Hashable DataFrames and Series for caching.py

import pandas as pd

from hashlib import sha256
from pandas.util import hash_pandas_object
from functools import lru_cache


class HashableDataFrame(pd.DataFrame):
    def __init__(self, obj):
        super().__init__(obj)

    def __hash__(self):
        hash_value = sha256(hash_pandas_object(self, index=True).values)
        hash_value = hash(hash_value.hexdigest())
        return hash_value

    def __eq__(self, other):
        return self.equals(other)


class HashableSeries(pd.Series):
    def __init__(self, obj):
        super().__init__(obj)

    def __hash__(self):
        hash_value = sha256(hash_pandas_object(self, index=True).values)
        hash_value = hash(hash_value.hexdigest())
        return hash_value

    def __eq__(self, other):
        return self.equals(other)

      
def make_hashable_pandas_object(obj):
    if type(obj) is pd.DataFrame:
        return HashableDataFrame(obj)
    elif type(obj) is pd.Series:
        return HashableSeries(obj)
    else:
        raise Exception(f'obj must be of type DataFrame or Series, '
                        f'but is {type(obj)}')

@lru_cache()
def cached_function_with_pandas(df, s):
    # heavy computation ...
    return ...

if __name__ == '__main__':
    df = ... # some pandas dataframe
    s = ... # some pandas series

    s_hashable = make_hashable_pandas_object(s)
    df_hashable = make_hashable_pandas_object(df)

    cached_function_with_pandas(df, s) # raises exception
    cached_function_with_pandas(df_hashable, s_hashable) # works!

Note that hash() is not deterministic across Python runs [src]—if you change:

hash_value = hash(hash_value.hexdigest())

to the following, it should be deterministic across Python runs:

hash_value = int.from_bytes(hash_value.digest(), 'big')