- Generated on Wed Jun 25 2014 21:01:58 for MediaWiki by
1.7.6.1
|
MediaWiki
REL1_21
|
00001 <?php 00036 class MWTidyWrapper { 00037 00041 protected $mTokens; 00042 00043 protected $mUniqPrefix; 00044 00045 protected $mMarkerIndex; 00046 00047 public function __construct() { 00048 $this->mTokens = null; 00049 $this->mUniqPrefix = null; 00050 } 00051 00056 public function getWrapped( $text ) { 00057 $this->mTokens = new ReplacementArray; 00058 $this->mUniqPrefix = "\x7fUNIQ" . 00059 dechex( mt_rand( 0, 0x7fffffff ) ) . dechex( mt_rand( 0, 0x7fffffff ) ); 00060 $this->mMarkerIndex = 0; 00061 00062 // Replace <mw:editsection> elements with placeholders 00063 $wrappedtext = preg_replace_callback( ParserOutput::EDITSECTION_REGEX, 00064 array( &$this, 'replaceEditSectionLinksCallback' ), $text ); 00065 00066 // Modify inline Microdata <link> and <meta> elements so they say <html-link> and <html-meta> so 00067 // we can trick Tidy into not stripping them out by including them in tidy's new-empty-tags config 00068 $wrappedtext = preg_replace( '!<(link|meta)([^>]*?)(/{0,1}>)!', '<html-$1$2$3', $wrappedtext ); 00069 00070 // Wrap the whole thing in a doctype and body for Tidy. 00071 $wrappedtext = '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN"' . 00072 ' "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd"><html>' . 00073 '<head><title>test</title></head><body>' . $wrappedtext . '</body></html>'; 00074 00075 return $wrappedtext; 00076 } 00077 00083 function replaceEditSectionLinksCallback( $m ) { 00084 $marker = "{$this->mUniqPrefix}-item-{$this->mMarkerIndex}" . Parser::MARKER_SUFFIX; 00085 $this->mMarkerIndex++; 00086 $this->mTokens->setPair( $marker, $m[0] ); 00087 return $marker; 00088 } 00089 00094 public function postprocess( $text ) { 00095 // Revert <html-{link,meta}> back to <{link,meta}> 00096 $text = preg_replace( '!<html-(link|meta)([^>]*?)(/{0,1}>)!', '<$1$2$3', $text ); 00097 00098 // Restore the contents of placeholder tokens 00099 $text = $this->mTokens->replace( $text ); 00100 00101 return $text; 00102 } 00103 00104 } 00105 00115 class MWTidy { 00124 public static function tidy( $text ) { 00125 global $wgTidyInternal; 00126 00127 $wrapper = new MWTidyWrapper; 00128 $wrappedtext = $wrapper->getWrapped( $text ); 00129 00130 $retVal = null; 00131 if ( $wgTidyInternal ) { 00132 $correctedtext = self::execInternalTidy( $wrappedtext, false, $retVal ); 00133 } else { 00134 $correctedtext = self::execExternalTidy( $wrappedtext, false, $retVal ); 00135 } 00136 00137 if ( $retVal < 0 ) { 00138 wfDebug( "Possible tidy configuration error!\n" ); 00139 return $text . "\n<!-- Tidy was unable to run -->\n"; 00140 } elseif ( is_null( $correctedtext ) ) { 00141 wfDebug( "Tidy error detected!\n" ); 00142 return $text . "\n<!