lzmaencoder.cpp

压缩软件源码

      UInt32 normalMatchPrice = matchPrice + 
        _isRep[state.Index].GetPrice0();
      while(lenEnd < cur + newLen)
        _optimum[++lenEnd].Price = kIfinityPrice;

      for(UInt32 lenTest = newLen; lenTest >= 2; lenTest--)
      {
        UInt32 curBack = _matchDistances[lenTest];
        UInt32 curAndLenPrice = normalMatchPrice + GetPosLenPrice(curBack, lenTest, posState);
        COptimal &optimum = _optimum[cur + lenTest];
        if (curAndLenPrice < optimum.Price) 
        {
          optimum.Price = curAndLenPrice;
          optimum.PosPrev = cur;
          optimum.BackPrev = curBack + kNumRepDistances;
          optimum.Prev1IsChar = false;
        }

        if (_maxMode)
        {
          UInt32 backOffset = curBack + 1;
          UInt32 temp;
          for (temp = lenTest + 1; temp < numAvailableBytes; temp++)
            if (data[temp] != data[(size_t)temp - backOffset])
              break;
          UInt32 lenTest2 = temp - (lenTest + 1);
          if (lenTest2 >= 2)
          {
            CState state2 = state;
            state2.UpdateMatch();
            UInt32 posStateNext = (position + lenTest) & _posStateMask;
            UInt32 curAndLenCharPrice = curAndLenPrice + 
                _isMatch[state2.Index][posStateNext].GetPrice0() +
                _literalEncoder.GetPrice(position + lenTest, data[(size_t)lenTest - 1], 
                true, data[(size_t)lenTest - backOffset], data[lenTest]);
            state2.UpdateChar();
            posStateNext = (position + lenTest + 1) & _posStateMask;
            UInt32 nextMatchPrice = curAndLenCharPrice + _isMatch[state2.Index][posStateNext].GetPrice1();
            UInt32 nextRepMatchPrice = nextMatchPrice + _isRep[state2.Index].GetPrice1();
            
            // for(; lenTest2 >= 2; lenTest2--)
            {
              UInt32 offset = lenTest + 1 + lenTest2;
              while(lenEnd < cur + offset)
                _optimum[++lenEnd].Price = kIfinityPrice;
              UInt32 curAndLenPrice = nextRepMatchPrice + GetRepPrice(
                  0, lenTest2, state2, posStateNext);
              COptimal &optimum = _optimum[cur + offset];
              if (curAndLenPrice < optimum.Price) 
              {
                optimum.Price = curAndLenPrice;
                optimum.PosPrev = cur + lenTest + 1;
                optimum.BackPrev = 0;
                optimum.Prev1IsChar = true;
                optimum.Prev2 = true;
                optimum.PosPrev2 = cur;
                optimum.BackPrev2 = curBack + kNumRepDistances;
              }
            }
          }
        }
      }
    }
  }
}

static inline bool ChangePair(UInt32 smallDist, UInt32 bigDist)
{
  const int kDif = 7;
  return (smallDist < (UInt32(1) << (32-kDif)) && bigDist >= (smallDist << kDif));
}


HRESULT CEncoder::ReadMatchDistances(UInt32 &lenRes)
{
  lenRes = _matchFinder->GetLongestMatch(_matchDistances);
  if (lenRes == _numFastBytes)
    lenRes += _matchFinder->GetMatchLen(lenRes, _matchDistances[lenRes], 
        kMatchMaxLen - lenRes);
  _additionalOffset++;
  return _matchFinder->MovePos();
}

HRESULT CEncoder::GetOptimumFast(UInt32 position, UInt32 &backRes, UInt32 &lenRes)
{
  UInt32 lenMain;
  if (!_longestMatchWasFound)
  {
    RINOK(ReadMatchDistances(lenMain));
  }
  else
  {
    lenMain = _longestMatchLength;
    _longestMatchWasFound = false;
  }
  UInt32 repLens[kNumRepDistances];
  UInt32 repMaxIndex = 0;
  for(UInt32 i = 0; i < kNumRepDistances; i++)
  {
    repLens[i] = _matchFinder->GetMatchLen(0 - 1, _repDistances[i], kMatchMaxLen);
    if (i == 0 || repLens[i] > repLens[repMaxIndex])
      repMaxIndex = i;
  }
  if(repLens[repMaxIndex] >= _numFastBytes)
  {
    backRes = repMaxIndex;
    lenRes = repLens[repMaxIndex];
    return MovePos(lenRes - 1);
  }
  if(lenMain >= _numFastBytes)
  {
    backRes = _matchDistances[_numFastBytes] + kNumRepDistances; 
    lenRes = lenMain;
    return MovePos(lenMain - 1);
  }
  while (lenMain > 2)
  {
    if (!ChangePair(_matchDistances[lenMain - 1], _matchDistances[lenMain]))
      break;
    lenMain--;
  }
  if (lenMain == 2 && _matchDistances[2] >= 0x80)
    lenMain = 1;

