如何在重新输入异常时有效地传播完整的堆栈跟踪？

Question

这是一个简化的例子：

class FooException(Exception): pass
class BarException(Exception): pass

def subprocess_one():
  # make lots of calls to external service Foo that might fail in all weird ways
  pass

def subprocess_two():
  # make lots of calls to external service Bar that might fail in all weird ways
  raise RuntimeError('hello world')
def api_call():
  try:
    subprocess_one()
  except Exception as e:
    raise FooException(e) from e
  try:
    print('woops')
    subprocess_two()
  except Exception as e:
    print('raising barexception')
    raise BarException(e) from e

api_call()

This raises the following stack trace：

Python 3.6.1 (default, Dec 2015, 13:05:11)
[GCC 4.8.2] on linux

woops
raising barexception
Traceback (most recent call last):
  File "python", line 23, in <module>
  File "python", line 21, in api_call
BarException: hello world

我希望看到更多的内容：

   Traceback (most recent call last):
  File "python", line 23, in <module>
  File "python", line 21, in api_call
  File "python", line 9, in subprocess_call
BarException: hello world

或以某种方式以其他方式清楚地显示由包装异常生成的历史跟踪（注意添加原始行和函数作为回溯的一部分）。

我的假设是raise BarException(e) from e方法可以做到这一点，但是，似乎并没有这样做。没有从exc_info打印堆栈有没有办法做到这一点？

我使用的是Python 3.5+，因此我对Python 2.7解决方案不太感兴趣。