hphp/test/slow/utf8_decode-invalid.php

   1 <?hh
   2
   3
   4 // https://github.com/facebook/hhvm/issues/5010
   5 // utf8_decode() targets latin1, therefore any character
   6 // greater than U+00FF should come out as
   7 // U+003F QUESTION MARK
   8
   9 // U+00DF LATIN SMALL LETTER SHARP S
  10 <<__EntryPoint>>
  11 function main_utf8_decode_invalid() {
  12 $ss = "\u{DF}";
  13 var_dump(urlencode($ss));
  14 var_dump(urlencode(utf8_decode($ss)));
  15 var_dump(json_encode(utf8_decode($ss)));
  16
  17 // U+044F CYRILLIC SMALL LETTER YA
  18 $ya = "\u{44F}";
  19 var_dump(urlencode($ya));
  20 var_dump(urlencode(utf8_decode($ya)));
  21 var_dump(json_encode(utf8_decode($ya)));
  22
  23 // Improperly formatted U+0041 LATIN CAPITAL LETTER A
  24 $A = "\xC1\x81";
  25 var_dump(urlencode($A));
  26 var_dump(urlencode(utf8_decode($A)));
  27 var_dump(json_encode(utf8_decode($A)));
  28 }