允许__main__中的全局变量访问pytest会话

Question

我很好奇，如果可以在__main__中初始化模块全局并在其使用pytest.main()机制运行时使pytest可以访问全局模块，这种做法是否可行。

我最初的尝试似乎表明是否将全局状态传递给pytest会话...

import pytest

globalList = []


def test_globalList() :
    global globalList
    print( "globalList: {}".format( globalList))
    assert globalList


if __name__ == '__main__':
    print('Main Start')
    globalList.append(1)
    globalList.append(2)
    globalList.append(3)

    pytest_options = [ __file__ ]
    print( "globalList: {}".format( globalList))
    print('\n\n\n\n')
    pytest.main( pytest_options  ) # run pytest for just this file
    print('Main Complete')

我产生以下控制台输出

Main Start
globalList: [1, 2, 3]


=================================== test session starts ================
platform darwin -- Python 2.7.10, pytest-4.0.2, py-1.7.0, pluggy-0.8.0
rootdir: /Users/USER/, inifile:
collected 1 item

test_temp.py F                                                    [100%]

=================================== FAILURES ===========================
___________________________________ test_globalList ____________________

    def test_globalList() :
        print( "globalList: {}".format( globalList))
>       assert globalList
E       assert []

test_temp.py:39: AssertionError
----------------------------------- Captured stdout call ---------------
globalList: []
=================================== 1 failed in 0.09 seconds ===========
Main Complete

我无法理解pytest中的哪个概念，这会提前提醒我这种尝试是不可能的。有没有一种方法可以在pytest.main调用中共享全局状态？

---

更新

证明globalList是一个全新的实例，但是在这两种情况下，进程和线程都是相同的。在与 main 上下文完全分开的pytest上下文下发生模块加载。引用documentation

中的注释

  

注意

     

调用pytest.main（）将导致导入您的测试及其导入的任何模块。

import pytest
import os
import threading

globalList = []

print( "\n__name__ : {}".format( __name__ ))
print("process ID : {}".format(hex(os.getpid())))
cur_thread = threading.current_thread()
print( "thread ID : {}".format( hex(id(cur_thread )) ))
print( "thread Name : {}".format( cur_thread.name ))
print( "globalList ID : {}".format(hex(id(globalList))))

def test_globalList() :
    global globalList
    print( "globalList: {}".format( globalList)) # TODO this will be empty in the pytest session, somehow this approach fails to bring this variable into scope.
    assert globalList

if __name__ == '__main__':
    print('Main Start')
    globalList.append(1)
    globalList.append(2)
    globalList.append(3)

    pytest_options = [ __file__ , '-s']
    print( "globalList: {}".format( globalList))
    print('\n\n\n\n')
    pytest.main( pytest_options  ) # run pytest for just this file
    print('Main Complete')

以及相应的控制台输出...

__name__ : __main__
process ID : 0xa43e
thread ID : 0x10ac9b190
thread Name : MainThread
globalList ID : 0x10b476320
Main Start
globalList: [1, 2, 3]




=================================== test session starts ================
platform darwin -- Python 2.7.10, pytest-4.0.2, py-1.7.0, pluggy-0.8.0
rootdir: /Users/USER, inifile:
plugins: lazy-fixture-0.5.1, html-1.20.0, metadata-1.8.0
collecting ...
__name__ : test_temp
process ID : 0xa43e
thread ID : 0x10ac9b190
thread Name : MainThread
globalList ID : 0x10bb46128
collected 1 item

test_temp.py globalList: []
F

=================================== FAILURES ===========================
___________________________________ test_globalList ____________________

    def test_globalList() :
        global globalList
        print( "globalList: {}".format( globalList))
>       assert globalList
E       assert []

test_temp.py:50: AssertionError
=================================== 1 failed in 0.10 seconds ===========
Main Complete

立即弹出一些差异，列表的模块名称和地址不同...

__name__      : __main__      test_temp   <--different
process ID    : 0xa43e        0xa43e
thread ID     : 0x10ac9b190   0x10ac9b190
thread Name   : MainThread    MainThread
globalList ID : 0x10b476320   0x10bb46128 <--different