includes/parser/Tidy.php

   1 <?php
   2 /**
   3  * HTML validation and correction
   4  *
   5  * @file
   6  */
   7
   8 /**
   9  * Class used to hide mw:editsection tokens from Tidy so that it doesn't break them
  10  * or break on them. This is a bit of a hack for now, but hopefully in the future
  11  * we may create a real postprocessor or something that will replace this.
  12  * It's called wrapper because for now it basically takes over MWTidy::tidy's task
  13  * of wrapping the text in a xhtml block
  14  *
  15  * This re-uses some of the parser's UNIQ tricks, though some of it is private so it's
  16  * duplicated. Perhaps we should create an abstract marker hiding class.
  17  */
  18 class MWTidyWrapper {
  19
  20         /**
  21          * @var ReplacementArray
  22          */
  23         protected $mTokens;
  24
  25         protected $mUniqPrefix;
  26
  27         protected $mMarkerIndex;
  28
  29         public function __construct() {
  30                 $this->mTokens = null;
  31                 $this->mUniqPrefix = null;
  32         }
  33
  34         /**
  35          * @param $text string
  36          * @return string
  37          */
  38         public function getWrapped( $text ) {
  39                 $this->mTokens = new ReplacementArray;
  40                 $this->mUniqPrefix = "\x7fUNIQ" .
  41                         dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) );
  42                 $this->mMarkerIndex = 0;
  43                 $wrappedtext = preg_replace_callback( ParserOutput::EDITSECTION_REGEX,
  44                         array( &$this, 'replaceEditSectionLinksCallback' ), $text );
  45
  46                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
  47                         ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
  48                         '<head><title>test</title></head><body>'.$wrappedtext.'</body></html>';
  49
  50                 return $wrappedtext;
  51         }
  52
  53         /**
  54          * @param $m array
  55          *
  56          * @return string
  57          */
  58         function replaceEditSectionLinksCallback( $m ) {
  59                 $marker = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}" . Parser::MARKER_SUFFIX;
  60                 $this->mMarkerIndex++;
  61                 $this->mTokens->setPair( $marker, $m[0] );
  62                 return $marker;
  63         }
  64
  65         /**
  66          * @param $text string
  67          * @return string
  68          */
  69         public function postprocess( $text ) {
  70                 return $this->mTokens->replace( $text );
  71         }
  72
  73 }
  74
  75 /**
  76  * Class to interact with HTML tidy
  77  *
  78  * Either the external tidy program or the in-process tidy extension
  79  * will be used depending on availability. Override the default
  80  * $wgTidyInternal setting to disable the internal if it's not working.
  81  *
  82  * @ingroup Parser
  83  */
  84 class MWTidy {
  85
  86         /**
  87          * Interface with html tidy, used if $wgUseTidy = true.
  88          * If tidy isn't able to correct the markup, the original will be
  89          * returned in all its glory with a warning comment appended.
  90          *
  91          * @param $text String: hideous HTML input
  92          * @return String: corrected HTML output
  93          */
  94         public static function tidy( $text ) {
  95                 global $wgTidyInternal;
  96
  97                 $wrapper = new MWTidyWrapper;
  98                 $wrappedtext = $wrapper->getWrapped( $text );
  99
 100                 if( $wgTidyInternal ) {
 101                         $correctedtext = self::execInternalTidy( $wrappedtext );
 102                 } else {
 103                         $correctedtext = self::execExternalTidy( $wrappedtext );
 104                 }
 105                 if( is_null( $correctedtext ) ) {
 106                         wfDebug( "Tidy error detected!\n" );
 107                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 108                 }
 109
 110                 $correctedtext = $wrapper->postprocess( $correctedtext ); // restore any hidden tokens
 111
 112                 return $correctedtext;
 113         }
 114
 115         function replaceEditSectionLinksCallback( $m ) {
 116
 117         }
 118
 119         /**
 120          * Check HTML for errors, used if $wgValidateAllHtml = true.
 121          *
 122          * @param $text String
 123          * @param &$errorStr String: return the error string
 124          * @return Boolean: whether the HTML is valid
 125          */
 126         public static function checkErrors( $text, &$errorStr = null ) {
 127                 global $wgTidyInternal;
 128
 129                 $retval = 0;
 130                 if( $wgTidyInternal ) {
 131                         $errorStr = self::execInternalTidy( $text, true, $retval );
 132                 } else {
 133                         $errorStr = self::execExternalTidy( $text, true, $retval );
 134                 }
 135                 return ( $retval < 0 && $errorStr == '' ) || $retval == 0;
 136         }
 137
 138         /**
 139          * Spawn an external HTML tidy process and get corrected markup back from it.
