Added a BitBufferWriter subclass that contains methods for writing bit and byte-sized data, along with exponential golomb encoded data.

This pattern (read-only base, writable subclass) was picked to maintain a *Buffer option that doesn't copy the source bits when parsing. ByteBuffer and Buffer both copy. I'm open to discussion on what the type relationship would be, though :)

Tests have been added to ensure the symmetric nature of read/write operations.

BUG=
R=bcornell@google.com, pthatcher@webrtc.org

Review URL: https://webrtc-codereview.appspot.com/45259005

Cr-Commit-Position: refs/heads/master@{#9107}
This commit is contained in:
Noah Richards 2015-04-28 15:13:44 -07:00
parent 80154f6b28
commit 86153c26a0
3 changed files with 307 additions and 20 deletions

View File

@ -10,6 +10,7 @@
#include "webrtc/base/bitbuffer.h"
#include <algorithm>
#include <limits>
#include "webrtc/base/checks.h"
@ -19,8 +20,7 @@ namespace {
// Returns the lowest (right-most) |bit_count| bits in |byte|.
uint8 LowestBits(uint8 byte, size_t bit_count) {
DCHECK_LE(bit_count, 8u);
uint8 mask_shift = 8 - static_cast<uint8>(bit_count);
return byte & (0xFF >> mask_shift);
return byte & ((1 << bit_count) - 1);
}
// Returns the highest (left-most) |bit_count| bits in |byte|, shifted to the
@ -32,6 +32,41 @@ uint8 HighestBits(uint8 byte, size_t bit_count) {
return (byte & mask) >> shift;
}
// Returns the highest byte of |val| in a uint8.
uint8 HighestByte(uint64 val) {
return static_cast<uint8>(val >> 56);
}
// Returns the result of writing partial data from |source|, of
// |source_bit_count| size in the highest bits, to |target| at
// |target_bit_offset| from the highest bit.
uint8 WritePartialByte(uint8 source, size_t source_bit_count,
uint8 target, size_t target_bit_offset) {
DCHECK(target_bit_offset < 8);
DCHECK(source_bit_count < 9);
DCHECK(source_bit_count <= (8 - target_bit_offset));
// Generate a mask for just the bits we're going to overwrite, so:
uint8 mask =
// The number of bits we want, in the most significant bits...
static_cast<uint8>(0xFF << (8 - source_bit_count))
// ...shifted over to the target offset from the most signficant bit.
>> target_bit_offset;
// We want the target, with the bits we'll overwrite masked off, or'ed with
// the bits from the source we want.
return (target & ~mask) | (source >> target_bit_offset);
}
// Counts the number of bits used in the binary representation of val.
size_t CountBits(uint64 val) {
size_t bit_count = 0;
while (val != 0) {
bit_count++;
val >>= 1;
}
return bit_count;
}
} // namespace
namespace rtc {
@ -143,12 +178,104 @@ bool BitBuffer::ReadExponentialGolomb(uint32* val) {
// read the value.
size_t value_bit_count = zero_bit_count + 1;
if (value_bit_count > 32 || !ReadBits(val, value_bit_count)) {
byte_offset_ = original_byte_offset;
bit_offset_ = original_bit_offset;
CHECK(Seek(original_byte_offset, original_bit_offset));
return false;
}
*val -= 1;
return true;
}
void BitBuffer::GetCurrentOffset(
size_t* out_byte_offset, size_t* out_bit_offset) {
CHECK(out_byte_offset != NULL);
CHECK(out_bit_offset != NULL);
*out_byte_offset = byte_offset_;
*out_bit_offset = bit_offset_;
}
bool BitBuffer::Seek(size_t byte_offset, size_t bit_offset) {
if (byte_offset > byte_count_ || bit_offset > 7 ||
(byte_offset == byte_count_ && bit_offset > 0)) {
return false;
}
byte_offset_ = byte_offset;
bit_offset_ = bit_offset;
return true;
}
BitBufferWriter::BitBufferWriter(uint8* bytes, size_t byte_count)
: BitBuffer(bytes, byte_count), writable_bytes_(bytes) {
}
bool BitBufferWriter::WriteUInt8(uint8 val) {
return WriteBits(val, sizeof(uint8) * 8);
}
bool BitBufferWriter::WriteUInt16(uint16 val) {
return WriteBits(val, sizeof(uint16) * 8);
}
bool BitBufferWriter::WriteUInt32(uint32 val) {
return WriteBits(val, sizeof(uint32) * 8);
}
bool BitBufferWriter::WriteBits(uint64 val, size_t bit_count) {
if (bit_count > RemainingBitCount()) {
return false;
}
size_t total_bits = bit_count;
// For simplicity, push the bits we want to read from val to the highest bits.
val <<= (sizeof(uint64) * 8 - bit_count);
uint8* bytes = writable_bytes_ + byte_offset_;
// The first byte is relatively special; the bit offset to write to may put us
// in the middle of the byte, and the total bit count to write may require we
// save the bits at the end of the byte.
size_t remaining_bits_in_current_byte = 8 - bit_offset_;
size_t bits_in_first_byte =
std::min(bit_count, remaining_bits_in_current_byte);
*bytes = WritePartialByte(
HighestByte(val), bits_in_first_byte, *bytes, bit_offset_);
if (bit_count <= remaining_bits_in_current_byte) {
// Nothing left to write, so quit early.
return ConsumeBits(total_bits);
}
// Subtract what we've written from the bit count, shift it off the value, and
// write the remaining full bytes.
val <<= bits_in_first_byte;
bytes++;
bit_count -= bits_in_first_byte;
while (bit_count >= 8) {
*bytes++ = HighestByte(val);
val <<= 8;
bit_count -= 8;
}
// Last byte may also be partial, so write the remaining bits from the top of
// val.
if (bit_count > 0) {
*bytes = WritePartialByte(HighestByte(val), bit_count, *bytes, 0);
}
// All done! Consume the bits we've written.
return ConsumeBits(total_bits);
}
bool BitBufferWriter::WriteExponentialGolomb(uint32 val) {
// We don't support reading UINT32_MAX, because it doesn't fit in a uint32
// when encoded, so don't support writing it either.
if (val == std::numeric_limits<uint32>::max()) {
return false;
}
uint64 val_to_encode = static_cast<uint64>(val) + 1;
// We need to write CountBits(val+1) 0s and then val+1. Since val (as a
// uint64) has leading zeros, we can just write the total golomb encoded size
// worth of bits, knowing the value will appear last.
return WriteBits(val_to_encode, CountBits(val_to_encode) * 2 - 1);
}
} // namespace rtc

