Why do you have to call .items() when iterating over a dictionary in Python?

Question

Why do you have to call items() to iterate over key, value pairs in a dictionary? ie.

dic = {\'one\': \'1\', \'two\': \'2\'}
for k, v in dic.items():
    pri         


        

Answer 1

My guess: Using the full tuple would be more intuitive for looping, but perhaps less so for testing for membership using in.

if key in counts:
    counts[key] += 1
else:
    counts[key] = 1

That code wouldn't really work if you had to specify both key and value for in. I am having a hard time imagining use case where you'd check if both the key AND value are in the dictionary. It is far more natural to only test the keys.

# When would you ever write a condition like this?
if (key, value) in dict:

Now it's not necessary that the in operator and for ... in operate over the same items. Implementation-wise they are different operations (__contains__ vs. __iter__). But that little inconsistency would be somewhat confusing and, well, inconsistent.

Answer 2

For every python container C, the expectation is that

for item in C:
    assert item in C

will pass just fine -- wouldn't you find it astonishing if one sense of in (the loop clause) had a completely different meaning from the other (the presence check)? I sure would! It naturally works that way for lists, sets, tuples, ...

So, when C is a dictionary, if in were to yield key/value tuples in a for loop, then, by the principle of least astonishment, in would also have to take such a tuple as its left-hand operand in the containment check.

How useful would that be? Pretty useless indeed, basically making if (key, value) in C a synonym for if C.get(key) == value -- which is a check I believe I may have performed, or wanted to perform, 100 times more rarely than what if k in C actually means, checking the presence of the key only and completely ignoring the value.

On the other hand, wanting to loop just on keys is quite common, e.g.:

for k in thedict:
    thedict[k] += 1

having the value as well would not help particularly:

for k, v in thedict.items():
    thedict[k] = v + 1

actually somewhat less clear and less concise. (Note that items was the original spelling of the "proper" methods to use to get key/value pairs: unfortunately that was back in the days when such accessors returned whole lists, so to support "just iterating" an alternative spelling had to be introduced, and iteritems it was -- in Python 3, where backwards compatibility constraints with previous Python versions were much weakened, it became items again).