includes/SearchEngine.php

   1 <?
   2 # See search.doc
   3
   4 class SearchEngine {
   5         /* private */ var $mUsertext, $mSearchterms;
   6         /* private */ var $mTitlecond, $mTextcond;
   7
   8         var $doSearchRedirects = true;
   9         var $addtoquery = array();
  10         var $namespacesToSearch = array();
  11         var $alternateTitle;
  12
  13         function SearchEngine( $text )
  14         {
  15                 # We display the query, so let's strip it for safety
  16                 #
  17                 global $wgDBmysql4;
  18                 $lc = SearchEngine::legalSearchChars() . "()";
  19                 if( $wgDBmysql4 ) $lc .= "\"~<>*+-";
  20                 $this->mUsertext = trim( preg_replace( "/[^{$lc}]/", " ", $text ) );
  21                 $this->mSearchterms = array();
  22         }
  23
  24         function queryNamespaces()
  25         {
  26                 $namespaces = implode( ",", $this->namespacesToSearch );
  27                 if ($namespaces == "") {
  28                         $namespaces = "0";
  29                 }
  30                 return "AND cur_namespace IN (" . $namespaces . ")";
  31                 #return "1";
  32         }
  33
  34         function searchRedirects()
  35         {
  36                 if ( $this->doSearchRedirects ) return "";
  37                 return "AND cur_is_redirect=0 ";
  38         }
  39
  40
  41
  42         /* private */ function initNamespaceCheckbox( $i )
  43         {
  44                 global $wgUser, $wgNamespacesToBeSearchedDefault;
  45
  46
  47                 if ($wgUser->getRights()) {
  48                         // User is logged in so we retrieve his default namespaces
  49                         return $wgUser->getOption( "searchNs".$i );
  50                 }
  51                 else {
  52                         // User is not logged in so we give him the global default namespaces
  53                         return $wgNamespacesToBeSearchedDefault[ $i ];
  54                 }
  55         }
  56
  57
  58
  59         function powersearch()
  60         {
  61                 global $wgUser, $wgOut, $wgLang, $wgTitle;
  62                 $nscb = array();
  63
  64                 $search                 = $_REQUEST['search'];
  65                 $searchx                = $_REQUEST['searchx'];
  66                 $listredirs             = $_REQUEST['redirs'];
  67
  68
  69                 if ( ! isset ( $searchx ) ) {   /* First time here */
  70                         $listredirs = 1;
  71                         for ($i = 0; ($i <= 7); $i++)
  72                         {
  73                                 $nscb[$i] = $this->initNamespaceCheckbox($i);
  74                         }
  75                 } else {
  76                         $nscb[0]                = $_REQUEST['ns0'];
  77                         $nscb[1]                = $_REQUEST['ns1'];
  78                         $nscb[2]                = $_REQUEST['ns2'];
  79                         $nscb[3]                = $_REQUEST['ns3'];
  80                         $nscb[4]                = $_REQUEST['ns4'];
  81                         $nscb[5]                = $_REQUEST['ns5'];
  82                         $nscb[6]                = $_REQUEST['ns6'];
  83                         $nscb[7]                = $_REQUEST['ns7'];
  84                 }
  85
  86                 $this->checkboxes["searchx"] = 1;
  87                 $ret = wfMsg("powersearchtext");
  88
  89                 # Determine namespace checkboxes
  90
  91                 $ns = $wgLang->getNamespaces();
  92                 array_shift( $ns ); /* Skip "Special" */
  93
  94                 $r1 = "";
  95                 for ( $i = 0; $i < count( $ns ); ++$i ) {
  96                         $checked = "";
  97                         if ( $nscb[$i] == 1 ) {
  98                                 $checked = " checked";
  99                                 $this->addtoquery["ns{$i}"] = 1;
 100                                 array_push( $this->namespacesToSearch, $i );
 101                         }
 102                         $name = str_replace( "_", " ", $ns[$i] );
 103                         if ( "" == $name ) { $name = wfMsg( "blanknamespace" ); }
 104
 105                         if ( 0 != $i ) { $r1 .= " "; }
 106                         $r1 .= "<input type=checkbox value=\"1\" name=\"" .
