Функции работы с UTF-8

Получение числа символов в строке

// returns the number of symbols in the UTF-8-formatted string
// on error (invalid UTF-8 formatted string) returns false
function utf8StrLen( $string )
{
 $length = 0;
 
 $i = 0;
 while ( $i < strlen( $string ) )
 {
 $code = ord( $string[$i] );
 if ( ( $code & 0x80 ) == 0 )
 $i = 1;
 else if ( ( $code & 0xe0 ) == 0xc0 )
 $i = 2;
 else if ( ( $code & 0xf0 ) == 0xe0 )
 $i = 3;
 else if ( ( $code & 0xf8 ) == 0xf0 )
 $i = 4;
 else return false;
 $length ;
 };
 
 // done
 return $length;
}

Получение символа с заданным смещением от начала строки

// returns the string with UTF-8 formatted character
// at the position, specified, or false on error (the position is outside the
// string boundaries or the string is not well-formatted UTF-8 string)
function utf8CharAt( $string, $position )
{
 // first checking the position to be inside the string
 if ( $position < 0 || $position >= utf8StrLen( $string ) )
 return false;
 
 $char = false;
 
 $i = 0;
 $curPosition = 0;
 $stringLength = strlen( $string );
 while ( $i < $stringLength )
 {
 $code = ord( $string[$i] );
 // retreaving the character, if at the position, required
 if ( $curPosition == $position )
 {
 if ( ( $code & 0x80 ) == 0 )
 $char = $string[$i];
 else if ( ( $code & 0xe0 ) == 0xc0 && $i 1 < $stringLength )
 $char = $string[$i].$string[$i 1];
 else if ( ( $code & 0xf0 ) == 0xe0 && $i 2 < $stringLength )
 $char = $string[$i].$string[$i 1].$string[$i 2];
 else if ( ( $code & 0xf8 ) == 0xf0 && $i 3 < $stringLength )
 $char = $string[$i].$string[$i 1].$string[$i 2].$string[$i 3];
 else
 return false;
 break;
 };
 // moving forward
 if ( ( $code & 0x80 ) == 0 )
 $i = 1;
 else if ( ( $code & 0xe0 ) == 0xc0 )
 $i = 2;
 else if ( ( $code & 0xf0 ) == 0xe0 )
 $i = 3;
 else if ( ( $code & 0xf8 ) == 0xf0 )
 $i = 4;
 else return false;
 $curPosition ;
 };
 
 // done
 return $char;
}