matchfinder.c

一个7ZIP的解压源码。比较详细。里面含有四种语言的实现代码。

        *ptr1 = curMatch;
        ptr1 = pair + 1;
        curMatch = *ptr1;
        len1 = len;
      }
      else
      {
        *ptr0 = curMatch;
        ptr0 = pair;
        curMatch = *ptr0;
        len0 = len;
      }
    }
  }
}

void SkipMatchesSpec(UInt32 lenLimit, UInt32 curMatch, UInt32 pos, const Byte *cur, CLzRef *son, 
    UInt32 _cyclicBufferPos, UInt32 _cyclicBufferSize, UInt32 cutValue)
{
  CLzRef *ptr0 = son + (_cyclicBufferPos << 1) + 1;
  CLzRef *ptr1 = son + (_cyclicBufferPos << 1);
  UInt32 len0 = 0, len1 = 0;
  for (;;)
  {
    UInt32 delta = pos - curMatch;
    if (cutValue-- == 0 || delta >= _cyclicBufferSize)
    {
      *ptr0 = *ptr1 = kEmptyHashValue;
      return;
    }
    {
      CLzRef *pair = son + ((_cyclicBufferPos - delta + ((delta > _cyclicBufferPos) ? _cyclicBufferSize : 0)) << 1);
      const Byte *pb = cur - delta;
      UInt32 len = (len0 < len1 ? len0 : len1);
      if (pb[len] == cur[len])
      {
        while(++len != lenLimit)
          if (pb[len] != cur[len])
            break;
        {
          if (len == lenLimit)
          {
            *ptr1 = pair[0];
            *ptr0 = pair[1];
            return;
          }
        }
      }
      if (pb[len] < cur[len])
      {
        *ptr1 = curMatch;
        ptr1 = pair + 1;
        curMatch = *ptr1;
        len1 = len;
      }
      else
      {
        *ptr0 = curMatch;
        ptr0 = pair;
        curMatch = *ptr0;
        len0 = len;
      }
    }
  }
}

#define MOVE_POS \
  ++p->cyclicBufferPos; \
  p->buffer++; \
  if (++p->pos == p->posLimit) MatchFinder_CheckLimits(p);

#define MOVE_POS_RET MOVE_POS return offset;

void MatchFinder_MovePos(CMatchFinder *p) { MOVE_POS; }

#define GET_MATCHES_HEADER2(minLen, ret_op) \
  UInt32 lenLimit; UInt32 hashValue; const Byte *cur; UInt32 curMatch; \
  lenLimit = p->lenLimit; { if (lenLimit < minLen) { MatchFinder_MovePos(p); ret_op; }} \
  cur = p->buffer;

#define GET_MATCHES_HEADER(minLen) GET_MATCHES_HEADER2(minLen, return 0)
#define SKIP_HEADER(minLen)        GET_MATCHES_HEADER2(minLen, continue)

#define MF_PARAMS(p) p->pos, p->buffer, p->son, p->cyclicBufferPos, p->cyclicBufferSize, p->cutValue

#define GET_MATCHES_FOOTER(offset, maxLen) \
  offset = (UInt32)(GetMatchesSpec1(lenLimit, curMatch, MF_PARAMS(p), \
  distances + offset, maxLen) - distances); MOVE_POS_RET;

#define SKIP_FOOTER \
  SkipMatchesSpec(lenLimit, curMatch, MF_PARAMS(p)); MOVE_POS;

UInt32 Bt2_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances)
{
  UInt32 offset;
  GET_MATCHES_HEADER(2)
  HASH2_CALC;
  curMatch = p->hash[hashValue];
  p->hash[hashValue] = p->pos;
  offset = 0;
  GET_MATCHES_FOOTER(offset, 1)
}

UInt32 Bt3Zip_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances)
{
  UInt32 offset;
  GET_MATCHES_HEADER(3)
  HASH_ZIP_CALC;
  curMatch = p->hash[hashValue];
  p->hash[hashValue] = p->pos;
  offset = 0;
  GET_MATCHES_FOOTER(offset, 2)
}

UInt32 Bt3_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances)
{
  UInt32 hash2Value, delta2, maxLen, offset;
  GET_MATCHES_HEADER(3)