 107                           "ns{$i}\"{$checked}>{$name}\n";
 108                 }
 109                 $ret = str_replace ( "$1", $r1, $ret );
 110
 111                 # List redirects checkbox
 112
 113                 $checked = "";
 114                 if ( $listredirs == 1 ) {
 115                         $this->addtoquery["redirs"] = 1;
 116                         $checked = " checked";
 117                 }
 118                 $r2 = "<input type=checkbox value=1 name=\"redirs\"{$checked}>\n";
 119                 $ret = str_replace( "$2", $r2, $ret );
 120
 121                 # Search field
 122
 123                 $r3 = "<input type=text name=\"search\" value=\"" .
 124                         htmlspecialchars( $search ) ."\" width=80>\n";
 125         $ret = str_replace( "$3", $r3, $ret );
 126
 127                 # Searchx button
 128
 129                 $r9 = "<input type=submit name=\"searchx\" value=\"" .
 130                   wfMsg("powersearch") . "\">\n";
 131                 $ret = str_replace( "$9", $r9, $ret );
 132
 133                 $ret = "<br><br>\n<form id=\"powersearch\" method=\"get\" " .
 134                   "action=\"" . wfLocalUrl( "" ) . "\">\n{$ret}\n</form>\n";
 135
 136                 if ( isset ( $searchx ) ) {
 137                         if ( ! $listredirs ) { $this->doSearchRedirects = false; }
 138                 }
 139                 return $ret;
 140         }
 141
 142         function showResults()
 143         {
 144                 global $wgUser, $wgTitle, $wgOut, $wgLang, $wgDisableTextSearch;
 145                 $fname = "SearchEngine::showResults";
 146
 147                 $search         = $_REQUEST['search'];
 148
 149                 $powersearch = $this->powersearch(); /* Need side-effects here? */
 150
 151                 $wgOut->setPageTitle( wfMsg( "searchresults" ) );
 152                 $q = wfMsg( "searchquery", htmlspecialchars( $this->mUsertext ) );
 153                 $wgOut->setSubtitle( $q );
 154                 $wgOut->setArticleFlag( false );
 155                 $wgOut->setRobotpolicy( "noindex,nofollow" );
 156
 157                 $sk = $wgUser->getSkin();
 158                 $text = wfMsg( "searchresulttext", $sk->makeKnownLink(
 159                   wfMsg( "searchhelppage" ), wfMsg( "searchingwikipedia" ) ) );
 160                 $wgOut->addHTML( $text );
 161
 162                 $this->parseQuery();
 163                 if ( "" == $this->mTitlecond || "" == $this->mTextcond ) {
 164                         $wgOut->addHTML( "<h2>" . wfMsg( "badquery" ) . "</h2>\n" .
 165                           "<p>" . wfMsg( "badquerytext" ) );
 166                         return;
 167                 }
 168                 list( $limit, $offset ) = wfCheckLimits( 20, "searchlimit" );
 169
 170                 $searchnamespaces = $this->queryNamespaces();
 171                 $redircond = $this->searchRedirects();
 172
 173                 $sql = "SELECT cur_id,cur_namespace,cur_title," .
 174                   "cur_text FROM cur,searchindex " .
 175                   "WHERE cur_id=si_page AND {$this->mTitlecond} " .
 176                   "{$searchnamespaces} {$redircond}" .
 177                   "LIMIT {$offset}, {$limit}";
 178                 $res1 = wfQuery( $sql, $fname );
 179                 $num = wfNumRows($res1);
 180
 181                 if ( $wgDisableTextSearch ) {
 182                         $res2 = 0;
 183                 } else {
 184                         $sql = "SELECT cur_id,cur_namespace,cur_title," .
