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     {
  10         $this->EntityParser = new HTMLPurifier_EntityParser();
  11         $this->_entity_lookup = HTMLPurifier_EntityLookup::instance();
  12     }
  13
  14     public function test_substituteNonSpecialEntities()
  15     {
  16         $char_theta = $this->_entity_lookup->table['theta'];
  17         $this->assertIdentical($char_theta,
  18             $this->EntityParser->substituteNonSpecialEntities('&theta;') );
  19         $this->assertIdentical($char_theta,
  20             $this->EntityParser->substituteTextEntities('&theta;') );
  21         $this->assertIdentical('"',
  22             $this->EntityParser->substituteNonSpecialEntities('"') );
  23         $this->assertIdentical('"',
  24             $this->EntityParser->substituteTextEntities('"') );
  25
  26         // numeric tests, adapted from Feyd
  27         $args = array();
  28         $args[] = array(1114112,false     );
  29         $args[] = array(1114111,'F48FBFBF'); // 0x0010FFFF
  30         $args[] = array(1048576,'F4808080'); // 0x00100000
  31         $args[] = array(1048575,'F3BFBFBF'); // 0x000FFFFF
  32         $args[] = array(262144, 'F1808080'); // 0x00040000
  33         $args[] = array(262143, 'F0BFBFBF'); // 0x0003FFFF
  34         $args[] = array(65536,  'F0908080'); // 0x00010000
  35         $args[] = array(65535,  'EFBFBF'  ); // 0x0000FFFF
  36         $args[] = array(57344,  'EE8080'  ); // 0x0000E000
  37         $args[] = array(57343,  false     ); // 0x0000DFFF  these are ill-formed
  38         $args[] = array(56040,  false     ); // 0x0000DAE8  these are ill-formed
  39         $args[] = array(55296,  false     ); // 0x0000D800  these are ill-formed
  40         $args[] = array(55295,  'ED9FBF'  ); // 0x0000D7FF
  41         $args[] = array(53248,  'ED8080'  ); // 0x0000D000
  42         $args[] = array(53247,  'ECBFBF'  ); // 0x0000CFFF
  43         $args[] = array(4096,   'E18080'  ); // 0x00001000
  44         $args[] = array(4095,   'E0BFBF'  ); // 0x00000FFF
  45         $args[] = array(2048,   'E0A080'  ); // 0x00000800
  46         $args[] = array(2047,   'DFBF'    ); // 0x000007FF
  47         $args[] = array(128,    'C280'    ); // 0x00000080  invalid SGML char
  48         $args[] = array(127,    '7F'      ); // 0x0000007F  invalid SGML char
  49         $args[] = array(0,      '00'      ); // 0x00000000  invalid SGML char
  50
  51         $args[] = array(20108,  'E4BA8C'  ); // 0x00004E8C
  52         $args[] = array(77,     '4D'      ); // 0x0000004D
  53         $args[] = array(66306,  'F0908C82'); // 0x00010302
  54         $args[] = array(1072,   'D0B0'    ); // 0x00000430
  55
  56         foreach ($args as $arg) {
  57             $string = '&#' . $arg[0] . ';' . // decimal
  58                       '&#x' . dechex($arg[0]) . ';'; // hex
  59             $expect = '';
  60             if ($arg[1] !== false) {
  61                 // this is only for PHP 5, the below is PHP 5 and PHP 4
  62                 //$chars = str_split($arg[1], 2);
  63                 $chars = array();
  64                 // strlen must be called in loop because strings size changes
  65                 for ($i = 0; strlen($arg[1]) > $i; $i += 2) {
  66                     $chars[] = $arg[1][$i] . $arg[1][$i+1];
  67                 }
  68                 foreach ($chars as $char) {
  69                     $expect .= chr(hexdec($char));
  70                 }
  71                 $expect .= $expect; // double it
  72             }
  73             $this->assertIdentical(
  74                 $this->EntityParser->substituteNonSpecialEntities($string),
  75                 $expect,
  76                 'Identical expectation [Hex: '. dechex($arg[0]) .']'
  77             );
  78             $this->assertIdentical(
  79                 $this->EntityParser->substituteTextEntities($string),
  80                 $expect,
  81                 'Identical expectation [Hex: '. dechex($arg[0]) .']'
  82             );
  83         }
  84
  85     }
  86
  87     public function test_substituteSpecialEntities()
  88     {
  89         $this->assertIdentical(
  90             "'",
  91             $this->EntityParser->substituteSpecialEntities('&#39;')
  92         );
  93         $this->assertIdentical(
  94             "'",
  95             $this->EntityParser->substituteTextEntities('&#39;')
  96         );
  97     }
  98
  99 }
 100
 101 // vim: et sw=4 sts=4