includes/parser/Tidy.php

   1 <?php
   2
   3 /**
   4  * Class to interact with HTML tidy
   5  *
   6  * Either the external tidy program or the in-process tidy extension
   7  * will be used depending on availability. Override the default
   8  * $wgTidyInternal setting to disable the internal if it's not working.
   9  *
  10  * @ingroup Parser
  11  */
  12 class MWTidy {
  13
  14         /**
  15          * Interface with html tidy, used if $wgUseTidy = true.
  16          * If tidy isn't able to correct the markup, the original will be
  17          * returned in all its glory with a warning comment appended.
  18          *
  19          * @param $text String: hideous HTML input
  20          * @return String: corrected HTML output
  21          */
  22         public static function tidy( $text ) {
  23                 global $wgTidyInternal;
  24
  25                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
  26 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
  27 '<head><title>test</title></head><body>'.$text.'</body></html>';
  28
  29                 # Tidy is known to clobber tabs; convert them to entities
  30                 $wrappedtext = str_replace( "\t", '&#9;', $wrappedtext );
  31
  32                 if( $wgTidyInternal ) {
  33                         $correctedtext = self::execInternalTidy( $wrappedtext );
  34                 } else {
  35                         $correctedtext = self::execExternalTidy( $wrappedtext );
  36                 }
  37                 if( is_null( $correctedtext ) ) {
  38                         wfDebug( "Tidy error detected!\n" );
  39                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
  40                 }
  41
  42                 # Convert the tabs back from entities
  43                 $correctedtext = str_replace( '&#9;', "\t", $correctedtext );
  44
  45                 return $correctedtext;
  46         }
  47
  48         /**
  49          * Check HTML for errors, used if $wgValidateAllHtml = true.
  50          *
  51          * @param $text String
  52          * @param &$errorStr String: return the error string
  53          * @return Boolean: whether the HTML is valid
  54          */
  55         public static function checkErrors( $text, &$errorStr = null ) {
  56                 global $wgTidyInternal;
  57
  58                 $retval = 0;
  59                 if( $wgTidyInternal ) {
  60                         $errorStr = self::execInternalTidy( $text, true, $retval );
  61                 } else {
  62                         $errorStr = self::execExternalTidy( $text, true, $retval );
  63                 }
  64                 return ( $retval < 0 && $errorStr == '' ) || $retval == 0;
  65         }
  66
  67         /**
  68          * Spawn an external HTML tidy process and get corrected markup back from it.
  69          * Also called in OutputHandler.php for full page validation
  70          *
  71          * @param $text String: HTML to check
  72          * @param $stderr Boolean: Whether to read from STDERR rather than STDOUT
  73          * @param &$retval Exit code (-1 on internal error)
  74          * @return mixed String or null
  75          */
  76         private static function execExternalTidy( $text, $stderr = false, &$retval = null ) {
  77                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
  78                 wfProfileIn( __METHOD__ );
  79
  80                 $cleansource = '';
  81                 $opts = ' -utf8';
  82
  83                 if( $stderr ) {
  84                         $descriptorspec = array(
  85                                 0 => array( 'pipe', 'r' ),
  86                                 1 => array( 'file', wfGetNull(), 'a' ),
  87                                 2 => array( 'pipe', 'w' )
  88                         );
  89                 } else {
  90                         $descriptorspec = array(
  91                                 0 => array( 'pipe', 'r' ),
  92                                 1 => array( 'pipe', 'w' ),
  93                                 2 => array( 'file', wfGetNull(), 'a' )
  94                         );
  95                 }
  96
  97                 $readpipe = $stderr ? 2 : 1;
  98                 $pipes = array();
  99
 100                 if( function_exists( 'proc_open' ) ) {
 101                         $process = proc_open( "$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes );
 102                         if ( is_resource( $process ) ) {
 103                                 // Theoretically, this style of communication could cause a deadlock
 104                                 // here. If the stdout buffer fills up, then writes to stdin could
 105                                 // block. This doesn't appear to happen with tidy, because tidy only
 106                                 // writes to stdout after it's finished reading from stdin. Search
 107                                 // for tidyParseStdin and tidySaveStdout in console/tidy.c
 108                                 fwrite( $pipes[0], $text );
 109                                 fclose( $pipes[0] );
 110                                 while ( !feof( $pipes[$readpipe] ) ) {
 111                                         $cleansource .= fgets( $pipes[$readpipe], 1024 );
 112                                 }
 113                                 fclose( $pipes[$readpipe] );
 114                                 $retval = proc_close( $process );
 115                         } else {
 116                                 $retval = -1;
 117                         }
 118                 } else {
 119                         $retval = -1;
 120                 }
 121
 122                 wfProfileOut( __METHOD__ );
 123
 124                 if( !$stderr && $cleansource == '' && $text != '' ) {
 125                         // Some kind of error happened, so we couldn't get the corrected text.
 126                         // Just give up; we'll use the source text and append a warning.
 127                         return null;
 128                 } else {
 129                         return $cleansource;
 130                 }
 131         }
 132
 133         /**
 134          * Use the HTML tidy PECL extension to use the tidy library in-process,
 135          * saving the overhead of spawning a new process.
 136          *
 137          * 'pear install tidy' should be able to compile the extension module.
 138          */
 139         private static function execInternalTidy( $text, $stderr = false, &$retval = null ) {
 140                 global $wgTidyConf, $wgDebugTidy;
 141                 wfProfileIn( __METHOD__ );
 142
 143                 $tidy = new tidy;
 144                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 145
 146                 if( $stderr ) {
 147                         $retval = $tidy->getStatus();
 148                         return $tidy->errorBuffer;
 149                 } else {
 150                         $tidy->cleanRepair();
 151                         $retval = $tidy->getStatus();
 152                         if( $retval == 2 ) {
 153                                 // 2 is magic number for fatal error
 154                                 // http://www.php.net/manual/en/function.tidy-get-status.php
 155                                 $cleansource = null;
 156                         } else {
 157                                 $cleansource = tidy_get_output( $tidy );
 158                         }
 159                         if ( $wgDebugTidy && $retval > 0 ) {
 160                                 $cleansource .= "<!--\nTidy reports:\n" .
 161                                         str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 162                                         "\n-->";
 163                         }
 164
 165                         wfProfileOut( __METHOD__ );
 166                         return $cleansource;
 167                 }
 168         }
 169
 170 }