 185                           "cur_text FROM cur,searchindex " .
 186                           "WHERE cur_id=si_page AND {$this->mTextcond} " .
 187                           "{$searchnamespaces} {$redircond} " .
 188                           "LIMIT {$offset}, {$limit}";
 189                         $res2 = wfQuery( $sql, $fname );
 190                         $num = $num + wfNumRows($res2);
 191                 }
 192
 193                 if ( $num == $limit ) {
 194                   $top = wfShowingResults( $offset, $limit);
 195                 } else {
 196                   $top = wfShowingResultsNum( $offset, $limit, $num );
 197                 }
 198                 $wgOut->addHTML( "<p>{$top}\n" );
 199
 200                 # For powersearch
 201
 202                 $a2l = "" ;
 203                 $akk = array_keys( $this->addtoquery ) ;
 204                 foreach ( $akk AS $ak ) {
 205                         $a2l .= "&{$ak}={$this->addtoquery[$ak]}" ;
 206                 }
 207
 208                 $sl = wfViewPrevNext( $offset, $limit, "",
 209                   "search=" . wfUrlencode( $this->mUsertext ) . $a2l );
 210                 $wgOut->addHTML( "<br>{$sl}\n" );
 211
 212                 $foundsome = false;
 213
 214                 if ( 0 == wfNumRows( $res1 ) ) {
 215                         $wgOut->addHTML( "<h2>" . wfMsg( "notitlematches" ) .
 216                           "</h2>\n" );
 217                 } else {
 218                         $foundsome = true;
 219                         $off = $offset + 1;
 220                         $wgOut->addHTML( "<h2>" . wfMsg( "titlematches" ) .
 221                           "</h2>\n<ol start='{$off}'>" );
 222
 223                         while ( $row = wfFetchObject( $res1 ) ) {
 224                                 $this->showHit( $row );
 225                         }
 226                         wfFreeResult( $res1 );
 227                         $wgOut->addHTML( "</ol>\n" );
 228                 }
 229
 230                 if ( $wgDisableTextSearch ) {
 231                         $wgOut->addHTML( str_replace( "$1",
 232                           htmlspecialchars( $search ), wfMsg( "searchdisabled" ) ) );
 233                 } else {
 234                         if ( 0 == wfNumRows( $res2 ) ) {
 235                                 $wgOut->addHTML( "<h2>" . wfMsg( "notextmatches" ) .
 236                                   "</h2>\n" );
 237                         } else {
 238                                 $foundsome = true;
 239                                 $off = $offset + 1;
 240                                 $wgOut->addHTML( "<h2>" . wfMsg( "textmatches" ) . "</h2>\n" .
 241                                   "<ol start='{$off}'>" );
 242                                 while ( $row = wfFetchObject( $res2 ) ) {
 243                                         $this->showHit( $row );
 244                                 }
 245                                 wfFreeResult( $res2 );
 246                                 $wgOut->addHTML( "</ol>\n" );
 247                         }
 248                 }
 249                 if ( ! $foundsome ) {
 250                         $wgOut->addHTML( "<p>" . wfMsg( "nonefound" ) . "\n" );
 251                 }
 252                 $wgOut->addHTML( "<p>{$sl}\n" );
 253                 $wgOut->addHTML( $powersearch );
 254         }
 255
 256         function legalSearchChars()
 257         {
 258                 $lc = "A-Za-z_'0-9\\x80-\\xFF\\-";
 259                 return $lc;
 260         }
 261
 262         function parseQuery()
 263         {
 264                 global $wgDBminWordLen, $wgLang, $wgDBmysql4;
 265
 266                 if( $wgDBmysql4 ) {
 267                         # Use cleaner boolean search if available
 268                         return $this->parseQuery4();
 269                 }
 270
 271                 $lc = SearchEngine::legalSearchChars() . "()";
 272                 $q = preg_replace( "/([()])/", " \\1 ", $this->mUsertext );
 273                 $q = preg_replace( "/\\s+/", " ", $q );
 274                 $w = explode( " ", strtolower( trim( $q ) ) );
 275
 276                 $last = $cond = "";
 277                 foreach ( $w as $word ) {
 278                         $word = $wgLang->stripForSearch( $word );
 279                         if ( "and" == $word || "or" == $word || "not" == $word
 280                           || "(" == $word || ")" == $word ) {
 281                                 $cond .= " " . strtoupper( $word );
 282                                 $last = "";
 283                         } else if ( strlen( $word ) < $wgDBminWordLen ) {
 284                                 continue;
 285                         } else if ( FulltextStoplist::inList( $word ) ) {
 286                                 continue;
 287                         } else {
 288                                 if ( "" != $last ) { $cond .= " AND"; }
 289                                 $cond .= " (MATCH (##field##) AGAINST ('" .