  HASH3_CALC;

  delta2 = p->pos - p->hash[hash2Value];
  curMatch = p->hash[kFix3HashSize + hashValue];
  
  p->hash[hash2Value] = 
  p->hash[kFix3HashSize + hashValue] = p->pos;


  maxLen = 2;
  offset = 0;
  if (delta2 < p->cyclicBufferSize && *(cur - delta2) == *cur)
  {
    for (; maxLen != lenLimit; maxLen++)
      if (cur[(ptrdiff_t)maxLen - delta2] != cur[maxLen])
        break;
    distances[0] = maxLen;
    distances[1] = delta2 - 1;
    offset = 2;
    if (maxLen == lenLimit)
    {
      SkipMatchesSpec(lenLimit, curMatch, MF_PARAMS(p));
      MOVE_POS_RET; 
    }
  }
  GET_MATCHES_FOOTER(offset, maxLen)
}

UInt32 Bt4_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances)
{
  UInt32 hash2Value, hash3Value, delta2, delta3, maxLen, offset;
  GET_MATCHES_HEADER(4)

  HASH4_CALC;

  delta2 = p->pos - p->hash[                hash2Value];
  delta3 = p->pos - p->hash[kFix3HashSize + hash3Value];
  curMatch = p->hash[kFix4HashSize + hashValue];
  
  p->hash[                hash2Value] =
  p->hash[kFix3HashSize + hash3Value] =
  p->hash[kFix4HashSize + hashValue] = p->pos;

  maxLen = 1;
  offset = 0;
  if (delta2 < p->cyclicBufferSize && *(cur - delta2) == *cur)
  {
    distances[0] = maxLen = 2;
    distances[1] = delta2 - 1;
    offset = 2;
  }
  if (delta2 != delta3 && delta3 < p->cyclicBufferSize && *(cur - delta3) == *cur)
  {
    maxLen = 3;
    distances[offset + 1] = delta3 - 1;
    offset += 2;
    delta2 = delta3;
  }
  if (offset != 0)
  {
    for (; maxLen != lenLimit; maxLen++)
      if (cur[(ptrdiff_t)maxLen - delta2] != cur[maxLen])
        break;
    distances[offset - 2] = maxLen;
    if (maxLen == lenLimit)
    {
      SkipMatchesSpec(lenLimit, curMatch, MF_PARAMS(p));
      MOVE_POS_RET; 
    }
  }
  if (maxLen < 3)
    maxLen = 3;
  GET_MATCHES_FOOTER(offset, maxLen)
}

UInt32 Hc4_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances)
{
  UInt32 hash2Value, hash3Value, delta2, delta3, maxLen, offset;
  GET_MATCHES_HEADER(4)

  HASH4_CALC;

  delta2 = p->pos - p->hash[                hash2Value];
  delta3 = p->pos - p->hash[kFix3HashSize + hash3Value];
  curMatch = p->hash[kFix4HashSize + hashValue];

  p->hash[                hash2Value] =
  p->hash[kFix3HashSize + hash3Value] =
  p->hash[kFix4HashSize + hashValue] = p->pos;

  maxLen = 1;
  offset = 0;
  if (delta2 < p->cyclicBufferSize && *(cur - delta2) == *cur)
  {
    distances[0] = maxLen = 2;
    distances[1] = delta2 - 1;
    offset = 2;
  }
  if (delta2 != delta3 && delta3 < p->cyclicBufferSize && *(cur - delta3) == *cur)
  {
    maxLen = 3;
    distances[offset + 1] = delta3 - 1;
    offset += 2;
    delta2 = delta3;
  }
  if (offset != 0)
  {
    for (; maxLen != lenLimit; maxLen++)
      if (cur[(ptrdiff_t)maxLen - delta2] != cur[maxLen])
        break;
    distances[offset - 2] = maxLen;
    if (maxLen == lenLimit)
    {
      p->son[p->cyclicBufferPos] = curMatch;
      MOVE_POS_RET; 
    }
  }
  if (maxLen < 3)
    maxLen = 3;
  offset = (UInt32)(Hc_GetMatchesSpec(lenLimit, curMatch, MF_PARAMS(p),
    distances + offset, maxLen) - (distances));
  MOVE_POS_RET
}

