view release on metacpan or  search on metacpan

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

            // 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;
        }
        // If   we filled up the out buffer before encoding the character
        if(to   == to_end && i != cont_octet_count) {
            from_next = from;
            to_next = to - (i+1);
            return std::codecvt_base::partial;
        }
        ++from;
    }
    from_next = from;
    to_next = to;
    // Were we done or did we run out of destination space
    if(from == from_end) return std::codecvt_base::ok;
    else return std::codecvt_base::partial;
}

// How many char objects can I process to get <= max_limit
// wchar_t objects?
int utf8_codecvt_facet::do_length(
    BOOST_CODECVT_DO_LENGTH_CONST std::mbstate_t &,
    const char * from,
    const char * from_end, 
    std::size_t max_limit
#if BOOST_WORKAROUND(__IBMCPP__, BOOST_TESTED_AT(600))
) const throw()
#else
) const
#endif
{ 
    // RG - this code is confusing!  I need a better way to express it.
    // and test cases.

    // Invariants:
    // 1) last_octet_count has the size of the last measured character
    // 2) char_count holds the number of characters shown to fit
    // within the bounds so far (no greater than max_limit)
    // 3) from_next points to the octet 'last_octet_count' before the
    // last measured character.  
    int last_octet_count=0;
    std::size_t char_count = 0;
    const char* from_next = from;
    // Use "<" because the buffer may represent incomplete characters
    while (from_next+last_octet_count <= from_end && char_count <= max_limit) {
        from_next += last_octet_count;