| 1617 | }; |
| 1618 | |
| 1619 | struct utf8_decoder |
| 1620 | { |
| 1621 | typedef uint8_t type; |
| 1622 | |
| 1623 | template <typename Traits> static inline typename Traits::value_type process(const uint8_t* data, size_t size, typename Traits::value_type result, Traits) |
| 1624 | { |
| 1625 | const uint8_t utf8_byte_mask = 0x3f; |
| 1626 | |
| 1627 | while (size) |
| 1628 | { |
| 1629 | uint8_t lead = *data; |
| 1630 | |
| 1631 | // 0xxxxxxx -> U+0000..U+007F |
| 1632 | if (lead < 0x80) |
| 1633 | { |
| 1634 | result = Traits::low(result, lead); |
| 1635 | data += 1; |
| 1636 | size -= 1; |
| 1637 | |
| 1638 | // process aligned single-byte (ascii) blocks |
| 1639 | if ((reinterpret_cast<uintptr_t>(data) & 3) == 0) |
| 1640 | { |
| 1641 | // round-trip through void* to silence 'cast increases required alignment of target type' warnings |
| 1642 | while (size >= 4 && (*static_cast<const uint32_t*>(static_cast<const void*>(data)) & 0x80808080) == 0) |
| 1643 | { |
| 1644 | result = Traits::low(result, data[0]); |
| 1645 | result = Traits::low(result, data[1]); |
| 1646 | result = Traits::low(result, data[2]); |
| 1647 | result = Traits::low(result, data[3]); |
| 1648 | data += 4; |
| 1649 | size -= 4; |
| 1650 | } |
| 1651 | } |
| 1652 | } |
| 1653 | // 110xxxxx -> U+0080..U+07FF |
| 1654 | else if (static_cast<unsigned int>(lead - 0xC0) < 0x20 && size >= 2 && (data[1] & 0xc0) == 0x80) |
| 1655 | { |
| 1656 | result = Traits::low(result, ((lead & ~0xC0) << 6) | (data[1] & utf8_byte_mask)); |
| 1657 | data += 2; |
| 1658 | size -= 2; |
| 1659 | } |
| 1660 | // 1110xxxx -> U+0800-U+FFFF |
| 1661 | else if (static_cast<unsigned int>(lead - 0xE0) < 0x10 && size >= 3 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80) |
| 1662 | { |
| 1663 | result = Traits::low(result, ((lead & ~0xE0) << 12) | ((data[1] & utf8_byte_mask) << 6) | (data[2] & utf8_byte_mask)); |
| 1664 | data += 3; |
| 1665 | size -= 3; |
| 1666 | } |
| 1667 | // 11110xxx -> U+10000..U+10FFFF |
| 1668 | else if (static_cast<unsigned int>(lead - 0xF0) < 0x08 && size >= 4 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80 && (data[3] & 0xc0) == 0x80) |
| 1669 | { |
| 1670 | result = Traits::high(result, ((lead & ~0xF0) << 18) | ((data[1] & utf8_byte_mask) << 12) | ((data[2] & utf8_byte_mask) << 6) | (data[3] & utf8_byte_mask)); |
| 1671 | data += 4; |
| 1672 | size -= 4; |
| 1673 | } |
| 1674 | // 10xxxxxx or 11111xxx -> invalid |
| 1675 | else |
| 1676 | { |
no outgoing calls
no test coverage detected