knarf knarf - 1 year ago 420
Python Question

zlib: Error -3 while decompressing data: invalid distance too far back

I'm trying to write code to write/read MS-ZIP compressed CAB files. MS-ZIP is using the same deflate algorithm implemented in zlib. I've simplified the problem to this simple script which only involves zlib.

  • I know the compress function is correct because other CAB extracting tools can read files produce this way.

  • Decompress function is not working. It fails when more than 1 chunk have to be decompressed.

I cannot get the decompression code that matches the compression code right.

import sys, struct, zlib


def main():

print("python", sys.version)
print("zlib version", zlib.ZLIB_VERSION)
print("zlib runtime version", zlib.ZLIB_RUNTIME_VERSION)

chunks = compress(uncomp)
decomp = decompress(chunks)
assert(uncomp == decomp)

def compress(data):
start_off = 0
data_size = len(data)
remaining = data_size
res = []
zdict = b''

while start_off < data_size:
print("compress chunk %d"%len(res))

chunk_size = min(MAX_CHUNK_SIZE, remaining)
chunk = data[start_off:start_off+chunk_size]
c = b''

z = zlib.compressobj(wbits=-15, zdict=zdict)
c += z.compress(chunk)
c += z.flush(zlib.Z_FINISH)


zdict = chunk
start_off += chunk_size
remaining -= chunk_size

return res

def decompress(chunks):
zdict = b''
res = []

for i, c in enumerate(chunks):
print("decompress chunk %d"%i)
out = b''

z = zlib.decompressobj(wbits=-15, zdict=zdict)
out += z.decompress(c)
out += z.flush()
zdict = out


return b''.join(res)

if __name__ == '__main__':

When I run this script I get:

python 3.4.5 (default, Jul 03 2016, 13:55:08) [GCC]
zlib version 1.2.8
zlib runtime version 1.2.8
compress chunk 0
compress chunk 1
decompress chunk 0
decompress chunk 1
Traceback (most recent call last):
File "", line 65, in <module>
File "", line 16, in main
decomp = decompress(chunks)
File "", line 55, in decompress
out += z.decompress(c)
zlib.error: Error -3 while decompressing data: invalid distance too far back

It also fails on which is running python 3.4.3+ zlib 1.2.8

Answer Source

It seems I was running into CPython issue #27164. As I'm writing this (2016-11-21), a fix is available in branches master and 3.5 only.

Recommended from our users: Dynamic Network Monitoring from WhatsUp Gold from IPSwitch. Free Download