Get exception description and stack trace which caused an exception, all as a string

Question

How to convert a caught Exception (its description and stack trace) into a str for external use?

try:
    method_that_can_raise_an_exception(params)
except Exception as e:
    print(complete_exception_description(e))

Bodgie

what is wrong with: sys.exc_info()? – Kissiah

@CharlieParker It has no trace. – Sproul

Answer 1

See the traceback module, specifically the format_exc() function. Here.

import traceback

try:
    raise ValueError
except ValueError:
    tb = traceback.format_exc()
else:
    tb = "No error"
finally:
    print tb

Epithelium

Does this only work with the last error? What happens if you start passing the error around to other bits of code? I'm writing a log_error(err) function. – Aborticide

It works with the error that was caught and handled. – Epithelium

In addition to what I was looking for, I just learned that try can take an else clause. – Meander

Can you show the output in case of an exception? – Suggestible

what does tb = traceback.format_exc() do or the type it returns? – Kissiah

what is wrong with: sys.exc_info()? – Kissiah

That's a useless "else", it would never go there... – Lambert

I mean, yes, it's useless in this toy example, but in a real situation you'd want to handle that possibility. – Epithelium

Answer 2

Let's create a decently complicated stacktrace, in order to demonstrate that we get the full stacktrace:

def raise_error():
    raise RuntimeError('something bad happened!')

def do_something_that_might_error():
    raise_error()

Logging the full stacktrace

A best practice is to have a logger set up for your module. It will know the name of the module and be able to change levels (among other attributes, such as handlers)

import logging
logging.basicConfig(level=logging.DEBUG)
logger = logging.getLogger(__name__)

And we can use this logger to get the error:

try:
    do_something_that_might_error()
except Exception as error:
    logger.exception(error)

Which logs:

ERROR:__main__:something bad happened!
Traceback (most recent call last):
  File "<stdin>", line 2, in <module>
  File "<stdin>", line 2, in do_something_that_might_error
  File "<stdin>", line 2, in raise_error
RuntimeError: something bad happened!

And so we get the same output as when we have an error:

>>> do_something_that_might_error()
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "<stdin>", line 2, in do_something_that_might_error
  File "<stdin>", line 2, in raise_error
RuntimeError: something bad happened!

Getting just the string

If you really just want the string, use the traceback.format_exc function instead, demonstrating logging the string here:

import traceback
try:
    do_something_that_might_error()
except Exception as error:
    just_the_string = traceback.format_exc()
    logger.debug(just_the_string)

Which logs:

DEBUG:__main__:Traceback (most recent call last):
  File "<stdin>", line 2, in <module>
  File "<stdin>", line 2, in do_something_that_might_error
  File "<stdin>", line 2, in raise_error
RuntimeError: something bad happened!

Cioffi

@Yunti I believe this API has been consistent across Python 2 and 3. – Cioffi

Formatting of this answer was discussed on meta: meta.#386977. – Resist

I sent an edit to the following but was not logged in so showing as anonymous: except Exception as e: logger.exception("<<clearly and distinctly describe what failed here>>", exc_info=e) – Kaitlinkaitlyn

traceback.format_exc() uses sys.exc_info(). In Python 2 it would return whatever exception was currently being handled and was buggy, sometimes returning exceptions being handled on other threads. Now the documentation currently asserts it's good, and I don't have time to look up the current state of the bug tracker... – Cioffi

Answer 3

With Python 3, the following code will format an Exception object exactly as would be obtained using traceback.format_exc():

import traceback

try: 
    method_that_can_raise_an_exception(params)
except Exception as ex:
    print(''.join(traceback.format_exception(etype=type(ex), value=ex, tb=ex.__traceback__)))

The advantage being that only the Exception object is needed (thanks to the recorded __traceback__ attribute), and can therefore be more easily passed as an argument to another function for further processing.

