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                         $hidden_vars .= '<input type="hidden" name="'.$v.'" value="'.$val.'"/>'."\n";
 190                         $hidden_search .= '<input type="hidden" name="'.$v.'" value="'.$val.'"/>'."\n" if ($v ne "rm");
 191                 }
 192         }
 193
 194         $tmpl->param('PAGER_HIDDEN', $hidden_vars);
 195         $tmpl->param('SEARCH_HIDDEN', $hidden_search);
 196         $tmpl->param('PAGER_JAVASCRIPT', qq#
 197 <SCRIPT LANGUAGE="Javascript">
 198 <!-- Begin
 199         // dummy emulator for HTML::Pager templates
 200         function PAGER_set_offset_and_submit() {
 201                 return true;
 202         }
 203 // End -->
 204 </script>
 205         #);
 206 }
 207
 208 #--------------------------------------------------------------------------
 209
 210 sub show_search_form {
 211         my $self = shift;
 212
 213         # Get the CGI.pm query object
 214         my $q = $self->query();
 215
 216         my $tmpl = $self->load_tmpl(url_ex($q,'search.html'));
 217         my $html = $tmpl->output;
 218
 219         my $fif = new HTML::FillInForm;
 220
 221         return in_template($q,$fif->fill(scalarref => \$html, fobject => $q,
 222                 target => 'search'));
 223 }
 224
 225 sub show_results_list {
 226         my $self = shift;
 227
 228         my $q = $self->query();
 229
 230         # submit was reset?
 231         if ($q->param('reset')) {
 232                 $q->delete_all;
 233                 return $self->show_search_form();
 234         }
 235
 236         # load template for this page
 237
 238         my @s_arr;      # all queries are located here
 239
 240         my @path_arr = $q->param('path');
 241         my $full = $q->param('full');
 242
 243         my @persist_vars = ( 'rm', 'persist_search' );
 244         my @url_params = ( 'rm=results', 'show_full=1', 'last_PAGER_offset='.($q->param('PAGER_offset') || 0) );
 245
 246         my @persist_search_vars;
 247         my @url_params_persist;
 248         if ($q->param("persist_search")) {
 249                 @persist_search_vars = split(/\s*,\s*/, $q->param("persist_search"));
 250                 push @url_params_persist, "persist_search=".$q->url_param("persist_search");
 251                 push @url_params,"persist_search=".$q->url_param("persist_search");
 252         }
 253
 254         # support parametars "f" and "v" for start
 255         for(my $i = 0; $i <=30; $i++) {
 256
 257                 $i = '' if ($i == 0);
 258
 259                 return show_index($self, $i) if ($q->param("f".$i."_index"));
 260
 261                 next if (! $q->param("v$i") || $q->param("v$i") eq '');
 262                 next if (! $q->param("f$i"));
 263
 264                 my $persist = grep(/^$i$/,@persist_search_vars);
 265
 266                 push @persist_vars, "f$i";
 267                 push @persist_vars, "v$i";
 268                 push @persist_vars, "e$i" if ($q->param("e$i"));
 269
 270                 # create url parametars (and persistent ones)
 271
 272                 push @url_params,"f$i=".$q->url_param("f$i");
 273                 push @url_params_persist,"f$i=".$q->url_param("f$i") if ($persist);
 274
 275                 foreach my $v ($q->url_param("v$i")) {
 276                         # escape quotes so that phrase search work
 277                         $v =~ s/"/%22/g;
 278                         push @url_params,"v$i=$v";
 279                         push @url_params_persist,"v$i=$v" if ($persist);
 280                 }
 281
 282                 if ($q->param("e$i")) {
 283                         push @url_params,"e$i=".$q->url_param("e$i");
 284 #                       push @url_params_persist,"e$i=".$q->url_param("e$i");
 285                 }
 286
 287                 my $wc="*";     # swish wildcard
 288                 $wc="" if ($i eq "");   # don't apply wildcard on field 0
 289
 290                 # re-write query from +/- to and/and not
 291                 my @param_vals = $q->param("v$i");
 292                 my @swish_q;
 293                 my ($pre,$post,$exact) = ('','','');
 294                 while (my $search = shift @param_vals) {
 295                         my $s;
 296                         # remove accents
 297                         $search = my_unac_string($CHARSET,$search);
 298                         while ($search =~ s/\s*("[^"]+")\s*/ /) {
 299                                 $s .= "$1 ";
 300                         }
 301                         $search =~ s/^\s+//;
 302                         $search =~ s/\s+$//;
 303
 304                         # filed e[nr] is exact match bitmask
 305                         # 1 = beginning, 2=end, 3=both
 306                         my $exact_flag = $q->param("e$i") || 0;
 307                         $pre = '"xxbxx ' if ($exact_flag & 1);
 308                         $post = ' xxexx"' if ($exact_flag & 2);
 309                         # add qotes on other side
 310                         if ($q->param("e$i")) {
 311                                 $pre = '"' if (! $pre);
 312                                 $post = '"' if (! $post);
 313                                 # what about wildcards?
