witness_lz4d.py

import struct
import sys

if (len(sys.argv) < 3):
  exit("need input and output file names")

infile = open(sys.argv[1],"rb")
outfile = open(sys.argv[2],"wb")

indata = infile.read()
inbytes = len(indata)
outlen = struct.unpack('<I',indata[8:12])[0]
outdata = bytearray(outlen)
inptr=12
outptr=0
printout=0
while(outptr < outlen):
  copybytes  = (indata[inptr] & 0xf0) >> 4
  matchbytes = (indata[inptr] & 0x0f)
  inptr += 1
  if (copybytes == 15):
    while( True ):
      copybytes += indata[inptr]
      inptr += 1
      if (indata[inptr-1] != 0xff):
        break

  #print("Copying %d bytes" % copybytes)
  if (copybytes > 0):
    outdata[outptr:outptr+copybytes] = indata[inptr:inptr+copybytes]
    inptr += copybytes
    outptr += copybytes

  if (inptr >= inbytes):
    break
  offset = indata[inptr] + (indata[inptr+1] << 8)
  inptr += 2
  if (matchbytes == 15):
    while( True ):
      matchbytes += indata[inptr]
      inptr += 1
      if (indata[inptr-1] != 0xff):
        break
  matchbytes += 4
  #print("Offset is %d, copying %d match bytes" % (offset, matchbytes))
  
  if (matchbytes >= offset): #overlapping copy
    for i in range(matchbytes):
      outdata[outptr+i] = outdata[outptr-offset+i] # slow
  else:
      outdata[outptr:outptr+matchbytes] = outdata[outptr-offset:outptr-offset+matchbytes]
  outptr += matchbytes  

  if (outptr > printout):
    print("%d/%d" % (outptr,outlen))
    printout += 256*1024

print("%d/%d" % (outptr,outlen))

outfile.write(outdata)