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                 # Tidy is known to clobber tabs; convert them to entities
  35                 $wrappedtext = str_replace( "\t", '&#9;', $wrappedtext );
  36
  37                 if( $wgTidyInternal ) {
  38                         $correctedtext = self::execInternalTidy( $wrappedtext );
  39                 } else {
  40                         $correctedtext = self::execExternalTidy( $wrappedtext );
  41                 }
  42                 if( is_null( $correctedtext ) ) {
  43                         wfDebug( "Tidy error detected!\n" );
  44                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
  45                 }
  46
  47                 # Convert the tabs back from entities
  48                 $correctedtext = str_replace( '&#9;', "\t", $correctedtext );
  49
  50                 return $correctedtext;
  51         }
  52
  53         /**
  54          * Check HTML for errors, used if $wgValidateAllHtml = true.
  55          *
  56          * @param $text String
  57          * @param &$errorStr String: return the error string
  58          * @return Boolean: whether the HTML is valid
  59          */
  60         public static function checkErrors( $text, &$errorStr = null ) {
  61                 global $wgTidyInternal;
  62
  63                 $retval = 0;
  64                 if( $wgTidyInternal ) {
  65                         $errorStr = self::execInternalTidy( $text, true, $retval );
  66                 } else {
  67                         $errorStr = self::execExternalTidy( $text, true, $retval );
  68                 }
  69                 return ( $retval < 0 && $errorStr == '' ) || $retval == 0;
  70         }
  71
  72         /**
  73          * Spawn an external HTML tidy process and get corrected markup back from it.
  74          * Also called in OutputHandler.php for full page validation
  75          *
  76          * @param $text String: HTML to check
  77          * @param $stderr Boolean: Whether to read from STDERR rather than STDOUT
  78          * @param &$retval Exit code (-1 on internal error)
  79          * @return mixed String or null
  80          */
  81         private static function execExternalTidy( $text, $stderr = false, &$retval = null ) {
  82                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
  83                 wfProfileIn( __METHOD__ );
  84
  85                 $cleansource = '';
  86                 $opts = ' -utf8';
  87
  88                 if( $stderr ) {
  89                         $descriptorspec = array(
  90                                 0 => array( 'pipe', 'r' ),
  91                                 1 => array( 'file', wfGetNull(), 'a' ),
  92                                 2 => array( 'pipe', 'w' )
  93                         );
  94                 } else {
  95                         $descriptorspec = array(
  96                                 0 => array( 'pipe', 'r' ),
  97                                 1 => array( 'pipe', 'w' ),
  98                                 2 => array( 'file', wfGetNull(), 'a' )
  99                         );
 100                 }
 101
 102                 $readpipe = $stderr ? 2 : 1;
 103                 $pipes = array();
 104
 105                 if( function_exists( 'proc_open' ) ) {
 106                         $process = proc_open( "$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes );
 107                         if ( is_resource( $process ) ) {
 108                                 // Theoretically, this style of communication could cause a deadlock
 109                                 // here. If the stdout buffer fills up, then writes to stdin could
 110                                 // block. This doesn't appear to happen with tidy, because tidy only
 111                                 // writes to stdout after it's finished reading from stdin. Search
 112                                 // for tidyParseStdin and tidySaveStdout in console/tidy.c
 113                                 fwrite( $pipes[0], $text );
 114                                 fclose( $pipes[0] );
 115                                 while ( !feof( $pipes[$readpipe] ) ) {
 116                                         $cleansource .= fgets( $pipes[$readpipe], 1024 );
 117                                 }
 118                                 fclose( $pipes[$readpipe] );
 119                                 $retval = proc_close( $process );
 120                         } else {
 121                                 $retval = -1;
 122                         }
 123                 } else {
 124                         $retval = -1;
 125                 }
 126
 127                 wfProfileOut( __METHOD__ );
 128
 129                 if( !$stderr && $cleansource == '' && $text != '' ) {
 130                         // Some kind of error happened, so we couldn't get the corrected text.
 131                         // Just give up; we'll use the source text and append a warning.
 132                         return null;
 133                 } else {
 134                         return $cleansource;
 135                 }
 136         }
 137
 138         /**
 139          * Use the HTML tidy PECL extension to use the tidy library in-process,
 140          * saving the overhead of spawning a new process.
 141          *
 142          * 'pear install tidy' should be able to compile the extension module.
 143          */
 144         private static function execInternalTidy( $text, $stderr = false, &$retval = null ) {
 145                 global $wgTidyConf, $wgDebugTidy;
 146                 wfProfileIn( __METHOD__ );
 147
 148                 $tidy = new tidy;
 149                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 150
 151                 if( $stderr ) {
 152                         $retval = $tidy->getStatus();
 153                         return $tidy->errorBuffer;
 154                 } else {
 155                         $tidy->cleanRepair();
 156                         $retval = $tidy->getStatus();
 157                         if( $retval == 2 ) {
 158                                 // 2 is magic number for fatal error
 159                                 // http://www.php.net/manual/en/function.tidy-get-status.php
 160                                 $cleansource = null;
 161                         } else {
 162                                 $cleansource = tidy_get_output( $tidy );
 163                         }
 164                         if ( $wgDebugTidy && $retval > 0 ) {
 165                                 $cleansource .= "<!--\nTidy reports:\n" .
 166                                         str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 167                                         "\n-->";
 168                         }
 169
 170                         wfProfileOut( __METHOD__ );
 171                         return $cleansource;
 172                 }
 173         }
 174
 175 }