view release on metacpan or  search on metacpan

src/boost/detail/utf8_codecvt_facet.ipp  view on Meta::CPAN

std::codecvt_base::result utf8_codecvt_facet::do_in(
    std::mbstate_t& /*state*/, 
    const char * from,
    const char * from_end, 
    const char * & from_next,
    wchar_t * to, 
    wchar_t * to_end, 
    wchar_t * & to_next
) const {
    // Basic algorithm:  The first octet determines how many
    // octets total make up the UCS-4 character.  The remaining
    // "continuing octets" all begin with "10". To convert, subtract
    // the amount that specifies the number of octets from the first
    // octet.  Subtract 0x80 (1000 0000) from each continuing octet,
    // then mash the whole lot together.  Note that each continuing
    // octet only uses 6 bits as unique values, so only shift by
    // multiples of 6 to combine.
    while (from != from_end && to != to_end) {

        // Error checking   on the first octet
        if (invalid_leading_octet(*from)){
            from_next = from;
            to_next = to;
            return std::codecvt_base::error;
        }

        // The first octet is   adjusted by a value dependent upon 
        // the number   of "continuing octets" encoding the character
        const   int cont_octet_count = get_cont_octet_count(*from);
        const   wchar_t octet1_modifier_table[] =   {
            0x00, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc
        };

        // The unsigned char conversion is necessary in case char is
        // signed   (I learned this the hard way)
        wchar_t ucs_result = 
            (unsigned char)(*from++) - octet1_modifier_table[cont_octet_count];

        // Invariants   : 
        //   1) At the start of the loop,   'i' continuing characters have been
        //    processed 
        //   2) *from   points to the next continuing character to be processed.
        int i   = 0;
        while(i != cont_octet_count && from != from_end) {

            // Error checking on continuing characters
            if (invalid_continuing_octet(*from)) {
                from_next   = from;
                to_next =   to;
                return std::codecvt_base::error;
            }

            ucs_result *= (1 << 6); 

            // each continuing character has an extra (10xxxxxx)b attached to 
            // it that must be removed.
            ucs_result += (unsigned char)(*from++) - 0x80;
            ++i;
        }

        // If   the buffer ends with an incomplete unicode character...
        if (from == from_end && i   != cont_octet_count) {
            // rewind "from" to before the current character translation
            from_next = from - (i+1); 
            to_next = to;
            return std::codecvt_base::partial;
        }
        *to++   = ucs_result;
    }
    from_next = from;
    to_next = to;

    // Were we done converting or did we run out of destination space?
    if(from == from_end) return std::codecvt_base::ok;
    else return std::codecvt_base::partial;
}

std::codecvt_base::result utf8_codecvt_facet::do_out(
    std::mbstate_t& /*state*/, 
    const wchar_t *   from,
    const wchar_t * from_end, 
    const wchar_t * & from_next,
    char * to, 
    char * to_end, 
    char * & to_next
) const
{
    // RG - consider merging this table with the other one
    const wchar_t octet1_modifier_table[] = {
        0x00, 0xc0, 0xe0, 0xf0, 0xf8, 0xfc
    };

    wchar_t max_wchar = (std::numeric_limits<wchar_t>::max)();
    while (from != from_end && to != to_end) {

        // Check for invalid UCS-4 character
        if (*from  > max_wchar) {
            from_next = from;
            to_next = to;
            return std::codecvt_base::error;
        }

        int cont_octet_count = get_cont_octet_out_count(*from);

        // RG  - comment this formula better
        int shift_exponent = (cont_octet_count) *   6;

        // Process the first character
        *to++ = static_cast<char>(octet1_modifier_table[cont_octet_count] +
            (unsigned char)(*from / (1 << shift_exponent)));

        // Process the continuation characters 
        // Invariants: At   the start of the loop:
        //   1) 'i' continuing octets   have been generated
        //   2) '*to'   points to the next location to place an octet
        //   3) shift_exponent is   6 more than needed for the next octet
        int i   = 0;
        while   (i != cont_octet_count && to != to_end) {
            shift_exponent -= 6;
            *to++ = static_cast<char>(0x80 + ((*from / (1 << shift_exponent)) % (1 << 6)));
            ++i;