UInt32 Hc3Zip_MatchFinder_GetMatches(CMatchFinder *p, UInt32 *distances)
{
  UInt32 offset;
  GET_MATCHES_HEADER(3)
  HASH_ZIP_CALC;
  curMatch = p->hash[hashValue];
  p->hash[hashValue] = p->pos;
  offset = (UInt32)(Hc_GetMatchesSpec(lenLimit, curMatch, MF_PARAMS(p),
    distances, 2) - (distances));
  MOVE_POS_RET
}

void Bt2_MatchFinder_Skip(CMatchFinder *p, UInt32 num)
{
  do
  {
    SKIP_HEADER(2) 
    HASH2_CALC;
    curMatch = p->hash[hashValue];
    p->hash[hashValue] = p->pos;
    SKIP_FOOTER
  }
  while (--num != 0);
}

void Bt3Zip_MatchFinder_Skip(CMatchFinder *p, UInt32 num)
{
  do
  {
    SKIP_HEADER(3)
    HASH_ZIP_CALC;
    curMatch = p->hash[hashValue];
    p->hash[hashValue] = p->pos;
    SKIP_FOOTER
  }
  while (--num != 0);
}

void Bt3_MatchFinder_Skip(CMatchFinder *p, UInt32 num)
{
  do
  {
    UInt32 hash2Value;
    SKIP_HEADER(3)
    HASH3_CALC;
    curMatch = p->hash[kFix3HashSize + hashValue];
    p->hash[hash2Value] =
    p->hash[kFix3HashSize + hashValue] = p->pos;
    SKIP_FOOTER
  }
  while (--num != 0);
}

void Bt4_MatchFinder_Skip(CMatchFinder *p, UInt32 num)
{
  do
  {
    UInt32 hash2Value, hash3Value;
    SKIP_HEADER(4) 
    HASH4_CALC;
    curMatch = p->hash[kFix4HashSize + hashValue];
    p->hash[                hash2Value] =
    p->hash[kFix3HashSize + hash3Value] = p->pos;
    p->hash[kFix4HashSize + hashValue] = p->pos;
    SKIP_FOOTER
  }
  while (--num != 0);
}

void Hc4_MatchFinder_Skip(CMatchFinder *p, UInt32 num)
{
  do
  {
    UInt32 hash2Value, hash3Value;
    SKIP_HEADER(4)
    HASH4_CALC;
    curMatch = p->hash[kFix4HashSize + hashValue];
    p->hash[                hash2Value] =
    p->hash[kFix3HashSize + hash3Value] =
    p->hash[kFix4HashSize + hashValue] = p->pos;
    p->son[p->cyclicBufferPos] = curMatch;
    MOVE_POS
  }
  while (--num != 0);
}

void Hc3Zip_MatchFinder_Skip(CMatchFinder *p, UInt32 num)
{
  do
  {
    SKIP_HEADER(3)
    HASH_ZIP_CALC;
    curMatch = p->hash[hashValue];
    p->hash[hashValue] = p->pos;
    p->son[p->cyclicBufferPos] = curMatch;
    MOVE_POS
  }
  while (--num != 0);
}

void MatchFinder_CreateVTable(CMatchFinder *p, IMatchFinder *vTable)
{
  vTable->Init = (Mf_Init_Func)MatchFinder_Init;
  vTable->GetIndexByte = (Mf_GetIndexByte_Func)MatchFinder_GetIndexByte;
  vTable->GetNumAvailableBytes = (Mf_GetNumAvailableBytes_Func)MatchFinder_GetNumAvailableBytes;
  vTable->GetPointerToCurrentPos = (Mf_GetPointerToCurrentPos_Func)MatchFinder_GetPointerToCurrentPos;
  if (!p->btMode)
  {
    vTable->GetMatches = (Mf_GetMatches_Func)Hc4_MatchFinder_GetMatches;
    vTable->Skip = (Mf_Skip_Func)Hc4_MatchFinder_Skip;
  }
  else if (p->numHashBytes == 2)
  {
    vTable->GetMatches = (Mf_GetMatches_Func)Bt2_MatchFinder_GetMatches;
    vTable->Skip = (Mf_Skip_Func)Bt2_MatchFinder_Skip;
  }
  else if (p->numHashBytes == 3)
  {
    vTable->GetMatches = (Mf_GetMatches_Func)Bt3_MatchFinder_GetMatches;
    vTable->Skip = (Mf_Skip_Func)Bt3_MatchFinder_Skip;
  }
  else
  {
    vTable->GetMatches = (Mf_GetMatches_Func)Bt4_MatchFinder_GetMatches;
    vTable->Skip = (Mf_Skip_Func)Bt4_MatchFinder_Skip;
  }
}