-- Tidy found serious XHTML errors -->\n"; 00143 } 00144 00145 $correctedtext = $wrapper->postprocess( $correctedtext ); // restore any hidden tokens 00146 00147 return $correctedtext; 00148 } 00149 00157 public static function checkErrors( $text, &$errorStr = null ) { 00158 global $wgTidyInternal; 00159 00160 $retval = 0; 00161 if( $wgTidyInternal ) { 00162 $errorStr = self::execInternalTidy( $text, true, $retval ); 00163 } else { 00164 $errorStr = self::execExternalTidy( $text, true, $retval ); 00165 } 00166 00167 return ( $retval < 0 && $errorStr == '' ) || $retval == 0; 00168 } 00169 00179 private static function execExternalTidy( $text, $stderr = false, &$retval = null ) { 00180 global $wgTidyConf, $wgTidyBin, $wgTidyOpts; 00181 wfProfileIn( __METHOD__ ); 00182 00183 $cleansource = ''; 00184 $opts = ' -utf8'; 00185 00186 if ( $stderr ) { 00187 $descriptorspec = array( 00188 0 => array( 'pipe', 'r' ), 00189 1 => array( 'file', wfGetNull(), 'a' ), 00190 2 => array( 'pipe', 'w' ) 00191 ); 00192 } else { 00193 $descriptorspec = array( 00194 0 => array( 'pipe', 'r' ), 00195 1 => array( 'pipe', 'w' ), 00196 2 => array( 'file', wfGetNull(), 'a' ) 00197 ); 00198 } 00199 00200 $readpipe = $stderr ? 2 : 1; 00201 $pipes = array(); 00202 00203 $process = proc_open( 00204 "$wgTidyBin -config $wgTidyConf $wgTidyOpts$opts", $descriptorspec, $pipes ); 00205 00206 if ( is_resource( $process ) ) { 00207 // Theoretically, this style of communication could cause a deadlock 00208 // here. If the stdout buffer fills up, then writes to stdin could 00209 // block. This doesn't appear to happen with tidy, because tidy only 00210 // writes to stdout after it's finished reading from stdin. Search 00211 // for tidyParseStdin and tidySaveStdout in console/tidy.c 00212 fwrite( $pipes[0], $text ); 00213 fclose( $pipes[0] ); 00214 while ( !feof( $pipes[$readpipe] ) ) { 00215 $cleansource .= fgets( $pipes[$readpipe], 1024 ); 00216 } 00217 fclose( $pipes[$readpipe] ); 00218 $retval = proc_close( $process ); 00219 } else { 00220 wfWarn( "Unable to start external tidy process" ); 00221 $retval = -1; 00222 } 00223 00224 if ( !$stderr && $cleansource == '' && $text != '' ) { 00225 // Some kind of error happened, so we couldn't get the corrected text. 00226 // Just give up; we'll use the source text and append a warning. 00227 $cleansource = null; 00228 } 00229 00230 wfProfileOut( __METHOD__ ); 00231 return $cleansource; 00232 } 00233 00243 private static function execInternalTidy( $text, $stderr = false, &$retval = null ) { 00244 global $wgTidyConf, $wgDebugTidy; 00245 wfProfileIn( __METHOD__ ); 00246 00247 if ( !MWInit::classExists( 'tidy' ) ) { 00248 wfWarn( "Unable to load internal tidy class." ); 00249 $retval = -1; 00250 00251 wfProfileOut( __METHOD__ ); 00252 return null; 00253 } 00254 00255 $tidy = new tidy; 00256 $tidy->parseString( $text, $wgTidyConf, 'utf8' ); 00257 00258 if ( $stderr ) { 00259 $retval = $tidy->getStatus(); 00260 00261 wfProfileOut( __METHOD__ ); 00262 return $tidy->errorBuffer; 00263 } 00264 00265 $tidy->cleanRepair(); 00266 $retval = $tidy->getStatus(); 00267 if ( $retval == 2 ) { 00268 // 2 is magic number for fatal error 00269 // http://www.php.net/manual/en/function.tidy-get-status.php 00270 $cleansource = null; 00271 } else { 00272 $cleansource = tidy_get_output( $tidy ); 00273 if ( $wgDebugTidy && $retval > 0 ) { 00274 $cleansource .= "<!--\nTidy reports:\n" . 00275 str_replace( '-->', '-->', $tidy->errorBuffer ) . 00276 "\n-->"; 00277 } 00278 } 00279 00280 wfProfileOut( __METHOD__ ); 00281 return $cleansource; 00282 } 00283 }
1.7.6.1