如何获取 python 中嵌套异常的堆栈跟踪？

Question

If an exception is raised I'd like to analyse the stack trace in python that tells about where exactly the problem is in the source code file.

Of course for that purpose the module traceback exists. And that works fine for regular exceptions. But how do you deal with this situation if nested exceptions occur?

Consider this example:

def test():
    try:
        a = 0
        b = 5 / a
    except Exception as ee1:
        assert False

test()

This example prints two exceptions:

Traceback (most recent call last):
  File "./test4d.py", line 12, in test
    b = 5 / a
ZeroDivisionError: division by zero

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "./test4d.py", line 18, in <module>
    test()
  File "./test4d.py", line 14, in test
    assert False
AssertionError

So information about both exceptions is known to the interpreter. I therefore like to retrieve these two pieces of information from Python: The stack trace of the assert statement (used as an example here to cause an exception) and the stack trace of the division by zero exception (used as an example here to cause an exception). How can I do that?

And the second part of the question: How can I do that in a structured way? The module traceback can be used to get more information about an existing exception. But I do not want to print the exception, I want to store it. Therefore I'd like to get the stack trace as a tuple of instances of some class where each instance represents the data within each stack frame. How can I do that ?

Answer 1

There is a variable named __context__ associated with an exception. This variable can be used to access nested exceptions. See this example:

import traceback

def test():
    try:
        a = 0
        b = 5 / a
    except Exception as ee1:
        assert False

try:
    test()
except Exception as ee:
    print(repr(ee))
    stackTraceList = traceback.extract_stack(ee.__traceback__.tb_frame)
    del stackTraceList[0]
    for frame in stackTraceList:
        print("\t", frame)

    if ee.__context__:
        print(repr(ee.__context__))
        stackTraceList = traceback.extract_stack(ee.__context__.__traceback__.tb_frame)
        del stackTraceList[0]
        for frame in stackTraceList:
            print("\t", frame)

This will output the following text:

AssertionError()
ZeroDivisionError('division by zero',)
     <FrameSummary file ./example.py, line 8 in test>

That indicates that both exceptions can be identified and their stack traces can be iterated through.

For convenience I implemented a simple helper module to process exceptions and stack traces named jk_exceptionhelper . You can install this module using pip . For details have a look at the GIT repository: https://github.com/jkpubsrc/python-module-jk-exceptionhelper