Using Python 'not in' on dict with tuple key where I don't have all tuple parts

user_439350 0
Viewed 267

I have a script that reads a list of users and reports from a MySQL database and runs a function for each of the values. The script checks the db every 5 minutes and adds these value pairs to a dictionary. As I loop through I don't want to re-add value pairs that are still in the dictionary. I've had no issues handling that with this code:

if (user, report) not in t1.reports_to_call:
    t1.add(report, user)

My issue is that I want to also check a dictionary t2.reports_requested where the key is a tuple of three parts user, report and some_unknown_id. If there a way to run not in on a dictionary where you only know 2 of three elements and want to wildcard the third?

I have been also looking to see if I can refactor this secondary dictionary and move some_unknown_id out of the key as I do think it may be the case that the user/report pair is unique in this dictionary. But if I determine I need to wildcard the third value what is best way to do this?

{('user', 'report', 'some_unknown_id'): {}}
dictionary python
5 Answers

You can use set and check if a new pair it is a subset of the t2.reports_requested.keys()

values_exist = [True for k in reports_requested.keys() if set(["user","report"]).issubset(k)]
if not values_exist:
    print("Add new tuple")
    pass

For example:

reports_requested = {('user', 'report', 'some_unknown_id'): {}, ('user1', 'report', 'some_unknown_id'): {}, ('user1', 'report', 'some_unknown_id2'): {}}
new_pair = {'user1', 'report2'}

values_exist = [True for k in reports_requested.keys() if new_pair.issubset(k)]
if not values_exist:
#     print("Add new tuple")
    pass

Performance check with timeit: 1.06 µs ± 184 ns per loop (mean ± std. dev. of 7 runs, 1000000 loops each)

user_125451020

No, you cannot check for the existence of keys in a dict using wild-cards.

Make a dict of dicts:

{('user', 'report'): {
    'some_unknown_id': {},
}

or even better:

{'user': {
    'report': {
        'some_unknown_id' : {},
    },
}

Then you have the best of all worlds.

You can first check whether a user occurs, and if so, whether in combination with a report, and if so, whether in combination with some other identifier.

user_152362200

You can use zip to extract "sub-keys" for testing against, as

if (user,report) not in zip(*list(zip(*t2.keys()))[0:2]):
  #etc

Here, the [0:2] extracts only the first two components of the key.

user_40147680

You can't wildcard the key but you could replace the third key with None if it isn't there:

x = {('user','report'): {'dummy': 1}, ('user','report', 'unknown_key'): {'dummy': 2}, }

new_dict = dict(zip({(k[0], k[1], k[2] if len(k) == 3 else None) for k in x.keys()}, x.values()))

new_dict

{('user', 'report', 'unknown_key'): {'dummy': 1},
 ('user', 'report', None): {'dummy': 2}}

This way you'll always know the length of the key and you can test the last value.

user_11789710

You can rely on the default get operation of the dictionary and add such element if None is returned as follow.

if not t1.get((user, report), default=None):
    t1.add(report, user)

for the t2, I'd suggest to do a two level dictionary, sort of:

t2 = {
    ('user', 'report'): {
        'some_unknown_id': {},
        'some_unknown_id_2': {},
    }
}

In that fashion you can handle several "third unkown keys" right under ('user', 'report').

user_35374300
Related