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                 $this->mStrictMatching = true; # Google-style, add '+' on all terms
  23         }
  24
  25         function queryNamespaces()
  26         {
  27                 $namespaces = implode( ",", $this->namespacesToSearch );
  28                 if ($namespaces == "") {
  29                         $namespaces = "0";
  30                 }
  31                 return "AND cur_namespace IN (" . $namespaces . ")";
  32                 #return "1";
  33         }
  34
  35         function searchRedirects()
  36         {
  37                 if ( $this->doSearchRedirects ) return "";
  38                 return "AND cur_is_redirect=0 ";
  39         }
  40
  41         /* private */ function initNamespaceCheckbox( $i )
  42         {
  43                 global $wgUser, $wgNamespacesToBeSearchedDefault;
  44
  45
  46                 if ($wgUser->getID()) {
  47                         // User is logged in so we retrieve his default namespaces
  48                         return $wgUser->getOption( "searchNs".$i );
  49                 }
  50                 else {
  51                         // User is not logged in so we give him the global default namespaces
  52                         return $wgNamespacesToBeSearchedDefault[ $i ];
  53                 }
  54         }
  55
  56         # Display the "power search" footer. Does not actually perform the search,
  57         # that is done by showResults()
  58         function powersearch()
  59         {
  60                 global $wgUser, $wgOut, $wgLang, $wgTitle;
  61
  62                 $search                 = $_REQUEST['search'];
  63                 $searchx                = $_REQUEST['searchx'];
  64                 $listredirs             = $_REQUEST['redirs'];
  65
  66                 $ret = wfMsg("powersearchtext"); # Text to be returned
  67                 $tempText = ""; # Temporary text, for substitution into $ret
  68
  69                 if( isset( $_REQUEST["searchx"] ) ) {
  70                         $this->addtoquery["searchx"] = "1";
  71                 }
  72
  73                 # Do namespace checkboxes
  74                 $namespaces = $wgLang->getNamespaces();
  75                 foreach ( $namespaces as $i => $namespace ) {
  76                         # Skip virtual namespaces
  77                         if ( $i < 0 ) {
  78                                 continue;
  79                         }
  80
  81                         $formVar = "ns$i";
  82
  83                         # Initialise checkboxValues, either from defaults or from
  84                         # a previous invocation
  85                         if ( !isset( $searchx ) ) {
  86                                 $checkboxValue = $this->initNamespaceCheckbox( $i );
  87                         } else {
  88                                 $checkboxValue = $_REQUEST[$formVar];
  89                         }
  90
  91                         $checked = "";
  92                         if ( $checkboxValue == 1 ) {
  93                                 $checked = " checked";
  94                                 $this->addtoquery["ns{$i}"] = 1;
  95                                 array_push( $this->namespacesToSearch, $i );
  96                         }
  97                         $name = str_replace( "_", " ", $namespaces[$i] );
  98                         if ( "" == $name ) {
  99                                 $name = wfMsg( "blanknamespace" );
 100                         }
 101
 102                         if ( $tempText !== "" ) {
 103                                 $tempText .= " ";
 104                         }
 105                         $tempText .= "<input type=checkbox value=\"1\" name=\"" .
 106                           "ns{$i}\"{$checked}>{$name}\n";
 107                 }
 108                 $ret = str_replace ( "$1", $tempText, $ret );
 109
 110                 # List redirects checkbox
 111
 112                 $checked = "";
 113                 if ( $listredirs == 1 ) {
 114                         $this->addtoquery["redirs"] = 1;
 115                         $checked = " checked";
 116                 }
 117                 $tempText = "<input type=checkbox value=1 name=\"redirs\"{$checked}>\n";
 118                 $ret = str_replace( "$2", $tempText, $ret );
 119
 120                 # Search field
 121
 122                 $tempText = "<input type=text name=\"search\" value=\"" .
 123                         htmlspecialchars( $search ) ."\" width=80>\n";
 124         $ret = str_replace( "$3", $tempText, $ret );
 125
 126                 # Searchx button
 127
 128                 $tempText = "<input type=submit name=\"searchx\" value=\"" .
