tests/phpunit/includes/libs/StringUtilsTest.php

   1 <?php
   2
   3 class StringUtilsTest extends PHPUnit_Framework_TestCase {
   4
   5         /**
   6          * @covers StringUtils::isUtf8
   7          * @dataProvider provideStringsForIsUtf8Check
   8          */
   9         public function testIsUtf8( $expected, $string ) {
  10                 $this->assertEquals( $expected, StringUtils::isUtf8( $string ),
  11                         'Testing string "' . $this->escaped( $string ) . '"' );
  12         }
  13
  14         /**
  15          * Print high range characters as a hexadecimal
  16          * @param string $string
  17          * @return string
  18          */
  19         function escaped( $string ) {
  20                 $escaped = '';
  21                 $length = strlen( $string );
  22                 for ( $i = 0; $i < $length; $i++ ) {
  23                         $char = $string[$i];
  24                         $val = ord( $char );
  25                         if ( $val > 127 ) {
  26                                 $escaped .= '\x' . dechex( $val );
  27                         } else {
  28                                 $escaped .= $char;
  29                         }
  30                 }
  31
  32                 return $escaped;
  33         }
  34
  35         /**
  36          * See also "UTF-8 decoder capability and stress test" by
  37          * Markus Kuhn:
  38          * http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt
  39          */
  40         public static function provideStringsForIsUtf8Check() {
  41                 // Expected return values for StringUtils::isUtf8()
  42                 $PASS = true;
  43                 $FAIL = false;
  44
  45                 return [
  46                         'some ASCII' => [ $PASS, 'Some ASCII' ],
  47                         'euro sign' => [ $PASS, "Euro sign €" ],
  48
  49                         'first possible sequence 1 byte' => [ $PASS, "\x00" ],
  50                         'first possible sequence 2 bytes' => [ $PASS, "\xc2\x80" ],
  51                         'first possible sequence 3 bytes' => [ $PASS, "\xe0\xa0\x80" ],
  52                         'first possible sequence 4 bytes' => [ $PASS, "\xf0\x90\x80\x80" ],
  53                         'first possible sequence 5 bytes' => [ $FAIL, "\xf8\x88\x80\x80\x80" ],
  54                         'first possible sequence 6 bytes' => [ $FAIL, "\xfc\x84\x80\x80\x80\x80" ],
  55
  56                         'last possible sequence 1 byte' => [ $PASS, "\x7f" ],
  57                         'last possible sequence 2 bytes' => [ $PASS, "\xdf\xbf" ],
  58                         'last possible sequence 3 bytes' => [ $PASS, "\xef\xbf\xbf" ],
  59                         'last possible sequence 4 bytes (U+1FFFFF)' => [ $FAIL, "\xf7\xbf\xbf\xbf" ],
  60                         'last possible sequence 5 bytes' => [ $FAIL, "\xfb\xbf\xbf\xbf\xbf" ],
  61                         'last possible sequence 6 bytes' => [ $FAIL, "\xfd\xbf\xbf\xbf\xbf\xbf" ],
  62
  63                         'boundary 1' => [ $PASS, "\xed\x9f\xbf" ],
  64                         'boundary 2' => [ $PASS, "\xee\x80\x80" ],
  65                         'boundary 3' => [ $PASS, "\xef\xbf\xbd" ],
  66                         'boundary 4' => [ $PASS, "\xf2\x80\x80\x80" ],
  67                         'boundary 5 (U+FFFFF)' => [ $PASS, "\xf3\xbf\xbf\xbf" ],
  68                         'boundary 6 (U+100000)' => [ $PASS, "\xf4\x80\x80\x80" ],
  69                         'boundary 7 (U+10FFFF)' => [ $PASS, "\xf4\x8f\xbf\xbf" ],
  70                         'boundary 8 (U+110000)' => [ $FAIL, "\xf4\x90\x80\x80" ],
  71
  72                         'malformed 1' => [ $FAIL, "\x80" ],
