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