 129                   wfMsg("powersearch") . "\">\n";
 130                 $ret = str_replace( "$9", $tempText, $ret );
 131
 132                 $ret = "<br><br>\n<form id=\"powersearch\" method=\"get\" " .
 133                   "action=\"" . wfLocalUrl( "" ) . "\">\n{$ret}\n</form>\n";
 134
 135                 if ( isset ( $searchx ) ) {
 136                         if ( ! $listredirs ) {
 137                                 $this->doSearchRedirects = false;
 138                         }
 139                 }
 140                 return $ret;
 141         }
 142
 143         # Perform the search and construct the results page
 144         function showResults()
 145         {
 146                 global $wgUser, $wgTitle, $wgOut, $wgLang, $wgDisableTextSearch;
 147                 global $wgInputEncoding;
 148                 $fname = "SearchEngine::showResults";
 149
 150                 $search         = $_REQUEST['search'];
 151
 152                 $powersearch = $this->powersearch(); /* Need side-effects here? */
 153
 154                 $wgOut->setPageTitle( wfMsg( "searchresults" ) );
 155                 $q = wfMsg( "searchquery", htmlspecialchars( $this->mUsertext ) );
 156                 $wgOut->setSubtitle( $q );
 157                 $wgOut->setArticleFlag( false );
 158                 $wgOut->setRobotpolicy( "noindex,nofollow" );
 159
 160                 $sk = $wgUser->getSkin();
 161                 $text = wfMsg( "searchresulttext", $sk->makeKnownLink(
 162                   wfMsg( "searchhelppage" ), wfMsg( "searchingwikipedia" ) ) );
 163                 $wgOut->addHTML( $text );
 164
 165                 $this->parseQuery();
 166                 if ( "" == $this->mTitlecond || "" == $this->mTextcond ) {
 167                         $wgOut->addHTML( "<h2>" . wfMsg( "badquery" ) . "</h2>\n" .
 168                           "<p>" . wfMsg( "badquerytext" ) );
 169                         return;
 170                 }
 171                 list( $limit, $offset ) = wfCheckLimits( 20, "searchlimit" );
 172
 173                 $searchnamespaces = $this->queryNamespaces();
 174                 $redircond = $this->searchRedirects();
 175
 176                 if ( $wgDisableTextSearch ) {
 177                         $wgOut->addHTML( wfMsg( "searchdisabled", htmlspecialchars( $search ), $wgInputEncoding ) );
 178                 } else {
 179                         $sql = "SELECT cur_id,cur_namespace,cur_title," .
 180                           "cur_text FROM cur,searchindex " .
 181                           "WHERE cur_id=si_page AND {$this->mTitlecond} " .
 182                           "{$searchnamespaces} {$redircond}" .
 183                           "LIMIT {$offset}, {$limit}";
 184                         $res1 = wfQuery( $sql, DB_READ, $fname );
 185                         $num = wfNumRows($res1);
 186
 187                         $sk = $wgUser->getSkin();
 188                         $text = wfMsg( "searchresulttext", $sk->makeKnownLink(
 189                           wfMsg( "searchhelppage" ), wfMsg( "searchingwikipedia" ) ) );
 190                         $wgOut->addHTML( $text );
 191
 192                         $this->parseQuery();
 193                         if ( "" == $this->mTitlecond || "" == $this->mTextcond ) {
 194                                 $wgOut->addHTML( "<h2>" . wfMsg( "badquery" ) . "</h2>\n" .
 195                                   "<p>" . wfMsg( "badquerytext" ) );
 196                                 return;
 197                         }
 198                         list( $limit, $offset ) = wfCheckLimits( 20, "searchlimit" );
 199
 200                         $searchnamespaces = $this->queryNamespaces();
 201                         $redircond = $this->searchRedirects();
 202
 203                         $sql = "SELECT cur_id,cur_namespace,cur_title," .
 204                           "cur_text FROM cur,searchindex " .
 205                           "WHERE cur_id=si_page AND {$this->mTitlecond} " .
 206                           "{$searchnamespaces} {$redircond}" .
