The standard hash('hello') function may generate different hashes on different machines, different versions of Python, and even different runs of the same program on the same Python version/machine.
What are decent pure Python (or built in) alternatives that have low chances of collision? The use case I'm after is checking uniqueness of a string in a database (note: it doesn't have to be 100% foolproof, just extremely unlikely to collide with another string).
Try the python hashlib. It has implementations of MD5, SHA1, SHA224, SHA256, SHA384, and SHA512 which should give you a good hash with low chance of collision.
Example Input:
import hashlib
hashed_string = hashlib.sha512("hello").hexdigest()
print hashed_string
Output:
9b71d224bd62f3785d96d46ad3ea3d73319bfbc2890caadae2dff72519673ca72323c3d99ba5c11d7c7acc6e14b8c5da0c4663475c2e5c3adef46f73bcdec043
hash which is in a different domain.
hashlib will do, but yes I would have preferred to find a pure Python solution that was not cryptographic. I'll also look into Murmurhash.Murmurhash is a good bet for non-cyrptographic uses (as opposed to hashlib, which uses slow crypto-hashes), for a variety of reasons:
very widely used
portable across not only Python versions and machines, but across different languages
Here is its Python bindings. Here is a question on the algorithm itself.
If, for some technical reason, a pure-Python roll-your-own implementation might be needed, there are a few feasible options:
hashlibbut was hoping to also find other alternatives.