includes/XmlTypeCheck.php

   1 <?php
   2
   3 class XmlTypeCheck {
   4         /**
   5          * Will be set to true or false to indicate whether the file is
   6          * well-formed XML. Note that this doesn't check schema validity.
   7          */
   8         public $wellFormed = false;
   9
  10         /**
  11          * Name of the document's root element, including any namespace
  12          * as an expanded URL.
  13          */
  14         public $rootElement = '';
  15
  16         private $softNamespaces;
  17         private $namespaces = array();
  18
  19         /**
  20          * @param $file string filename
  21          * @param $softNamespaces bool
  22          *        If set to true, use of undeclared XML namespaces will be ignored.
  23          *        This matches the behavior of rsvg, but more compliant consumers
  24          *        such as Firefox will reject such files.
  25          *        Leave off for the default, stricter checks.
  26          */
  27         function __construct( $file, $softNamespaces=false ) {
  28                 $this->softNamespaces = $softNamespaces;
  29                 $this->run( $file );
  30         }
  31
  32         private function run( $fname ) {
  33                 if( $this->softNamespaces ) {
  34                         $parser = xml_parser_create( 'UTF-8' );
  35                 } else {
  36                         $parser = xml_parser_create_ns( 'UTF-8' );
  37                 }
  38
  39                 // case folding violates XML standard, turn it off
  40                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
  41
  42                 xml_set_element_handler( $parser, array( $this, 'elementOpen' ), false );
  43
  44                 $file = fopen( $fname, "rb" );
  45                 do {
  46                         $chunk = fread( $file, 32768 );
  47                         $ret = xml_parse( $parser, $chunk, feof( $file ) );
  48                         if( $ret == 0 ) {
  49                                 // XML isn't well-formed!
  50                                 fclose( $file );
  51                                 xml_parser_free( $parser );
  52                                 return;
  53                         }
  54                 } while( !feof( $file ) );
  55
  56                 $this->wellFormed = true;
  57
  58                 fclose( $file );
  59                 xml_parser_free( $parser );
  60         }
  61
  62         private function elementOpen( $parser, $name, $attribs ) {
  63                 if( $this->softNamespaces ) {
  64                         // Check namespaces manually, so expat doesn't throw
  65                         // errors on use of undeclared namespaces.
  66                         foreach( $attribs as $attrib => $val ) {
  67                                 if( $attrib == 'xmlns' ) {
  68                                         $this->namespaces[''] = $val;
  69                                 } elseif( substr( $attrib, 0, strlen( 'xmlns:' ) ) == 'xmlns:' ) {
  70                                         $this->namespaces[substr( $attrib, strlen( 'xmlns:' ) )] = $val;
  71                                 }
  72                         }
  73
  74                         if( strpos( $name, ':' ) === false ) {
  75                                 $ns = '';
  76                                 $subname = $name;
  77                         } else {
  78                                 list( $ns, $subname ) = explode( ':', $name, 2 );
  79                         }
  80
  81                         if( isset( $this->namespaces[$ns] ) ) {
  82                                 $name = $this->namespaces[$ns] . ':' . $subname;
  83                         } else {
  84                                 // Technically this is invalid for XML with Namespaces.
  85                                 // But..... we'll just let it slide in soft mode.
  86                         }
  87                 }
  88
  89                 // We only need the first open element
  90                 $this->rootElement = $name;
  91                 xml_set_element_handler( $parser, false, false );
  92         }
  93 }