如何在python中没有全局变量的情况下安全地将代码分为函数

Question

It's somewhat too general. Suppose I'm writing something long and then I want to cut the code into several small parts with elementary function.

So the code would be something like:

def g():
    # do sth with var1,var2
    pass
def h():
    # do sth with changed var1,var2, and original var3
    pass
def f():
    global var1,var2,var3
    g()
    h()

And in this form, usually, variables are in-place defined or changed in g() or h() .

As it's warned I should minimize the use of global variables, is there anyway to share (enable in-place/dynamic editing of some vars) data without global statement?

Answer 1

Well, actually there're two options, but both require shifting the paradigm from imperative to something else.

Create a class, have all those functions be methods of the class, shared vars - the (instance or class-level) attributes - shifts to object-oriented paradigm

class Something(object):
    def __init__(self, var1_init_val, var2_init_val, var3_init_val):
        self.var1 = var1_init_val
        self.var2 = var2_init_val
        self.var3 = var3_init_val

   def g(self):
       self.var1 = 'qwe'

   def h(self):
       self.var2 = 'asd'

   def f(self):
       self.g()
       self.h()

#use as
something = Something(var1, var2, var3)
something.f()

Make those vars parameters of the function and restructure the code that they are not modified in place - shifts to functional paradigm

def g(var1, var2):
    return 'qwe', var2

def h(var1, var2):
    return var1, 'asd'

def f(var1, var2, var3):
    var1, var2 = g(var1, var2)
    var1, var2 = h(var1, var2)  # h here sees "updated" values

# use as
f(var1, var2, var3)

Both examples achieve exactly hte same result - var1 becomes 'qwe', var2 becomes 'asd' and var3 stays unchanged

So, choose your Paradigm :)