Idiotism

It is better than sys.exc_info() which is not good OO style and use global variable. – Sollars

This asks specifically how to get the traceback from the exception object as you have done here: #11415394 – Carney

There is a simpler Python3 way without using .__traceback__ and type, see https://mcmap.net/q/36591/-get-exception-description-and-stack-trace-which-caused-an-exception-all-as-a-string – Jessabell

Answer 4

For Python 3.5+
Use traceback.TracebackException, it can handle exceptions caught anywhere.

def print_trace(ex: BaseException):
    print(''.join(traceback.TracebackException.from_exception(ex).format()))

---

Example

import traceback

try:
    1/0
except Exception as ex:
    print(''.join(traceback.TracebackException.from_exception(ex).format()))

>> Output

Traceback (most recent call last):
  File "your_file_name_here.py", line 29, in <module>
    1/0
ZeroDivisionError: division by zero

---

It is identical to format_exc() and format_exception():

    a = ''.join(traceback.TracebackException.from_exception(ex).format())
    b = traceback.format_exc()
    c = ''.join(traceback.format_exception(type(ex), ex, ex.__traceback__))
    print(a == b == c)  # This is True !!

Jessabell

Answer 5

>>> import sys
>>> import traceback
>>> try:
...   5 / 0
... except ZeroDivisionError as e:
...   type_, value_, traceback_ = sys.exc_info()
>>> traceback.format_tb(traceback_)
['  File "<stdin>", line 2, in <module>\n']
>>> value_
ZeroDivisionError('integer division or modulo by zero',)
>>> type_
<type 'exceptions.ZeroDivisionError'>
>>>
>>> 5 / 0
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
ZeroDivisionError: integer division or modulo by zero

You use sys.exc_info() to collect the information and the functions in the traceback module to format it. Here are some examples for formatting it.

The whole exception string is at:

>>> ex = traceback.format_exception(type_, value_, traceback_)
>>> ex
['Traceback (most recent call last):\n', '  File "<stdin>", line 2, in <module>\n', 'ZeroDivisionError: integer division or modulo by zero\n']

Blueness

Answer 6

For those using Python-3

Using traceback module and exception.__traceback__ one can extract the stack-trace as follows:

• grab the current stack-trace using traceback.extract_stack()
• remove the last three elements (as those are entries in the stack that got me to my debug function)
• append the __traceback__ from the exception object using traceback.extract_tb()
• format the whole thing using traceback.format_list()

import traceback
def exception_to_string(excp):
   stack = traceback.extract_stack()[:-3] + traceback.extract_tb(excp.__traceback__)  # add limit=?? 
   pretty = traceback.format_list(stack)
   return ''.join(pretty) + '\n  {} {}'.format(excp.__class__,excp)

A simple demonstration:

def foo():
    try:
        something_invalid()
    except Exception as e:
        print(exception_to_string(e))

def bar():
    return foo()

We get the following output when we call bar():

  File "./test.py", line 57, in <module>
    bar()
  File "./test.py", line 55, in bar
    return foo()
  File "./test.py", line 50, in foo
    something_invalid()

  <class 'NameError'> name 'something_invalid' is not defined

Mada

It looks like an unreadable complicated code. In Python 3.5+ there is a more elegant and simple way: https://mcmap.net/q/36591/-get-exception-description-and-stack-trace-which-caused-an-exception-all-as-a-string – Jessabell

In my application this method is better to the one linked to by don_vanchos in another answer. The stack trace produced is longer and more complete, even in Python 3.5+. – Cohabit

Answer 7

You might also consider using the built-in Python module, cgitb, to get some really good, nicely formatted exception information including local variable values, source code context, function parameters etc..

For instance for this code...

import cgitb
cgitb.enable(format='text')

def func2(a, divisor):
    return a / divisor

def func1(a, b):
    c = b - 5
    return func2(a, c)

func1(1, 5)

we get this exception output...

ZeroDivisionError
Python 3.4.2: C:\tools\python\python.exe
Tue Sep 22 15:29:33 2015

A problem occurred in a Python script.  Here is the sequence of
function calls leading up to the error, in the order they occurred.

 c:\TEMP\cgittest2.py in <module>()
    7 def func1(a, b):
    8   c = b - 5
    9   return func2(a, c)
   10
   11 func1(1, 5)
func1 = <function func1>

 c:\TEMP\cgittest2.py in func1(a=1, b=5)
    7 def func1(a, b):
    8   c = b - 5
    9   return func2(a, c)
   10
   11 func1(1, 5)
global func2 = <function func2>
a = 1
c = 0

