Python flatten multilevel/nested JSON

I used the following function (details can be found here):

def flatten_data(y):
    out = {}

    def flatten(x, name=''):
        if type(x) is dict:
            for a in x:
                flatten(x[a], name + a + '_')
        elif type(x) is list:
            i = 0
            for a in x:
                flatten(a, name + str(i) + '_')
                i += 1
        else:
            out[name[:-1]] = x

    flatten(y)
    return out

This unfortunately completely flattens whole JSON, meaning that if you have multi-level JSON (many nested dictionaries), it might flatten everything into single line with tons of columns.

What I used, in the end, was json_normalize() and specified structure that I required. A nice example of how to do it that way can be found here.


IMO accepted answer doesn't properly handle JSON array.

If JSON object has array as value then it should be flattened to array of objects like

{'a': [1, 2]} -> [{'a': 1}, {'a': 2}]

instead of adding index to key.

And nested objects should be flattened by concatenating keys (e.g. with dot as separator) like

{'a': {'b': 1}} -> {'a.b': 1}

(and this is done correctly in accepted one).

With all these requirements I've ended up with following (developed and used in CPython3.5.3):

from functools import (partial,
                       singledispatch)
from itertools import chain
from typing import (Dict,
                    List,
                    TypeVar)

Serializable = TypeVar('Serializable', None, int, bool, float, str, 
                       dict, list, tuple)
Array = List[Serializable]
Object = Dict[str, Serializable]


def flatten(object_: Object,
            *,
            path_separator: str = '.') -> Array[Object]:
    """
    Flattens given JSON object into list of objects with non-nested values.

    >>> flatten({'a': 1})
    [{'a': 1}]
    >>> flatten({'a': [1, 2]})
    [{'a': 1}, {'a': 2}]
    >>> flatten({'a': {'b': None}})
    [{'a.b': None}]
    """
    keys = set(object_)
    result = [dict(object_)]
    while keys:
        key = keys.pop()
        new_result = []
        for index, record in enumerate(result):
            try:
                value = record[key]
            except KeyError:
                new_result.append(record)
            else:
                if isinstance(value, dict):
                    del record[key]
                    new_value = flatten_nested_objects(
                            value,
                            prefix=key + path_separator,
                            path_separator=path_separator)
                    keys.update(new_value.keys())
                    new_result.append({**new_value, **record})
                elif isinstance(value, list):
                    del record[key]
                    new_records = [
                        flatten_nested_objects(sub_value,
                                               prefix=key + path_separator,
                                               path_separator=path_separator)
                        for sub_value in value]
                    keys.update(chain.from_iterable(map(dict.keys,
                                                        new_records)))
                    new_result.extend({**new_record, **record}
                                      for new_record in new_records)
                else:
                    new_result.append(record)
        result = new_result
    return result


@singledispatch
def flatten_nested_objects(object_: Serializable,
                           *,
                           prefix: str = '',
                           path_separator: str) -> Object:
    return {prefix[:-len(path_separator)]: object_}


@flatten_nested_objects.register(dict)
def _(object_: Object,
      *,
      prefix: str = '',
      path_separator: str) -> Object:
    result = dict(object_)
    for key in list(result):
        result.update(flatten_nested_objects(result.pop(key),
                                             prefix=(prefix + key
                                                     + path_separator),
                                             path_separator=path_separator))
    return result


@flatten_nested_objects.register(list)
def _(object_: Array,
      *,
      prefix: str = '',
      path_separator: str) -> Object:
    return {prefix[:-len(path_separator)]: list(map(partial(
            flatten_nested_objects,
            path_separator=path_separator),
            object_))}

Cross-posting (but then adapting further) from https://stackoverflow.com/a/62186053/4355695 : In this repo: https://github.com/ScriptSmith/socialreaper/blob/master/socialreaper/tools.py#L8 , I found an implementation of the list-inclusion comment by @roneo to the answer posted by @Imran.

