| 1 | #pragma once |
| 2 | |
| 3 | #include <IO/ReadBuffer.h> |
| 4 | #include <IO/WriteBuffer.h> |
| 5 | #include <Core/Types.h> |
| 6 | #include <Common/BitHelpers.h> |
| 7 | |
| 8 | #if defined(__OpenBSD__) || defined(__FreeBSD__) |
| 9 | # include <sys/endian.h> |
| 10 | #elif defined(__APPLE__) |
| 11 | # include <libkern/OSByteOrder.h> |
| 12 | |
| 13 | # define htobe64(x) OSSwapHostToBigInt64(x) |
| 14 | # define be64toh(x) OSSwapBigToHostInt64(x) |
| 15 | #endif |
| 16 | |
| 17 | namespace DB |
| 18 | { |
| 19 | |
| 20 | /** Reads data from underlying ReadBuffer bit by bit, max 64 bits at once. |
| 21 | * |
| 22 | * reads MSB bits first, imagine that you have a data: |
| 23 | * 11110000 10101010 00100100 11111110 |
| 24 | * |
| 25 | * Given that r is BitReader created with a ReadBuffer that reads from data above: |
| 26 | * r.readBits(3) => 0b111 |
| 27 | * r.readBit() => 0b1 |
| 28 | * r.readBits(8) => 0b1010 // 4 leading zero-bits are not shown |
| 29 | * r.readBit() => 0b1 |
| 30 | * r.readBit() => 0b0 |
| 31 | * r.readBits(15) => 0b10001001001111111 |
| 32 | * r.readBit() => 0b0 |
| 33 | **/ |
| 34 | |
| 35 | class BitReader |
| 36 | { |
| 37 | ReadBuffer & buf; |
| 38 | |
| 39 | UInt64 bits_buffer; |
| 40 | UInt8 bits_count; |
| 41 | static constexpr UInt8 BIT_BUFFER_SIZE = sizeof(bits_buffer) * 8; |
| 42 | |
| 43 | public: |
| 44 | BitReader(ReadBuffer & buf_) |
| 45 | : buf(buf_), |
| 46 | bits_buffer(0), |
| 47 | bits_count(0) |
| 48 | {} |
| 49 | |
| 50 | ~BitReader() |
| 51 | {} |
| 52 | |
| 53 | inline UInt64 readBits(UInt8 bits) |
| 54 | { |
| 55 | UInt64 result = 0; |
| 56 | bits = std::min(static_cast<UInt8>(sizeof(result) * 8), bits); |
| 57 | |
| 58 | while (bits != 0) |
| 59 | { |
| 60 | if (bits_count == 0) |
| 61 | { |
| 62 | fillBuffer(); |
| 63 | if (bits_count == 0) |
| 64 | { |
| 65 | // EOF. |
| 66 | break; |
| 67 | } |
| 68 | } |
| 69 | |
| 70 | const auto to_read = std::min(bits, bits_count); |
| 71 | |
| 72 | const UInt64 v = bits_buffer >> (bits_count - to_read); |
| 73 | const UInt64 mask = maskLowBits<UInt64>(to_read); |
| 74 | const UInt64 value = v & mask; |
| 75 | result |= value; |
| 76 | |
| 77 | // unset bits that were read |
| 78 | bits_buffer &= ~(mask << (bits_count - to_read)); |
| 79 | bits_count -= to_read; |
| 80 | bits -= to_read; |
| 81 | |
| 82 | result <<= std::min(bits, BIT_BUFFER_SIZE); |
| 83 | } |
| 84 | |
| 85 | return result; |
| 86 | } |
| 87 | |
| 88 | inline UInt64 peekBits(UInt8 /*bits*/) |
| 89 | { |
| 90 | return 0; |
| 91 | } |
| 92 | |
| 93 | inline UInt8 readBit() |
| 94 | { |
| 95 | return static_cast<UInt8>(readBits(1)); |
| 96 | } |
| 97 | |
| 98 | inline bool eof() const |
| 99 | { |
| 100 | return bits_count == 0 && buf.eof(); |
| 101 | } |
| 102 | |
| 103 | private: |
| 104 | void fillBuffer() |
| 105 | { |
| 106 | auto read = buf.read(reinterpret_cast<char *>(&bits_buffer), BIT_BUFFER_SIZE / 8); |
| 107 | bits_buffer = be64toh(bits_buffer); |
| 108 | bits_buffer >>= BIT_BUFFER_SIZE - read * 8; |
| 109 | |
| 110 | bits_count = static_cast<UInt8>(read) * 8; |
| 111 | } |
| 112 | }; |
| 113 | |
| 114 | class BitWriter |
| 115 | { |
| 116 | WriteBuffer & buf; |
| 117 | |
| 118 | UInt64 bits_buffer; |
| 119 | UInt8 bits_count; |
| 120 | |
| 121 | static constexpr UInt8 BIT_BUFFER_SIZE = sizeof(bits_buffer) * 8; |
| 122 | |
| 123 | public: |
| 124 | BitWriter(WriteBuffer & buf_) |
| 125 | : buf(buf_), |
| 126 | bits_buffer(0), |
| 127 | bits_count(0) |
| 128 | {} |
| 129 | |
| 130 | ~BitWriter() |
| 131 | { |
| 132 | flush(); |
| 133 | } |
| 134 | |
| 135 | inline void writeBits(UInt8 bits, UInt64 value) |
| 136 | { |
| 137 | bits = std::min(static_cast<UInt8>(sizeof(value) * 8), bits); |
| 138 | |
| 139 | while (bits > 0) |
| 140 | { |
| 141 | auto v = value; |
| 142 | auto to_write = bits; |
| 143 | |
| 144 | const UInt8 capacity = BIT_BUFFER_SIZE - bits_count; |
| 145 | if (capacity < bits) |
| 146 | { |
| 147 | v >>= bits - capacity; |
| 148 | to_write = capacity; |
| 149 | } |
| 150 | |
| 151 | const UInt64 mask = maskLowBits<UInt64>(to_write); |
| 152 | v &= mask; |
| 153 | |
| 154 | bits_buffer <<= to_write; |
| 155 | bits_buffer |= v; |
| 156 | bits_count += to_write; |
| 157 | |
| 158 | if (bits_count < BIT_BUFFER_SIZE) |
| 159 | break; |
| 160 | |
| 161 | doFlush(); |
| 162 | bits -= to_write; |
| 163 | } |
| 164 | } |
| 165 | |
| 166 | inline void flush() |
| 167 | { |
| 168 | if (bits_count != 0) |
| 169 | { |
| 170 | bits_buffer <<= (BIT_BUFFER_SIZE - bits_count); |
| 171 | doFlush(); |
| 172 | } |
| 173 | } |
| 174 | |
| 175 | private: |
| 176 | void doFlush() |
| 177 | { |
| 178 | bits_buffer = htobe64(bits_buffer); |
| 179 | buf.write(reinterpret_cast<const char *>(&bits_buffer), (bits_count + 7) / 8); |
| 180 | |
| 181 | bits_count = 0; |
| 182 | bits_buffer = 0; |
| 183 | } |
| 184 | }; |
| 185 | |
| 186 | } |
| 187 | |