includes/tidy/RaggettExternal.php

   1 <?php
   2
   3 namespace MediaWiki\Tidy;
   4
   5 /**
   6  * @deprecated since 1.32, use RemexDriver
   7  */
   8 class RaggettExternal extends RaggettBase {
   9         /**
  10          * Spawn an external HTML tidy process and get corrected markup back from it.
  11          * Also called in OutputHandler.php for full page validation
  12          *
  13          * @param string $text HTML to check
  14          * @param bool $stderr Whether to read result from STDERR rather than STDOUT
  15          * @param int|null &$retval Exit code (-1 on internal error)
  16          * @return string|null
  17          */
  18         protected function cleanWrapped( $text, $stderr = false, &$retval = null ) {
  19                 $cleansource = '';
  20                 $opts = ' -utf8';
  21
  22                 if ( $stderr ) {
  23                         $descriptorspec = [
  24                                 0 => [ 'pipe', 'r' ],
  25                                 1 => [ 'file', wfGetNull(), 'a' ],
  26                                 2 => [ 'pipe', 'w' ]
  27                         ];
  28                 } else {
  29                         $descriptorspec = [
  30                                 0 => [ 'pipe', 'r' ],
  31                                 1 => [ 'pipe', 'w' ],
  32                                 2 => [ 'file', wfGetNull(), 'a' ]
  33                         ];
  34                 }
  35
  36                 $readpipe = $stderr ? 2 : 1;
  37                 $pipes = [];
  38
  39                 $process = proc_open(
  40                         "{$this->config['tidyBin']} -config {$this->config['tidyConfigFile']} " .
  41                         $this->config['tidyCommandLine'] . $opts, $descriptorspec, $pipes );
  42
  43                 // NOTE: At least on linux, the process will be created even if tidy is not installed.
  44                 //      This means that missing tidy will be treated as a validation failure.
  45
  46                 if ( is_resource( $process ) ) {
  47                         // Theoretically, this style of communication could cause a deadlock
  48                         // here. If the stdout buffer fills up, then writes to stdin could
  49                         // block. This doesn't appear to happen with tidy, because tidy only
  50                         // writes to stdout after it's finished reading from stdin. Search
  51                         // for tidyParseStdin and tidySaveStdout in console/tidy.c
  52                         fwrite( $pipes[0], $text );
  53                         fclose( $pipes[0] );
  54                         while ( !feof( $pipes[$readpipe] ) ) {
  55                                 $cleansource .= fgets( $pipes[$readpipe], 1024 );
  56                         }
  57                         fclose( $pipes[$readpipe] );
  58                         $retval = proc_close( $process );
  59                 } else {
  60                         wfWarn( "Unable to start external tidy process" );
  61                         $retval = -1;
  62                 }
  63
  64                 if ( !$stderr && $cleansource == '' && $text != '' ) {
  65                         // Some kind of error happened, so we couldn't get the corrected text.
  66                         // Just give up; we'll use the source text and append a warning.
  67                         $cleansource = null;
  68                 }
  69
  70                 return $cleansource;
  71         }
  72
  73         public function supportsValidate() {
  74                 return true;
  75         }
  76 }