Python: Should I put my data in lists or object attributes?

I am looking for an appropriate data structure in Python for processing variably structured forms. By variably structured forms I mean that the number of form fields and the types of the form's contents are not known in advance. They are defined by the user who populates the forms with his input.

What are the pros and cons of putting data in A) object attributes (e.g. of an otherwise empty "form"-class) or B) simply lists/dicts? Con开发者_高级运维sider that I have to preserve the sequence of form fields, the form field names and the types.

(Strangely, it has been difficult to find conclusive information on this topic. As I am still new to Python, it's possible that I have searched for the wrong terms. If my question is not clear enough, please ask in the comments and I will try to clarify.)

In Python, as in all object-oriented languages, the purpose of classes is to associate data and closely-related methods that act on that data. If there's no real encapsulation going on (i.e. the methods help define the ways you can interact with the data), the best choice is a conglomeration of builtin types like lists and dictionaries as you mention and perhaps some utility functions that act on those sorts of data structures.

Python classes are literally just two dicts (one for functions, one for data), a name and the rules how Python looks for keys. When you access existing keys, there is absolutely no difference to a dict (unless you overwrote the access rules of cause).

That means that there is no drawback (besides more code) to using classes at all and you should never be afraid to write a class.

In your particular case I think you should go with classes, for one simple reason: You might want to extend them later. Maybe you want to add constraints on the name (length, allowed letters, uniqueness, ...) or the value (not empty, length, type, ...) of a field one day. Maybe you want to validate all fields in a form. If you use a class you can do this without changing any code outside the class! And as I said before, even if you don't, there are no drawbacks!

I guess my rule of thumb for classes is: Don't use a class if you're absolutely sure that there is nothing to add to it. If not just write those few extra lines.

It's not very Pythonic to randomly add members to an object. It would be more Pythonic if you used member methods to do it, but still not the way things are usually done.

Every library I've seen for this kind of thing uses dictionaries or lists. So that is the idiomatically Python way to handle the problem. Sometimes they use an object that overrides __getitem__ so it can behave like a dictionary or list, but it's still dictionary syntax that's used to access the fields.

I think all the pros and cons have to do with people understanding your code, and since I've never seen code that handles this by having an object with members that can appear I don't think many people will find code that does do that to be very understandable.

A list of dictionaries (e.g. [{"type": "text", "name": "field_name", "value": "test value"}, ...]) would be a usable structure, if I understand your requirement correctly.

Whether object are better in this case depends on what you're doing later. If you use the objects just as data storage, you don't gain anything. Maybe a list of field objects, which implement some appropriate methods to deal with your data, would also be a good choice.

maybe if you set up an object to use for each field and store those in a list, but that is practically ending up like a glorified dictionary

then you could access it like

fields[2].name fields[2].value

ect