/* Bcj2.c -- Converter for x86 code (BCJ2)
 2008-10-04 : Igor Pavlov : Public domain */
 
 #include "Bcj2.h"
 
 #ifdef _LZMA_PROB32
 #define CProb UInt32
 #else
 #define CProb UInt16
 #endif
 
 #define IsJcc(b0, b1) ((b0) == 0x0F && ((b1) & 0xF0) == 0x80)
 #define IsJ(b0, b1) ((b1 & 0xFE) == 0xE8 || IsJcc(b0, b1))
 
 #define kNumTopBits 24
 #define kTopValue ((UInt32)1 << kNumTopBits)
 
 #define kNumBitModelTotalBits 11
 #define kBitModelTotal (1 << kNumBitModelTotalBits)
 #define kNumMoveBits 5
 
 #define RC_READ_BYTE (*buffer++)
 #define RC_TEST { if (buffer == bufferLim) return SZ_ERROR_DATA; }
 #define RC_INIT2 code = 0; range = 0xFFFFFFFF; \
   { int i; for (i = 0; i < 5; i++) { RC_TEST; code = (code << 8) | RC_READ_BYTE; }}
 
 #define NORMALIZE if (range < kTopValue) { RC_TEST; range <<= 8; code = (code << 8) | RC_READ_BYTE; }
 
 #define IF_BIT_0(p) ttt = *(p); bound = (range >> kNumBitModelTotalBits) * ttt; if (code < bound)
 #define UPDATE_0(p) range = bound; *(p) = (CProb)(ttt + ((kBitModelTotal - ttt) >> kNumMoveBits)); NORMALIZE;
 #define UPDATE_1(p) range -= bound; code -= bound; *(p) = (CProb)(ttt - (ttt >> kNumMoveBits)); NORMALIZE;
 
 int Bcj2_Decode(
     const Byte *buf0, SizeT size0,
     const Byte *buf1, SizeT size1,
     const Byte *buf2, SizeT size2,
     const Byte *buf3, SizeT size3,
     Byte *outBuf, SizeT outSize)
 {
   CProb p[256 + 2];
   SizeT inPos = 0, outPos = 0;
 
   const Byte *buffer, *bufferLim;
   UInt32 range, code;
   Byte prevByte = 0;
 
   unsigned int i;
   for (i = 0; i < sizeof(p) / sizeof(p[0]); i++)
     p[i] = kBitModelTotal >> 1;
 
   buffer = buf3;
   bufferLim = buffer + size3;
   RC_INIT2
 
   if (outSize == 0)
     return SZ_OK;
 
   for (;;)
   {
     Byte b;
     CProb *prob;
     UInt32 bound;
     UInt32 ttt;
 
     SizeT limit = size0 - inPos;
     if (outSize - outPos < limit)
       limit = outSize - outPos;
     while (limit != 0)
     {
       Byte b = buf0[inPos];
       outBuf[outPos++] = b;
       if (IsJ(prevByte, b))
         break;
       inPos++;
       prevByte = b;
       limit--;
     }
 
     if (limit == 0 || outPos == outSize)
       break;
 
     b = buf0[inPos++];
 
     if (b == 0xE8)
       prob = p + prevByte;
     else if (b == 0xE9)
       prob = p + 256;
     else
       prob = p + 257;
 
     IF_BIT_0(prob)
     {
       UPDATE_0(prob)
       prevByte = b;
     }
     else
     {
       UInt32 dest;
       const Byte *v;
       UPDATE_1(prob)
       if (b == 0xE8)
       {
         v = buf1;
         if (size1 < 4)
           return SZ_ERROR_DATA;
         buf1 += 4;
         size1 -= 4;
       }
       else
       {
         v = buf2;
         if (size2 < 4)
           return SZ_ERROR_DATA;
         buf2 += 4;
         size2 -= 4;
       }
       dest = (((UInt32)v[0] << 24) | ((UInt32)v[1] << 16) |
           ((UInt32)v[2] << 8) | ((UInt32)v[3])) - ((UInt32)outPos + 4);
       outBuf[outPos++] = (Byte)dest;
       if (outPos == outSize)
         break;
       outBuf[outPos++] = (Byte)(dest >> 8);
       if (outPos == outSize)
         break;
       outBuf[outPos++] = (Byte)(dest >> 16);
       if (outPos == outSize)
         break;
       outBuf[outPos++] = prevByte = (Byte)(dest >> 24);
     }
   }
   return (outPos == outSize) ? SZ_OK : SZ_ERROR_DATA;
 }