Matching dictionary key tuple items in a comprehension

Question

When using a comprehension expression that scans a dictionary, if I know the keys of the dictionary are simple 2 items tuples, is it possible to "expand" them to named variables?

eg if I have:

d = { ('car','blue') : 24,
      ('car',    'red'  ): 5,
      ('plant',     'green'): 12,
      ('box','blue' ): 3
}

I know I can do

[d[key] for key in d.keys() if key[1]=='blue']

and get

[24, 3]

But this loses a lot of the semantics and I wanted to know if I can somehow do something more like:

[count for {(object,color):count} in d if color=='blue']

(which I know does not work). Also I don't understand why the error returned for the last expression is:

SyntaxError: can't assign to literal

Answer 1

I would go with:

[count for (object, color), count in d.items() if color == 'blue']

as the clearest solution.

The error you are getting is because when using the in keyword python is trying to assign an item from d dictionary to {(object,color):count} which is a literal means create a new dictionary. instead, use (object, color), count as this is how python expands dict items.

Answer 2

You can't quite do that, but you can get close if you "upgrade" your keys to namedtuples . I'm using 'kind' as the name of the first item in the namedtuple because object is a built-in type, and although we can safely use it here I think it'd make the code a bit confusing.

from collections import namedtuple

d = { ('car', 'blue') : 24,
      ('car', 'red'): 5,
      ('plant', 'green'): 12,
      ('box', 'blue'): 3
}

Thing = namedtuple('Thing', ('kind', 'color'))
d = {Thing(*key): count for key, count in d.items()}
print(d)

lst = [count for key, count in d.items() if key.color == 'blue']
print(lst)
lst = [count for key, count in d.items() if key.kind == 'car']
print(lst)

output

{Thing(kind='car', color='blue'): 24, Thing(kind='car', color='red'): 5, Thing(kind='plant', color='green'): 12, Thing(kind='box', color='blue'): 3}
[24, 3]
[24, 5]

---

Actually, you don't really need namedtuples, although I think they make it a bit nicer. ;)

d = { ('car', 'blue') : 24,
      ('car', 'red'): 5,
      ('plant', 'green'): 12,
      ('box', 'blue'): 3
}

lst = [count for (kind, color), count in d.items() if color == 'blue']
print(lst)

Answer 3

It is possible. Try:

  [d.get((object,color)) for object,color in d.keys() if color == 'blue']

this returns

  [24, 3]