Python通过多个其他模块重用同一模块

Question

Is it possible to achieve this in python? Basically, shared_fun() is a very frequently used utility function, used in multiple modules (user_module1 and user_module2). However, each user wants a slightly different parameter SHARE_GLOBAL_VAR.

Some motivation why I would want it this way:

A. Imagine there are a lot of shared_fun(), shared_fun1(), shared_fun2() ...... and all of them rely on the same SHARE_GLOBAL_VAR but do different things with it. Therefore I really don't want to make SHARE_GLOBAL_VAR an argument for every one of those shared_fun(SHARE_GLOBAL_VAR)

B. I want to make the maintenance of shared_fun() easy so that I don't want to copy the code into each user_module.py file individually.

I think the task boils down to make a copy of share module namespace inside each user_module namespace, which I'm not sure if it's legal in python.

#share.py:
SHARE_GLOBAL_VAR = 0
def shared_fun():
    return SHARE_GLOBAL_VAR

#user_module1.py:
import share
share.SHARE_GLOBAL_VAR = 1
def user1_fun():
    return share.shared_fun()

#user_module2.py:
import share
share.SHARE_GLOBAL_VAR = 2
def user2_fun():
    return share.shared_fun()

#main.py:
import user_module1
import user_module2
# expecting a result of 1
print(user_module1.user1_fun())
# expecting a result of 2
print(user_module2.user2_fun())

Answer 1

You can make use of the concept of closure , for example

def sharedFun():
    shared_var=0
    def getSharedVariable(shared_value=None):
        if shared_value:
            return shared_value
        else:
            return shared_var
    return getSharedVariable


shared=sharedFun()
print shared(1) # output 1
print shared(2) # output 2
print shared(3) # output 3
print shared() # output 0

you can find more about closure here

Answer 2

For a given file on disk, there is at most one in-memory module object per process, so what you show above, with multiple imports in one process, will not work. This is why it is 'impolite' for 3rd party library modules to 'monkey patch' standard library modules and why global data variables are generally discouraged.

Answer 3

pushpendra's answer of closure is fantastic!!! I know very little about functional programming but this is really inspiring me to dig more into it.

On the other hand, I just came up with an OOP solution for my specific issue: make the shared.py a class and all the shared_fun as class method while SHARE_GLOBAL_VAR is now a class data member. Each user_module1/2 will now just instantiate an object of the share class with different SHARE_GLOBAL_VAR.

This achieves the similar idea of creating multiple copies of share_fun, via a class instead of duplicating module importing (which is impossible per Terry's answer ).

My very naive comparison between OOP and Functional in this case: Closure solution provides much more freedom in creating different versions of functions. Since closure provides function duplicate per function instead of per GLOBAL_SHARE_VAR, therefore it saves the overhead of creating unnecessary methods that are never used by the OOP solution.

Time to get serious about functional programming!

Answer 4

Claim down guy. Let me explain step by step for you.

First, take a careful look at your accepted answer. Actually, there is no essential difference between this answer and a modified version of your original question.

#share.py:
SHARE_GLOBAL_VAR = 0
def getSharedVariable(shared_value=None):
    if shared_value:
        return shared_value
    else:
        return SHARE_GLOBAL_VAR

In accepted answer, the scope is function sharedFun while the scope is module in the above snippet. There are almost the same. You can use the above snippet like this:

import share
shared = share.getSharedVariable
print shared(1) # output 1
print shared(2) # output 2
print shared(3) # output 3
print shared() # output 0

You are still passing an identical value to getSharedVariable .

Further, let's talk about your OOP solution. It is just meaningless. If each object will have their own different SHARE_GLOBAL_VAR , why this variable called SHARE_GLOBAL_VAR ? It is just a normal local variable.

Finally, let's modify the accepted answer:

# assume SHARE_GLOBAL_VAR equals to 0.
def getSharedVariable(shared_value=SHARE_GLOBAL_VAR):
    return shared_value

shared = getSharedVariable
print shared(1) # output 1
print shared(2) # output 2
print shared(3) # output 3
print shared() # output 0

Do you notice what you are actually want? Just a default parameter.