utf8_decoder Struct Reference

Public Types
typedef uint8_t	type

Static Public Member Functions
template<typename Traits>
static Traits::value_type	process (const uint8_t *data, size_t size, typename Traits::value_type result, Traits)

Detailed Description

Definition at line 1636 of file pugixml.cc.

Member Typedef Documentation

type

typedef uint8_t utf8_decoder::type

Definition at line 1638 of file pugixml.cc.

Member Function Documentation

process()

template<typename Traits>

Traits::value_type utf8_decoder::process ( const uint8_t * data, size_t size, typename Traits::value_type result, Traits  )

inlinestatic

Definition at line 1640 of file pugixml.cc.

        {
            const uint8_t utf8_byte_mask = 0x3f;
 
            while (size)
            {
                uint8_t lead = *data;
 
                // 0xxxxxxx -> U+0000..U+007F
                if (lead < 0x80)
                {
                    result = Traits::low(result, lead);
                    data += 1;
                    size -= 1;
 
                    // process aligned single-byte (ascii) blocks
                    if ((reinterpret_cast<uintptr_t>(data) & 3) == 0)
                    {
                        // round-trip through void* to silence 'cast increases required alignment of target type' warnings
                        while (size >= 4 && (*static_cast<const uint32_t*>(static_cast<const void*>(data)) & 0x80808080) == 0)
                        {
                            result = Traits::low(result, data[0]);
                            result = Traits::low(result, data[1]);
                            result = Traits::low(result, data[2]);
                            result = Traits::low(result, data[3]);
                            data += 4;
                            size -= 4;
                        }
                    }
                }
                // 110xxxxx -> U+0080..U+07FF
                else if (static_cast<unsigned int>(lead - 0xC0) < 0x20 && size >= 2 && (data[1] & 0xc0) == 0x80)
                {
                    result = Traits::low(result, ((lead & ~0xC0) << 6) | (data[1] & utf8_byte_mask));
                    data += 2;
                    size -= 2;
                }
                // 1110xxxx -> U+0800-U+FFFF
                else if (static_cast<unsigned int>(lead - 0xE0) < 0x10 && size >= 3 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80)
                {
                    result = Traits::low(result, ((lead & ~0xE0) << 12) | ((data[1] & utf8_byte_mask) << 6) | (data[2] & utf8_byte_mask));
                    data += 3;
                    size -= 3;
                }
                // 11110xxx -> U+10000..U+10FFFF
                else if (static_cast<unsigned int>(lead - 0xF0) < 0x08 && size >= 4 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80 && (data[3] & 0xc0) == 0x80)
                {
                    result = Traits::high(result, ((lead & ~0xF0) << 18) | ((data[1] & utf8_byte_mask) << 12) | ((data[2] & utf8_byte_mask) << 6) | (data[3] & utf8_byte_mask));
                    data += 4;
                    size -= 4;
                }
                // 10xxxxxx or 11111xxx -> invalid
                else
                {
                    data += 1;
                    size -= 1;
                }
            }
 
            return result;
        }

Referenced by as_wide_impl().

---

The documentation for this struct was generated from the following file:

• pugixml.cc