Organize heavy python imports

About 25% of my code depends on the modules: Traits, tvtk, ... which are quite 开发者_开发知识库heavy to import. It typically takes a good 2 seconds on my machine (and more on other).

My modules are organized as the following

mainmodule
|--submodule1
|--submodule2
   |--subsubmodule1
   |--subsubmodule2
|--submodule3
|--submodule4
   |--subsubmodule1
   |--subsubmodule2

In these, the submodule1 and submodule2 use Traits. That means 75% of the time, if I call import mainmodule, I will have to wait for the heavy modules to be imported but then they won't be used.

How do I organize my imports so that I can lower my import time?

Maybe there is a way to do something like:

import mainmodule

and have

mainmodule
|--submodule3
|--submodule4
   |--subsubmodule1
   |--subsubmodule2

And only call:

import mainmodule.heavy

to have everything

It sounds like what you want is a way so that importing mainmodule doesn't automatically import submodule1 and submodule2, which take a long time to load.

That's pretty easy, actually. You can import submodule1 and submodule2 only in functions that need them. Or move those functions into a separate module called mainmodule_heavy.py.

(Or you could hack the Python module system to load modules lazily. But that kind of hack tends to cause problems, and it sounds unnecessary for your case.)

You can put some code like this within a function / module:-

def heavy():
    global x
    global y
    import x, y

def mainmodule():
    if heavy not in globals():
        import heavy

Actually, this wouldn't work within the same program, as a function cannot be imported. Also, you'd want to check for a string within globals, not the module itself. So, instead:-

def heavy():
    global x
    global y
    import x, y

def mainmodule():
    if 'x' not in globals() or 'y' not in globals():
        heavy()