tests/HTMLPurifier/EntityParserTest.php

   1 <?php
   2
   3 class HTMLPurifier_EntityParserTest extends HTMLPurifier_Harness
   4 {
   5
   6     protected $EntityParser;
   7
   8     public function setUp() {
   9         $this->EntityParser = new HTMLPurifier_EntityParser();
  10         $this->_entity_lookup = HTMLPurifier_EntityLookup::instance();
  11     }
  12
  13     function test_substituteNonSpecialEntities() {
  14         $char_theta = $this->_entity_lookup->table['theta'];
  15         $this->assertIdentical($char_theta,
  16             $this->EntityParser->substituteNonSpecialEntities('&theta;') );
  17         $this->assertIdentical('"',
  18             $this->EntityParser->substituteNonSpecialEntities('"') );
  19
  20         // numeric tests, adapted from Feyd
  21         $args = array();
  22         $args[] = array(1114112,false     );
  23         $args[] = array(1114111,'F48FBFBF'); // 0x0010FFFF
  24         $args[] = array(1048576,'F4808080'); // 0x00100000
  25         $args[] = array(1048575,'F3BFBFBF'); // 0x000FFFFF
  26         $args[] = array(262144, 'F1808080'); // 0x00040000
  27         $args[] = array(262143, 'F0BFBFBF'); // 0x0003FFFF
  28         $args[] = array(65536,  'F0908080'); // 0x00010000
  29         $args[] = array(65535,  'EFBFBF'  ); // 0x0000FFFF
  30         $args[] = array(57344,  'EE8080'  ); // 0x0000E000
  31         $args[] = array(57343,  false     ); // 0x0000DFFF  these are ill-formed
  32         $args[] = array(56040,  false     ); // 0x0000DAE8  these are ill-formed
  33         $args[] = array(55296,  false     ); // 0x0000D800  these are ill-formed
  34         $args[] = array(55295,  'ED9FBF'  ); // 0x0000D7FF
  35         $args[] = array(53248,  'ED8080'  ); // 0x0000D000
  36         $args[] = array(53247,  'ECBFBF'  ); // 0x0000CFFF
  37         $args[] = array(4096,   'E18080'  ); // 0x00001000
  38         $args[] = array(4095,   'E0BFBF'  ); // 0x00000FFF
  39         $args[] = array(2048,   'E0A080'  ); // 0x00000800
  40         $args[] = array(2047,   'DFBF'    ); // 0x000007FF
  41         $args[] = array(128,    'C280'    ); // 0x00000080  invalid SGML char
  42         $args[] = array(127,    '7F'      ); // 0x0000007F  invalid SGML char
  43         $args[] = array(0,      '00'      ); // 0x00000000  invalid SGML char
  44
  45         $args[] = array(20108,  'E4BA8C'  ); // 0x00004E8C
  46         $args[] = array(77,     '4D'      ); // 0x0000004D
  47         $args[] = array(66306,  'F0908C82'); // 0x00010302
  48         $args[] = array(1072,   'D0B0'    ); // 0x00000430
  49
  50         foreach ($args as $arg) {
  51             $string = '&#' . $arg[0] . ';' . // decimal
  52                       '&#x' . dechex($arg[0]) . ';'; // hex
  53             $expect = '';
  54             if ($arg[1] !== false) {
  55                 // this is only for PHP 5, the below is PHP 5 and PHP 4
  56                 //$chars = str_split($arg[1], 2);
  57                 $chars = array();
  58                 // strlen must be called in loop because strings size changes
  59                 for ($i = 0; strlen($arg[1]) > $i; $i += 2) {
  60                     $chars[] = $arg[1][$i] . $arg[1][$i+1];
  61                 }
  62                 foreach ($chars as $char) {
  63                     $expect .= chr(hexdec($char));
  64                 }
  65                 $expect .= $expect; // double it
  66             }
  67             $this->assertIdentical(
  68                 $this->EntityParser->substituteNonSpecialEntities($string),
  69                 $expect,
  70                 'Identical expectation [Hex: '. dechex($arg[0]) .']'
  71             );
  72         }
  73
  74     }
  75
  76     function test_substituteSpecialEntities() {
  77         $this->assertIdentical(
  78             "'",
  79             $this->EntityParser->substituteSpecialEntities('&#39;')
  80         );
  81     }
  82
  83 }
  84