 314                                 $wc = '';
 315                                 $wc = '*' if ($q->param("e$i") & 4);
 316                                 $exact = '_exact';
 317                         }
 318
 319                         foreach (split(/\s+/,$search)) {
 320                                 if (m/^([+-])(\S+)/) {
 321                                         $s.= ($s) ? "and " : "";
 322                                         $s.="not " if ($1 eq "-");
 323                                         $s.=$2.$wc." ";
 324                                 } elsif (m/^\s*(and|or|not)\s*$/i) {
 325                                         $s.=$_." ";
 326                                 # don't add * to words with less than x chars
 327                                 } elsif (length($_) <= $MIN_WILDCARD) {
 328                                         $s.=$_." ";
 329                                 } else {
 330                                         $s.=$_.$wc." ";
 331                                 }
 332                         }
 333                         $s =~ s/\*+/*/g;
 334                         $s = $pre.$s.$post if ($q->param("e$i"));
 335                         push @swish_q,$s;
 336                 }
 337                 # FIXME default operator for multi-value fields is or. There is
 338                 # no way to change it, except here for now. Is there need?
 339                 push @s_arr, $q->param("f$i")."_swish".$exact."=(".join(" or ",@swish_q).")";
 340         }
 341
 342         my $tmpl = $self->load_tmpl(url_ex($q,'results.html'), global_vars => 1, die_on_bad_params => 0);
 343
 344         $tmpl->param('url_params',"?".join("&",@url_params));
 345
 346         sub esc_html {
 347                 my $html = shift;
 348                 $html =~ s/</&lt;/g;
 349                 $html =~ s/>/&gt;/g;
 350                 return $html;
 351         }
 352
 353         my $sort = 'swishrank';
 354         if ($q->param("sort")) {
 355                 $sort = 'headline';
 356                 push @persist_vars, "sort";
 357         }
 358
 359         my $sortby = $q->param("sortby");
 360         if ($sortby) {
 361                 $sort = $sortby;
 362                 push @persist_vars, "sortby";
 363         }
 364         $tmpl->param('url_params_paths',"?".join("&",@url_params).'&'.join("&",map { my $t = $_; $t =~ s/\#/%23/g; "path=$t"; } @path_arr));
 365
 366         # construct swish query
 367         my $sw_q = join(" and ",@s_arr);
 368         if (@path_arr && $q->param('show_full')) {
 369                 $sw_q .= "and (swishdocpath=\"";
 370                 $sw_q .= join("\" or swishdocpath=\"",@path_arr);
 371                 $sw_q .= "\")";
 372                 $tmpl->param('full',1); # show full records
 373         } elsif ($q->param('show_full')) {
 374                 # just show full path, no path defined
 375                 $tmpl->param('full',1);
 376         } else {
 377                 $tmpl->param('full',0);
 378         }
 379
 380         my $swish_msg = ' ';
 381
 382         # create new swish instance
 383         my $swish = SWISH::API->new($INDEX);
 384         $swish_msg .= $swish->ErrorString." ".$swish->LastErrorMsg if $swish->Error;
 385
 386         # execute query and get number of results from SWISH-E
 387         my $search = $swish->New_Search_Object;
 388
 389         $search->SetSort($sort);
 390
 391         my $results = $search->Execute($sw_q);
 392         $swish_msg .= $swish->ErrorString." ".$swish->LastErrorMsg if $swish->Error;
 393
 394         my $hits = $results->Hits;
 395
 396         $tmpl->param('hits',$hits);
 397         my $search_msg = $sw_q;
 398         $search_msg .= '<em>'.$swish_msg.'</em>' if ($swish_msg);
 399         $tmpl->param('search', $search_msg);
 400
 401         $tmpl->param('PAGER_offset',$q->param("PAGER_offset") || 0);
