wc-duck/pymmh3

a pure python MurmurHash3 implementation.

pymmh3's Issues

Issues in Python3.4

Hi

I was trying out pymmh3 with python3.4 and saw that on calling hash it gave me this error:

TypeError: string argument without an encoding on trying to do key = bytearray( key ) (line 30)

The fix there is to call encode() on the string before trying to call bytearray on it. I was wondering if that is the expectation for every string. It only seems to be a problem on python 3.4. Once you call encode on the string and then try it works fine.

Difference from mmh3

Hi

I was trying out pymmh3 vs mmh3 (https://pypi.python.org/pypi/mmh3/2.0) and noticed that for certain inputs the value was different.
It seems to me that pymmh3.hash returns an unsigned 32 integer vs mmh3.hash which returns a signed 32 bit integer. Is this difference intentional?

You could try with the value 'abc' and would notice that pymmh3.hash('abc') returns 3017643002 whereas mmh3.hash('abc') returns -1277324294. The 2 differ by 4294967296 (which is 2^32) and so I am assuming that for negative values some addition is being done, but just wanted to confirm.

Thanks

How to run using unitest_solution_gen.cpp

I am getting segmentation fault. Are there any flags used to complile?

Weired sys line

Could you explain these two lines to me please https://github.com/wc-duck/pymmh3/blob/master/pymmh3.py#L20 + https://github.com/wc-duck/pymmh3/blob/master/pymmh3.py#L32 - what is the purpose of this?

Can't we simply use import sys and use it? Why rename and delete it?

Also in line https://github.com/wc-duck/pymmh3/blob/master/pymmh3.py#L451 sys is used although it doesn't exist.. so running it obviously give this error:

$: python3 pymmh3.py "foobar"
Traceback (most recent call last):
  File "pymmh3.py", line 451, in <module>
    sys.stdout.write( '"%s" = 0x%08X\n' % ( str_to_hash, hash( str_to_hash ) ) )
NameError: name 'sys' is not defined