 207                           "LIMIT {$offset}, {$limit}";
 208                         $res1 = wfQuery( $sql, DB_READ, $fname );
 209                         $num = wfNumRows($res1);
 210
 211                         $sql = "SELECT cur_id,cur_namespace,cur_title," .
 212                           "cur_text FROM cur,searchindex " .
 213                           "WHERE cur_id=si_page AND {$this->mTextcond} " .
 214                           "{$searchnamespaces} {$redircond} " .
 215                           "LIMIT {$offset}, {$limit}";
 216                         $res2 = wfQuery( $sql, DB_READ, $fname );
 217                         $num = $num + wfNumRows($res2);
 218
 219                         if ( $num == $limit ) {
 220                           $top = wfShowingResults( $offset, $limit);
 221                         } else {
 222                           $top = wfShowingResultsNum( $offset, $limit, $num );
 223                         }
 224                         $wgOut->addHTML( "<p>{$top}\n" );
 225
 226                         # For powersearch
 227
 228                         $a2l = "" ;
 229                         $akk = array_keys( $this->addtoquery ) ;
 230                         foreach ( $akk AS $ak ) {
 231                                 $a2l .= "&{$ak}={$this->addtoquery[$ak]}" ;
 232                         }
 233
 234                         $sl = wfViewPrevNext( $offset, $limit, "",
 235                           "search=" . wfUrlencode( $this->mUsertext ) . $a2l );
 236                         $wgOut->addHTML( "<br>{$sl}\n" );
 237
 238                         $foundsome = false;
 239
 240                         if ( 0 == wfNumRows( $res1 ) ) {
 241                                 $wgOut->addHTML( "<h2>" . wfMsg( "notitlematches" ) .
 242                                   "</h2>\n" );
 243                         } else {
 244                                 $foundsome = true;
 245                                 $off = $offset + 1;
 246                                 $wgOut->addHTML( "<h2>" . wfMsg( "titlematches" ) .
 247                                   "</h2>\n<ol start='{$off}'>" );
 248
 249                                 while ( $row = wfFetchObject( $res1 ) ) {
 250                                         $this->showHit( $row );
 251                                 }
 252                                 wfFreeResult( $res1 );
 253                                 $wgOut->addHTML( "</ol>\n" );
 254                         }
 255
 256                         if ( 0 == wfNumRows( $res2 ) ) {
 257                                 $wgOut->addHTML( "<h2>" . wfMsg( "notextmatches" ) .
 258                                   "</h2>\n" );
 259                         } else {
 260                                 $foundsome = true;
 261                                 $off = $offset + 1;
 262                                 $wgOut->addHTML( "<h2>" . wfMsg( "textmatches" ) . "</h2>\n" .
 263                                   "<ol start='{$off}'>" );
 264                                 while ( $row = wfFetchObject( $res2 ) ) {
 265                                         $this->showHit( $row );
 266                                 }
 267                                 wfFreeResult( $res2 );
 268                                 $wgOut->addHTML( "</ol>\n" );
 269                         }
 270                         if ( ! $foundsome ) {
 271                                 $wgOut->addHTML( "<p>" . wfMsg( "nonefound" ) . "\n" );
 272                         }
 273                         $wgOut->addHTML( "<p>{$sl}\n" );
 274                         $wgOut->addHTML( $powersearch );
 275                 }
 276         }
 277
 278         function legalSearchChars()
 279         {
 280                 $lc = "A-Za-z_'0-9\\x80-\\xFF\\-";
 281                 return $lc;
 282         }
 283
 284         function parseQuery()
 285         {
 286                 global $wgDBminWordLen, $wgLang, $wgDBmysql4;
 287
 288                 if( $wgDBmysql4 ) {
 289                         # Use cleaner boolean search if available
 290                         return $this->parseQuery4();
 291                 }
 292
 293                 $lc = SearchEngine::legalSearchChars() . "()";
 294                 $q = preg_replace( "/([()])/", " \\1 ", $this->mUsertext );
 295                 $q = preg_replace( "/\\s+/", " ", $q );
 296                 $w = explode( " ", strtolower( trim( $q ) ) );
 297
 298                 $last = $cond = "";
 299                 foreach ( $w as $word ) {
 300                         $word = $wgLang->stripForSearch( $word );
 301                         if ( "and" == $word || "or" == $word || "not" == $word
 302                           || "(" == $word || ")" == $word ) {
 303                                 $cond .= " " . strtoupper( $word );
 304                                 $last = "";
 305                         } else if ( strlen( $word ) < $wgDBminWordLen ) {
 306                                 continue;
 307                         } else if ( FulltextStoplist::inList( $word ) ) {
 308                                 continue;
 309                         } else {
 310                                 if ( "" != $last ) { $cond .= " AND"; }
 311                                 $cond .= " (MATCH (##field##) AGAINST ('" .
