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