feat: Allow compatibility with other ShortUUID libraries (#104)

[Thiagomrfs]

My attempt to fix: #104.

I've added should_sort kwarg to set_alphabet, whenever this is set to False, the code will use an ordered set rather than the default behaviour of sorting an unordered set.

Other than that, I've tried to integrate this behaviour into the Django Field. I don't have any experience working with custom Django fields, so I'd appreciate if someone could review commit 077ed4.

[skorokithakis]

This is getting the causality wrong, as this is the original ShortUUID library, which means that the other libraries aren't compatible with this one. I'm not sure whether this library needs to change because someone copied it and made it incompatible, but maybe it won't hurt to be practical. I'll review the PR now and see, thank you for the contribution!

[skorokithakis]

This looks ok, I guess it can't hurt anything. However, could you rename the argument to dont_sort_alphabet and make it default False? Also, could you add some documentation to explain what the argument does, and why it's required?

Thanks again!

[Thiagomrfs]

Thanks for the response, I'll get at it shortly.

[Thiagomrfs]

Just finished the requested changes!

[skorokithakis]

Looks good, thank you! Do you know whether libraries with repeated characters work? They should, since you're using a dict, but I just want to make sure.

[skorokithakis]

This looks good, thank you!

[Thiagomrfs]

Sorry, I didn't quite understand your question.

You meant to ask if libraries with non-unique characters in the alphabet would work?

Since dictionaries preserve insertion order, but can't have duplicated keys, I used it to eliminate duplicates while maintaining the original order of the characters in the alphabet string.

I mainly use this library when working with shortUUIDs, so I don't really have a grasp of other libraries. Maybe the issue's OP could answer that.

[skorokithakis]

I was asking if character deduplication works as required or if, for example, it leaves the last character it sees, so 1a1 would become a1 instead of 1a. I'm sure it's fine, but also it removes compatibility with any Python before 3.7, which is probably OK.

[Thiagomrfs]

Understood. It stores the first appearance of the character.

Using your example:

>>> string = "1a1"
>>> list(dict.fromkeys(string))
['1',  'a']

and

>>> string = "a1a1"
>>> list(dict.fromkeys(string))
['a', '1']

[skorokithakis]

Excellent, thank you!