includes/SearchUpdate.php

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