Re-raise Python exception and preserve stack trace

roskakori picture roskakori · Jan 6, 2012 · Viewed 14k times · Source

I'm trying to catch an exception in a thread and re-raise it in the main thread:

import threading
import sys

class FailingThread(threading.Thread):
    def run(self):
        try:
            raise ValueError('x')
        except ValueError:
            self.exc_info = sys.exc_info()

failingThread = FailingThread()
failingThread.start()
failingThread.join()

print failingThread.exc_info
raise failingThread.exc_info[1]

This basically works and yields the following output:

(<type 'exceptions.ValueError'>, ValueError('x',), <traceback object at 0x1004cc320>)
Traceback (most recent call last):
  File "test.py", line 16, in <module>
    raise failingThread.exc_info[1]

However, the source of the exception points to line 16, where the re-raise occurred. The original exception comes from line 7. How do I have to modify the main thread so that the output reads:

Traceback (most recent call last):
  File "test.py", line 7, in <module>

Answer

Duncan picture Duncan · Jan 6, 2012

In Python 2 you need to use all three arguments to raise:

raise failingThread.exc_info[0], failingThread.exc_info[1], failingThread.exc_info[2]

passing the traceback object in as the third argument preserves the stack.

From help('raise'):

If a third object is present and not None, it must be a traceback object (see section The standard type hierarchy), and it is substituted instead of the current location as the place where the exception occurred. If the third object is present and not a traceback object or None, a TypeError exception is raised. The three-expression form of raise is useful to re-raise an exception transparently in an except clause, but raise with no expressions should be preferred if the exception to be re-raised was the most recently active exception in the current scope.

In this particular case you cannot use the no expression version.

For Python 3 (as per the comments):

raise failingThread.exc_info[1].with_traceback(failingThread.exc_info[2])

or you can simply chain the exceptions using raise ... from ... but that raises a chained exception with the original context attached in the cause attribute and that may or may not be what you want.