tests/HTMLPurifier/EncoderTest.php

   1 <?php
   2
   3 class HTMLPurifier_EncoderTest extends HTMLPurifier_Harness
   4 {
   5
   6     protected $_entity_lookup;
   7
   8     function setUp() {
   9         $this->_entity_lookup = HTMLPurifier_EntityLookup::instance();
  10         parent::setUp();
  11     }
  12
  13     function assertCleanUTF8($string, $expect = null) {
  14         if ($expect === null) $expect = $string;
  15         $this->assertIdentical(HTMLPurifier_Encoder::cleanUTF8($string), $expect, 'iconv: %s');
  16         $this->assertIdentical(HTMLPurifier_Encoder::cleanUTF8($string, true), $expect, 'PHP: %s');
  17     }
  18
  19     function test_cleanUTF8() {
  20         $this->assertCleanUTF8('Normal string.');
  21         $this->assertCleanUTF8("Test\tAllowed\nControl\rCharacters");
  22         $this->assertCleanUTF8("null byte: \0", 'null byte: ');
  23         $this->assertCleanUTF8("\1\2\3\4\5\6\7", '');
  24         $this->assertCleanUTF8("\x7F", ''); // one byte invalid SGML char
  25         $this->assertCleanUTF8("\xC2\x80", ''); // two byte invalid SGML
  26         $this->assertCleanUTF8("\xF3\xBF\xBF\xBF"); // valid four byte
  27         $this->assertCleanUTF8("\xDF\xFF", ''); // malformed UTF8
  28         // invalid codepoints
  29         $this->assertCleanUTF8("\xED\xB0\x80", '');
  30     }
  31
  32     function test_convertToUTF8_noConvert() {
  33         // UTF-8 means that we don't touch it
  34         $this->assertIdentical(
  35             HTMLPurifier_Encoder::convertToUTF8("\xF6", $this->config, $this->context),
  36             "\xF6", // this is invalid
  37             'Expected identical [Binary: F6]'
  38         );
  39     }
  40
  41     function test_convertToUTF8_spuriousEncoding() {
  42         // We don't support this as UTF-8, because UTF-8 is the default and
  43         // shouldn't be set if not necessary.
  44         $this->config->set('Core', 'Encoding', 'utf8');
  45         $this->expectError('Invalid encoding utf8');
  46         $this->assertIdentical(
  47             HTMLPurifier_Encoder::convertToUTF8("\xF6", $this->config, $this->context),
  48             ''
  49         );
  50     }
  51
  52     function test_convertToUTF8_iso8859_1() {
  53         $this->config->set('Core', 'Encoding', 'ISO-8859-1');
  54         $this->assertIdentical(
  55             HTMLPurifier_Encoder::convertToUTF8("\xF6", $this->config, $this->context),
  56             "\xC3\xB6"
  57         );
  58     }
  59
  60     function test_convertToUTF8_withoutIconv() {
  61         $this->config->set('Core', 'Encoding', 'ISO-8859-1');
  62         $this->config->set('Test', 'ForceNoIconv', true);
  63         $this->assertIdentical(
  64             HTMLPurifier_Encoder::convertToUTF8("\xF6", $this->config, $this->context),
  65             "\xC3\xB6"
  66         );
  67
  68     }
  69
  70     function getZhongWen() {
  71         return "\xE4\xB8\xAD\xE6\x96\x87 (Chinese)";
  72     }
  73
  74     function test_convertFromUTF8_utf8() {
  75         // UTF-8 means that we don't touch it
  76         $this->assertIdentical(
  77             HTMLPurifier_Encoder::convertFromUTF8("\xC3\xB6", $this->config, $this->context),
  78             "\xC3\xB6"
  79         );
  80     }
  81
  82     function test_convertFromUTF8_iso8859_1() {
  83         $this->config->set('Core', 'Encoding', 'ISO-8859-1');
  84         $this->assertIdentical(
  85             HTMLPurifier_Encoder::convertFromUTF8("\xC3\xB6", $this->config, $this->context),
  86             "\xF6",
  87             'Expected identical [Binary: F6]'
  88         );
  89     }
  90
  91     function test_convertFromUTF8_iconvNoChars() {
  92         if (!function_exists('iconv')) return;