  73                         'malformed 2' => [ $FAIL, "\xbf" ],
  74                         'malformed 3' => [ $FAIL, "\x80\xbf" ],
  75                         'malformed 4' => [ $FAIL, "\x80\xbf\x80" ],
  76                         'malformed 5' => [ $FAIL, "\x80\xbf\x80\xbf" ],
  77                         'malformed 6' => [ $FAIL, "\x80\xbf\x80\xbf\x80" ],
  78                         'malformed 7' => [ $FAIL, "\x80\xbf\x80\xbf\x80\xbf" ],
  79                         'malformed 8' => [ $FAIL, "\x80\xbf\x80\xbf\x80\xbf\x80" ],
  80
  81                         'last byte missing 1' => [ $FAIL, "\xc0" ],
  82                         'last byte missing 2' => [ $FAIL, "\xe0\x80" ],
  83                         'last byte missing 3' => [ $FAIL, "\xf0\x80\x80" ],
  84                         'last byte missing 4' => [ $FAIL, "\xf8\x80\x80\x80" ],
  85                         'last byte missing 5' => [ $FAIL, "\xfc\x80\x80\x80\x80" ],
  86                         'last byte missing 6' => [ $FAIL, "\xdf" ],
  87                         'last byte missing 7' => [ $FAIL, "\xef\xbf" ],
  88                         'last byte missing 8' => [ $FAIL, "\xf7\xbf\xbf" ],
  89                         'last byte missing 9' => [ $FAIL, "\xfb\xbf\xbf\xbf" ],
  90                         'last byte missing 10' => [ $FAIL, "\xfd\xbf\xbf\xbf\xbf" ],
  91
  92                         'extra continuation byte 1' => [ $FAIL, "e\xaf" ],
  93                         'extra continuation byte 2' => [ $FAIL, "\xc3\x89\xaf" ],
  94                         'extra continuation byte 3' => [ $FAIL, "\xef\xbc\xa5\xaf" ],
  95                         'extra continuation byte 4' => [ $FAIL, "\xf0\x9d\x99\xb4\xaf" ],
  96
  97                         'impossible bytes 1' => [ $FAIL, "\xfe" ],
  98                         'impossible bytes 2' => [ $FAIL, "\xff" ],
  99                         'impossible bytes 3' => [ $FAIL, "\xfe\xfe\xff\xff" ],
 100
 101                         'overlong sequences 1' => [ $FAIL, "\xc0\xaf" ],
 102                         'overlong sequences 2' => [ $FAIL, "\xc1\xaf" ],
 103                         'overlong sequences 3' => [ $FAIL, "\xe0\x80\xaf" ],
 104                         'overlong sequences 4' => [ $FAIL, "\xf0\x80\x80\xaf" ],
 105                         'overlong sequences 5' => [ $FAIL, "\xf8\x80\x80\x80\xaf" ],
 106                         'overlong sequences 6' => [ $FAIL, "\xfc\x80\x80\x80\x80\xaf" ],
 107
 108                         'maximum overlong sequences 1' => [ $FAIL, "\xc1\xbf" ],
 109                         'maximum overlong sequences 2' => [ $FAIL, "\xe0\x9f\xbf" ],
 110                         'maximum overlong sequences 3' => [ $FAIL, "\xf0\x8f\xbf\xbf" ],
 111                         'maximum overlong sequences 4' => [ $FAIL, "\xf8\x87\xbf\xbf" ],
 112                         'maximum overlong sequences 5' => [ $FAIL, "\xfc\x83\xbf\xbf\xbf\xbf" ],
 113
 114                         'surrogates 1 (U+D799)' => [ $PASS, "\xed\x9f\xbf" ],
 115                         'surrogates 2 (U+E000)' => [ $PASS, "\xee\x80\x80" ],
 116                         'surrogates 3 (U+D800)' => [ $FAIL, "\xed\xa0\x80" ],
 117                         'surrogates 4 (U+DBFF)' => [ $FAIL, "\xed\xaf\xbf" ],
 118                         'surrogates 5 (U+DC00)' => [ $FAIL, "\xed\xb0\x80" ],
 119                         'surrogates 6 (U+DFFF)' => [ $FAIL, "\xed\xbf\xbf" ],
 120                         'surrogates 7 (U+D800 U+DC00)' => [ $FAIL, "\xed\xa0\x80\xed\xb0\x80" ],
 121
 122                         'noncharacters 1' => [ $PASS, "\xef\xbf\xbe" ],
 123                         'noncharacters 2' => [ $PASS, "\xef\xbf\xbf" ],
 124                 ];
 125         }
 126 }