Combining 2 dictionaries with common key

Question

I have two dictionaries and need to combine the values of similar keys in them. Here's an example:

dict1 = {'key1':[value11,value12,value13] , 'key2':[value21,value22,value23]}
dict2 = {'key1':[value14,value15] , 'key2':[value24,value25]}

I used :

dict3 = {}
for key in (dict1.viewkeys() | dict2.keys()):
    if key in dict1: dict3.setdefault(key, []).append(dict1[key])
    if key in dict2: dict3.setdefault(key, []).append(dict2[key])

which gives me:

dict3 = {'key1':[[value11,value12,value13],[value14,value15]] , 'key2':[[value21,value22,value23],[value24,value25]]}

What I want is a simple one like:

Desired output :

 dict3 = {'key1':[value11,value12,value13,value14,value15] , 'key2':[value21,value22,value23,value24,value25]}

Solution 1

All you need to do is to modify append to extend which will then add the elements of the list rather than adding the list itself. See the list docs for more details on the difference between append and extend.

dict1 = {'key1':['value11','value12','value13'] , 'key2':['value21','value22','value23']}
dict2 = {'key1':['value14','value15'] , 'key2':['value24','value25']}

dict3 = {}
for key in set().union(dict1, dict2):
    if key in dict1: dict3.setdefault(key, []).extend(dict1[key])
    if key in dict2: dict3.setdefault(key, []).extend(dict2[key])

print(dict3)
# {'key2': ['value21', 'value22', 'value23', 'value24', 'value25'], 'key1': ['value11', 'value12', 'value13', 'value14', 'value15']}

Alternatively you could use a collections.defaultdict with the default set to list as shown below.

from collections import defaultdict
dict3 = defaultdict(list)

for key in set().union(dict1, dict2):
  for dic in [dict1, dict2]:
    if key in dic:
      dict3[key] += dic[key]

Solution 2

Here is a generic method on which you can pass as many dict as you want as parameter.

>>> def mix_dict(*args):
       res = {}
       for d in args:
           if not isinstance(d, dict):
               continue
           for k, v in d.iteritems():
               res.setdefault(k, [])
               if isinstance(v, list):
                   res[k].extend(v)
               else:
                   res[k].append(v)
       return res
>>> dict1 = {'key1':['value11','value12','value13'] , 'key2':['value21','value22','value23']}
>>> dict2 = {'key1':['value14','value15'] , 'key2':['value24','value25']}
>>> dict3 = mix_dict(dict1, dict2)
>>> print dict3
... {'key1': ['value11', 'value12', 'value13', 'value14', 'value15'],
     'key2': ['value21', 'value22', 'value23', 'value24', 'value25']}

Solution 3

You can do it much simpler but if you want to use your code just change append to extend

dict1 = {'key1':['value11','value12','value13'] , 'key2':['value21','value22','value23']}
dict2 = {'key1':['value14','value15'] , 'key2':['value24','value25']}

dict3 = {}
for key in (dict1.viewkeys() | dict2.keys()):
    if key in dict1: dict3.setdefault(key, []).extend(dict1[key])
    if key in dict2: dict3.setdefault(key, []).extend(dict2[key])

print dict3

output:

{'key2': ['value21', 'value22', 'value23', 'value24', 'value25'], 'key1': ['value11', 'value12', 'value13', 'value14', 'value15']}

You can read this post about the difference ov append to extend

Solution 4

Here is another way to do this.

You can support merging N dicts of lists into a single dict of lists with this function:

def mergeDoLs(*dicts):
    def flatten(LoL):
        return [e for l in LoL for e in l]

    rtr={k:[] for k in set(flatten(d.keys() for d in dicts))}
    for k, v in flatten(d.items() for d in dicts):
        rtr[k].extend(v)
    return rtr

To use:

>>> dict1 = {'key1':['value11','value12','value13'] , 'key2':['value21','value22','value23'], 'key3':[1]}
>>> dict2 = {'key1':['value14','value15'] , 'key2':['value24','value25']}
>>> dict3 = {'key3':[2]}
>>> mergeDoLs(dict1, dict2, dict3) 
{'key3': [1, 2], 'key2': ['value21', 'value22', 'value23', 'value24', 'value25'], 'key1': ['value11', 'value12', 'value13', 'value14', 'value15']}

View more solutions

Comments

[njzk2]

probably a simple print {k:dict1.get(k, []) + dict2.get(k, []) for k in set(dict1.keys() + dict2.keys())} would do

[Padraic Cunningham]

do you still want the averages?

[Hypothetical Ninja]

Yes , but others told me to ask it as a separate question

[Padraic Cunningham]

lol and when you did you were asked to edit this question, I had an answer to the question you deleted

[Hypothetical Ninja]

Here is the other question.

[Hypothetical Ninja]

what does defaultdict(list) do??

[Ffisegydd]

defaultdict(list) is similar to using setdefault in your example. Basically if the key isn't already in dict3 then it will set the value as default []. This way is just a bit neater (IMHO).

[Quentin THEURET]

I dislike the except KeyError: pass − It's a bad management of Exceptions.

[Quentin THEURET]

@Ffisegydd you can use a condition (if key in dic) instead of an Exception management with a pass statement.

[Quentin THEURET]

Plus, I tried to test your answer and (dict1.keys() | dict2.keys()) returns an error in python 2.7

[Ffisegydd]

@QuentinTHEURET whilst I disagree with you I've modified it as it is probably more readable for people who aren't too versed in try...except... loops. And yes I changed it to keys for Python 3.4 (which is what I used).