  93         $this->config->set('Core', 'Encoding', 'ISO-8859-1');
  94         $this->assertIdentical(
  95             HTMLPurifier_Encoder::convertFromUTF8($this->getZhongWen(), $this->config, $this->context),
  96             " (Chinese)"
  97         );
  98     }
  99
 100     function test_convertFromUTF8_phpNormal() {
 101         // Plain PHP implementation has slightly different behavior
 102         $this->config->set('Core', 'Encoding', 'ISO-8859-1');
 103         $this->config->set('Test', 'ForceNoIconv', true);
 104         $this->assertIdentical(
 105             HTMLPurifier_Encoder::convertFromUTF8("\xC3\xB6", $this->config, $this->context),
 106             "\xF6",
 107             'Expected identical [Binary: F6]'
 108         );
 109     }
 110
 111     function test_convertFromUTF8_phpNoChars() {
 112         $this->config->set('Core', 'Encoding', 'ISO-8859-1');
 113         $this->config->set('Test', 'ForceNoIconv', true);
 114         $this->assertIdentical(
 115             HTMLPurifier_Encoder::convertFromUTF8($this->getZhongWen(), $this->config, $this->context),
 116             "?? (Chinese)"
 117         );
 118     }
 119
 120     function test_convertFromUTF8_withProtection() {
 121         // Preserve the characters!
 122         $this->config->set('Core', 'Encoding', 'ISO-8859-1');
 123         $this->config->set('Core', 'EscapeNonASCIICharacters', true);
 124         $this->assertIdentical(
 125             HTMLPurifier_Encoder::convertFromUTF8($this->getZhongWen(), $this->config, $this->context),
 126             "&#20013;&#25991; (Chinese)"
 127         );
 128
 129     }
 130
 131     function test_convertToASCIIDumbLossless() {
 132
 133         // Uppercase thorn letter
 134         $this->assertIdentical(
 135             HTMLPurifier_Encoder::convertToASCIIDumbLossless("\xC3\x9Eorn"),
 136             "&#222;orn"
 137         );
 138
 139         $this->assertIdentical(
 140             HTMLPurifier_Encoder::convertToASCIIDumbLossless("an"),
 141             "an"
 142         );
 143
 144         // test up to four bytes
 145         $this->assertIdentical(
 146             HTMLPurifier_Encoder::convertToASCIIDumbLossless("\xF3\xA0\x80\xA0"),
 147             "&#917536;"
 148         );
 149
 150     }
 151
 152     function assertASCIISupportCheck($enc, $ret) {
 153         $test = HTMLPurifier_Encoder::testEncodingSupportsASCII($enc, true);
 154         if ($test === false) return;
 155         $this->assertIdentical(
 156             HTMLPurifier_Encoder::testEncodingSupportsASCII($enc),
 157             $ret
 158         );
 159         $this->assertIdentical(
 160             HTMLPurifier_Encoder::testEncodingSupportsASCII($enc, true),
 161             $ret
 162         );
 163     }
 164
 165     function test_testEncodingSupportsASCII() {
 166         $this->assertASCIISupportCheck('Shift_JIS', array("\xC2\xA5" => '\\', "\xE2\x80\xBE" => '~'));
 167         $this->assertASCIISupportCheck('JOHAB', array("\xE2\x82\xA9" => '\\'));
 168         $this->assertASCIISupportCheck('ISO-8859-1', array());
 169         $this->assertASCIISupportCheck('dontexist', array()); // canary
 170     }
 171
 172     function testShiftJIS() {
 173         if (!function_exists('iconv')) return;
 174         $this->config->set('Core', 'Encoding', 'Shift_JIS');
 175         // This actually looks like a Yen, but we're going to treat it differently
 176         $this->assertIdentical(
 177             HTMLPurifier_Encoder::convertFromUTF8('\\~', $this->config, $this->context),
 178             '\\~'
 179         );
 180         $this->assertIdentical(
 181             HTMLPurifier_Encoder::convertToUTF8('\\~', $this->config, $this->context),
 182             '\\~'
 183         );
 184     }
 185
 186 }
 187