  UInt32 backMain = _matchDistances[lenMain];
  if (repLens[repMaxIndex] >= 2)
  {
    if (repLens[repMaxIndex] + 1 >= lenMain || 
        repLens[repMaxIndex] + 2 >= lenMain && (backMain > (1<<12)))
    {
      backRes = repMaxIndex;
      lenRes = repLens[repMaxIndex];
      return MovePos(lenRes - 1);
    }
  }
  

  if (lenMain >= 2)
  {
    RINOK(ReadMatchDistances(_longestMatchLength));
    if (_longestMatchLength >= 2 &&
      (
        (_longestMatchLength >= lenMain && _matchDistances[lenMain] < backMain) || 
        _longestMatchLength == lenMain + 1 && 
          !ChangePair(backMain, _matchDistances[_longestMatchLength]) ||
        _longestMatchLength > lenMain + 1 ||
        _longestMatchLength + 1 >= lenMain && lenMain >= 3 &&
          ChangePair(_matchDistances[lenMain - 1], backMain)
      )
      )
    {
      _longestMatchWasFound = true;
      backRes = UInt32(-1);
      lenRes = 1;
      return S_OK;
    }
    for(UInt32 i = 0; i < kNumRepDistances; i++)
    {
      UInt32 repLen = _matchFinder->GetMatchLen(0 - 1, _repDistances[i], kMatchMaxLen);
      if (repLen >= 2 && repLen + 1 >= lenMain)
      {
        _longestMatchWasFound = true;
        backRes = UInt32(-1);
        lenRes = 1;
        return S_OK;
      }
    }
    backRes = backMain + kNumRepDistances; 
    lenRes = lenMain;
    return MovePos(lenMain - 2);
  }
  backRes = UInt32(-1);
  lenRes = 1;
  return S_OK;
}

STDMETHODIMP CEncoder::InitMatchFinder(IMatchFinder *matchFinder)
{
  _matchFinder = matchFinder;
  return S_OK;
}

HRESULT CEncoder::Flush(UInt32 nowPos)
{
  ReleaseMFStream();
  WriteEndMarker(nowPos & _posStateMask);
  _rangeEncoder.FlushData();
  return _rangeEncoder.FlushStream();
}

void CEncoder::WriteEndMarker(UInt32 posState)
{
  // This function for writing End Mark for stream version of LZMA. 
  // In current version this feature is not used.
  if (!_writeEndMark)
    return;

  _isMatch[_state.Index][posState].Encode(&_rangeEncoder, 1);
  _isRep[_state.Index].Encode(&_rangeEncoder, 0);
  _state.UpdateMatch();
  UInt32 len = kMatchMinLen; // kMatchMaxLen;
  _lenEncoder.Encode(&_rangeEncoder, len - kMatchMinLen, posState);
  UInt32 posSlot = (1 << kNumPosSlotBits)  - 1;
  UInt32 lenToPosState = GetLenToPosState(len);
  _posSlotEncoder[lenToPosState].Encode(&_rangeEncoder, posSlot);
  UInt32 footerBits = 30;
  UInt32 posReduced = (UInt32(1) << footerBits) - 1;
  _rangeEncoder.EncodeDirectBits(posReduced >> kNumAlignBits, footerBits - kNumAlignBits);
  _posAlignEncoder.ReverseEncode(&_rangeEncoder, posReduced & kAlignMask);
}

HRESULT CEncoder::CodeReal(ISequentialInStream *inStream,
      ISequentialOutStream *outStream, 
      const UInt64 *inSize, const UInt64 *outSize,
      ICompressProgressInfo *progress)
{
  _needReleaseMFStream = false;
  CCoderReleaser coderReleaser(this);
  RINOK(SetStreams(inStream, outStream, inSize, outSize));
  while(true)
  {
    UInt64 processedInSize;
    UInt64 processedOutSize;
    Int32 finished;
    RINOK(CodeOneBlock(&processedInSize, &processedOutSize, &finished));
    if (finished != 0)
      return S_OK;
    if (progress != 0)
    {
      RINOK(progress->SetRatioInfo(&processedInSize, &processedOutSize));
    }
  }
}

HRESULT CEncoder::SetStreams(ISequentialInStream *inStream,
      ISequentialOutStream *outStream, 
      const UInt64 *inSize, const UInt64 *outSize)
{
  _inStream = inStream;
  _finished = false;
  RINOK(Create());
  RINOK(SetOutStream(outStream));
  RINOK(Init());
  
  // CCoderReleaser releaser(this);

