7zin.c

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

/* 7zIn.c -- 7z Input functions
2008-12-31 : Igor Pavlov : Public domain */

#include "../../7zCrc.h"
#include "../../CpuArch.h"

#include "7zDecode.h"
#include "7zIn.h"

#define RINOM(x) { if ((x) == 0) return SZ_ERROR_MEM; }

#define NUM_FOLDER_CODERS_MAX 32
#define NUM_CODER_STREAMS_MAX 32

void SzArEx_Init(CSzArEx *p)
{
  SzAr_Init(&p->db);
  p->FolderStartPackStreamIndex = 0;
  p->PackStreamStartPositions = 0;
  p->FolderStartFileIndex = 0;
  p->FileIndexToFolderIndexMap = 0;
}

void SzArEx_Free(CSzArEx *p, ISzAlloc *alloc)
{
  IAlloc_Free(alloc, p->FolderStartPackStreamIndex);
  IAlloc_Free(alloc, p->PackStreamStartPositions);
  IAlloc_Free(alloc, p->FolderStartFileIndex);
  IAlloc_Free(alloc, p->FileIndexToFolderIndexMap);
  SzAr_Free(&p->db, alloc);
  SzArEx_Init(p);
}

/*
UInt64 GetFolderPackStreamSize(int folderIndex, int streamIndex) const
{
  return PackSizes[FolderStartPackStreamIndex[folderIndex] + streamIndex];
}

UInt64 GetFilePackSize(int fileIndex) const
{
  int folderIndex = FileIndexToFolderIndexMap[fileIndex];
  if (folderIndex >= 0)
  {
    const CSzFolder &folderInfo = Folders[folderIndex];
    if (FolderStartFileIndex[folderIndex] == fileIndex)
    return GetFolderFullPackSize(folderIndex);
  }
  return 0;
}
*/

#define MY_ALLOC(T, p, size, alloc) { if ((size) == 0) p = 0; else \
  if ((p = (T *)IAlloc_Alloc(alloc, (size) * sizeof(T))) == 0) return SZ_ERROR_MEM; }

static SRes SzArEx_Fill(CSzArEx *p, ISzAlloc *alloc)
{
  UInt32 startPos = 0;
  UInt64 startPosSize = 0;
  UInt32 i;
  UInt32 folderIndex = 0;
  UInt32 indexInFolder = 0;
  MY_ALLOC(UInt32, p->FolderStartPackStreamIndex, p->db.NumFolders, alloc);
  for (i = 0; i < p->db.NumFolders; i++)
  {
    p->FolderStartPackStreamIndex[i] = startPos;
    startPos += p->db.Folders[i].NumPackStreams;
  }

  MY_ALLOC(UInt64, p->PackStreamStartPositions, p->db.NumPackStreams, alloc);

  for (i = 0; i < p->db.NumPackStreams; i++)
  {
    p->PackStreamStartPositions[i] = startPosSize;
    startPosSize += p->db.PackSizes[i];
  }

  MY_ALLOC(UInt32, p->FolderStartFileIndex, p->db.NumFolders, alloc);
  MY_ALLOC(UInt32, p->FileIndexToFolderIndexMap, p->db.NumFiles, alloc);

  for (i = 0; i < p->db.NumFiles; i++)
  {
    CSzFileItem *file = p->db.Files + i;
    int emptyStream = !file->HasStream;
    if (emptyStream && indexInFolder == 0)
    {
      p->FileIndexToFolderIndexMap[i] = (UInt32)-1;
      continue;
    }
    if (indexInFolder == 0)
    {
      /*
      v3.13 incorrectly worked with empty folders
      v4.07: Loop for skipping empty folders
      */
      for (;;)
      {
        if (folderIndex >= p->db.NumFolders)
          return SZ_ERROR_ARCHIVE;
        p->FolderStartFileIndex[folderIndex] = i;
        if (p->db.Folders[folderIndex].NumUnpackStreams != 0)
          break;
        folderIndex++;
      }
    }
    p->FileIndexToFolderIndexMap[i] = folderIndex;
    if (emptyStream)
      continue;
    indexInFolder++;
    if (indexInFolder >= p->db.Folders[folderIndex].NumUnpackStreams)
    {
      folderIndex++;
      indexInFolder = 0;
    }
  }
  return SZ_OK;
}


