list approach due nested order

[3Pinter]

Horrible thread name, but let me try and explain it.

I put lists (with 2 (always) values) in a main-list and I want to cleanup the duplicates in that main-list. The order of the 2 values don't matter to me and are therefor the same.

mainlist = []
listA = ["a", "b"]
listB = ["b", "a"]
listC = [1,2]

mainlist = [ listA, listB ]
unique-mainlist = [ ["a", "b"], [1,2] ]

[stranac]

Since the order doesn't matter, you might want to consider using a set of frozensets:

>>> unique = set()
>>> unique.add(frozenset(['a', 'b']))
>>> unique.add(frozenset(['b', 'a']))
>>> unique.add(frozenset([1, 2]))
>>> unique
{frozenset({'a', 'b'}), frozenset({1, 2})}

[3Pinter]

frozenset... awesome. New functionality for me, but your suggestion certainly works Thanks!.

Tiny side question:
from a pythonic point of view is it better to
- check if a list (or reversed) exists in main-list
- if so, skip, else add

Or like my current approach:
- dump all in main-list
- frozenset them

Thinking in large numbers here.

[Gribouillis]

Combining frozenset with unique_everseen yields my preferred solution

from more_itertools import unique_everseen

def create():
    yield ['a', 'b']
    yield ['b', 'a']
    yield [1, 2]
    
unique = list(unique_everseen(create(), key=frozenset))
print(unique)

Output:
[['a', 'b'], [1, 2]]

[3Pinter]

Hmmm, more_itertools isn't installed.

Unsure if I can install this.

[Gribouillis]

Unsure if I can install this.

Try

Output:
python -m pip install more_itertools

[3Pinter]

# mainlist is a massive list 

joinlist = []
unique = set()
for m in mainlist:
	unique.add(frozenset(m))

for u in unique:
	joinlist.append(list(u))

if joinlist:
	for join_pair in joinlist:
		#do something
		joinelements(join_pair)

This is working, but feels a bit clumpsy. And looking at the yield functionality ... I think my coding could be improved?