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