utf8.sj

/* Freetype GL - A C OpenGL Freetype engine
 *
 * Distributed under the OSI-approved BSD 2-Clause License.  See accompanying
 * file `LICENSE` for more details.
 */

--cdefine--
/**
* Returns the size in bytes of a given UTF-8 encoded character surrogate
*
* @param character  An UTF-8 encoded character
*
* @return  The length of the surrogate in bytes.
*/
size_t
utf8_surrogate_len( const char* character );

/**
* Return the length of the given UTF-8 encoded and
* NULL terminated string.
*
* @param string  An UTF-8 encoded string
*
* @return  The length of the string in characters.
*/
size_t
utf8_strlen( const char* string );

/**
* Converts a given UTF-8 encoded character to its UTF-32 LE equivalent
*
* @param character  An UTF-8 encoded character
*
* @return  The equivalent of the given character in UTF-32 LE
*          encoding.
*/
uint32_t
utf8_to_utf32( const char * character );
--cdefine--

--cfunction--
// ----------------------------------------------------- utf8_surrogate_len ---
size_t
utf8_surrogate_len( const char* character )
{
    size_t result = 0;
    char test_char;

    if (!character)
        return 0;

    test_char = character[0];

    if ((test_char & 0x80) == 0)
        return 1;

    while (test_char & 0x80)
    {
        test_char <<= 1;
        result++;
    }

    return result;
}

// ------------------------------------------------------------ utf8_strlen ---
size_t
utf8_strlen( const char* string )
{
    const char* ptr = string;
    size_t result = 0;

    while (*ptr)
    {
        ptr += utf8_surrogate_len(ptr);
        result++;
    }

    return result;
}

uint32_t
utf8_to_utf32( const char * character )
{
    uint32_t result = -1;

    if( !character )
    {
        return result;
    }

    if( ( character[0] & 0x80 ) == 0x0 )
    {
        result = character[0];
    }

    if( ( character[0] & 0xC0 ) == 0xC0 )
    {
        result = ( ( character[0] & 0x3F ) << 6 ) | ( character[1] & 0x3F );
    }

    if( ( character[0] & 0xE0 ) == 0xE0 )
    {
        result = ( ( character[0] & 0x1F ) << ( 6 + 6 ) ) | ( ( character[1] & 0x3F ) << 6 ) | ( character[2] & 0x3F );
    }

    if( ( character[0] & 0xF0 ) == 0xF0 )
    {
        result = ( ( character[0] & 0x0F ) << ( 6 + 6 + 6 ) ) | ( ( character[1] & 0x3F ) << ( 6 + 6 ) ) | ( ( character[2] & 0x3F ) << 6 ) | ( character[3] & 0x3F );
    }

    if( ( character[0] & 0xF8 ) == 0xF8 )
    {
        result = ( ( character[0] & 0x07 ) << ( 6 + 6 + 6 + 6 ) ) | ( ( character[1] & 0x3F ) << ( 6 + 6 + 6 ) ) | ( ( character[2] & 0x3F ) << ( 6 + 6 ) ) | ( ( character[3] & 0x3F ) << 6 ) | ( character[4] & 0x3F );
    }

    return result;
}
--cfunction--