I've added checks to it for catching empty lists and empty dicts. And also added print lines that will help one understand precisely how this function works. You can turn off those print statemenents by setting crumbs=False

from collections import MutableMapping
crumbs = True
def flatten(dictionary, parent_key=False, separator='.'):
    """
    Turn a nested dictionary into a flattened dictionary
    :param dictionary: The dictionary to flatten
    :param parent_key: The string to prepend to dictionary's keys
    :param separator: The string used to separate flattened keys
    :return: A flattened dictionary
    """

    items = []
    for key, value in dictionary.items():
        if crumbs: print('checking:',key)
        new_key = str(parent_key) + separator + key if parent_key else key
        if isinstance(value, MutableMapping):
            if crumbs: print(new_key,': dict found')
            if not value.items():
                if crumbs: print('Adding key-value pair:',new_key,None)
                items.append((new_key,None))
            else:
                items.extend(flatten(value, new_key, separator).items())
        elif isinstance(value, list):
            if crumbs: print(new_key,': list found')
            if len(value):
                for k, v in enumerate(value):
                    items.extend(flatten({str(k): v}, new_key).items())
            else:
                if crumbs: print('Adding key-value pair:',new_key,None)
                items.append((new_key,None))
        else:
            if crumbs: print('Adding key-value pair:',new_key,value)
            items.append((new_key, value))
    return dict(items)

Test it:

ans = flatten({'a': 1, 'c': {'a': 2, 'b': {'x': 5, 'y' : 10}}, 'd': [1, 2, 3], 'e':{'f':[], 'g':{}} })
print('\nflattened:',ans)

Output:

checking: a
Adding key-value pair: a 1
checking: c
c : dict found
checking: a
Adding key-value pair: c.a 2
checking: b
c.b : dict found
checking: x
Adding key-value pair: c.b.x 5
checking: y
Adding key-value pair: c.b.y 10
checking: d
d : list found
checking: 0
Adding key-value pair: d.0 1
checking: 1
Adding key-value pair: d.1 2
checking: 2
Adding key-value pair: d.2 3
checking: e
e : dict found
checking: f
e.f : list found
Adding key-value pair: e.f None
checking: g
e.g : dict found
Adding key-value pair: e.g None

flattened: {'a': 1, 'c.a': 2, 'c.b.x': 5, 'c.b.y': 10, 'd.0': 1, 'd.1': 2, 'd.2': 3, 'e.f': None, 'e.g': None}

Annd that does the job I need done: I throw any complicated json at this and it flattens it out for me. I added a check to the original code to handle empty lists too

Credits to https://github.com/ScriptSmith whose repo I found the intial flatten function in.

Testing OP's sample json, here's the output:

{'count': 13,
 'virtualmachine.0.id': '1082e2ed-ff66-40b1-a41b-26061afd4a0b',
 'virtualmachine.0.name': 'test-2',
 'virtualmachine.0.displayname': 'test-2',
 'virtualmachine.0.securitygroup.0.id': '9e649fbc-3e64-4395-9629-5e1215b34e58',
 'virtualmachine.0.securitygroup.0.name': 'test',
 'virtualmachine.0.securitygroup.0.tags': None,
 'virtualmachine.0.nic.0.id': '79568b14-b377-4d4f-b024-87dc22492b8e',
 'virtualmachine.0.nic.0.networkid': '05c0e278-7ab4-4a6d-aa9c-3158620b6471',
 'virtualmachine.0.nic.1.id': '3d7f2818-1f19-46e7-aa98-956526c5b1ad',
 'virtualmachine.0.nic.1.networkid': 'b4648cfd-0795-43fc-9e50-6ee9ddefc5bd',
 'virtualmachine.0.nic.1.traffictype': 'Guest',
 'virtualmachine.0.hypervisor': 'KVM',
 'virtualmachine.0.affinitygroup': None,
 'virtualmachine.0.isdynamicallyscalable': False}

So you'll see that 'tags' and 'affinitygroup' keys are also handled and added to output. Original code was omitting them.

2021-05-30 : Updated: collections.MutableMapping is changed to collections.abc.MutableMapping