includes/SearchUpdate.php

   1 <?php
   2 # See deferred.doc
   3
   4 class SearchUpdate {
   5
   6         /* private */ var $mId = 0, $mNamespace, $mTitle, $mText;
   7         /* private */ var $mTitleWords;
   8
   9         function SearchUpdate( $id, $title, $text = false )
  10         {
  11                 $nt = Title::newFromText( $title );
  12                 if( $nt ) {
  13                         $this->mId = $id;
  14                         $this->mText = $text;
  15
  16                         $this->mNamespace = $nt->getNamespace();
  17                         $this->mTitle = $nt->getText(); # Discard namespace
  18
  19                         $this->mTitleWords = $this->mTextWords = array();
  20                 } else {
  21                         wfDebug( "SearchUpdate object created with invalid title '$title'\n" );
  22                 }
  23         }
  24
  25         function doUpdate()
  26         {
  27                 global $wgDBminWordLen, $wgLang, $wgDisableSearchUpdate;
  28
  29                 if( $wgDisableSearchUpdate || !$this->mId ) {
  30                         return false;
  31                 }
  32                 $lc = SearchEngine::legalSearchChars() . "&#;";
  33                 if( $this->mText == false ) {
  34                         # Just update the title
  35                         $sql = "UPDATE LOW_PRIORITY searchindex SET si_title='" .
  36                           wfStrencode( Title::indexTitle( $this->mNamespace, $this->mTitle ) ) .
  37                           "' WHERE si_page={$this->mId}";
  38                         wfQuery( $sql, DB_WRITE, "SearchUpdate::doUpdate" );
  39                         return;
  40                 }
  41
  42                 # Language-specific strip/conversion
  43                 $text = $wgLang->stripForSearch( $this->mText );
  44
  45                 $text = preg_replace( "/<\\/?\\s*[A-Za-z][A-Za-z0-9]*\\s*([^>]*?)>/",
  46                   " ", strtolower( " " . $text /*$this->mText*/ . " " ) ); # Strip HTML markup
  47                 $text = preg_replace( "/(^|\\n)\\s*==\\s+([^\\n]+)\\s+==\\s/sD",
  48                   "\\2 \\2 \\2 ", $text ); # Emphasize headings
  49
  50                 # Strip external URLs
  51                 $uc = "A-Za-z0-9_\\/:.,~%\\-+&;#?!=()@\\xA0-\\xFF";
  52                 $protos = "http|https|ftp|mailto|news|gopher";
  53                 $pat = "/(^|[^\\[])({$protos}):[{$uc}]+([^{$uc}]|$)/";
  54                 $text = preg_replace( $pat, "\\1 \\3", $text );
  55
  56                 $p1 = "/([^\\[])\\[({$protos}):[{$uc}]+]/";
  57                 $p2 = "/([^\\[])\\[({$protos}):[{$uc}]+\\s+([^\\]]+)]/";
  58                 $text = preg_replace( $p1, "\\1 ", $text );
  59                 $text = preg_replace( $p2, "\\1 \\3 ", $text );
  60
  61                 # Internal image links
  62                 $pat2 = "/\\[\\[image:([{$uc}]+)\\.(gif|png|jpg|jpeg)([^{$uc}])/i";
  63                 $text = preg_replace( $pat2, " \\1 \\3", $text );
  64
  65                 $text = preg_replace( "/([^{$lc}])([{$lc}]+)]]([a-z]+)/",
  66                   "\\1\\2 \\2\\3", $text ); # Handle [[game]]s
  67
  68                 # Strip all remaining non-search characters
  69                 $text = preg_replace( "/[^{$lc}]+/", " ", $text );
  70
  71                 # Handle 's, s'
  72                 $text = preg_replace( "/([{$lc}]+)'s /", "\\1 \\1's ", $text );
  73                 $text = preg_replace( "/([{$lc}]+)s' /", "\\1s ", $text );
  74
  75                 # Strip wiki '' and '''
  76                 $text = preg_replace( "/''[']*/", " ", $text );
  77
  78                 $sql = "REPLACE DELAYED INTO searchindex (si_page,si_title,si_text) VALUES ({$this->mId},'" .
  79                   wfStrencode( Title::indexTitle( $this->mNamespace, $this->mTitle ) ) . "','" .
  80                   wfStrencode( $text ) . "')";
  81                 wfQuery( $sql, DB_WRITE, "SearchUpdate::doUpdate" );
  82         }
  83 }
  84
  85 ?>