Change seed to accept unsigned long for c compatibility

[pik]

When parsing arguments with the i format string ParseTupleAndKeywords returns an OverflowError on Uint32 values. I will correctly accept both Int32 and Uint32 seeds.

[hajimes]

I'm grateful to you for commit. But, for two reasons, I opt for the format unit i.

1. Security concern. The format unit I does not check overflow. In realty it may be not a problem, since a variable seed is statically allocated and the memory will be never overflown. But because C/C++ is very tricky, I'm still not sure.
2. Interface concern. By changing the format unit i to I, I place the burden of checking inputs on users. This might cause unfavorable issues which are hard to detect in some cases, especially if the integer is overflown. For example, hash('foo', 2 ** 33) and hash('foo', 2 ** 34) return the same results without any notices.

If seeds within the range [2 ** 31, 2 ** 32) have imminent and important use cases, I will change the format unit. Until needed, though, I think it's safer to leave them untouched. It is just a design choice so your commit itself is quite reasonable. Again, I appreciate for your help.

[paulie-g]

Integer overflow and stack/heap overflow are two different things. The C standard guarantees that the storage size and alignment requirements are the same for any given signed integer type and its corresponding unsigned integer type. There can be no memory overflow here, it's not tricky at all and exactly why the C standards are useful. All integer overflow related security bugs stem from the use of integers in pointer arithmetic, eg to calculate offsets which are then used to index into arrays. None of this happens to this value and none of this will ever happen to this value.

The use case for this patch is correctness. Murmurhash3's function signature tells you it wants an unsigned int, the int you give it will be treated as such anyway. Document that the seed parameter is a 32-bit unsigned int and it is the user's responsibility to make sure he passes in a correct seed.

[dan-blanchard]

I was hoping to move some of our code from https://github.com/phensley/python-smhasher over to mmh3 and this is literally the only thing stopping me.

[hajimes]

Because there is a growing need for the feature, I added unit tests/CI to ensure stability and merged this one. Thank you very much for your commit.