 312                                   wfStrencode( $word ). "'))";
 313                                 $last = $word;
 314                                 array_push( $this->mSearchterms, "\\b" . $word . "\\b" );
 315                         }
 316                 }
 317                 if ( 0 == count( $this->mSearchterms ) ) { return; }
 318
 319                 $this->mTitlecond = "(" . str_replace( "##field##",
 320                   "si_title", $cond ) . " )";
 321
 322                 $this->mTextcond = "(" . str_replace( "##field##",
 323                   "si_text", $cond ) . " AND (cur_is_redirect=0) )";
 324         }
 325
 326         function parseQuery4()
 327         {
 328                 global $wgLang;
 329                 $lc = SearchEngine::legalSearchChars();
 330                 $searchon = "";
 331                 $this->mSearchterms = array();
 332
 333                 # FIXME: This doesn't handle parenthetical expressions.
 334                 if( preg_match_all( '/([-+<>~]?)(([' . $lc . ']+)(\*?)|"[^"]*")/',
 335                           $this->mUsertext, $m, PREG_SET_ORDER ) ) {
 336                         foreach( $m as $terms ) {
 337                                 if( $searchon !== "" ) $searchon .= " ";
 338                                 if( $this->mStrictMatching && ($terms[1] == "") ) {
 339                                         $terms[1] = "+";
 340                                 }
 341                                 $searchon .= $terms[1] . $wgLang->stripForSearch( $terms[2] );
 342                                 if( $terms[3] ) {
 343                                         $regexp = preg_quote( $terms[3] );
 344                                         if( $terms[4] ) $regexp .= "[0-9A-Za-z_]+";
 345                                 } else {
 346                                         $regexp = preg_quote( str_replace( '"', '', $terms[2] ) );
 347                                 }
 348                                 $this->mSearchterms[] = $regexp;
 349                         }
 350                         wfDebug( "Would search with '$searchon'\n" );
 351                         wfDebug( "Match with /\b" . implode( '\b|\b', $this->mSearchterms ) . "\b/\n" );
 352                 } else {
 353                         wfDebug( "Can't understand search query '$this->mUsertext'\n" );
 354                 }
 355
 356                 $searchon = wfStrencode( $searchon );
 357                 $this->mTitlecond = " MATCH(si_title) AGAINST('$searchon' IN BOOLEAN MODE)";
 358                 $this->mTextcond = " (MATCH(si_text) AGAINST('$searchon' IN BOOLEAN MODE) AND cur_is_redirect=0)";
 359         }
 360
 361         function showHit( $row )
 362         {
 363                 global $wgUser, $wgOut;
 364
 365                 $t = Title::makeName( $row->cur_namespace, $row->cur_title );
 366                 $sk = $wgUser->getSkin();
 367
 368                 $contextlines = $wgUser->getOption( "contextlines" );
 369                 if ( "" == $contextlines ) { $contextlines = 5; }
 370                 $contextchars = $wgUser->getOption( "contextchars" );
 371                 if ( "" == $contextchars ) { $contextchars = 50; }
 372
 373                 $link = $sk->makeKnownLink( $t, "" );
 374                 $size = wfMsg( "nbytes", strlen( $row->cur_text ) );
 375                 $wgOut->addHTML( "<li>{$link} ({$size})" );
 376
 377                 $lines = explode( "\n", $row->cur_text );
 378                 $pat1 = "/(.*)(" . implode( "|", $this->mSearchterms ) . ")(.*)/i";
 379                 $lineno = 0;
 380
 381                 foreach ( $lines as $line ) {
 382                         if ( 0 == $contextlines ) { break; }
 383                         --$contextlines;
 384                         ++$lineno;
 385                         if ( ! preg_match( $pat1, $line, $m ) ) { continue; }
 386
 387                         $pre = $m[1];
 388                         if ( 0 == $contextchars ) { $pre = "..."; }
 389                         else {
 390                                 if ( strlen( $pre ) > $contextchars ) {
 391                                         $pre = "..." . substr( $pre, -$contextchars );
 392                                 }
 393                         }
