从Python中的配对键字典中检索值

Question

I have the need for a data structure where the keys are essentially a combination of 2 strings. The overall combination is unique, and the strings are separated by a comma (so a mini csv). Something like this:

paired_dict = {
    '123,abc': [1, 2, 3],
    '456,abc': [4, 5, 6],
    '123,def': [7, 8, 9],
    ...
}

Now, my question is this. What is the most optimal way to retrieve all the values given only one part of the pair? So lets say I give the dictionary ('abc', 1) , the following objects are returned:

[1, 2, 3], [4, 5, 6]

Or if I give the dictionary ('456', 0) , then only the following object is retrieved:

[4, 5, 6]

Note, the first argument is the pair of the key, and the second argument is the key index within that pair. Is the only way to do this by running a loop through the entire dictionary till all the patterns are found?

Answer 1

You can create an optimised lookup table by traversing the dictionary once. Afterwards, you can directly look up partial keys:

split_dicts = [{}, {}]
for key, value in paired_dict.items():
    head, tail = key.split(',')
    split_dicts[0].setdefault(head, []).append(value)
    split_dicts[1].setdefault(tail, []).append(value)

def lookup(key, index):
    return split_dicts[index][key]

This fetches all corresponding values without iteration:

>>> lookup('abc', 1)
[[1, 2, 3], [4, 5, 6]]
>>> lookup('456', 0)
[[4, 5, 6]]

Answer 2

I would suggest making your keys tuples:

paired_dict = {
    ('123', 'abc'): [1, 2, 3],
    ('456', 'abc'): [4, 5, 6],
    ('123', 'def'): [7, 8, 9],
    ...
}

You can retrieve values with a list comprehension:

[v for k, v in paired_dict.items() if k[0] == '456']

Answer 3

Edit: This answer is kinda the same idea as @MisterMiyagi 's but not as Pythonic, I'm keeping the answer since I have some other stuff that maybe useful for you.

If you can't modify the paired_dict structure you could preprocess the keys in it to create two more dicts mapping the two partial_keys to the overall_key.

Something like this should get you started assuming the paired_dict will always be valid ie the part before the ',' will always be a valid int :

from collections import defaultdict

paired_dict = {
    '123,abc': [1, 2, 3],
    '456,abc': [4, 5, 6],
    '123,def': [7, 8, 9],
}

first_key_mappings = defaultdict(list)
second_key_mappings = defaultdict(list)
for key in paired_dict.keys():
    first_key, second_key = key.split(",")
    first_key_mappings[int(first_key)].append(key)
    second_key_mappings[second_key].append(key)


def paired_dict_search(search_term):
    partial_key, index = search_term
    res = []
    if index == 0:
        for key in first_key_mappings[partial_key]:
            res.append(paired_dict[key])
    elif index == 1:
        for key in second_key_mappings[partial_key]:
            res.append(paired_dict[key])
    else:
        raise ValueError(f"Invalid index provided: {index}")
    return res


test_search_terms = [('123', 0), (123, 0), (456, 0), ('abc', 1), ('def', 1),
                     ('123', 3)]
for search_term in test_search_terms:
    print(f"{search_term} => ")
    print(f"\t{paired_dict_search(search_term)}")
    print()

Output:

('123', 0) =>
    []

(123, 0) =>
    [[1, 2, 3], [7, 8, 9]]

(456, 0) =>
    [[4, 5, 6]]

('abc', 1) =>
    [[1, 2, 3], [4, 5, 6]]

('def', 1) =>
    [[7, 8, 9]]

('123', 3) =>
Traceback (most recent call last):
  File "main.py", line 32, in <module>
    print(f"\t{paired_dict_search(search_term)}")
  File "main.py", line 26, in paired_dict_search
    raise ValueError(f"Invalid index provided: {index}")
ValueError: Invalid index provided: 3