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         /**
  33          * Get the root element. Simple accessor to $rootElement
  34          */
  35         public function getRootElement() {
  36                 return $this->rootElement;
  37         }
  38
  39         private function run( $fname ) {
  40                 if( $this->softNamespaces ) {
  41                         $parser = xml_parser_create( 'UTF-8' );
  42                 } else {
  43                         $parser = xml_parser_create_ns( 'UTF-8' );
  44                 }
  45
  46                 // case folding violates XML standard, turn it off
  47                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
  48
  49                 xml_set_element_handler( $parser, array( $this, 'elementOpen' ), false );
  50
  51                 $file = fopen( $fname, "rb" );
  52                 do {
  53                         $chunk = fread( $file, 32768 );
  54                         $ret = xml_parse( $parser, $chunk, feof( $file ) );
  55                         if( $ret == 0 ) {
  56                                 // XML isn't well-formed!
  57                                 fclose( $file );
  58                                 xml_parser_free( $parser );
  59                                 return;
  60                         }
  61                 } while( !feof( $file ) );
  62
  63                 $this->wellFormed = true;
  64
  65                 fclose( $file );
  66                 xml_parser_free( $parser );
  67         }
  68
  69         private function elementOpen( $parser, $name, $attribs ) {
  70                 if( $this->softNamespaces ) {
  71                         // Check namespaces manually, so expat doesn't throw
  72                         // errors on use of undeclared namespaces.
  73                         foreach( $attribs as $attrib => $val ) {
  74                                 if( $attrib == 'xmlns' ) {
  75                                         $this->namespaces[''] = $val;
  76                                 } elseif( substr( $attrib, 0, strlen( 'xmlns:' ) ) == 'xmlns:' ) {
  77                                         $this->namespaces[substr( $attrib, strlen( 'xmlns:' ) )] = $val;
  78                                 }
  79                         }
  80
  81                         if( strpos( $name, ':' ) === false ) {
  82                                 $ns = '';
  83                                 $subname = $name;
  84                         } else {
  85                                 list( $ns, $subname ) = explode( ':', $name, 2 );
  86                         }
  87
  88                         if( isset( $this->namespaces[$ns] ) ) {
  89                                 $name = $this->namespaces[$ns] . ':' . $subname;
  90                         } else {
  91                                 // Technically this is invalid for XML with Namespaces.
  92                                 // But..... we'll just let it slide in soft mode.
  93                         }
  94                 }
  95
  96                 // We only need the first open element
  97                 $this->rootElement = $name;
  98                 xml_set_element_handler( $parser, false, false );
  99         }
 100 }