View File

@ -16,14 +16,20 @@
namespace rtc {
// A class, similar to ByteBuffer, that can parse bit-sized data out of a set of
// bytes. Has a similar API to the read-only parts of ByteBuffer, plus methods
// for reading bit-sized data and processing exponential golomb encoded data.
// bytes. Has a similar API to ByteBuffer, plus methods for reading bit-sized
// and exponential golomb encoded data. For a writable version, use
// BitBufferWriter. Unlike ByteBuffer, this class doesn't make a copy of the
// source bytes, so it can be used on read-only data.
// Sizes/counts specify bits/bytes, for clarity.
// Byte order is assumed big-endian/network.
class BitBuffer {
public:
BitBuffer(const uint8* bytes, size_t byte_count);
// Gets the current offset, in bytes/bits, from the start of the buffer. The
// bit offset is the offset into the current byte, in the range [0,7].
void GetCurrentOffset(size_t* out_byte_offset, size_t* out_bit_offset);
// The remaining bits in the byte buffer.
uint64 RemainingBitCount() const;
@ -34,14 +40,15 @@ class BitBuffer {
bool ReadUInt32(uint32* val);
// Reads bit-sized values from the buffer. Returns false if there isn't enough
// data left for the specified type.
// data left for the specified bit count..
bool ReadBits(uint32* val, size_t bit_count);
// Peeks bit-sized values from the buffer. Returns false if there isn't enough
// data left for the specified type. Doesn't move the current read offset.
// data left for the specified number of bits. Doesn't move the current
// offset.
bool PeekBits(uint32* val, size_t bit_count);
// Reads the exponential golomb encoded value at the current bit offset.
// Reads the exponential golomb encoded value at the current offset.
// Exponential golomb values are encoded as:
// 1) x = source val + 1
// 2) In binary, write [countbits(x) - 1] 0s, then x
@ -58,7 +65,11 @@ class BitBuffer {
// there aren't enough bits left in the buffer.
bool ConsumeBits(size_t bit_count);
private:
// Sets the current offset to the provied byte/bit offsets. The bit
// offset is from the given byte, in the range [0,7].
bool Seek(size_t byte_offset, size_t bit_offset);
protected:
const uint8* const bytes_;
// The total size of |bytes_|.
size_t byte_count_;
@ -70,6 +81,35 @@ class BitBuffer {
DISALLOW_COPY_AND_ASSIGN(BitBuffer);
};
// A BitBuffer API for write operations. Supports symmetric write APIs to the
// reading APIs of BitBuffer. Note that the read/write offset is shared with the
// BitBuffer API, so both reading and writing will consume bytes/bits.
class BitBufferWriter : public BitBuffer {
public:
// Constructs a bit buffer for the writable buffer of |bytes|.
BitBufferWriter(uint8* bytes, size_t byte_count);
// Writes byte-sized values from the buffer. Returns false if there isn't
// enough data left for the specified type.
bool WriteUInt8(uint8 val);
bool WriteUInt16(uint16 val);
bool WriteUInt32(uint32 val);
// Writes bit-sized values to the buffer. Returns false if there isn't enough
// room left for the specified number of bits.
bool WriteBits(uint64 val, size_t bit_count);
// Writes the exponential golomb encoded version of the supplied value.
// Returns false if there isn't enough room left for the value.
bool WriteExponentialGolomb(uint32 val);
private:
// The buffer, as a writable array.
uint8* const writable_bytes_;
DISALLOW_COPY_AND_ASSIGN(BitBufferWriter);
};
} // namespace rtc
#endif // WEBRTC_BASE_BITBUFFER_H_

