tests/phpunit/includes/libs/StringUtilsTest.php

   1 <?php
   2
   3 class StringUtilsTest extends PHPUnit_Framework_TestCase {
   4
   5         /**
   6          * This tests StringUtils::isUtf8 whenever we have the mbstring extension
   7          * loaded.
   8          *
   9          * @covers StringUtils::isUtf8
  10          * @dataProvider provideStringsForIsUtf8Check
  11          */
  12         public function testIsUtf8WithMbstring( $expected, $string ) {
  13                 if ( !function_exists( 'mb_check_encoding' ) ) {
  14                         $this->markTestSkipped( 'Test requires the mbstring PHP extension' );
  15                 }
  16                 $this->assertEquals( $expected,
  17                         StringUtils::isUtf8( $string ),
  18                         'Testing string "' . $this->escaped( $string ) . '" with mb_check_encoding'
  19                 );
  20         }
  21
  22         /**
  23          * This tests StringUtils::isUtf8 making sure we use the pure PHP
  24          * implementation used as a fallback when mb_check_encoding() is
  25          * not available.
  26          *
  27          * @covers StringUtils::isUtf8
  28          * @dataProvider provideStringsForIsUtf8Check
  29          */
  30         public function testIsUtf8WithPhpFallbackImplementation( $expected, $string ) {
  31                 $this->assertEquals( $expected,
  32                         StringUtils::isUtf8( $string, /** disable mbstring: */true ),
  33                         'Testing string "' . $this->escaped( $string ) . '" with pure PHP implementation'
  34                 );
  35         }
  36
  37         /**
  38          * Print high range characters as a hexadecimal
  39          * @param string $string
  40          * @return string
  41          */
  42         function escaped( $string ) {
  43                 $escaped = '';
  44                 $length = strlen( $string );
  45                 for ( $i = 0; $i < $length; $i++ ) {
  46                         $char = $string[$i];
  47                         $val = ord( $char );
  48                         if ( $val > 127 ) {
  49                                 $escaped .= '\x' . dechex( $val );
  50                         } else {
  51                                 $escaped .= $char;
  52                         }
  53                 }
  54
  55                 return $escaped;
  56         }
  57
  58         /**
  59          * See also "UTF-8 decoder capability and stress test" by
  60          * Markus Kuhn:
  61          * http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt
  62          */
  63         public static function provideStringsForIsUtf8Check() {
  64                 // Expected return values for StringUtils::isUtf8()
  65                 $PASS = true;
  66                 $FAIL = false;
  67
  68                 return [
  69                         'some ASCII' => [ $PASS, 'Some ASCII' ],
  70                         'euro sign' => [ $PASS, "Euro sign €" ],
  71
  72                         'first possible sequence 1 byte' => [ $PASS, "\x00" ],
  73                         'first possible sequence 2 bytes' => [ $PASS, "\xc2\x80" ],
  74                         'first possible sequence 3 bytes' => [ $PASS, "\xe0\xa0\x80" ],
  75                         'first possible sequence 4 bytes' => [ $PASS, "\xf0\x90\x80\x80" ],
  76                         'first possible sequence 5 bytes' => [ $FAIL, "\xf8\x88\x80\x80\x80" ],
  77                         'first possible sequence 6 bytes' => [ $FAIL, "\xfc\x84\x80\x80\x80\x80" ],
  78
  79                         'last possible sequence 1 byte' => [ $PASS, "\x7f" ],
  80                         'last possible sequence 2 bytes' => [ $PASS, "\xdf\xbf" ],
  81                         'last possible sequence 3 bytes' => [ $PASS, "\xef\xbf\xbf" ],
  82                         'last possible sequence 4 bytes (U+1FFFFF)' => [ $FAIL, "\xf7\xbf\xbf\xbf" ],
  83                         'last possible sequence 5 bytes' => [ $FAIL, "\xfb\xbf\xbf\xbf\xbf" ],
  84                         'last possible sequence 6 bytes' => [ $FAIL, "\xfd\xbf\xbf\xbf\xbf\xbf" ],
  85
  86                         'boundary 1' => [ $PASS, "\xed\x9f\xbf" ],
  87                         'boundary 2' => [ $PASS, "\xee\x80\x80" ],