  /*
  if (_matchFinder->GetNumAvailableBytes() == 0)
    return Flush();
  */

  if (!_fastMode)
  {
    FillPosSlotPrices();
    FillDistancesPrices();
    FillAlignPrices();
  }

  _lenEncoder.SetTableSize(_numFastBytes);
  _lenEncoder.UpdateTables(1 << _posStateBits);
  _repMatchLenEncoder.SetTableSize(_numFastBytes);
  _repMatchLenEncoder.UpdateTables(1 << _posStateBits);

  lastPosSlotFillingPos = 0;
  nowPos64 = 0;
  return S_OK;
}

HRESULT CEncoder::CodeOneBlock(UInt64 *inSize, UInt64 *outSize, Int32 *finished)
{
  if (_inStream != 0)
  {
    RINOK(_matchFinder->Init(_inStream));
    _needReleaseMFStream = true;
    _inStream = 0;
  }


  *finished = 1;
  if (_finished)
    return S_OK;
  _finished = true;


  UInt64 progressPosValuePrev = nowPos64;
  if (nowPos64 == 0)
  {
    if (_matchFinder->GetNumAvailableBytes() == 0)
      return Flush(UInt32(nowPos64));
    UInt32 len; // it's not used
    RINOK(ReadMatchDistances(len));
    UInt32 posState = UInt32(nowPos64) & _posStateMask;
    _isMatch[_state.Index][posState].Encode(&_rangeEncoder, 0);
    _state.UpdateChar();
    Byte curByte = _matchFinder->GetIndexByte(0 - _additionalOffset);
    _literalEncoder.GetSubCoder(UInt32(nowPos64), _previousByte)->Encode(&_rangeEncoder, curByte);
    _previousByte = curByte;
    _additionalOffset--;
    nowPos64++;
  }
  if (_matchFinder->GetNumAvailableBytes() == 0)
    return Flush(UInt32(nowPos64));
  while(true)
  {
    #ifdef _NO_EXCEPTIONS
    if (_rangeEncoder.Stream.ErrorCode != S_OK)
      return _rangeEncoder.Stream.ErrorCode;
    #endif
    UInt32 pos;
    UInt32 posState = UInt32(nowPos64) & _posStateMask;

    UInt32 len;
    HRESULT result;
    if (_fastMode)
      result = GetOptimumFast(UInt32(nowPos64), pos, len);
    else
      result = GetOptimum(UInt32(nowPos64), pos, len);
    RINOK(result);

    if(len == 1 && pos == 0xFFFFFFFF)
    {
      _isMatch[_state.Index][posState].Encode(&_rangeEncoder, 0);
      Byte curByte = _matchFinder->GetIndexByte(0 - _additionalOffset);
      CLiteralEncoder2 *subCoder = _literalEncoder.GetSubCoder(UInt32(nowPos64), _previousByte);
      if(!_state.IsCharState())
      {
        Byte matchByte = _matchFinder->GetIndexByte(0 - _repDistances[0] - 1 - _additionalOffset);
        subCoder->EncodeMatched(&_rangeEncoder, matchByte, curByte);
      }
      else
        subCoder->Encode(&_rangeEncoder, curByte);
      _state.UpdateChar();
      _previousByte = curByte;
    }
    else
    {
      _isMatch[_state.Index][posState].Encode(&_rangeEncoder, 1);
      if(pos < kNumRepDistances)
      {
        _isRep[_state.Index].Encode(&_rangeEncoder, 1);
        if(pos == 0)
        {
          _isRepG0[_state.Index].Encode(&_rangeEncoder, 0);
          if(len == 1)
            _isRep0Long[_state.Index][posState].Encode(&_rangeEncoder, 0);
          else
            _isRep0Long[_state.Index][posState].Encode(&_rangeEncoder, 1);
        }
        else
        {
          _isRepG0[_state.Index].Encode(&_rangeEncoder, 1);
          if (pos == 1)
            _isRepG1[_state.Index].Encode(&_rangeEncoder, 0);
          else
          {
            _isRepG1[_state.Index].Encode(&_rangeEncoder, 1);
            _isRepG2[_state.Index].Encode(&_rangeEncoder, pos - 2);
          }
        }
        if (len == 1)
          _state.UpdateShortRep();
        else
        {
          _repMatchLenEncoder.Encode(&_rangeEncoder, len - kMatchMinLen, posState);
          _state.UpdateRep();
        }


