How to update a dict nested in another?

Question

I have the below original dict:

{'data_extraction': {'if_extraction': False,
  'path_data': 'data_extraction/extractions.sql',
  'set_params': {'START_DT': "'201901'",
   'END_DT': "'202104'"
                         },
  'calibration': True,
  'target_variable': 'unsure'}}

I'd like to replace the set params bit with the following dict:

dict1= {'set_params': {'START_DT': "'201001'",
   'END_DT': "'201004'"
}}

I took this code from a similar question:

import collections.abc

def update(d, u):
    for k, v in u.items():
        if isinstance(v, collections.abc.Mapping):
            d[k] = update(d.get(k, {}), v)
        else:
            d[k] = v
    return d

update(original,dict1)

Applied it to the above and I got:

 {'data_extraction': {'if_extraction': False,
          'path_data': 'data_extraction/extractions.sql',
          'set_params': {'START_DT': "'201901'",
           'END_DT': "'202104'"
                                 },
          'calibration': True,
          'target_variable': 'unsure'},
           'set_params': {'START_DT': "'201001'",
            'END_DT': "'201004'" }}

So rather than replace, it appended the replacement dict at the end. How can I update values in a dict such that if I pass another dict as I have with dict1 it replaces it? What am I doing wrong?

Answer 1

These days Python dicts have an update method to update from another dict. Be sure to specify the appropriate dicts that you want to update and update from!

original['data_extraction']['set_params'].update(dict1['set_params'])

Or possibly

original['data_extraction'].update(dict1)

if there can be other keys besides 'set_params' in dict1 and you want to update those as well.

Answer 2

Use recursive function with dict.update() . This way you don't have to sepcify the keys in case of multilevel dictionary.

def update_dict(source_dict, dict1):
    for key in source_dict:
        if key in dict1.keys():
            source_dict.update(dict1)
        elif isinstance(source_dict[key],dict):
            update_dict(source_dict[key], dict1)


source_dict = {'data_extraction': {'if_extraction': False,
      'path_data': 'data_extraction/extractions.sql',
      'set_params': {'START_DT': "'201901'",
       'END_DT': "'202104'"},
      'calibration': True,
      'target_variable': 'unsure'}}

dict1= {'set_params': {'START_DT': "'201001'",
   'END_DT': "'201004'"
}}

update_dict(source_dict, dict1)

print(source_dict)

>> 
{'data_extraction': {'if_extraction': False,
  'path_data': 'data_extraction/extractions.sql',
  'set_params': {'START_DT': "'201001'", 'END_DT': "'201004'"},
  'calibration': True,
  'target_variable': 'unsure'}}

Answer 3

You were almost there... In this implementation, I use deepcopy() to avoid overwriting the original dictionary.

from copy import deepcopy

def update(d, u):
    r = deepcopy(d)
    for k, v in r.items():
        if type(v) is dict:
            for _k, _v in v.items():
                if _k in u:
                    r[k][_k] = u[_k]
        elif k in u:
            r[k] = u[k]
    return r


a = {'data_extraction':{'if_extraction': False,
                         'path_data': 'data_extraction/extractions.sql',
                         'set_params': {'START_DT': "'201901'", 'END_DT': "'202104'"},
                         'calibration': True,
                         'target_variable': 'unsure'}
     }

b = {'set_params': {'START_DT': "'201001'", 'END_DT': "'201004'"}}

c = update(a, b)

print(a, '\n')
print(b, '\n')
print(c, '\n')

Output:

{'data_extraction': {'if_extraction': False, 'path_data': 'data_extraction/extractions.sql', 'set_params': {'START_DT': "'201901'", 'END_DT': "'202104'"}, 'calibration': True, 'target_variable': 'unsure'}} 

{'set_params': {'START_DT': "'201001'", 'END_DT': "'201004'"}} 

{'data_extraction': {'if_extraction': False, 'path_data': 'data_extraction/extractions.sql', 'set_params': {'START_DT': "'201001'", 'END_DT': "'201004'"}, 'calibration': True, 'target_variable': 'unsure'}}

Answer 4

In a dict when you asing something to a key that doesnt exists, it is appended and then the content is added. If you want do substitute some key you have to delete it first.
Use pop for that ( yourdict.pop ("key to delete") ), then you can add the other key normally.

Answer 5

You need to have the recursive function check each (nested) dictionary to see if it contains the key that needs to be replaced (the target_key ). In the code below this is retrieved from the replacement dictionary argument new_dict that's passed to the update() function. This only needs to be done once, so there's a nested_update() function that does actual replacing.

import collections.abc

original = {
  "data_extraction": {
    "if_extraction": False,
    "path_data": "data_extraction/extractions.sql",
    "set_params": {
      "START_DT": "'201901'",
      "END_DT": "'202104'"
    },
    "calibration": True,
    "target_key_variable": "unsure"
  }
}

dict1 = {
  "set_params": {
    "START_DT": "'201001'",
    "END_DT": "'201004'"
  }
}


def update(d, new_dict):
    target_key, new_value = list(new_dict.items())[0]

    def nested_update(d, new_dict):
        for key, value in d.items():
            if key == target_key:
                d[key] = new_value
            elif isinstance(value, collections.abc.Mapping):
                d[key] = nested_update(value, new_dict)
        return d

    nested_update(d, new_dict)
    return d

update(original, dict1)

import json
print(json.dumps(original, indent=2))  # Pretty-print result.

Output:

{
  "data_extraction": {
    "if_extraction": false,
    "path_data": "data_extraction/extractions.sql",
    "set_params": {
      "START_DT": "'201001'",
      "END_DT": "'201004'"
    },
    "calibration": true,
    "target_key_variable": "unsure"
  }
}