UInt64 SzArEx_GetFolderStreamPos(const CSzArEx *p, UInt32 folderIndex, UInt32 indexInFolder)
{
  return p->dataPos +
    p->PackStreamStartPositions[p->FolderStartPackStreamIndex[folderIndex] + indexInFolder];
}

int SzArEx_GetFolderFullPackSize(const CSzArEx *p, UInt32 folderIndex, UInt64 *resSize)
{
  UInt32 packStreamIndex = p->FolderStartPackStreamIndex[folderIndex];
  CSzFolder *folder = p->db.Folders + folderIndex;
  UInt64 size = 0;
  UInt32 i;
  for (i = 0; i < folder->NumPackStreams; i++)
  {
    UInt64 t = size + p->db.PackSizes[packStreamIndex + i];
    if (t < size) /* check it */
      return SZ_ERROR_FAIL;
    size = t;
  }
  *resSize = size;
  return SZ_OK;
}


/*
SRes SzReadTime(const CObjectVector<CBuf> &dataVector,
    CObjectVector<CSzFileItem> &files, UInt64 type)
{
  CBoolVector boolVector;
  RINOK(ReadBoolVector2(files.Size(), boolVector))

  CStreamSwitch streamSwitch;
  RINOK(streamSwitch.Set(this, &dataVector));

  for (int i = 0; i < files.Size(); i++)
  {
    CSzFileItem &file = files[i];
    CArchiveFileTime fileTime;
    bool defined = boolVector[i];
    if (defined)
    {
      UInt32 low, high;
      RINOK(SzReadUInt32(low));
      RINOK(SzReadUInt32(high));
      fileTime.dwLowDateTime = low;
      fileTime.dwHighDateTime = high;
    }
    switch(type)
    {
      case k7zIdCTime: file.IsCTimeDefined = defined; if (defined) file.CTime = fileTime; break;
      case k7zIdATime: file.IsATimeDefined = defined; if (defined) file.ATime = fileTime; break;
      case k7zIdMTime: file.IsMTimeDefined = defined; if (defined) file.MTime = fileTime; break;
    }
  }
  return SZ_OK;
}
*/

static int TestSignatureCandidate(Byte *testBytes)
{
  size_t i;
  for (i = 0; i < k7zSignatureSize; i++)
    if (testBytes[i] != k7zSignature[i])
      return 0;
  return 1;
}

typedef struct _CSzState
{
  Byte *Data;
  size_t Size;
}CSzData;

static SRes SzReadByte(CSzData *sd, Byte *b)
{
  if (sd->Size == 0)
    return SZ_ERROR_ARCHIVE;
  sd->Size--;
  *b = *sd->Data++;
  return SZ_OK;
}

static SRes SzReadBytes(CSzData *sd, Byte *data, size_t size)
{
  size_t i;
  for (i = 0; i < size; i++)
  {
    RINOK(SzReadByte(sd, data + i));
  }
  return SZ_OK;
}

static SRes SzReadUInt32(CSzData *sd, UInt32 *value)
{
  int i;
  *value = 0;
  for (i = 0; i < 4; i++)
  {
    Byte b;
    RINOK(SzReadByte(sd, &b));
    *value |= ((UInt32)(b) << (8 * i));
  }
  return SZ_OK;
}

static SRes SzReadNumber(CSzData *sd, UInt64 *value)
{
  Byte firstByte;
  Byte mask = 0x80;
  int i;
  RINOK(SzReadByte(sd, &firstByte));
  *value = 0;
  for (i = 0; i < 8; i++)
  {
    Byte b;
    if ((firstByte & mask) == 0)
    {
      UInt64 highPart = firstByte & (mask - 1);
      *value += (highPart << (8 * i));
      return SZ_OK;
    }
    RINOK(SzReadByte(sd, &b));
    *value |= ((UInt64)b << (8 * i));
    mask >>= 1;
  }
  return SZ_OK;
}

static SRes SzReadNumber32(CSzData *sd, UInt32 *value)
{
  UInt64 value64;
  RINOK(SzReadNumber(sd, &value64));
  if (value64 >= 0x80000000)
    return SZ_ERROR_UNSUPPORTED;
  if (value64 >= ((UInt64)(1) << ((sizeof(size_t) - 1) * 8 + 2)))
    return SZ_ERROR_UNSUPPORTED;
  *value = (UInt32)value64;
  return SZ_OK;
}

static SRes SzReadID(CSzData *sd, UInt64 *value)
{
  return SzReadNumber(sd, value);
}

