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          * @cover 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( $string ) . '" with mb_check_encoding'
  19                 );
  20         }
  21
  22         /**
  23          * This test StringUtils::isUtf8 making sure we use the pure PHP
  24          * implementation used as a fallback when mb_check_encoding() is
  25          * not available.
  26          *
  27          * @cover StringUtils::isUtf8
  28          * @dataProvider provideStringsForIsUtf8Check
  29          */
  30         function testIsUtf8WithPhpFallbackImplementation($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                 $escaped = '';
  42                 for($i=0; $i<strlen($string);$i++) {
  43                         $char = $string[$i];
  44                         $val = ord($char);
  45                         if( $val > 127 ) {
  46                                 $escaped .='\x' . dechex($val);
  47                         } else {
  48                                 $escaped .= $char;
  49                         }
  50                 }
  51                 return $escaped;
  52         }
  53
  54         /**
  55          * See also "UTF-8 decoder capability and stress test" by
  56          * Markus Kuhn:
  57          * http://www.cl.cam.ac.uk/~mgk25/ucs/examples/UTF-8-test.txt
  58          */
  59         function provideStringsForIsUtf8Check() {
  60                 // Expected return values for StringUtils::isUtf8()
  61                 $PASS = true;
  62                 $FAIL = false;
  63
  64                 return array(
  65                         array( $PASS, 'Some ASCII' ),
  66                         array( $PASS, "Euro sign €" ),
  67
  68                         # First possible sequences
  69                         array( $PASS, "\x00" ),
  70                         array( $PASS, "\xc2\x80" ),
  71                         array( $PASS, "\xe0\xa0\x80" ),
  72                         array( $PASS, "\xf0\x90\x80\x80" ),
  73                         array( $PASS, "\xf8\x88\x80\x80\x80" ),
  74                         array( $PASS, "\xfc\x84\x80\x80\x80\x80" ),
  75
  76                         # Last possible sequence
  77                         array( $PASS, "\x7f" ),
  78                         array( $PASS, "\xdf\xbf" ),
  79                         array( $PASS, "\xef\xbf\xbf" ),
  80                         array( $PASS, "\xf7\xbf\xbf\xbf" ),
  81                         array( $PASS, "\xfb\xbf\xbf\xbf\xbf" ),
  82                         array( $FAIL, "\xfd\xbf\xbf\xbf\xbf\xbf" ),
  83
  84                         # boundaries:
  85                         array( $PASS, "\xed\x9f\xbf" ),
  86                         array( $PASS, "\xee\x80\x80" ),
  87                         array( $PASS, "\xef\xbf\xbd" ),
  88                         array( $PASS, "\xf4\x8f\xbf\xbf" ),
  89                         array( $PASS, "\xf4\x90\x80\x80" ),
  90
  91                         # Malformed
  92                         array( $FAIL, "\x80" ),
  93                         array( $FAIL, "\xBF" ),
  94                         array( $FAIL, "\x80\xbf" ),
  95                         array( $FAIL, "\x80\xbf\x80" ),
  96                         array( $FAIL, "\x80\xbf\x80\xbf" ),
  97                         array( $FAIL, "\x80\xbf\x80\xbf\x80" ),
  98                         array( $FAIL, "\x80\xbf\x80\xbf\x80\xbf" ),
  99                         array( $FAIL, "\x80\xbf\x80\xbf\x80\xbf\x80" ),
 100
 101                         # last byte missing
 102                         array( $FAIL, "\xc0" ),
 103                         array( $FAIL, "\xe0\x80" ),
 104                         array( $FAIL, "\xf0\x80\x80" ),
 105                         array( $FAIL, "\xf8\x80\x80\x80" ),
 106                         array( $FAIL, "\xfc\x80\x80\x80\x80" ),
 107                         array( $FAIL, "\xdf" ),
 108                         array( $FAIL, "\xef\xbf" ),
 109                         array( $FAIL, "\xf7\xbf\xbf" ),
 110                         array( $FAIL, "\xfb\xbf\xbf\xbf" ),
 111                         array( $FAIL, "\xfd\xbf\xbf\xbf\xbf" ),
 112
 113                         # impossible bytes
 114                         array( $FAIL, "\xfe" ),
 115                         array( $FAIL, "\xff" ),
 116                         array( $FAIL, "\xfe\xfe\xff\xff" ),
 117
 118                         /**
 119                         # The PHP implementation does not handle characters
 120                         # being represented in a form which is too long :(
 121
 122                         # overlong sequences
 123                         array( $FAIL, "\xc0\xaf" ),
 124                         array( $FAIL, "\xe0\x80\xaf" ),
 125                         array( $FAIL, "\xf0\x80\x80\xaf" ),
 126                         array( $FAIL, "\xf8\x80\x80\x80\xaf" ),
 127                         array( $FAIL, "\xfc\x80\x80\x80\x80\xaf" ),
 128
 129                         # Maximum overlong sequences
 130                         array( $FAIL, "\xc1\xbf" ),
 131                         array( $FAIL, "\xe0\x9f\xbf" ),
 132                         array( $FAIL, "\xf0\x8F\xbf\xbf" ),
 133                         array( $FAIL, "\xf8\x87\xbf\xbf" ),
 134                         array( $FAIL, "\xfc\x83\xbf\xbf\xbf\xbf" ),
 135                         **/
 136
 137                         # non characters
 138                         array( $PASS, "\xef\xbf\xbe" ),
 139                         array( $PASS, "\xef\xbf\xbf" ),
 140                 );
 141         }
 142 }