"Collecting" classes/variables from various files to simplify importing?

Suppose several classes are defined in multiple different files across a Python project, such as

mylib.somefile.Class1
mylib.somefile.Class2
mylib.anotherfile.Class3
mylib.athirdfile.Class4
...

What would be the best way to aggregate these classes so that a user could hypothetically just do something like:

from mylib.models import Class1, Class2, Class3, Class4

without simply moving all these classes into the same file? (Would result in a far too long unreadable unmaintanable file.

Project I'm working on currently accomplishes this by having a names.py file which itself imports everything from everywhere, including statements such as from .somemodule import *, which I have mixed feelings on.

Thoughts, suggestions, etc?

You just need to setup a init file with all the imports that you may need.

If you could, move all your files/functions to a folder.

EDIT: Just to clarify that you don't need to move things around, it's just to keep things more organized. In fact, you just need __init__.py in a folder with the imports defined. You can import classes and functions from anywhere into your init, just remember to add them to the __all__ list.

Then setup a __init__.py file with all your imports, should like this:

"""
Module definition...

"""

from somefile import Class1, Class2
from anotherfile import Class3
from athirdfile import Class4

#If you want some version control, use this to it
from distutils.version import LooseVersion
__version__ = "0.0.1"
__version_info__ = tuple(LooseVersion(__version__).version)

__all__ = [
    "Class1",
    "Class2",
    "Class3",
    "Class4"
]

Your directory tree should look like this:

|--mylib/
|    |--__init__.py
|    |--somefile.py
|    |--anotherfile.py
|    |--athirdfile.py
|--main.py
|--...stuff...

Then just call This module where you need it: from mylib import Class1, Class2, Class3, Class4