static SRes SzSkeepDataSize(CSzData *sd, UInt64 size)
{
  if (size > sd->Size)
    return SZ_ERROR_ARCHIVE;
  sd->Size -= (size_t)size;
  sd->Data += (size_t)size;
  return SZ_OK;
}

static SRes SzSkeepData(CSzData *sd)
{
  UInt64 size;
  RINOK(SzReadNumber(sd, &size));
  return SzSkeepDataSize(sd, size);
}

static SRes SzReadArchiveProperties(CSzData *sd)
{
  for (;;)
  {
    UInt64 type;
    RINOK(SzReadID(sd, &type));
    if (type == k7zIdEnd)
      break;
    SzSkeepData(sd);
  }
  return SZ_OK;
}

static SRes SzWaitAttribute(CSzData *sd, UInt64 attribute)
{
  for (;;)
  {
    UInt64 type;
    RINOK(SzReadID(sd, &type));
    if (type == attribute)
      return SZ_OK;
    if (type == k7zIdEnd)
      return SZ_ERROR_ARCHIVE;
    RINOK(SzSkeepData(sd));
  }
}

static SRes SzReadBoolVector(CSzData *sd, size_t numItems, Byte **v, ISzAlloc *alloc)
{
  Byte b = 0;
  Byte mask = 0;
  size_t i;
  MY_ALLOC(Byte, *v, numItems, alloc);
  for (i = 0; i < numItems; i++)
  {
    if (mask == 0)
    {
      RINOK(SzReadByte(sd, &b));
      mask = 0x80;
    }
    (*v)[i] = (Byte)(((b & mask) != 0) ? 1 : 0);
    mask >>= 1;
  }
  return SZ_OK;
}

static SRes SzReadBoolVector2(CSzData *sd, size_t numItems, Byte **v, ISzAlloc *alloc)
{
  Byte allAreDefined;
  size_t i;
  RINOK(SzReadByte(sd, &allAreDefined));
  if (allAreDefined == 0)
    return SzReadBoolVector(sd, numItems, v, alloc);
  MY_ALLOC(Byte, *v, numItems, alloc);
  for (i = 0; i < numItems; i++)
    (*v)[i] = 1;
  return SZ_OK;
}

static SRes SzReadHashDigests(
    CSzData *sd,
    size_t numItems,
    Byte **digestsDefined,
    UInt32 **digests,
    ISzAlloc *alloc)
{
  size_t i;
  RINOK(SzReadBoolVector2(sd, numItems, digestsDefined, alloc));
  MY_ALLOC(UInt32, *digests, numItems, alloc);
  for (i = 0; i < numItems; i++)
    if ((*digestsDefined)[i])
    {
      RINOK(SzReadUInt32(sd, (*digests) + i));
    }
  return SZ_OK;
}

static SRes SzReadPackInfo(
    CSzData *sd,
    UInt64 *dataOffset,
    UInt32 *numPackStreams,
    UInt64 **packSizes,
    Byte **packCRCsDefined,
    UInt32 **packCRCs,
    ISzAlloc *alloc)
{
  UInt32 i;
  RINOK(SzReadNumber(sd, dataOffset));
  RINOK(SzReadNumber32(sd, numPackStreams));

  RINOK(SzWaitAttribute(sd, k7zIdSize));

  MY_ALLOC(UInt64, *packSizes, (size_t)*numPackStreams, alloc);

  for (i = 0; i < *numPackStreams; i++)
  {
    RINOK(SzReadNumber(sd, (*packSizes) + i));
  }

  for (;;)
  {
    UInt64 type;
    RINOK(SzReadID(sd, &type));
    if (type == k7zIdEnd)
      break;
    if (type == k7zIdCRC)
    {
      RINOK(SzReadHashDigests(sd, (size_t)*numPackStreams, packCRCsDefined, packCRCs, alloc));
      continue;
    }
    RINOK(SzSkeepData(sd));
  }
  if (*packCRCsDefined == 0)
  {
    MY_ALLOC(Byte, *packCRCsDefined, (size_t)*numPackStreams, alloc);
    MY_ALLOC(UInt32, *packCRCs, (size_t)*numPackStreams, alloc);
    for (i = 0; i < *numPackStreams; i++)
    {
      (*packCRCsDefined)[i] = 0;
      (*packCRCs)[i] = 0;
    }
  }