ziphandlerout.cpp

7z一个高压缩比的压缩程序源代码,重要的是里面的算法值得学习

// ZipHandlerOut.cpp

#include "StdAfx.h"

#include "Common/ComTry.h"
#include "Common/StringConvert.h"
#include "Common/StringToInt.h"

#include "Windows/PropVariant.h"
#include "Windows/Time.h"

#include "../../IPassword.h"

#include "../../Common/OutBuffer.h"

#include "../../Crypto/WzAes.h"

#include "../Common/ItemNameUtils.h"
#include "../Common/ParseProperties.h"

#include "ZipHandler.h"
#include "ZipUpdate.h"

using namespace NWindows;
using namespace NCOM;
using namespace NTime;

namespace NArchive {
namespace NZip {

static const UInt32 kLzAlgoX1 = 0;
static const UInt32 kLzAlgoX5 = 1;

static const UInt32 kDeflateNumPassesX1  = 1;
static const UInt32 kDeflateNumPassesX7  = 3;
static const UInt32 kDeflateNumPassesX9  = 10;

static const UInt32 kDeflateNumFastBytesX1 = 32;
static const UInt32 kDeflateNumFastBytesX7 = 64;
static const UInt32 kDeflateNumFastBytesX9 = 128;

static const wchar_t *kLzmaMatchFinderX1 = L"HC4";
static const wchar_t *kLzmaMatchFinderX5 = L"BT4";

static const UInt32 kLzmaNumFastBytesX1 = 32;
static const UInt32 kLzmaNumFastBytesX7 = 64;

static const UInt32 kLzmaDicSizeX1 = 1 << 16;
static const UInt32 kLzmaDicSizeX3 = 1 << 20;
static const UInt32 kLzmaDicSizeX5 = 1 << 24;
static const UInt32 kLzmaDicSizeX7 = 1 << 25;
static const UInt32 kLzmaDicSizeX9 = 1 << 26;

static const UInt32 kBZip2NumPassesX1 = 1;
static const UInt32 kBZip2NumPassesX7 = 2;
static const UInt32 kBZip2NumPassesX9 = 7;

static const UInt32 kBZip2DicSizeX1 = 100000;
static const UInt32 kBZip2DicSizeX3 = 500000;
static const UInt32 kBZip2DicSizeX5 = 900000;

STDMETHODIMP CHandler::GetFileTimeType(UInt32 *timeType)
{
  *timeType = NFileTimeType::kDOS;
  return S_OK;
}

static bool IsAsciiString(const UString &s)
{
  for (int i = 0; i < s.Length(); i++)
  {
    wchar_t c = s[i];
    if (c < 0x20 || c > 0x7F)
      return false;
  }
  return true;
}

#define COM_TRY_BEGIN2 try {
#define COM_TRY_END2 } \
catch(const CSystemException &e) { return e.ErrorCode; } \
catch(...) { return E_OUTOFMEMORY; }

static HRESULT GetTime(IArchiveUpdateCallback *callback, int index, PROPID propID, FILETIME &filetime)
{
  filetime.dwHighDateTime = filetime.dwLowDateTime = 0;
  NCOM::CPropVariant prop;
  RINOK(callback->GetProperty(index, propID, &prop));
  if (prop.vt == VT_FILETIME)
    filetime = prop.filetime;
  else if (prop.vt != VT_EMPTY)
    return E_INVALIDARG;
  return S_OK;
}

STDMETHODIMP CHandler::UpdateItems(ISequentialOutStream *outStream, UInt32 numItems,
    IArchiveUpdateCallback *callback)
{
  COM_TRY_BEGIN2
  CObjectVector<CUpdateItem> updateItems;
  bool thereAreAesUpdates = false;
  for (UInt32 i = 0; i < numItems; i++)
  {
    CUpdateItem ui;
    Int32 newData;
    Int32 newProperties;
    UInt32 indexInArchive;
    if (!callback)
      return E_FAIL;
    RINOK(callback->GetUpdateItemInfo(i, &newData, &newProperties, &indexInArchive));
    ui.NewProperties = IntToBool(newProperties);
    ui.NewData = IntToBool(newData);
    ui.IndexInArchive = indexInArchive;
    ui.IndexInClient = i;
    bool existInArchive = (indexInArchive != UInt32(-1));
    if (existInArchive && newData)
      if (m_Items[indexInArchive].IsAesEncrypted())
        thereAreAesUpdates = true;

    if (IntToBool(newProperties))
    {
      UString name;
      {
        NCOM::CPropVariant prop;
        RINOK(callback->GetProperty(i, kpidAttrib, &prop));
        if (prop.vt == VT_EMPTY)
          ui.Attributes = 0;
        else if (prop.vt != VT_UI4)
          return E_INVALIDARG;
        else
          ui.Attributes = prop.ulVal;
      }