 402         $tmpl->param('last_PAGER_offset',$q->param("last_PAGER_offset") || 0);
 403
 404         #
 405         # build pager
 406         #
 407
 408         my $current_page = $q->param('PAGER_offset') || 1;
 409
 410         my $pager = Data::Pageset->new({
 411                 'total_entries' => $hits,
 412                 'entries_per_page' => $ON_PAGE,
 413                 'current_page' => $current_page,
 414                 'pages_per_set' => $pages_per_set,
 415         });
 416
 417         $results->SeekResult( $pager->first - 1 );
 418
 419         # get number of entries on this page
 420         my $i = $pager->entries_on_this_page;
 421
 422         # results from swish for template
 423         my @pager_data_list;
 424
 425         for(my $i=$pager->first; $i<=$pager->last; $i++) {
 426
 427                 my $result = $results->NextResult;
 428                 last if (! $result);
 429
 430                 my $r = {
 431                         nr => $i,
 432                         path => $result->Property('swishdocpath'),
 433                         headline => esc_html($from_utf8->convert($result->Property('headline'))),
 434                         rank => $result->Property('swishrank')
 435                 };
 436
 437                 $r->{html} = back2html($from_utf8->convert($result->Property('html')), join("&",@url_params_persist)) if ($q->param('show_full'));
 438
 439                 push @pager_data_list, $r;
 440         }
 441
 442
 443
 444         # put something in template
 445         make_pager($q, $tmpl, $pager);
 446         make_pager_vars($q, $tmpl, @persist_vars);
 447         $tmpl->param('PAGER_DATA_LIST', \@pager_data_list);
 448
 449         my $html = $tmpl->output;
 450
 451         return in_template($q,$html);
 452 }
 453
 454 sub show_index {
 455         my $self = shift;
 456         my $i = shift;          # field number
 457
 458         my $q = $self->query();
 459
 460         my $field = $q->param("f$i");
 461         my $limit = $q->param("v$i");
 462
 463         my $filter = $q->param("filter");
 464
 465         my $html;
 466
 467         my $index = new index_DBI(
 468                 $cfg_global->val('global', 'dbi_dbd'),
 469                 $cfg_global->val('global', 'dbi_dsn'),
 470                 $cfg_global->val('global', 'dbi_user'),
 471                 $cfg_global->val('global', 'dbi_passwd') || ''
 472         );
 473
 474         my $total = $index->count($field,$limit,$filter);
 475
 476         if (! defined($total)) {
 477                 my $tmpl = $self->load_tmpl(url_ex($q,'no_index.html'));
 478                 $tmpl->param('field',$field);
 479                 $html = $tmpl->output;
 480                 return $html;
 481         }
 482
 483         my $tmpl = $self->load_tmpl(url_ex($q,'index_res.html'), global_vars => 1, die_on_bad_params => 0);
 484         $tmpl->param('field',$field);
 485         $tmpl->param('limit',$limit);
 486         $tmpl->param('total',$total);
 487
 488 # FIXME I should set offset and leave out limit from fetch!!
 489 #       if (! $q->param("PAGER_offset") {
 490 #               $q->param("Pager_offet)
 491 #       }
 492
 493
 494         #
 495         # build pager
 496         #
 497         my $pager = Data::Pageset->new({
 498                 'total_entries' => $total,
 499                 'entries_per_page' => $ON_PAGE,
 500                 'current_page' => $q->param('PAGER_offset') || 1,
 501                 'pages_per_set' => $pages_per_set
 502         });
 503
 504         my @persist_vars = qw{rm f$i v$i f$i_index offset};
 505
 506         make_pager($q, $tmpl, $pager);
 507         make_pager_vars($q, $tmpl, @persist_vars);
 508
 509         my @pager_data_list = $index->fetch($field,$limit, $pager->first - 1, $pager->entries_on_this_page, $filter);
 510         $tmpl->param('PAGER_DATA_LIST', \@pager_data_list);
 511
 512         return in_template($q,$tmpl->output);
 513 }
 514
 515 1;