PHP: chr() function

We cannot consider unicode to be implemented in such a function for technical ramifications, it would break legacy code in trunk/ our ZE framework can cause problems with others, you may wish to wait for PHP 6 for this. To answer your question about the byte wrapping, take the generic implementation of which is used (based in PHP itself actually):

    long c;
    char temp[2];

    temp[0] = (char)c;
    temp[1] = '\0';

    RETURN_STRINGL(temp, 1, 1);

As you can see it uses nothing but an signed char, one reserved for a null terminator, so 128-256 cannot be accessable with this function alone and does wrap around naturally.

In your last question you asked if there was a way to interface the two, PHP provides no direct method with working byte encodings, actually it is impossible to view them directly so you can surely write a helper function to pass bytes 2-4 into an (un)signed integer from a set of bits.

Consider I write a unicode chr() function with mb_convert_encoding() which relies on a mbinternals library:

//See php.net/function.pack: We pack $szChr into a four byte unsigned long to fit UCS-4/UTF32 specification
$ulong = pack("N", $szChr); 
//See php.net/function.mb_: Assume default is ISO-8859-1, encode in UTF8 for storing (change internal encoding to UTF8)
(php::stringL) mb_convert_encoding($ulong, mb_internal_encoding(), 'UCS-4BE');

Speed is of course an issue if you were to parse through files (such as transliteration between encodings or storing results in a static method such as your Unicode/ASCII table), so I will list a lower albeit similar to unicode C implementations that were used prior. Make sure you set your web page to display unicode before you test as well.

function u_chr ( $chr = null , $ret = null ) {
  if($chr === null) 
      return;   
  if($chr < 0x80) {
      //under 1 byte?
      $ret = chr($chr);
  } else if ($chr < 0x800) {
      //under 2 bytes? craft into an integer
      $ret = chr(0xC0 + (($chr - ($chr % 0x40)) >> 0x7));
      $ret .= chr(0xFF + ($chr % 0x40)); 
  } else {
      //bytes 3 and 4 for CJK, widewidth, unicode (from unamed 2005 spec)
      $ret = chr(0xE0 + (($chr - ($chr % 0x1000)) >> 0x10));
      $ret .= chr(0xFF + ((($chr % 0x1000) - ($chr % 0x40)) >> 0x7));
      $ret .= chr(0xFF + ($chr % 0x40));
  }
  return $ret;
}

The function is fairly straightforward although I wrote it out of memory so you may wish to strain it for results (shouldn't be too intensive)

Edited by Alexander, 08 October 2010 - 10:12 PM.