 140          * Also called in OutputHandler.php for full page validation
 141          *
 142          * @param $text String: HTML to check
 143          * @param $stderr Boolean: Whether to read from STDERR rather than STDOUT
 144          * @param &$retval Exit code (-1 on internal error)
 145          * @return mixed String or null
 146          */
 147         private static function execExternalTidy( $text, $stderr = false, &$retval = null ) {
 148                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 149                 wfProfileIn( __METHOD__ );
 150
 151                 $cleansource = '';
 152                 $opts = ' -utf8';
 153
 154                 if( $stderr ) {
 155                         $descriptorspec = array(
 156                                 0 => array( 'pipe', 'r' ),
 157                                 1 => array( 'file', wfGetNull(), 'a' ),
 158                                 2 => array( 'pipe', 'w' )
 159                         );
 160                 } else {
 161                         $descriptorspec = array(
 162                                 0 => array( 'pipe', 'r' ),
 163                                 1 => array( 'pipe', 'w' ),
 164                                 2 => array( 'file', wfGetNull(), 'a' )
 165                         );
 166                 }
 167
 168                 $readpipe = $stderr ? 2 : 1;
 169                 $pipes = array();
 170
 171                 if( function_exists( 'proc_open' ) ) {
 172                         $process = proc_open( "$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes );
 173                         if ( is_resource( $process ) ) {
 174                                 // Theoretically, this style of communication could cause a deadlock
 175                                 // here. If the stdout buffer fills up, then writes to stdin could
 176                                 // block. This doesn't appear to happen with tidy, because tidy only
 177                                 // writes to stdout after it's finished reading from stdin. Search
 178                                 // for tidyParseStdin and tidySaveStdout in console/tidy.c
 179                                 fwrite( $pipes[0], $text );
 180                                 fclose( $pipes[0] );
 181                                 while ( !feof( $pipes[$readpipe] ) ) {
 182                                         $cleansource .= fgets( $pipes[$readpipe], 1024 );
 183                                 }
 184                                 fclose( $pipes[$readpipe] );
 185                                 $retval = proc_close( $process );
 186                         } else {
 187                                 $retval = -1;
 188                         }
 189                 } else {
 190                         $retval = -1;
 191                 }
 192
 193                 if( !$stderr && $cleansource == '' && $text != '' ) {
 194                         // Some kind of error happened, so we couldn't get the corrected text.
 195                         // Just give up; we'll use the source text and append a warning.
 196                         $cleansource = null;
 197                 }
 198                 wfProfileOut( __METHOD__ );
 199                 return $cleansource;
 200         }
 201
 202         /**
 203          * Use the HTML tidy PECL extension to use the tidy library in-process,
 204          * saving the overhead of spawning a new process.
 205          *
 206          * 'pear install tidy' should be able to compile the extension module.
 207          *
 208          * @param $text
 209          * @param $stderr
 210          * @param $retval
 211          *
 212          * @return string
 213          */
 214         private static function execInternalTidy( $text, $stderr = false, &$retval = null ) {
 215                 global $wgTidyConf, $wgDebugTidy;
 216                 wfProfileIn( __METHOD__ );
 217
 218                 $tidy = new tidy;
 219                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 220
 221                 if( $stderr ) {
 222                         $retval = $tidy->getStatus();
 223                         wfProfileOut( __METHOD__ );
 224                         return $tidy->errorBuffer;
 225                 } else {
 226                         $tidy->cleanRepair();
 227                         $retval = $tidy->getStatus();
 228                         if( $retval == 2 ) {
 229                                 // 2 is magic number for fatal error
 230                                 // http://www.php.net/manual/en/function.tidy-get-status.php
 231                                 $cleansource = null;
 232                         } else {
 233                                 $cleansource = tidy_get_output( $tidy );
 234                         }
 235                         if ( $wgDebugTidy && $retval > 0 ) {
 236                                 $cleansource .= "<!--\nTidy reports:\n" .
 237                                         str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 238                                         "\n-->";
 239                         }
 240
 241                         wfProfileOut( __METHOD__ );
 242                         return $cleansource;
 243                 }
 244         }
 245
 246 }