      {
        NCOM::CPropVariant prop;
        RINOK(callback->GetProperty(i, kpidPath, &prop));
        if (prop.vt == VT_EMPTY)
          name.Empty();
        else if (prop.vt != VT_BSTR)
          return E_INVALIDARG;
        else
          name = prop.bstrVal;
      }
      {
        NCOM::CPropVariant prop;
        RINOK(callback->GetProperty(i, kpidIsDir, &prop));
        if (prop.vt == VT_EMPTY)
          ui.IsDir = false;
        else if (prop.vt != VT_BOOL)
          return E_INVALIDARG;
        else
          ui.IsDir = (prop.boolVal != VARIANT_FALSE);
      }

      {
        CPropVariant prop;
        RINOK(callback->GetProperty(i, kpidTimeType, &prop));
        if (prop.vt == VT_UI4)
          ui.NtfsTimeIsDefined = (prop.ulVal == NFileTimeType::kWindows);
        else
          ui.NtfsTimeIsDefined = m_WriteNtfsTimeExtra;
      }
      RINOK(GetTime(callback, i, kpidMTime, ui.NtfsMTime));
      RINOK(GetTime(callback, i, kpidATime, ui.NtfsATime));
      RINOK(GetTime(callback, i, kpidCTime, ui.NtfsCTime));

      {
        FILETIME localFileTime = { 0, 0 };
        if (ui.NtfsMTime.dwHighDateTime != 0 ||
            ui.NtfsMTime.dwLowDateTime != 0)
          if (!FileTimeToLocalFileTime(&ui.NtfsMTime, &localFileTime))
            return E_INVALIDARG;
        FileTimeToDosTime(localFileTime, ui.Time);
      }

      name = NItemName::MakeLegalName(name);
      bool needSlash = ui.IsDir;
      const wchar_t kSlash = L'/';
      if (!name.IsEmpty())
      {
        if (name[name.Length() - 1] == kSlash)
        {
          if (!ui.IsDir)
            return E_INVALIDARG;
          needSlash = false;
        }
      }
      if (needSlash)
        name += kSlash;

      bool tryUtf8 = true;
      if (m_ForseLocal || !m_ForseUtf8)
      {
        bool defaultCharWasUsed;
        ui.Name = UnicodeStringToMultiByte(name, CP_OEMCP, '_', defaultCharWasUsed);
        tryUtf8 = (!m_ForseLocal && (defaultCharWasUsed ||
          MultiByteToUnicodeString(ui.Name, CP_OEMCP) != name));
      }

      if (tryUtf8)
      {
        int i;
        for (i = 0; i < name.Length() && (unsigned)name[i] < 0x80; i++);
        ui.IsUtf8 = (i != name.Length());
        if (!ConvertUnicodeToUTF8(name, ui.Name))
          return E_INVALIDARG;
      }

      if (ui.Name.Length() >= (1 << 16))
        return E_INVALIDARG;

      ui.IndexInClient = i;
      /*
      if (existInArchive)
      {
        const CItemEx &itemInfo = m_Items[indexInArchive];
        // ui.Commented = itemInfo.IsCommented();
        ui.Commented = false;
        if (ui.Commented)
        {
          ui.CommentRange.Position = itemInfo.GetCommentPosition();
          ui.CommentRange.Size  = itemInfo.CommentSize;
        }
      }
      else
        ui.Commented = false;
      */
    }
    if (IntToBool(newData))
    {
      UInt64 size;
      {
        NCOM::CPropVariant prop;
        RINOK(callback->GetProperty(i, kpidSize, &prop));
        if (prop.vt != VT_UI8)
          return E_INVALIDARG;
        size = prop.uhVal.QuadPart;
      }
      ui.Size = size;
    }
    updateItems.Add(ui);
  }

  CMyComPtr<ICryptoGetTextPassword2> getTextPassword;
  {
    CMyComPtr<IArchiveUpdateCallback> udateCallBack2(callback);
    udateCallBack2.QueryInterface(IID_ICryptoGetTextPassword2, &getTextPassword);
  }
  CCompressionMethodMode options;

  if (getTextPassword)
  {