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          * Will be set to true if the optional element filter returned
  12          * a match at some point.
  13          */
  14         public $filterMatch = false;
  15
  16         /**
  17          * Name of the document's root element, including any namespace
  18          * as an expanded URL.
  19          */
  20         public $rootElement = '';
  21
  22         /**
  23          * @param $file string filename
  24          * @param $filterCallback callable (optional)
  25          *        Function to call to do additional custom validity checks from the
  26          *        SAX element handler event. This gives you access to the element
  27          *        namespace, name, and attributes, but not to text contents.
  28          *        Filter should return 'true' to toggle on $this->filterMatch
  29          */
  30         function __construct( $file, $filterCallback=null ) {
  31                 $this->filterCallback = $filterCallback;
  32                 $this->run( $file );
  33         }
  34
  35         /**
  36          * Get the root element. Simple accessor to $rootElement
  37          */
  38         public function getRootElement() {
  39                 return $this->rootElement;
  40         }
  41
  42         private function run( $fname ) {
  43                 $parser = xml_parser_create_ns( 'UTF-8' );
  44
  45                 // case folding violates XML standard, turn it off
  46                 xml_parser_set_option( $parser, XML_OPTION_CASE_FOLDING, false );
  47
  48                 xml_set_element_handler( $parser, array( $this, 'rootElementOpen' ), false );
  49
  50                 $file = fopen( $fname, "rb" );
  51                 do {
  52                         $chunk = fread( $file, 32768 );
  53                         $ret = xml_parse( $parser, $chunk, feof( $file ) );
  54                         if( $ret == 0 ) {
  55                                 // XML isn't well-formed!
  56                                 fclose( $file );
  57                                 xml_parser_free( $parser );
  58                                 return;
  59                         }
  60                 } while( !feof( $file ) );
  61
  62                 $this->wellFormed = true;
  63
  64                 fclose( $file );
  65                 xml_parser_free( $parser );
  66         }
  67
  68         private function rootElementOpen( $parser, $name, $attribs ) {
  69                 $this->rootElement = $name;
  70
  71                 if( is_callable( $this->filterCallback ) ) {
  72                         xml_set_element_handler( $parser, array( $this, 'elementOpen' ), false );
  73                         $this->elementOpen( $parser, $name, $attribs );
  74                 } else {
  75                         // We only need the first open element
  76                         xml_set_element_handler( $parser, false, false );
  77                 }
  78         }
  79
  80         private function elementOpen( $parser, $name, $attribs ) {
  81                 if( call_user_func( $this->filterCallback, $name, $attribs ) ) {
  82                         // Filter hit!
  83                         $this->filterMatch = true;
  84                 }
  85         }
  86 }