View File

@ -135,6 +135,39 @@ TEST(BitBufferTest, ReadBits) {
EXPECT_FALSE(buffer.ReadBits(&val, 1));
}
TEST(BitBufferTest, SetOffsetValues) {
uint8 bytes[4] = {0};
BitBufferWriter buffer(bytes, 4);
size_t byte_offset, bit_offset;
// Bit offsets are [0,7].
EXPECT_TRUE(buffer.Seek(0, 0));
EXPECT_TRUE(buffer.Seek(0, 7));
buffer.GetCurrentOffset(&byte_offset, &bit_offset);
EXPECT_EQ(0u, byte_offset);
EXPECT_EQ(7u, bit_offset);
EXPECT_FALSE(buffer.Seek(0, 8));
buffer.GetCurrentOffset(&byte_offset, &bit_offset);
EXPECT_EQ(0u, byte_offset);
EXPECT_EQ(7u, bit_offset);
// Byte offsets are [0,length]. At byte offset length, the bit offset must be
// 0.
EXPECT_TRUE(buffer.Seek(0, 0));
EXPECT_TRUE(buffer.Seek(2, 4));
buffer.GetCurrentOffset(&byte_offset, &bit_offset);
EXPECT_EQ(2u, byte_offset);
EXPECT_EQ(4u, bit_offset);
EXPECT_TRUE(buffer.Seek(4, 0));
EXPECT_FALSE(buffer.Seek(5, 0));
buffer.GetCurrentOffset(&byte_offset, &bit_offset);
EXPECT_EQ(4u, byte_offset);
EXPECT_EQ(0u, bit_offset);
EXPECT_FALSE(buffer.Seek(4, 1));
// Passing a NULL out parameter is death.
EXPECT_DEATH(buffer.GetCurrentOffset(&byte_offset, NULL), "");
}
uint64 GolombEncoded(uint32 val) {
val++;
uint32 bit_counter = val;
@ -146,19 +179,23 @@ uint64 GolombEncoded(uint32 val) {
return static_cast<uint64>(val) << (64 - (bit_count * 2 - 1));
}
TEST(BitBufferTest, GolombString) {
char test_string[] = "my precious";
for (size_t i = 0; i < ARRAY_SIZE(test_string); ++i) {
uint64 encoded_val = GolombEncoded(test_string[i]);
// Use ByteBuffer to convert to bytes, to account for endianness (BitBuffer
// requires network order).
ByteBuffer byteBuffer;
TEST(BitBufferTest, GolombUint32Values) {
ByteBuffer byteBuffer;
byteBuffer.Resize(16);
BitBuffer buffer(reinterpret_cast<const uint8*>(byteBuffer.Data()),
byteBuffer.Capacity());
// Test over the uint32 range with a large enough step that the test doesn't
// take forever. Around 20,000 iterations should do.
const int kStep = std::numeric_limits<uint32>::max() / 20000;
for (uint32 i = 0; i < std::numeric_limits<uint32>::max() - kStep;
i += kStep) {
uint64 encoded_val = GolombEncoded(i);
byteBuffer.Clear();
byteBuffer.WriteUInt64(encoded_val);
BitBuffer buffer(reinterpret_cast<const uint8*>(byteBuffer.Data()),
byteBuffer.Length());
uint32 decoded_val;
EXPECT_TRUE(buffer.Seek(0, 0));
EXPECT_TRUE(buffer.ReadExponentialGolomb(&decoded_val));
EXPECT_EQ(test_string[i], static_cast<char>(decoded_val));
EXPECT_EQ(i, decoded_val);
}
}
