includes/SearchUpdate.php

   1 <?php
   2 # $Id$
   3 # See deferred.doc
   4
   5 class SearchUpdate {
   6
   7         /* private */ var $mId = 0, $mNamespace, $mTitle, $mText;
   8         /* private */ var $mTitleWords;
   9
  10         function SearchUpdate( $id, $title, $text = false )
  11         {
  12                 $nt = Title::newFromText( $title );
  13                 if( $nt ) {
  14                         $this->mId = $id;
  15                         $this->mText = $text;
  16
  17                         $this->mNamespace = $nt->getNamespace();
  18                         $this->mTitle = $nt->getText(); # Discard namespace
  19
  20                         $this->mTitleWords = $this->mTextWords = array();
  21                 } else {
  22                         wfDebug( "SearchUpdate object created with invalid title '$title'\n" );
  23                 }
  24         }
  25
  26         function doUpdate()
  27         {
  28                 global $wgDBminWordLen, $wgLang, $wgDisableSearchUpdate, $wgIsMySQL;
  29
  30                 if( $wgDisableSearchUpdate || !$this->mId ) {
  31                         return false;
  32                 }
  33                 $lc = SearchEngine::legalSearchChars() . "&#;";
  34                 if( $this->mText == false ) {
  35                         # Just update the title
  36                         $lowpri=$wgIsMySQL?"LOW_PRIORITY":"";
  37                         $sql = "UPDATE $lowpri searchindex SET si_title='" .
  38                           wfStrencode( Title::indexTitle( $this->mNamespace, $this->mTitle ) ) .
  39                           "' WHERE si_page={$this->mId}";
  40                         wfQuery( $sql, DB_WRITE, "SearchUpdate::doUpdate" );
  41                         return;
  42                 }
  43
  44                 # Language-specific strip/conversion
  45                 $text = $wgLang->stripForSearch( $this->mText );
  46
  47                 $text = preg_replace( "/<\\/?\\s*[A-Za-z][A-Za-z0-9]*\\s*([^>]*?)>/",
  48                   " ", strtolower( " " . $text /*$this->mText*/ . " " ) ); # Strip HTML markup
  49                 $text = preg_replace( "/(^|\\n)\\s*==\\s+([^\\n]+)\\s+==\\s/sD",
  50                   "\\2 \\2 \\2 ", $text ); # Emphasize headings
  51
  52                 # Strip external URLs
  53                 $uc = "A-Za-z0-9_\\/:.,~%\\-+&;#?!=()@\\xA0-\\xFF";
  54                 $protos = "http|https|ftp|mailto|news|gopher";
  55                 $pat = "/(^|[^\\[])({$protos}):[{$uc}]+([^{$uc}]|$)/";
  56                 $text = preg_replace( $pat, "\\1 \\3", $text );
  57
  58                 $p1 = "/([^\\[])\\[({$protos}):[{$uc}]+]/";
  59                 $p2 = "/([^\\[])\\[({$protos}):[{$uc}]+\\s+([^\\]]+)]/";
  60                 $text = preg_replace( $p1, "\\1 ", $text );
  61                 $text = preg_replace( $p2, "\\1 \\3 ", $text );
  62
  63                 # Internal image links
  64                 $pat2 = "/\\[\\[image:([{$uc}]+)\\.(gif|png|jpg|jpeg)([^{$uc}])/i";
  65                 $text = preg_replace( $pat2, " \\1 \\3", $text );
  66
  67                 $text = preg_replace( "/([^{$lc}])([{$lc}]+)]]([a-z]+)/",
  68                   "\\1\\2 \\2\\3", $text ); # Handle [[game]]s
  69
  70                 # Strip all remaining non-search characters
  71                 $text = preg_replace( "/[^{$lc}]+/", " ", $text );
  72
  73                 # Handle 's, s'
  74                 $text = preg_replace( "/([{$lc}]+)'s /", "\\1 \\1's ", $text );
  75                 $text = preg_replace( "/([{$lc}]+)s' /", "\\1s ", $text );
  76
  77                 # Strip wiki '' and '''
  78                 $text = preg_replace( "/''[']*/", " ", $text );
  79
  80                 $sql = "REPLACE  INTO searchindex (si_page,si_title,si_text) VALUES ({$this->mId},'" .
  81                   wfStrencode( Title::indexTitle( $this->mNamespace, $this->mTitle ) ) . "','" .
  82                   wfStrencode( $text ) . "')";
  83                 wfQuery( $sql, DB_WRITE, "SearchUpdate::doUpdate" );
  84         }
  85 }
  86
  87 ?>