webpac##/$html/gsi;
 116                 return $template_html;
 117         } else {
 118                 return "Can't read template '$master_tpl'";
 119         }
 120 }
 121
 122 #--------------------------------------------------------------------------
 123
 124 #
 125 # make pager navigation and fill template variables
 126 # compatibile with HTML::Pager
 127 #
 128
 129 sub make_pager($$$) {
 130         my ($q,$tmpl,$pager) = @_;
 131
 132         #
 133         # pager navigation
 134         #
 135         my ($pager_prev,$pager_next, $pager_jump) = ('','','');
 136
 137         my $nav_fmt=qq{ <a href="%s">%s</a> };
 138
 139         if ($pager->current_page() > $pager->first_page) {
 140                 $q->param('PAGER_offset', $pager->current_page - 1);
 141                 $pager_prev .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'&lt;&lt;');
 142         }
 143
 144         if ($pager->previous_set) {
 145                 $q->param('PAGER_offset', $pager->previous_set);
 146                 $pager_prev .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'..');
 147         }
 148
 149
 150         foreach my $p (@{$pager->pages_in_set()}) {
 151                 next if ($p <= 0);
 152                 if($p == $pager->current_page()) {
 153                         $pager_jump .= "<b>$p</b> ";
 154                 } else {
 155                         $q->param('PAGER_offset', $p);
 156                         $pager_jump .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),$p);
 157                 }
 158         }
 159
 160         if ($pager->next_set) {
 161                 $q->param('PAGER_offset', $pager->next_set);
 162                 $pager_next .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'..');
 163         }
 164
 165         if ($pager->current_page() < $pager->last_page) {
 166                 $q->param('PAGER_offset', $pager->current_page + 1);
 167                 $pager_next .= sprintf($nav_fmt,$q->url(-relative=>1, -query=>1),'&gt;&gt;');
 168         }
 169
 170         $tmpl->param('PAGER_PREV', $pager_prev);
 171         $tmpl->param('PAGER_JUMP', $pager_jump);
 172         $tmpl->param('PAGER_NEXT', $pager_next);
 173
 174 }
 175
 176 #
 177 # put persisten variables in template
 178 #
 179
 180 sub make_pager_vars {
 181         my $q = shift @_;
 182         my $tmpl = shift @_;
 183         my @persist_vars = @_;
 184         my $hidden_vars = '';
 185         my $hidden_search = '';
 186         foreach my $v (@persist_vars) {
 187                 foreach my $val ($q->param($v)) {
 188                         next if (! $val || $val eq '');
 189                         $val =~ s/"/&quot;/g;
 190                         $hidden_vars .= '<input type="hidden" name="'.$v.'" value="'.$val.'"/>'."\n";
 191                         $hidden_search .= '<input type="hidden" name="'.$v.'" value="'.$val.'"/>'."\n" if ($v ne "rm");
 192                 }
 193         }
 194
 195         $tmpl->param('PAGER_HIDDEN', $hidden_vars);
 196         $tmpl->param('SEARCH_HIDDEN', $hidden_search);
 197         $tmpl->param('PAGER_JAVASCRIPT', qq#
 198 <SCRIPT LANGUAGE="Javascript">
 199 <!-- Begin
 200         // dummy emulator for HTML::Pager templates
 201         function PAGER_set_offset_and_submit() {
 202                 return true;
 203         }
 204 // End -->
 205 </script>
 206         #);
 207 }
 208
 209 #--------------------------------------------------------------------------
 210
 211 sub show_search_form {
 212         my $self = shift;
 213
 214         # Get the CGI.pm query object
 215         my $q = $self->query();
 216
 217         my $tmpl = $self->load_tmpl(url_ex($q,'search.html'));
 218         my $html = $tmpl->output;
 219
 220         my $fif = new HTML::FillInForm;
 221
 222         return in_template($q,$fif->fill(scalarref => \$html, fobject => $q,
 223                 target => 'search'));
 224 }
 225
 226 sub show_results_list {
 227         my $self = shift;
 228
 229         my $q = $self->query();
 230
 231         # submit was reset?
