I just wrote a merge that handles multi-layer dictionaries

from itertools import chain from functools import reduce def deep_merge_func(acc, item): if not acc: acc = {} k, v = item if isinstance(v, dict): original_v = acc.get(k) if original_v and isinstance(original_v, dict): return {**acc, k: deep_merge(original_v, v)} return {**acc, k: v} def deep_merge(origin, target): return reduce(deep_merge_func, chain(origin.items(), target.items()), None) a = {1: {"171": True}} b = {1: {"171": False}} print(deep_merge(a, b)) c = {1: {"171": True}} d = {1: {"172": False}} print(deep_merge(c ,d))

Only tested python3.6.1, you only need to call deep_merge

The writing is more functional, don’t blame me

For case2 it is relatively simple:

>>> a={1:{"171": True}} >>> b={1:{"171":False}} >>> a.update(b) >>> a {1: {'171': False}}

For case1 you can do this:

>>> c={} >>> for k,v in a.items(): ... if k in b: ... v.update(b[k]) ... c[k] = v ... >>> c {1: {'172': False, '171': True}}

You can encapsulate the above operations into functions.