simpleml.persistables.hashing
Mixin classes to handle hashing
Module Contents
Classes
Mixin class to hash any object |
Functions
|
Helper for hashing pandas - outside local context to be pickleable between threads |
Attributes
- class simpleml.persistables.hashing.CustomHasherMixin[source]
Bases:
objectMixin class to hash any object
- classmethod custom_hasher(cls, object_to_hash, raise_on_nonprimitive=False)[source]
Adapted from: https://stackoverflow.com/questions/5884066/hashing-a-dictionary Makes a hash from a dictionary, list, tuple or set to any level, that contains only other hashable types (including any lists, tuples, sets, and dictionaries). In the case where other kinds of objects (like classes) need to be hashed, pass in a collection of object attributes that are pertinent. For example, a class can be hashed in this fashion:
custom_hasher([cls.__dict__, cls.__name__])
A function can be hashed like so:
custom_hasher([fn.__dict__, fn.__code__])
python 3.3+ changes the default hash method to add an additional random seed. Need to set the global PYTHONHASHSEED=0 or use a different hash function
reduces to primitive dtypes and then calls cls.md5_hasher for a consistent hash value. falls back to joblib pickle hashing for other dtypes
If raise_on_nonprimitive is True, raises a ValueError if object_to_hash will be hashed based on inconsistent-across-instantiations object identity rather than primitives/consistent-across-instantiations content values.