 290                                   wfStrencode( $word ). "'))";
 291                                 $last = $word;
 292                                 array_push( $this->mSearchterms, "\\b" . $word . "\\b" );
 293                         }
 294                 }
 295                 if ( 0 == count( $this->mSearchterms ) ) { return; }
 296
 297                 $this->mTitlecond = "(" . str_replace( "##field##",
 298                   "si_title", $cond ) . " )";
 299
 300                 $this->mTextcond = "(" . str_replace( "##field##",
 301                   "si_text", $cond ) . " AND (cur_is_redirect=0) )";
 302         }
 303
 304         function parseQuery4()
 305         {
 306                 # FIXME: not ready yet! Do not use.
 307
 308                 global $wgLang;
 309                 $lc = SearchEngine::legalSearchChars();
 310                 #$q = preg_replace( "/([+-]?)([$lc]+)/e",
 311                 #       "\"$1\" . \$wgLang->stripForSearch(\"$2\")",
 312                 #       $this->mUsertext );
 313
 314                 $q = $this->mUsertext;
 315                 $qq = wfStrencode( $wgLang->stripForSearch( $q ) );
 316                 $this->mSearchterms = preg_split( '/\s+/', $q );
 317                 $this->mTitlecond = " MATCH(si_title) AGAINST('$qq' IN BOOLEAN MODE)";
 318                 $this->mTextcond = " (MATCH(si_text) AGAINST('$qq' IN BOOLEAN MODE) AND cur_is_redirect=0)";
 319         }
 320
 321         function showHit( $row )
 322         {
 323                 global $wgUser, $wgOut;
 324
 325                 $t = Title::makeName( $row->cur_namespace, $row->cur_title );
 326                 $sk = $wgUser->getSkin();
 327
 328                 $contextlines = $wgUser->getOption( "contextlines" );
 329                 if ( "" == $contextlines ) { $contextlines = 5; }
 330                 $contextchars = $wgUser->getOption( "contextchars" );
 331                 if ( "" == $contextchars ) { $contextchars = 50; }
 332
 333                 $link = $sk->makeKnownLink( $t, "" );
 334                 $size = str_replace( "$1", strlen( $row->cur_text ), WfMsg( "nbytes" ) );
 335                 $wgOut->addHTML( "<li>{$link} ({$size})" );
 336
 337                 $lines = explode( "\n", $row->cur_text );
 338                 $pat1 = "/(.*)(" . implode( "|", $this->mSearchterms ) . ")(.*)/i";
 339                 $lineno = 0;
 340
 341                 foreach ( $lines as $line ) {
 342                         if ( 0 == $contextlines ) { break; }
 343                         --$contextlines;
 344                         ++$lineno;
 345                         if ( ! preg_match( $pat1, $line, $m ) ) { continue; }
 346
 347                         $pre = $m[1];
 348                         if ( 0 == $contextchars ) { $pre = "..."; }
 349                         else {
 350                                 if ( strlen( $pre ) > $contextchars ) {
 351                                         $pre = "..." . substr( $pre, -$contextchars );
 352                                 }
 353                         }
 354                         $pre = wfEscapeHTML( $pre );
 355
 356                         if ( count( $m ) < 3 ) { $post = ""; }
 357                         else { $post = $m[3]; }
 358
 359                         if ( 0 == $contextchars ) { $post = "..."; }
 360                         else {
 361                                 if ( strlen( $post ) > $contextchars ) {
 362                                         $post = substr( $post, 0, $contextchars ) . "...";
