AkelPad

Code: Select all

typedef int bool_t;

// Check up to 6 utf-8 bytes.
// FYI, the BOM [0xEF, 0xBB, 0xBF] satisfies these requirements
bool_t IsStringUtf8(const char* s)
{
    bool_t containsUtf8; // contains utf-8 leading byte(s)
    bool_t errorUtf8;    // error: unexpected byte in utf-8 sequence
    unsigned int n;      // number of continuation bytes
    unsigned char uch;

    containsUtf8 = 0;
    errorUtf8 = 0;
    n = 0;

    while ( (!errorUtf8) && ((uch = (unsigned char) *s) != 0) )
    {
        if ( n == 0 )
        {
            if ( uch >= 0xC0 )
            {
                containsUtf8 = 1;
                if ( uch < 0xE0 )
                    n = 1;
                else if ( uch < 0xF0 )
                    n = 2;
                else if ( uch < 0xF8 )
                    n = 3;
                else if ( uch < 0xFC )
                    n = 4;
                else if ( uch < 0xFE )
                    n = 5;
                else
                    n = 6;
            }
            else if ( uch >= 0x80 )
                errorUtf8 = 1;
        }
        else
        {
            if ( uch >= 0x80 && uch < 0xC0 )
                --n;
            else
                errorUtf8 = 1;
        }
        ++s;
    }

    return (containsUtf8 && (!errorUtf8) && (!n)) ? 1 : 0;
}