Python: how does a dict with mixed key type work?

Question

I understand that the following is valid in Python: foo = {'a': 0, 1: 2, some_obj: 'c'}

However, I wonder how the internal works. Does it treat everything (object, string, number, etc.) as object? Does it type check to determine how to compute the hash code given a key?

Answer 1

Types aren't used the same way in Python as statically types languages. A hashable object is simply one with a valid hash method. The interpreter simply calls that method, no type checking or anything. From there on out, standard hash map principles apply: for an object to fulfill the contract, it must implement both hash and equals methods.

Answer 2

You can answer this by opening a Python interactive prompt and trying several of these keys:

>>> hash('a')
12416037344
>>> hash(2)
2
>>> hash(object())
8736272801544

Does it treat everything (object, string, number, etc.) as object?

You are simply using the hash function to represent each dictionary key as an integer. This integer is simply used to index in the underlying dictionary array. Assuming a dictionary starts of with a pre-allocated size of 8, we use the modulus operator (the remainder) to fit it into an appropriate location:

>>> hash('a')
12416037344
>>> hash(object()) % 8
2

So in this particular case, the hashed object is placed in index 2 of the underlying array. Of course there can be collisions, and so depending on the underlying implementation, the underlying array may actually be an array of arrays.

Note that items that aren't hashable cannot be used as dictionary keys:

>>> hash({})
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: unhashable type: 'dict'

Proof:

>>> d = {}
>>> d[{}] = 5
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: unhashable type: 'dict'