 394                         $pre = wfEscapeHTML( $pre );
 395
 396                         if ( count( $m ) < 3 ) { $post = ""; }
 397                         else { $post = $m[3]; }
 398
 399                         if ( 0 == $contextchars ) { $post = "..."; }
 400                         else {
 401                                 if ( strlen( $post ) > $contextchars ) {
 402                                         $post = substr( $post, 0, $contextchars ) . "...";
 403                                 }
 404                         }
 405                         $post = wfEscapeHTML( $post );
 406                         $found = wfEscapeHTML( $m[2] );
 407
 408                         $line = "{$pre}{$found}{$post}";
 409                         $pat2 = "/(" . implode( "|", $this->mSearchterms ) . ")/i";
 410                         $line = preg_replace( $pat2,
 411                           "<font color='red'>\\1</font>", $line );
 412
 413                         $wgOut->addHTML( "<br><small>{$lineno}: {$line}</small>\n" );
 414                 }
 415                 $wgOut->addHTML( "</li>\n" );
 416         }
 417
 418         function goResult()
 419         {
 420                 global $wgOut, $wgDisableTextSearch;
 421                 $fname = "SearchEngine::goResult";
 422
 423                 $search         = $_REQUEST['search'];
 424
 425                 # First try to go to page as entered.
 426                 #
 427                 $t = Title::newFromText( $search );
 428
 429                 # If the string cannot be used to create a title
 430                 if( false == $t ){
 431                         $this->showResults();
 432                         return;
 433                 }
 434
 435                 if ( 0 != $t->getArticleID() ) {
 436                         $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
 437                         return;
 438                 }
 439
 440                 # Now try all lower case (i.e. first letter capitalized)
 441                 #
 442                 $t = Title::newFromText( strtolower( $search ) );
 443                 if ( 0 != $t->getArticleID() ) {
 444                         $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
 445                         return;
 446                 }
 447
 448                 # Now try capitalized string
 449                 #
 450                 $t = Title::newFromText( ucwords( strtolower( $search ) ) );
 451                 if ( 0 != $t->getArticleID() ) {
 452                         $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
 453                         return;
 454                 }
 455
 456                 # Now try all upper case
 457                 #
 458                 $t = Title::newFromText( strtoupper( $search ) );
 459                 if ( 0 != $t->getArticleID() ) {
 460                         $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
 461                         return;
 462                 }
 463
 464                 # Try a near match
 465                 #
 466                 if( !$wgDisableTextSearch ) {
 467                         $this->parseQuery();
 468                         $sql = "SELECT cur_id,cur_title,cur_namespace,si_page FROM cur,searchindex " .
 469                           "WHERE cur_id=si_page AND {$this->mTitlecond} ORDER BY cur_namespace LIMIT 1";
 470
 471                         if ( "" != $this->mTitlecond ) {
 472                                 $res = wfQuery( $sql, DB_READ, $fname );
 473                         }
 474                         if ( isset( $res ) && 0 != wfNumRows( $res ) ) {
 475                                 $s = wfFetchObject( $res );
 476
 477                                 $t = Title::makeTitle( $s->cur_namespace, $s->cur_title );
 478                                 $wgOut->redirect( wfLocalUrl( $t->getPrefixedURL() ) );
 479                                 return;
 480                         }
 481                 }
 482                 $wgOut->addHTML( wfMsg("nogomatch",
 483                   htmlspecialchars( wfLocalUrl( ucfirst($this->mUsertext), "action=edit") ) )
 484                   . "\n<p>" );
 485                 $this->showResults();
 486         }
 487 }
 488
 489 ?>