 232         if ($q->param('reset')) {
 233                 $q->delete_all;
 234                 return $self->show_search_form();
 235         }
 236
 237         # load template for this page
 238
 239         my @s_arr;      # all queries are located here
 240
 241         my @path_arr = $q->param('path');
 242         my $full = $q->param('full');
 243
 244         my @persist_vars = ( 'rm', 'persist_search' );
 245         my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.($q->param('PAGER_offset') || 0) );
 246
 247         my @persist_search_vars;
 248         my @url_params_persist;
 249         if ($q->param("persist_search")) {
 250                 @persist_search_vars = split(/\s*,\s*/, $q->param("persist_search"));
 251                 push @url_params_persist, "persist_search=".$q->url_param("persist_search");
 252                 push @url_params,"persist_search=".$q->url_param("persist_search");
 253         }
 254
 255         # support parametars "f" and "v" for start
 256         for(my $i = 0; $i <=30; $i++) {
 257
 258                 $i = '' if ($i == 0);
 259
 260                 return show_index($self, $i) if ($q->param("f".$i."_index"));
 261
 262                 next if (! $q->param("v$i") || $q->param("v$i") eq '');
 263                 next if (! $q->param("f$i"));
 264
 265                 my $persist = grep(/^$i$/,@persist_search_vars);
 266
 267                 push @persist_vars, "f$i";
 268                 push @persist_vars, "v$i";
 269                 push @persist_vars, "e$i" if ($q->param("e$i"));
 270
 271                 # create url parametars (and persistent ones)
 272
 273                 push @url_params,"f$i=".$q->url_param("f$i");
 274                 push @url_params_persist,"f$i=".$q->url_param("f$i") if ($persist);
 275
 276                 foreach my $v ($q->url_param("v$i")) {
 277                         # escape quotes so that phrase search work
 278                         $v =~ s/"/%22/g;
 279                         push @url_params,"v$i=$v";
 280                         push @url_params_persist,"v$i=$v" if ($persist);
 281                 }
 282
 283                 if ($q->param("e$i")) {
 284                         push @url_params,"e$i=".$q->url_param("e$i");
 285 #                       push @url_params_persist,"e$i=".$q->url_param("e$i");
 286                 }
 287
 288                 my $wc="*";     # swish wildcard
 289                 $wc="" if ($i eq "");   # don't apply wildcard on field 0
 290
 291                 # re-write query from +/- to and/and not
 292                 my @param_vals = $q->param("v$i");
 293                 my @swish_q;
 294                 my ($pre,$post,$exact) = ('','','');
 295                 while (my $search = shift @param_vals) {
 296                         my $s;
 297                         # remove accents
 298                         $search = my_unac_string($CHARSET,$search);
 299                         while ($search =~ s/\s*("[^"]+")\s*/ /) {
 300                                 $s .= "$1 ";
 301                         }
 302                         $search =~ s/^\s+//;
 303                         $search =~ s/\s+$//;
 304
 305                         # filed e[nr] is exact match bitmask
 306                         # 1 = beginning, 2=end, 3=both
 307                         my $exact_flag = $q->param("e$i") || 0;
 308                         $pre = '"xxbxx ' if ($exact_flag & 1);
 309                         $post = ' xxexx"' if ($exact_flag & 2);
 310                         # add qotes on other side
 311                         if ($q->param("e$i")) {
 312                                 $pre = '"' if (! $pre);
 313                                 $post = '"' if (! $post);
 314                                 # what about wildcards?
 315                                 $wc = '';
 316                                 $wc = '*' if ($q->param("e$i") & 4);
 317                                 $exact = '_exact';
 318                         }
 319
 320                         foreach (split(/\s+/,$search)) {
 321                                 if (m/^([+-])(\S+)/) {
 322                                         $s.= ($s) ? "and " : "";
 323                                         $s.="not " if ($1 eq "-");
 324                                         $s.=$2.$wc." ";
 325                                 } elsif (m/^\s*(and|or|not)\s*$/i) {
 326                                         $s.=$_." ";
 327                                 # don't add * to words with less than x chars
 328                                 } elsif (length($_) <= $MIN_WILDCARD) {
 329                                         $s.=$_." ";
 330                                 } else {
 331                                         $s.=$_.$wc." ";
 332                                 }
 333                         }
 334                         $s =~ s/\*+/*/g;
 335                         $s = $pre.$s.$post if ($q->param("e$i"));
 336                         push @swish_q,$s;
 337                 }
 338                 # FIXME default operator for multi-value fields is or. There is
 339                 # no way to change it, except here for now. Is there need?
