bzip2decoder.cpp

著名的7zip的压缩算法,压缩比最高,但是压缩时间也比较长!

      } 
      if (nextSym <= (UInt32)numInUse) 
      {
        Byte b = mtf.GetAndMove((int)nextSym - 1);
        if (blockSize >= blockSizeMax) 
          return S_FALSE;
        state.CharCounters[b]++;
        state.tt[blockSize++] = (UInt32)b;
      }
      else if (nextSym == (UInt32)numInUse + 1) 
        break;
      else 
        return S_FALSE;
    }
  }
  if (state.OrigPtr >= blockSize)
    return S_FALSE;
  state.BlockSize = blockSize;
  return S_OK;
}

void CState::DecodeBlock1()
{
  UInt32 *charCounters = this->CharCounters;
  {
    UInt32 sum = 0;
    for (UInt32 i = 0; i < 256; i++) 
    {
      sum += charCounters[i];
      charCounters[i] = sum - charCounters[i];
    }
  }
  
  // Compute the T^(-1) vector
  UInt32 blockSize = this->BlockSize;
  UInt32 i = 0;
  do
    tt[charCounters[tt[i] & 0xFF]++] |= (i << 8);
  while(++i < blockSize);
}

UInt32 CState::DecodeBlock2(COutBuffer &m_OutStream)
{
  UInt32 blockSize = this->BlockSize;

  CBZip2CRC crc;
  
  UInt32 randIndex = 1;
  UInt32 randToGo = kRandNums[0] - 2;
  
  int numReps = 0;

  // it's for speed optimization: prefetch & prevByte_init;
  UInt32 tPos = tt[tt[OrigPtr] >> 8];
  Byte prevByte = (Byte)(tPos & 0xFF);
  
  do
  {
    Byte b = (Byte)(tPos & 0xFF);
    tPos = tt[tPos >> 8];
    
    if (BlockRandomised) 
    {
      if (randToGo == 0) 
      {
        b ^= 1;
        randToGo = kRandNums[randIndex++];
        randIndex &= 0x1FF;
      }
      randToGo--;
    }
    
    if (numReps == kRleModeRepSize)
    {
      for (; b > 0; b--)
      {
        crc.UpdateByte(prevByte);
        m_OutStream.WriteByte(prevByte);
      }
      numReps = 0;
      continue;
    }
    if (prevByte == b)
      numReps++;
    else
    {
      numReps = 1;
      prevByte = b;
    }
    crc.UpdateByte(b);
    m_OutStream.WriteByte(b);
  }
  while(--blockSize != 0);
  return crc.GetDigest();
}

#ifdef COMPRESS_BZIP2_MT
CDecoder::CDecoder():
  m_States(0)
{
  m_NumThreadsPrev = 0;
  NumThreads = 1;
  CS.Enter();
}

CDecoder::~CDecoder()
{
  Free();
}

bool CDecoder::Create()
{
  try 
  { 
    if (m_States != 0 && m_NumThreadsPrev == NumThreads)
      return true;
    Free();
    MtMode = (NumThreads > 1);
    m_NumThreadsPrev = NumThreads;
    m_States = new CState[NumThreads];
    if (m_States == 0)
      return false;
    #ifdef COMPRESS_BZIP2_MT
    for (UInt32 t = 0; t < NumThreads; t++)
    {
      CState &ti = m_States[t];
      ti.Decoder = this;
      if (MtMode)
        if (!ti.Thread.Create(MFThread, &ti))
        {
          NumThreads = t;
          Free();
          return false; 
        }
    }
    #endif
  }
  catch(...) { return false; }
  return true;
}

void CDecoder::Free()
{
  if (!m_States)
    return;
  CloseThreads = true;
  CS.Leave();
  for (UInt32 t = 0; t < NumThreads; t++)
  {
    CState &s = m_States[t];
    if (MtMode)
      s.Thread.Wait();
    s.Free();
  }
  delete []m_States;
  m_States = 0;
}
#endif

HRESULT CDecoder::ReadSignatures(bool &wasFinished, UInt32 &crc)
{
  wasFinished = false;
  Byte s[6];
  for (int i = 0; i < 6; i++)
    s[i] = ReadByte();
  crc = ReadCRC();
  if (s[0] == kFinSig0)
  {
    if (s[1] != kFinSig1 || 
        s[2] != kFinSig2 || 
        s[3] != kFinSig3 || 
        s[4] != kFinSig4 || 
        s[5] != kFinSig5)
      return S_FALSE;
    
    wasFinished = true;
    return (crc == CombinedCRC.GetDigest()) ? S_OK : S_FALSE;
  }
  if (s[0] != kBlockSig0 || 
      s[1] != kBlockSig1 || 
      s[2] != kBlockSig2 || 
      s[3] != kBlockSig3 || 
      s[4] != kBlockSig4 || 
      s[5] != kBlockSig5)
    return S_FALSE;
  CombinedCRC.Update(crc);
  return S_OK;
}

