Repeated import path patterns in python

There is no miracle cure for this problem.

There are, however, a few strategies you can use to make your imports a little bit nicer:

  • Import all names from a module at the same time

    Instead of

    from a.b.c.d.e.f.g import g1
    from a.b.c.d.e.f.g import g2
    

    use

    from a.b.c.d.e.f.g import g1, g2
    
  • Use relative imports (if you're importing within your own package)

    Assuming that this import is happening in a.b.c.d.e, you can replace

    from a.b.c.d.e.f.g.h import h1
    

    with

    from .f.g.h import h1
    

    This also works if you're in a sibling (sub-)module. For example, if this import is taking place in a.b.c.d.e.x.y, you can use

    from ...f.g.h import h1
    

    For more details about relative imports, see

    • Intra-package references
    • How to do relative imports in Python?
    • Relative imports for the billionth time

  • Refactor your package (if it's your own code)

    If your package has more than 4 levels of submodules (like a.b.c.d.e), there's a pretty high chance that you should rethink your package structure. At that point you're really just shooting yourself in the foot. Perhaps c could be a standalone package, outside of a. Or perhaps e doesn't really need to be inside d and can be moved up a level or two. To quote the Zen of Python: Flat is better than nested.

Critique of other suggestions

  • importlib.import_module

    Someone suggested this:

    import importlib
    ref_path = 'a.b.c.d.e.f.g'
    
    g1 = importlib.import_module(ref_path).g1
    h1 = importlib.import_module(ref_path).h.h1
    

    This serves absolutely no purpose. Now you have to write importlib.import_module(ref_path) instead of from a.b.c.d.e.f import. It's not shorter. It's not more readable. It's nothing but a more verbose version of the next suggestion. (Read on...)

  • Assigning a.b.c.d.e.f to a variable

    The next suggestion was this:

    import a.b.c.d.e.f.g as ref
    
    g1 = ref.g1
    h1 = ref.h.h1
    

    This looks nice, but it doesn't always work. If the g module doesn't automatically import the h submodule, this code will throw an AttributeError. To demonstrate:

    >>> import urllib as ref
    >>> urlopen = ref.request.urlopen
    Traceback (most recent call last):
      File "<stdin>", line 1, in <module>
    AttributeError: module 'urllib' has no attribute 'request'
    >>>
    >>> from urllib.request import urlopen  # this works, though
    >>>
    

    If this is inside your own package, there's an additional disadvantage to this solution: Your IDE most likely won't understand your imports, and won't automatically update them for you if you ever refactor your package structure. In the long run, old-fashioned import statements are more maintainable.


You can try to use importlib as follows:

import importlib
ref_path = 'a.b.c.d.e.f.g'

g1 = importlib.import_module(ref_path).g1
h1 = importlib.import_module(ref_path).h.h1

Edit: Another way of doing it could be to import the ref path as a package and then assigning the sub-modules to local variables as follows:

import a.b.c.d.e.f as ref

g1 = ref.g1
h1 = ref.h.h1