includes/parser/Tidy.php

   1 <?php
   2 /**
   3  * HTML validation and correction
   4  *
   5  * @file
   6  */
   7
   8 /**
   9  * Class to interact with HTML tidy
  10  *
  11  * Either the external tidy program or the in-process tidy extension
  12  * will be used depending on availability. Override the default
  13  * $wgTidyInternal setting to disable the internal if it's not working.
  14  *
  15  * @ingroup Parser
  16  */
  17 class MWTidy {
  18
  19         /**
  20          * Interface with html tidy, used if $wgUseTidy = true.
  21          * If tidy isn't able to correct the markup, the original will be
  22          * returned in all its glory with a warning comment appended.
  23          *
  24          * @param $text String: hideous HTML input
  25          * @return String: corrected HTML output
  26          */
  27         public static function tidy( $text ) {
  28                 global $wgTidyInternal;
  29
  30                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
  31 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
  32 '<head><title>test</title></head><body>'.$text.'</body></html>';
  33
  34                 if( $wgTidyInternal ) {
  35                         $correctedtext = self::execInternalTidy( $wrappedtext );
  36                 } else {
  37                         $correctedtext = self::execExternalTidy( $wrappedtext );
  38                 }
  39                 if( is_null( $correctedtext ) ) {
  40                         wfDebug( "Tidy error detected!\n" );
  41                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
  42                 }
  43
  44                 return $correctedtext;
  45         }
  46
  47         /**
  48          * Check HTML for errors, used if $wgValidateAllHtml = true.
  49          *
  50          * @param $text String
  51          * @param &$errorStr String: return the error string
  52          * @return Boolean: whether the HTML is valid
  53          */
  54         public static function checkErrors( $text, &$errorStr = null ) {
  55                 global $wgTidyInternal;
  56
  57                 $retval = 0;
  58                 if( $wgTidyInternal ) {
  59                         $errorStr = self::execInternalTidy( $text, true, $retval );
  60                 } else {
  61                         $errorStr = self::execExternalTidy( $text, true, $retval );
  62                 }
  63                 return ( $retval < 0 && $errorStr == '' ) || $retval == 0;
  64         }
  65
  66         /**
  67          * Spawn an external HTML tidy process and get corrected markup back from it.
  68          * Also called in OutputHandler.php for full page validation
  69          *
  70          * @param $text String: HTML to check
  71          * @param $stderr Boolean: Whether to read from STDERR rather than STDOUT
  72          * @param &$retval Exit code (-1 on internal error)
  73          * @return mixed String or null
  74          */
  75         private static function execExternalTidy( $text, $stderr = false, &$retval = null ) {
  76                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
  77                 wfProfileIn( __METHOD__ );
  78
  79                 $cleansource = '';
  80                 $opts = ' -utf8';
  81
  82                 if( $stderr ) {
  83                         $descriptorspec = array(
  84                                 0 => array( 'pipe', 'r' ),
  85                                 1 => array( 'file', wfGetNull(), 'a' ),
  86                                 2 => array( 'pipe', 'w' )
  87                         );
  88                 } else {
  89                         $descriptorspec = array(
  90                                 0 => array( 'pipe', 'r' ),
  91                                 1 => array( 'pipe', 'w' ),
  92                                 2 => array( 'file', wfGetNull(), 'a' )
  93                         );
  94                 }
  95
  96                 $readpipe = $stderr ? 2 : 1;
  97                 $pipes = array();
  98
  99                 if( function_exists( 'proc_open' ) ) {
 100                         $process = proc_open( "$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes );
 101                         if ( is_resource( $process ) ) {
 102                                 // Theoretically, this style of communication could cause a deadlock
 103                                 // here. If the stdout buffer fills up, then writes to stdin could
 104                                 // block. This doesn't appear to happen with tidy, because tidy only
 105                                 // writes to stdout after it's finished reading from stdin. Search
 106                                 // for tidyParseStdin and tidySaveStdout in console/tidy.c
 107                                 fwrite( $pipes[0], $text );
 108                                 fclose( $pipes[0] );
 109                                 while ( !feof( $pipes[$readpipe] ) ) {
 110                                         $cleansource .= fgets( $pipes[$readpipe], 1024 );
 111                                 }
 112                                 fclose( $pipes[$readpipe] );
 113                                 $retval = proc_close( $process );
 114                         } else {
 115                                 $retval = -1;
 116                         }
 117                 } else {
 118                         $retval = -1;
 119                 }
 120
 121                 wfProfileOut( __METHOD__ );
 122
 123                 if( !$stderr && $cleansource == '' && $text != '' ) {
 124                         // Some kind of error happened, so we couldn't get the corrected text.
 125                         // Just give up; we'll use the source text and append a warning.
 126                         return null;
 127                 } else {
 128                         return $cleansource;
 129                 }
 130         }
 131
 132         /**
 133          * Use the HTML tidy PECL extension to use the tidy library in-process,
 134          * saving the overhead of spawning a new process.
 135          *
 136          * 'pear install tidy' should be able to compile the extension module.
 137          */
 138         private static function execInternalTidy( $text, $stderr = false, &$retval = null ) {
 139                 global $wgTidyConf, $wgDebugTidy;
 140                 wfProfileIn( __METHOD__ );
 141
 142                 $tidy = new tidy;
 143                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 144
 145                 if( $stderr ) {
 146                         $retval = $tidy->getStatus();
 147                         return $tidy->errorBuffer;
 148                 } else {
 149                         $tidy->cleanRepair();
 150                         $retval = $tidy->getStatus();
 151                         if( $retval == 2 ) {
 152                                 // 2 is magic number for fatal error
 153                                 // http://www.php.net/manual/en/function.tidy-get-status.php
 154                                 $cleansource = null;
 155                         } else {
 156                                 $cleansource = tidy_get_output( $tidy );
 157                         }
 158                         if ( $wgDebugTidy && $retval > 0 ) {
 159                                 $cleansource .= "<!--\nTidy reports:\n" .
 160                                         str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 161                                         "\n-->";
 162                         }
 163
 164                         wfProfileOut( __METHOD__ );
 165                         return $cleansource;
 166                 }
 167         }
 168
 169 }