        UInt32 distance = _repDistances[pos];
        if (pos != 0)
        {
          for(UInt32 i = pos; i >= 1; i--)
            _repDistances[i] = _repDistances[i - 1];
          _repDistances[0] = distance;
        }
      }
      else
      {
        _isRep[_state.Index].Encode(&_rangeEncoder, 0);
        _state.UpdateMatch();
        _lenEncoder.Encode(&_rangeEncoder, len - kMatchMinLen, posState);
        pos -= kNumRepDistances;
        UInt32 posSlot = GetPosSlot(pos);
        UInt32 lenToPosState = GetLenToPosState(len);
        _posSlotEncoder[lenToPosState].Encode(&_rangeEncoder, posSlot);
        
        if (posSlot >= kStartPosModelIndex)
        {
          UInt32 footerBits = ((posSlot >> 1) - 1);
          UInt32 base = ((2 | (posSlot & 1)) << footerBits);
          UInt32 posReduced = pos - base;

          if (posSlot < kEndPosModelIndex)
            NRangeCoder::ReverseBitTreeEncode(_posEncoders + base - posSlot - 1, 
                &_rangeEncoder, footerBits, posReduced);
          else
          {
            _rangeEncoder.EncodeDirectBits(posReduced >> kNumAlignBits, footerBits - kNumAlignBits);
            _posAlignEncoder.ReverseEncode(&_rangeEncoder, posReduced & kAlignMask);
            if (!_fastMode)
              if (--_alignPriceCount == 0)
                FillAlignPrices();
          }
        }
        UInt32 distance = pos;
        for(UInt32 i = kNumRepDistances - 1; i >= 1; i--)
          _repDistances[i] = _repDistances[i - 1];
        _repDistances[0] = distance;
      }
      _previousByte = _matchFinder->GetIndexByte(len - 1 - _additionalOffset);
    }
    _additionalOffset -= len;
    nowPos64 += len;
    if (!_fastMode)
      if (nowPos64 - lastPosSlotFillingPos >= (1 << 9))
      {
        FillPosSlotPrices();
        FillDistancesPrices();
        lastPosSlotFillingPos = nowPos64;
      }
    if (_additionalOffset == 0)
    {
      *inSize = nowPos64;
      *outSize = _rangeEncoder.GetProcessedSize();
      if (_matchFinder->GetNumAvailableBytes() == 0)
        return Flush(UInt32(nowPos64));
      if (nowPos64 - progressPosValuePrev >= (1 << 12))
      {
        _finished = false;
        *finished = 0;
        return S_OK;
      }
    }
  }
}

STDMETHODIMP CEncoder::Code(ISequentialInStream *inStream,
    ISequentialOutStream *outStream, const UInt64 *inSize, const UInt64 *outSize,
    ICompressProgressInfo *progress)
{
  #ifndef _NO_EXCEPTIONS
  try 
  { 
  #endif
    return CodeReal(inStream, outStream, inSize, outSize, progress); 
  #ifndef _NO_EXCEPTIONS
  }
  catch(const COutBufferException &e) { return e.ErrorCode; }
  catch(...) { return E_FAIL; }
  #endif
}
  
void CEncoder::FillPosSlotPrices()
{
  for (UInt32 lenToPosState = 0; lenToPosState < kNumLenToPosStates; lenToPosState++)
  {
	  UInt32 posSlot;
    for (posSlot = 0; posSlot < kEndPosModelIndex && posSlot < _distTableSize; posSlot++)
      _posSlotPrices[lenToPosState][posSlot] = _posSlotEncoder[lenToPosState].GetPrice(posSlot);
    for (; posSlot < _distTableSize; posSlot++)
      _posSlotPrices[lenToPosState][posSlot] = _posSlotEncoder[lenToPosState].GetPrice(posSlot) + 
      ((((posSlot >> 1) - 1) - kNumAlignBits) << NRangeCoder::kNumBitPriceShiftBits);
  }
}

void CEncoder::FillDistancesPrices()
{
  for (UInt32 lenToPosState = 0; lenToPosState < kNumLenToPosStates; lenToPosState++)
  {
	  UInt32 i;
    for (i = 0; i < kStartPosModelIndex; i++)
      _distancesPrices[lenToPosState][i] = _posSlotPrices[lenToPosState][i];
    for (; i < kNumFullDistances; i++)
    { 
      UInt32 posSlot = GetPosSlot(i);
      UInt32 footerBits = ((posSlot >> 1) - 1);
      UInt32 base = ((2 | (posSlot & 1)) << footerBits);

      _distancesPrices[lenToPosState][i] = _posSlotPrices[lenToPosState][posSlot] +
          NRangeCoder::ReverseBitTreeGetPrice(_posEncoders + 
              base - posSlot - 1, footerBits, i - base);
            
    }
  }
}

void CEncoder::FillAlignPrices()
{
  for (UInt32 i = 0; i < kAlignTableSize; i++)
    _alignPrices[i] = _posAlignEncoder.ReverseGetPrice(i);
  _alignPriceCount = kAlignTableSize;
}

}}