 363                                 }
 364                         }
 365                         $post = wfEscapeHTML( $post );
 366                         $found = wfEscapeHTML( $m[2] );
 367
 368                         $line = "{$pre}{$found}{$post}";
 369                         $pat2 = "/(" . implode( "|", $this->mSearchterms ) . ")/i";
 370                         $line = preg_replace( $pat2,
 371                           "<font color='red'>\\1</font>", $line );
 372
 373                         $wgOut->addHTML( "<br><small>{$lineno}: {$line}</small>\n" );
 374                 }
 375                 $wgOut->addHTML( "</li>\n" );
 376         }
 377
 378         function goResult()
 379         {
 380                 global $wgOut, $wgArticle, $wgTitle;
 381                 $fname = "SearchEngine::goResult";
 382
 383                 $search         = $_REQUEST['search'];
 384
 385                 # First try to go to page as entered
 386                 #
 387                 $wgArticle = new Article();
 388                 $wgTitle = Title::newFromText( $search );
 389
 390                 if ( 0 != $wgArticle->getID() ) {
 391                         $wgOut->redirect( wfLocalUrl( $wgTitle->getPrefixedURL() ) );
 392                         return;
 393                 }
 394
 395                 # Now try all lower case (i.e. first letter capitalized)
 396                 #
 397                 $wgTitle = Title::newFromText( strtolower( $search ) );
 398                 if ( 0 != $wgArticle->getID() ) {
 399                         $wgOut->redirect( wfLocalUrl( $wgTitle->getPrefixedURL() ) );
 400                         return;
 401                 }
 402
 403                 # Now try capitalized string
 404                 #
 405                 $wgTitle=Title::newFromText( ucwords( strtolower( $search ) ) );
 406                 if ( 0 != $wgArticle->getID() ) {
 407                         $wgOut->redirect( wfLocalUrl( $wgTitle->getPrefixedURL() ) );
 408                         return;
 409                 }
 410
 411                 # Now try all upper case
 412                 #
 413                 $wgTitle = Title::newFromText( strtoupper( $search ) );
 414                 if ( 0 != $wgArticle->getID() ) {
 415                         $wgOut->redirect( wfLocalUrl( $wgTitle->getPrefixedURL() ) );
 416                         return;
 417                 }
 418
 419                 # Try a near match
 420                 #
 421                 $this->parseQuery();
 422                 $sql = "SELECT cur_id,cur_title,cur_namespace,si_page FROM cur,searchindex " .
 423                   "WHERE cur_id=si_page AND {$this->mTitlecond} ORDER BY cur_namespace LIMIT 1";
 424
 425                 if ( "" != $this->mTitlecond ) {
 426                         $res = wfQuery( $sql, $fname );
 427                 }
 428                 if ( isset( $res ) && 0 != wfNumRows( $res ) ) {
 429                         $s = wfFetchObject( $res );
 430
 431                         $wgTitle = Title::newFromDBkey( $s->cur_title );
 432                         $wgTitle->setNamespace( $s->cur_namespace );
 433                         $wgOut->redirect( wfLocalUrl( $wgTitle->getPrefixedURL() ) );
 434                         return;
 435                 }
 436                 $wgOut->addHTML( wfMsg("nogomatch",
 437                   htmlspecialchars( wfLocalUrl( ucfirst($this->mUsertext), "action=edit") ) )
 438                   . "\n<p>" );
 439                 $this->showResults();
 440         }
 441 }
 442