@ -180,4 +217,87 @@ TEST(BitBufferTest, NoGolombOverread) {
EXPECT_EQ(0x01FEu, decoded_val);
}
TEST(BitBufferWriterTest, SymmetricReadWrite) {
uint8 bytes[16] = {0};
BitBufferWriter buffer(bytes, 4);
// Write some bit data at various sizes.
EXPECT_TRUE(buffer.WriteBits(0x2u, 3));
EXPECT_TRUE(buffer.WriteBits(0x1u, 2));
EXPECT_TRUE(buffer.WriteBits(0x53u, 7));
EXPECT_TRUE(buffer.WriteBits(0x0u, 2));
EXPECT_TRUE(buffer.WriteBits(0x1u, 1));
EXPECT_TRUE(buffer.WriteBits(0x1ABCDu, 17));
// That should be all that fits in the buffer.
EXPECT_FALSE(buffer.WriteBits(1, 1));
EXPECT_TRUE(buffer.Seek(0, 0));
uint32 val;
EXPECT_TRUE(buffer.ReadBits(&val, 3));
EXPECT_EQ(0x2u, val);
EXPECT_TRUE(buffer.ReadBits(&val, 2));
EXPECT_EQ(0x1u, val);
EXPECT_TRUE(buffer.ReadBits(&val, 7));
EXPECT_EQ(0x53u, val);
EXPECT_TRUE(buffer.ReadBits(&val, 2));
EXPECT_EQ(0x0u, val);
EXPECT_TRUE(buffer.ReadBits(&val, 1));
EXPECT_EQ(0x1u, val);
EXPECT_TRUE(buffer.ReadBits(&val, 17));
EXPECT_EQ(0x1ABCDu, val);
// And there should be nothing left.
EXPECT_FALSE(buffer.ReadBits(&val, 1));
}
TEST(BitBufferWriterTest, SymmetricBytesMisaligned) {
uint8 bytes[16] = {0};
BitBufferWriter buffer(bytes, 16);
// Offset 3, to get things misaligned.
EXPECT_TRUE(buffer.ConsumeBits(3));
EXPECT_TRUE(buffer.WriteUInt8(0x12u));
EXPECT_TRUE(buffer.WriteUInt16(0x3456u));
EXPECT_TRUE(buffer.WriteUInt32(0x789ABCDEu));
buffer.Seek(0, 3);
uint8 val8;
uint16 val16;
uint32 val32;
EXPECT_TRUE(buffer.ReadUInt8(&val8));
EXPECT_EQ(0x12u, val8);
EXPECT_TRUE(buffer.ReadUInt16(&val16));
EXPECT_EQ(0x3456u, val16);
EXPECT_TRUE(buffer.ReadUInt32(&val32));
EXPECT_EQ(0x789ABCDEu, val32);
}
TEST(BitBufferWriterTest, SymmetricGolomb) {
char test_string[] = "my precious";
uint8 bytes[64] = {0};
BitBufferWriter buffer(bytes, 64);
for (size_t i = 0; i < ARRAY_SIZE(test_string); ++i) {
EXPECT_TRUE(buffer.WriteExponentialGolomb(test_string[i]));
}
buffer.Seek(0, 0);
for (size_t i = 0; i < ARRAY_SIZE(test_string); ++i) {
uint32 val;
EXPECT_TRUE(buffer.ReadExponentialGolomb(&val));
EXPECT_LE(val, std::numeric_limits<uint8>::max());
EXPECT_EQ(test_string[i], static_cast<char>(val));
}
}
TEST(BitBufferWriterTest, WriteClearsBits) {
uint8 bytes[] = {0xFF, 0xFF};
BitBufferWriter buffer(bytes, 2);
EXPECT_TRUE(buffer.ConsumeBits(3));
EXPECT_TRUE(buffer.WriteBits(0, 1));
EXPECT_EQ(0xEFu, bytes[0]);
EXPECT_TRUE(buffer.WriteBits(0, 3));
EXPECT_EQ(0xE1u, bytes[0]);
EXPECT_TRUE(buffer.WriteBits(0, 2));
EXPECT_EQ(0xE0u, bytes[0]);
EXPECT_EQ(0x7F, bytes[1]);
}
} // namespace rtc