unhashable type: 'dict' Type Error [duplic

Question

This question already has answers here:

TypeError: unhashable type: 'dict', when dict used as a key for another dict [duplicate] (2 answers)

Closed 3 years ago.

Suppose I have this dictionary:

items = {1: {'title': u'testing123', 'description': u'testing456'},
2: {'description': u'testing123', 'description': u'testing456'},
3: {'description': u'testing123', 'description': u'testing456'},
4: {'description': u'testing123', 'description': u'testing456'},
5: {'description': u'testing123', 'description': u'testing456'},
6: {'description': u'somethingelse', 'description': u'somethingelse'}}

I want to filter out the duplicate values, so that in the end I'd get

{1: {'title': u'testing123', 'description': u'testing456'}, 6: {'title': u'something', 'description': u'somethingelse'}}

I wrote this code:

dic = {}

for key, value in items.items():
    if not set(value.values()).issubset(set(dic.values())):
        dic[key] = value

however I get the error message TypeError: unhashable type: 'dict'. I am not sure why this happens and how to fix it.

This is inspired by another question and my failed attempt to solve it.

Answer 1

dic.values() return list of dict

>>> for key, value in items.items():
...     print dic.values()
...
[{'description': u'testing456', 'title': u'testing123'}]
[{'description': u'testing456', 'title': u'testing123'}]
[{'description': u'testing456', 'title': u'testing123'}]
[{'description': u'testing456', 'title': u'testing123'}]
[{'description': u'testing456', 'title': u'testing123'}]
[{'description': u'testing456', 'title': u'testing123'}]
>>>

So, you can't apply set on dict as dict is not hashable.

Btw you can fix it by:

>>> dic = {}
>>> for key, value in items.items():
...     if not set(value.values()).issubset(set(sum([x.values() for x in dic.values()],[]))):
...         dic[key] = value
...
>>> dic
{1: {'description': u'testing456', 'title': u'testing123'}, 6: {'description':     u'somethingelse', 'title': u'somethingelse'}}
>>>

For python > 3.x

if not set(value.values()).issubset(set(sum([list(x.values()) for x in list(dic.values())],[]))):

Answer 2

You are trying to create a set of dicts, but that's not possible, since dicts are unhashable (because they are mutable -- whether they are equal can change as you modify/add/remove pairs in the dict).

Perhaps instead of using the dicts, you can use tuples of their values for your set, a la if not set((v['description_a'], v['description_b]) for v in value.values()).issubset((v['description_a'], v['description_b]) for v in set(dic.values())): or similar?

Answer 3

Edit: If you must use a set, as others have noted you have to use a hashable object like a tuple:

unique_items = set()
for k, v in items.items():
    sorted_v = tuple(sorted((k2, v2) for k2, v2 in v.items()))
    unique_items.add(sorted_v)
unique_items = dict(unique_items)

gives for unique_items:

{1: {'description': u'testing456', 'title': u'testing123'},
 6: {'description': u'somethingelse', 'title': u'somethingelse'}}

If items is not huge (or, at least if the output dictionary isn't expected to be huge):

items = {1: {'title': u'testing123', 'description': u'testing456'},
2: {'title': u'testing123', 'description': u'testing456'},
3: {'title': u'testing123', 'description': u'testing456'},
4: {'title': u'testing123', 'description': u'testing456'},
5: {'title': u'testing123', 'description': u'testing456'},
6: {'title': u'somethingelse', 'description': u'somethingelse'}}

unique_items = {}
for k, v in items.items():
    if v not in unique_items.values():
        unique_items[k] = v

(assuming the first key in your dictionary example should have been title). But you can't predict what the keys to this dictionary will be if duplicates do exist in items.