  88                         'boundary 3' => [ $PASS, "\xef\xbf\xbd" ],
  89                         'boundary 4' => [ $PASS, "\xf2\x80\x80\x80" ],
  90                         'boundary 5 (U+FFFFF)' => [ $PASS, "\xf3\xbf\xbf\xbf" ],
  91                         'boundary 6 (U+100000)' => [ $PASS, "\xf4\x80\x80\x80" ],
  92                         'boundary 7 (U+10FFFF)' => [ $PASS, "\xf4\x8f\xbf\xbf" ],
  93                         'boundary 8 (U+110000)' => [ $FAIL, "\xf4\x90\x80\x80" ],
  94
  95                         'malformed 1' => [ $FAIL, "\x80" ],
  96                         'malformed 2' => [ $FAIL, "\xbf" ],
  97                         'malformed 3' => [ $FAIL, "\x80\xbf" ],
  98                         'malformed 4' => [ $FAIL, "\x80\xbf\x80" ],
  99                         'malformed 5' => [ $FAIL, "\x80\xbf\x80\xbf" ],
 100                         'malformed 6' => [ $FAIL, "\x80\xbf\x80\xbf\x80" ],
 101                         'malformed 7' => [ $FAIL, "\x80\xbf\x80\xbf\x80\xbf" ],
 102                         'malformed 8' => [ $FAIL, "\x80\xbf\x80\xbf\x80\xbf\x80" ],
 103
 104                         'last byte missing 1' => [ $FAIL, "\xc0" ],
 105                         'last byte missing 2' => [ $FAIL, "\xe0\x80" ],
 106                         'last byte missing 3' => [ $FAIL, "\xf0\x80\x80" ],
 107                         'last byte missing 4' => [ $FAIL, "\xf8\x80\x80\x80" ],
 108                         'last byte missing 5' => [ $FAIL, "\xfc\x80\x80\x80\x80" ],
 109                         'last byte missing 6' => [ $FAIL, "\xdf" ],
 110                         'last byte missing 7' => [ $FAIL, "\xef\xbf" ],
 111                         'last byte missing 8' => [ $FAIL, "\xf7\xbf\xbf" ],
 112                         'last byte missing 9' => [ $FAIL, "\xfb\xbf\xbf\xbf" ],
 113                         'last byte missing 10' => [ $FAIL, "\xfd\xbf\xbf\xbf\xbf" ],
 114
 115                         'extra continuation byte 1' => [ $FAIL, "e\xaf" ],
 116                         'extra continuation byte 2' => [ $FAIL, "\xc3\x89\xaf" ],
 117                         'extra continuation byte 3' => [ $FAIL, "\xef\xbc\xa5\xaf" ],
 118                         'extra continuation byte 4' => [ $FAIL, "\xf0\x9d\x99\xb4\xaf" ],
 119
 120                         'impossible bytes 1' => [ $FAIL, "\xfe" ],
 121                         'impossible bytes 2' => [ $FAIL, "\xff" ],
 122                         'impossible bytes 3' => [ $FAIL, "\xfe\xfe\xff\xff" ],
 123
 124                         'overlong sequences 1' => [ $FAIL, "\xc0\xaf" ],
 125                         'overlong sequences 2' => [ $FAIL, "\xc1\xaf" ],
 126                         'overlong sequences 3' => [ $FAIL, "\xe0\x80\xaf" ],
 127                         'overlong sequences 4' => [ $FAIL, "\xf0\x80\x80\xaf" ],
 128                         'overlong sequences 5' => [ $FAIL, "\xf8\x80\x80\x80\xaf" ],
 129                         'overlong sequences 6' => [ $FAIL, "\xfc\x80\x80\x80\x80\xaf" ],
 130
 131                         'maximum overlong sequences 1' => [ $FAIL, "\xc1\xbf" ],
 132                         'maximum overlong sequences 2' => [ $FAIL, "\xe0\x9f\xbf" ],
 133                         'maximum overlong sequences 3' => [ $FAIL, "\xf0\x8f\xbf\xbf" ],
 134                         'maximum overlong sequences 4' => [ $FAIL, "\xf8\x87\xbf\xbf" ],
 135                         'maximum overlong sequences 5' => [ $FAIL, "\xfc\x83\xbf\xbf\xbf\xbf" ],
 136
 137                         'surrogates 1 (U+D799)' => [ $PASS, "\xed\x9f\xbf" ],
 138                         'surrogates 2 (U+E000)' => [ $PASS, "\xee\x80\x80" ],
 139                         'surrogates 3 (U+D800)' => [ $FAIL, "\xed\xa0\x80" ],
 140                         'surrogates 4 (U+DBFF)' => [ $FAIL, "\xed\xaf\xbf" ],
 141                         'surrogates 5 (U+DC00)' => [ $FAIL, "\xed\xb0\x80" ],
 142                         'surrogates 6 (U+DFFF)' => [ $FAIL, "\xed\xbf\xbf" ],
 143                         'surrogates 7 (U+D800 U+DC00)' => [ $FAIL, "\xed\xa0\x80\xed\xb0\x80" ],
 144
 145                         'noncharacters 1' => [ $PASS, "\xef\xbf\xbe" ],
 146                         'noncharacters 2' => [ $PASS, "\xef\xbf\xbf" ],
 147                 ];
 148         }
 149 }