json_sem_hash.py

"""
Compute a hash for a JSON data structure, such that semantically equivalent 
JSON structures get the same hash. The notion of "semantic equivalence" is 
currently rather basic and informal. Eg, the following are semantically 
equivalent, and this is reflected in the computed hashes: 
```
d3 = {'d1': {'a': 1, 'b': [1,2]}, 'd2': {'b': [1,2], 'a': 1}, 'L': [1, 2, 3]}
d4 = {'d2': {'b': [1,2], 'a': 1}, 'L': [1, 2, 3], 'd1': {'a': 1, 'b': [1,2]}}
print('d3 hash:', get_json_sem_hash(d3))
print('d4 hash:', get_json_sem_hash(d4))
assert get_json_sem_hash(d3) == get_json_sem_hash(d4)
```
This prints hash value 'e17246aa9136a25581fb859fdeb2dd1da4cda9a221124cd27208646749b85cd7'
for both d3 and d4. 

If you find that `get_json_sem_hash()` doesn't return the same hash for 2 json structures 
that *you* think are in fact "semantically equivalent", please raise an issue!

The idea for this approach was inspired by https://github.com/fraunhoferfokus/JSum. It 
probably needs Python 3.5+ due to type hinting. Clearly the type hinting is not essential, 
so if it's presence is a hindrance for lots of people, please raise an issue. 

(C) Oliver Schoenborn
License: modified MIT, ie MIT plus the following restriction: This code 
can be included in your code base only as the complete file, and this 
license comment cannot be removed or changed. This code was taken from 
https://github.com/schollii/sandals/blob/master/json_sem_hash.py. If you 
find modifications necessary, please contribute a PR so that the open-source
community can benefit the same way you benefit from this file.
GIT_COMMIT: <REPLACE WHEN FILE COPIED FROM GITHUB>
"""

from typing import Union, Dict, List
import hashlib

JsonType = Union[str, int, float, List['JsonType'], 'JsonTree']
JsonTree = Dict[str, JsonType]
StrTreeType = Union[str, List['StrTreeType'], 'StrTree']
StrTree = Dict[str, StrTreeType]


def sorted_dict_str(data: JsonType) -> StrTreeType:
    if type(data) == dict:
        return {k: sorted_dict_str(data[k]) for k in sorted(data.keys())}
    elif type(data) == list:
        return [sorted_dict_str(val) for val in data]
    else:
        return str(data)


def get_json_sem_hash(data: JsonTree, hasher=hashlib.sha256) -> str:
    return hasher(bytes(repr(sorted_dict_str(data)), 'UTF-8')).hexdigest()