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
  44                 $wrappedtext = preg_replace_callback( ParserOutput::EDITSECTION_REGEX,
  45                         array( &$this, 'replaceEditSectionLinksCallback' ), $text );
  46
  47                 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"'.
  48                         ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>'.
  49                         '<head><title>test</title></head><body>'.$wrappedtext.'</body></html>';
  50
  51                 return $wrappedtext;
  52         }
  53
  54         /**
  55          * @param $m array
  56          *
  57          * @return string
  58          */
  59         function replaceEditSectionLinksCallback( $m ) {
  60                 $marker = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}" . Parser::MARKER_SUFFIX;
  61                 $this->mMarkerIndex++;
  62                 $this->mTokens->setPair( $marker, $m[0] );
  63                 return $marker;
  64         }
  65
  66         /**
  67          * @param $text string
  68          * @return string
  69          */
  70         public function postprocess( $text ) {
  71                 return $this->mTokens->replace( $text );
  72         }
  73
  74 }
  75
  76 /**
  77  * Class to interact with HTML tidy
  78  *
  79  * Either the external tidy program or the in-process tidy extension
  80  * will be used depending on availability. Override the default
  81  * $wgTidyInternal setting to disable the internal if it's not working.
  82  *
  83  * @ingroup Parser
  84  */
  85 class MWTidy {
  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                 $retVal = null;
 101                 if ( $wgTidyInternal ) {
 102                         $correctedtext = self::execInternalTidy( $wrappedtext, false, $retVal );
 103                 } else {
 104                         $correctedtext = self::execExternalTidy( $wrappedtext, false, $retVal );
 105                 }
 106
 107                 if ( $retVal < 0 ) {
 108                         wfDebug( "Possible tidy configuration error!\n" );
 109                         return $text . "\n<!-- Tidy was unable to run -->\n";
 110                 } elseif ( is_null( $correctedtext ) ) {
 111                         wfDebug( "Tidy error detected!\n" );
 112                         return $text . "\n<!-- Tidy found serious XHTML errors -->\n";
 113                 }
 114
 115                 $correctedtext = $wrapper->postprocess( $correctedtext ); // restore any hidden tokens
 116
 117                 return $correctedtext;
 118         }
 119
 120         /**
 121          * Check HTML for errors, used if $wgValidateAllHtml = true.
 122          *
 123          * @param $text String
 124          * @param &$errorStr String: return the error string
 125          * @return Boolean: whether the HTML is valid
 126          */
 127         public static function checkErrors( $text, &$errorStr = null ) {
 128                 global $wgTidyInternal;
 129
 130                 $retval = 0;
 131                 if( $wgTidyInternal ) {
 132                         $errorStr = self::execInternalTidy( $text, true, $retval );
 133                 } else {
 134                         $errorStr = self::execExternalTidy( $text, true, $retval );
 135                 }
 136
 137                 return ( $retval < 0 && $errorStr == '' ) || $retval == 0;
 138         }
 139
 140         /**
 141          * Spawn an external HTML tidy process and get corrected markup back from it.
 142          * Also called in OutputHandler.php for full page validation
 143          *
 144          * @param $text String: HTML to check
 145          * @param $stderr Boolean: Whether to read result from STDERR rather than STDOUT
 146          * @param &$retval int Exit code (-1 on internal error)
 147          * @return mixed String or null
 148          */
 149         private static function execExternalTidy( $text, $stderr = false, &$retval = null ) {
 150                 global $wgTidyConf, $wgTidyBin, $wgTidyOpts;
 151                 wfProfileIn( __METHOD__ );
 152
 153                 $cleansource = '';
 154                 $opts = ' -utf8';
 155
 156                 if ( $stderr ) {
 157                         $descriptorspec = array(
 158                                 0 => array( 'pipe', 'r' ),
 159                                 1 => array( 'file', wfGetNull(), 'a' ),
 160                                 2 => array( 'pipe', 'w' )
 161                         );
 162                 } else {
 163                         $descriptorspec = array(
 164                                 0 => array( 'pipe', 'r' ),
 165                                 1 => array( 'pipe', 'w' ),
 166                                 2 => array( 'file', wfGetNull(), 'a' )
 167                         );
 168                 }
 169
 170                 $readpipe = $stderr ? 2 : 1;
 171                 $pipes = array();
 172
 173                 $process = proc_open(
 174                         "$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes );
 175
 176                 if ( is_resource( $process ) ) {
 177                         // Theoretically, this style of communication could cause a deadlock
 178                         // here. If the stdout buffer fills up, then writes to stdin could
 179                         // block. This doesn't appear to happen with tidy, because tidy only
 180                         // writes to stdout after it's finished reading from stdin. Search
 181                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
 182                         fwrite( $pipes[0], $text );
 183                         fclose( $pipes[0] );
 184                         while ( !feof( $pipes[$readpipe] ) ) {
 185                                 $cleansource .= fgets( $pipes[$readpipe], 1024 );
 186                         }
 187                         fclose( $pipes[$readpipe] );
 188                         $retval = proc_close( $process );
 189                 } else {
 190                         wfWarn( "Unable to start external tidy process" );
 191                         $retval = -1;
 192                 }
 193
 194                 if ( !$stderr && $cleansource == '' && $text != '' ) {
 195                         // Some kind of error happened, so we couldn't get the corrected text.
 196                         // Just give up; we'll use the source text and append a warning.
 197                         $cleansource = null;
 198                 }
 199
 200                 wfProfileOut( __METHOD__ );
 201                 return $cleansource;
 202         }
 203
 204         /**
 205          * Use the HTML tidy extension to use the tidy library in-process,
 206          * saving the overhead of spawning a new process.
 207          *
 208          * @param $text String: HTML to check
 209          * @param $stderr Boolean: Whether to read result from error status instead of output
 210          * @param &$retval int Exit code (-1 on internal error)
 211          * @return mixed String or null
 212          */
 213         private static function execInternalTidy( $text, $stderr = false, &$retval = null ) {
 214                 global $wgTidyConf, $wgDebugTidy;
 215                 wfProfileIn( __METHOD__ );
 216
 217                 if ( !MWInit::classExists( 'tidy' ) ) {
 218                         wfWarn( "Unable to load internal tidy class." );
 219                         $retval = -1;
 220
 221                         wfProfileOut( __METHOD__ );
 222                         return null;
 223                 }
 224
 225                 $tidy = new tidy;
 226                 $tidy->parseString( $text, $wgTidyConf, 'utf8' );
 227
 228                 if ( $stderr ) {
 229                         $retval = $tidy->getStatus();
 230
 231                         wfProfileOut( __METHOD__ );
 232                         return $tidy->errorBuffer;
 233                 } else {
 234                         $tidy->cleanRepair();
 235                         $retval = $tidy->getStatus();
 236                         if ( $retval == 2 ) {
 237                                 // 2 is magic number for fatal error
 238                                 // http://www.php.net/manual/en/function.tidy-get-status.php
 239                                 $cleansource = null;
 240                         } else {
 241                                 $cleansource = tidy_get_output( $tidy );
 242                                 if ( $wgDebugTidy && $retval > 0 ) {
 243                                         $cleansource .= "<!--\nTidy reports:\n" .
 244                                                 str_replace( '-->', '--&gt;', $tidy->errorBuffer ) .
 245                                                 "\n-->";
 246                                 }
 247                         }
 248
 249                         wfProfileOut( __METHOD__ );
 250                         return $cleansource;
 251                 }
 252         }
 253 }