HRESULT CDecoder::DecodeFile(bool &isBZ, ICompressProgressInfo *progress)
{
  #ifdef COMPRESS_BZIP2_MT
  Progress = progress;
  if (!Create())
    return E_FAIL;
  for (UInt32 t = 0; t < NumThreads; t++)
  {
    CState &s = m_States[t];
    if (!s.Alloc())
      return E_OUTOFMEMORY;
    s.StreamWasFinishedEvent.Reset();
    s.WaitingWasStartedEvent.Reset();
    s.CanWriteEvent.Reset();
  }
  #else
  if (!m_States[0].Alloc())
    return E_OUTOFMEMORY;
  #endif

  isBZ = false;
  Byte s[6];
  int i;
  for (i = 0; i < 4; i++)
    s[i] = ReadByte();
  if (s[0] != kArSig0 || 
      s[1] != kArSig1 || 
      s[2] != kArSig2 || 
      s[3] <= kArSig3 || 
      s[3] > kArSig3 + kBlockSizeMultMax)
    return S_OK;
  isBZ = true;
  UInt32 dicSize = (UInt32)(s[3] - kArSig3) * kBlockSizeStep;

  CombinedCRC.Init();
  #ifdef COMPRESS_BZIP2_MT
  if (MtMode)
  {
    NextBlockIndex = 0;
    StreamWasFinished = false;
    CloseThreads = false;
    CanStartWaitingEvent.Reset();
    m_States[0].CanWriteEvent.Set();
    BlockSizeMax = dicSize;
    Result = S_OK;
    CS.Leave();
    UInt32 t;
    for (t = 0; t < NumThreads; t++)
      m_States[t].StreamWasFinishedEvent.Lock();
    CS.Enter();
    CanStartWaitingEvent.Set();
    for (t = 0; t < NumThreads; t++)
      m_States[t].WaitingWasStartedEvent.Lock();
    CanStartWaitingEvent.Reset();
    RINOK(Result);
  }
  else
  #endif
  {
    CState &state = m_States[0];
    while (true)
    {
      if (progress)
      {
        UInt64 packSize = m_InStream.GetProcessedSize();
        UInt64 unpackSize = m_OutStream.GetProcessedSize();
        RINOK(progress->SetRatioInfo(&packSize, &unpackSize));
      }
      bool wasFinished;
      UInt32 crc;
      RINOK(ReadSignatures(wasFinished, crc));
      if (wasFinished)
        return S_OK;

      RINOK(ReadBlock(dicSize, state));
      state.DecodeBlock1();
      if (state.DecodeBlock2(m_OutStream) != crc)
        return S_FALSE;
    }
  }
  return S_OK;
}

HRESULT CDecoder::CodeReal(ISequentialInStream *inStream,
    ISequentialOutStream *outStream, const UInt64 *inSize, const UInt64 *outSize,
    ICompressProgressInfo *progress)
{
  if (!m_InStream.Create(kBufferSize))
    return E_OUTOFMEMORY;
  if (!m_OutStream.Create(kBufferSize))
    return E_OUTOFMEMORY;

  m_InStream.SetStream(inStream);
  m_InStream.Init();

  m_OutStream.SetStream(outStream);
  m_OutStream.Init();

  CDecoderFlusher flusher(this);

  bool isBZ;
  RINOK(DecodeFile(isBZ, progress));
  return isBZ ? S_OK: S_FALSE;
}

STDMETHODIMP CDecoder::Code(ISequentialInStream *inStream,
    ISequentialOutStream *outStream, const UInt64 *inSize, const UInt64 *outSize,
    ICompressProgressInfo *progress)
{
  try { return CodeReal(inStream, outStream, inSize, outSize, progress); }
  catch(...) { return S_FALSE; }
}

STDMETHODIMP CDecoder::GetInStreamProcessedSize(UInt64 *value)
{
  if (value == NULL)
    return E_INVALIDARG;
  *value = m_InStream.GetProcessedSize();
  return S_OK;
}

#ifdef COMPRESS_BZIP2_MT
STDMETHODIMP CDecoder::SetNumberOfThreads(UInt32 numThreads)
{
  NumThreads = numThreads;
  if (NumThreads < 1)
    NumThreads = 1;
  if (NumThreads > kNumThreadsMax)
    NumThreads = kNumThreadsMax;
  return S_OK;
}
#endif

}}