tests/phpunit/includes/StringUtilsTest.php

   1 <?php
   2
   3 class StringUtilsTest extends MediaWikiTestCase {
   4
   5         /**
   6          * This test StringUtils::isUtf8 whenever we have mbstring extension
   7          * loaded.
   8          *
   9          * @covers StringUtils::isUtf8
  10          * @dataProvider provideStringsForIsUtf8Check
  11          */
  12         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 test 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         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 an hexadecimal
  39          */
  40         function escaped( $string ) {
  41                 $escaped = '';
  42                 $length = strlen( $string );
  43                 for ( $i = 0; $i < $length; $i++ ) {
  44                         $char = $string[$i];
  45                         $val = ord( $char );
  46                         if ( $val > 127 ) {
  47                                 $escaped .= '\x' . dechex( $val );
  48                         } else {
  49                                 $escaped .= $char;
  50                         }
  51                 }
  52
  53                 return $escaped;
  54         }
  55
  56         /**
  57          * See also "UTF-8 decoder capability and stress test" by
  58          * Markus Kuhn:
  59          * http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt
  60          */
  61         public static function provideStringsForIsUtf8Check() {
  62                 // Expected return values for StringUtils::isUtf8()
  63                 $PASS = true;
  64                 $FAIL = false;
  65
  66                 return array(
  67                         array( $PASS, 'Some ASCII' ),
  68                         array( $PASS, "Euro sign €" ),
  69
  70                         // First possible sequences
  71                         array( $PASS, "\x00" ),
  72                         array( $PASS, "\xc2\x80" ),
  73                         array( $PASS, "\xe0\xa0\x80" ),
  74                         array( $PASS, "\xf0\x90\x80\x80" ),
  75                         array( $FAIL, "\xf8\x88\x80\x80\x80" ),
  76                         array( $FAIL, "\xfc\x84\x80\x80\x80\x80" ),
  77
  78                         // Last possible sequence
  79                         array( $PASS, "\x7f" ),
  80                         array( $PASS, "\xdf\xbf" ),
  81                         array( $PASS, "\xef\xbf\xbf" ),
  82                         array( $FAIL, "\xf7\xbf\xbf\xbf" ), // U+1FFFFF
  83                         array( $FAIL, "\xfb\xbf\xbf\xbf\xbf" ),
  84                         array( $FAIL, "\xfd\xbf\xbf\xbf\xbf\xbf" ),
  85
  86                         // Boundaries
  87                         array( $PASS, "\xed\x9f\xbf" ),
  88                         array( $PASS, "\xee\x80\x80" ),
  89                         array( $PASS, "\xef\xbf\xbd" ),
  90                         array( $PASS, "\xf2\x80\x80\x80" ),
  91                         array( $PASS, "\xf3\xbf\xbf\xbf" ), // U+FFFFF
  92                         array( $PASS, "\xf4\x80\x80\x80" ), // U+100000
  93                         array( $PASS, "\xf4\x8f\xbf\xbf" ), // U+10FFFF
  94                         array( $FAIL, "\xf4\x90\x80\x80" ), // U+110000
  95
  96                         // Malformed
  97                         array( $FAIL, "\x80" ),
  98                         array( $FAIL, "\xbf" ),
  99                         array( $FAIL, "\x80\xbf" ),
 100                         array( $FAIL, "\x80\xbf\x80" ),
 101                         array( $FAIL, "\x80\xbf\x80\xbf" ),
 102                         array( $FAIL, "\x80\xbf\x80\xbf\x80" ),
 103                         array( $FAIL, "\x80\xbf\x80\xbf\x80\xbf" ),
 104                         array( $FAIL, "\x80\xbf\x80\xbf\x80\xbf\x80" ),
 105
 106                         // Last byte missing
 107                         array( $FAIL, "\xc0" ),
 108                         array( $FAIL, "\xe0\x80" ),
 109                         array( $FAIL, "\xf0\x80\x80" ),
 110                         array( $FAIL, "\xf8\x80\x80\x80" ),
 111                         array( $FAIL, "\xfc\x80\x80\x80\x80" ),
 112                         array( $FAIL, "\xdf" ),
 113                         array( $FAIL, "\xef\xbf" ),
 114                         array( $FAIL, "\xf7\xbf\xbf" ),
 115                         array( $FAIL, "\xfb\xbf\xbf\xbf" ),
 116                         array( $FAIL, "\xfd\xbf\xbf\xbf\xbf" ),
 117
 118                         // Extra continuation byte
 119                         array( $FAIL, "e\xaf" ),
 120                         array( $FAIL, "\xc3\x89\xaf" ),
 121                         array( $FAIL, "\xef\xbc\xa5\xaf" ),
 122                         array( $FAIL, "\xf0\x9d\x99\xb4\xaf" ),
 123
 124                         // Impossible bytes
 125                         array( $FAIL, "\xfe" ),
 126                         array( $FAIL, "\xff" ),
 127                         array( $FAIL, "\xfe\xfe\xff\xff" ),
 128
 129                         // Overlong sequences
 130                         array( $FAIL, "\xc0\xaf" ),
 131                         array( $FAIL, "\xc1\xaf" ),
 132                         array( $FAIL, "\xe0\x80\xaf" ),
 133                         array( $FAIL, "\xf0\x80\x80\xaf" ),
 134                         array( $FAIL, "\xf8\x80\x80\x80\xaf" ),
 135                         array( $FAIL, "\xfc\x80\x80\x80\x80\xaf" ),
 136
 137                         // Maximum overlong sequences
 138                         array( $FAIL, "\xc1\xbf" ),
 139                         array( $FAIL, "\xe0\x9f\xbf" ),
 140                         array( $FAIL, "\xf0\x8f\xbf\xbf" ),
 141                         array( $FAIL, "\xf8\x87\xbf\xbf" ),
 142                         array( $FAIL, "\xfc\x83\xbf\xbf\xbf\xbf" ),
 143
 144                         // Surrogates
 145                         array( $PASS, "\xed\x9f\xbf" ), // U+D799
 146                         array( $PASS, "\xee\x80\x80" ), // U+E000
 147                         array( $FAIL, "\xed\xa0\x80" ), // U+D800
 148                         array( $FAIL, "\xed\xaf\xbf" ), // U+DBFF
 149                         array( $FAIL, "\xed\xb0\x80" ), // U+DC00
 150                         array( $FAIL, "\xed\xbf\xbf" ), // U+DFFF
 151                         array( $FAIL, "\xed\xa0\x80\xed\xb0\x80" ), // U+D800 U+DC00
 152
 153                         // Noncharacters
 154                         array( $PASS, "\xef\xbf\xbe" ),
 155                         array( $PASS, "\xef\xbf\xbf" ),
 156                 );
 157         }
 158 }