 340                 push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")";
 341         }
 342
 343         my $tmpl = $self->load_tmpl(url_ex($q,'results.html'), global_vars => 1, die_on_bad_params => 0);
 344
 345         $tmpl->param('url_params',"?".join("&",@url_params));
 346
 347         sub esc_html {
 348                 my $html = shift;
 349                 $html =~ s/</&lt;/g;
 350                 $html =~ s/>/&gt;/g;
 351                 return $html;
 352         }
 353
 354         my $sort = 'swishrank';
 355         if ($q->param("sort")) {
 356                 $sort = 'headline';
 357                 push @persist_vars, "sort";
 358         }
 359
 360         my $sortby = $q->param("sortby");
 361         if ($sortby) {
 362                 $sort = $sortby;
 363                 push @persist_vars, "sortby";
 364         }
 365         $tmpl->param('url_params_paths',"?".join("&",@url_params).'&'.join("&",map { my $t = $_; $t =~ s/\#/%23/g; "path=$t"; } @path_arr));
 366
 367         # construct swish query
 368         my $sw_q = join(" and ",@s_arr);
 369         if (@path_arr && $q->param('show_full')) {
 370                 $sw_q .= "and (swishdocpath=\"";
 371                 $sw_q .= join("\" or swishdocpath=\"",@path_arr);
 372                 $sw_q .= "\")";
 373                 $tmpl->param('full',1); # show full records
 374         } elsif ($q->param('show_full')) {
 375                 # just show full path, no path defined
 376                 $tmpl->param('full',1);
 377         } else {
 378                 $tmpl->param('full',0);
 379         }
 380
 381         my $swish_msg = ' ';
 382
 383         # create new swish instance
 384         my $swish = SWISH::API->new($INDEX);
 385         $swish_msg .= $swish->ErrorString." ".$swish->LastErrorMsg if $swish->Error;
 386
 387         # execute query and get number of results from SWISH-E
 388         my $search = $swish->New_Search_Object;
 389
 390         $search->SetSort($sort);
 391
 392         my $results = $search->Execute($sw_q);
 393         $swish_msg .= $swish->ErrorString." ".$swish->LastErrorMsg if $swish->Error;
 394
 395         my $hits = $results->Hits;
 396
 397         $tmpl->param('hits',$hits);
 398         my $search_msg = $sw_q;
 399         $search_msg .= '<em>'.$swish_msg.'</em>' if ($swish_msg);
 400         $tmpl->param('search', $search_msg);
 401
 402         $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
 403         $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
 404
 405         #
 406         # build pager
 407         #
 408
 409         my $current_page = $q->param('PAGER_offset') || 1;
 410
 411         my $pager = Data::Pageset->new({
 412                 'total_entries' => $hits,
 413                 'entries_per_page' => $ON_PAGE,
 414                 'current_page' => $current_page,
 415                 'pages_per_set' => $pages_per_set,
 416         });
 417
 418         $results->SeekResult( $pager->first - 1 );
 419
 420         # get number of entries on this page
 421         my $i = $pager->entries_on_this_page;
 422
 423         # results from swish for template
 424         my @pager_data_list;
 425
 426         for(my $i=$pager->first; $i<=$pager->last; $i++) {
 427
 428                 my $result = $results->NextResult;
 429                 last if (! $result);
 430
 431                 my $r = {
 432                         nr => $i,
 433                         path => $result->Property('swishdocpath'),
 434                         headline => esc_html($from_utf8->convert($result->Property('headline'))),
 435                         rank => $result->Property('swishrank')
 436                 };
 437
 438                 $r->{html} = back2html($from_utf8->convert($result->Property('html')), join("&",@url_params_persist)) if ($q->param('show_full'));
 439
 440                 push @pager_data_list, $r;
 441         }
 442
 443
 444
 445         # put something in template
 446         make_pager($q, $tmpl, $pager);
 447         make_pager_vars($q, $tmpl, @persist_vars);
 448         $tmpl->param('PAGER_DATA_LIST', \@pager_data_list);
 449
 450         my $html = $tmpl->output;
 451
 452         return in_template($q,$html);
 453 }
 454
 455 sub show_index {
 456         my $self = shift;
 457         my $i = shift;          # field number
 458
 459         my $q = $self->query();
 460
 461         my $field = $q->param("f$i");
 462         my $limit = $q->param("v$i");
 463
 464         my $filter = $q->param("filter");
 465
 466         my $html;
 467
 468         my $index = new index_DBI(
 469                 $cfg_global->val('global', 'dbi_dbd'),
 470                 $cfg_global->val('global', 'dbi_dsn'),
 471                 $cfg_global->val('global', 'dbi_user'),
 472                 $cfg_global->val('global', 'dbi_passwd') || ''
 473         );
 474
 475         my $total = $index->count($field,$limit,$filter);
 476
 477         if (! defined($total)) {
 478                 my $tmpl = $self->load_tmpl(url_ex($q,'no_index.html'));
 479                 $tmpl->param('field',$field);
 480                 $html = $tmpl->output;
 481                 return $html;
 482         }
 483
 484         my $tmpl = $self->load_tmpl(url_ex($q,'index_res.html'), global_vars => 1, die_on_bad_params => 0);
 485         $tmpl->param('field',$field);
 486         $tmpl->param('limit',$limit);
 487         $tmpl->param('total',$total);
 488         $tmpl->param('filter',$filter);
 489
 490 # FIXME I should set offset and leave out limit from fetch!!
 491 #       if (! $q->param("PAGER_offset") {
 492 #               $q->param("Pager_offet)
 493 #       }
 494
 495
 496         #
 497         # build pager
 498         #
 499         my $pager = Data::Pageset->new({
 500                 'total_entries' => $total,
 501                 'entries_per_page' => $ON_PAGE,
 502                 'current_page' => $q->param('PAGER_offset') || 1,
 503                 'pages_per_set' => $pages_per_set
 504         });
 505
 506         my @persist_vars = qw{rm f$i v$i f$i_index offset};
 507
 508         make_pager($q, $tmpl, $pager);
 509         make_pager_vars($q, $tmpl, @persist_vars);
 510
 511         my @pager_data_list = $index->fetch($field,$limit, $pager->first - 1, $pager->entries_on_this_page, $filter);
 512         $tmpl->param('PAGER_DATA_LIST', \@pager_data_list);
 513
 514         return in_template($q,$tmpl->output);
 515 }
 516
 517 1;