includes/tidy/RaggettExternal.php

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