 c:\TEMP\cgittest2.py in func2(a=1, divisor=0)
    3
    4 def func2(a, divisor):
    5   return a / divisor
    6
    7 def func1(a, b):
a = 1
divisor = 0
ZeroDivisionError: division by zero
    __cause__ = None
    __class__ = <class 'ZeroDivisionError'>
    __context__ = None
    __delattr__ = <method-wrapper '__delattr__' of ZeroDivisionError object>
    __dict__ = {}
    __dir__ = <built-in method __dir__ of ZeroDivisionError object>
    __doc__ = 'Second argument to a division or modulo operation was zero.'
    __eq__ = <method-wrapper '__eq__' of ZeroDivisionError object>
    __format__ = <built-in method __format__ of ZeroDivisionError object>
    __ge__ = <method-wrapper '__ge__' of ZeroDivisionError object>
    __getattribute__ = <method-wrapper '__getattribute__' of ZeroDivisionError object>
    __gt__ = <method-wrapper '__gt__' of ZeroDivisionError object>
    __hash__ = <method-wrapper '__hash__' of ZeroDivisionError object>
    __init__ = <method-wrapper '__init__' of ZeroDivisionError object>
    __le__ = <method-wrapper '__le__' of ZeroDivisionError object>
    __lt__ = <method-wrapper '__lt__' of ZeroDivisionError object>
    __ne__ = <method-wrapper '__ne__' of ZeroDivisionError object>
    __new__ = <built-in method __new__ of type object>
    __reduce__ = <built-in method __reduce__ of ZeroDivisionError object>
    __reduce_ex__ = <built-in method __reduce_ex__ of ZeroDivisionError object>
    __repr__ = <method-wrapper '__repr__' of ZeroDivisionError object>
    __setattr__ = <method-wrapper '__setattr__' of ZeroDivisionError object>
    __setstate__ = <built-in method __setstate__ of ZeroDivisionError object>
    __sizeof__ = <built-in method __sizeof__ of ZeroDivisionError object>
    __str__ = <method-wrapper '__str__' of ZeroDivisionError object>
    __subclasshook__ = <built-in method __subclasshook__ of type object>
    __suppress_context__ = False
    __traceback__ = <traceback object>
    args = ('division by zero',)
    with_traceback = <built-in method with_traceback of ZeroDivisionError object>

The above is a description of an error in a Python program.  Here is
the original traceback:

Traceback (most recent call last):
  File "cgittest2.py", line 11, in <module>
    func1(1, 5)
  File "cgittest2.py", line 9, in func1
    return func2(a, c)
  File "cgittest2.py", line 5, in func2
    return a / divisor
ZeroDivisionError: division by zero

Almond

Kudos for coming up with a module I did not even know exists in the standard library. – Flawy

Answer 8

If you would like to get the same information given when an exception isn't handled you can do something like this. Do import traceback and then:

try:
    ...
except Exception as e:
    print(traceback.print_tb(e.__traceback__))

I'm using Python 3.7.

Patsy

Answer 9

If your goal is to make the exception and stacktrace message look exactly like when python throws an error, the following works in both python 2+3:

import sys, traceback


def format_stacktrace():
    parts = ["Traceback (most recent call last):\n"]
    parts.extend(traceback.format_stack(limit=25)[:-2])
    parts.extend(traceback.format_exception(*sys.exc_info())[1:])
    return "".join(parts)

# EXAMPLE BELOW...

def a():
    b()


def b():
    c()


def c():
    d()


def d():
    assert False, "Noooh don't do it."


print("THIS IS THE FORMATTED STRING")
print("============================\n")

try:
    a()
except:
    stacktrace = format_stacktrace()
    print(stacktrace)

print("THIS IS HOW PYTHON DOES IT")
print("==========================\n")
a()

It works by removing the last format_stacktrace() call from the stack and joining the rest. When run, the example above gives the following output:

THIS IS THE FORMATTED STRING
============================

Traceback (most recent call last):
  File "test.py", line 31, in <module>
    a()
  File "test.py", line 12, in a
    b()
  File "test.py", line 16, in b
    c()
  File "test.py", line 20, in c
    d()
  File "test.py", line 24, in d
    assert False, "Noooh don't do it."
AssertionError: Noooh don't do it.

THIS IS HOW PYTHON DOES IT
==========================

Traceback (most recent call last):
  File "test.py", line 38, in <module>
    a()
  File "test.py", line 12, in a
    b()
  File "test.py", line 16, in b
    c()
  File "test.py", line 20, in c
    d()
  File "test.py", line 24, in d
    assert False, "Noooh don't do it."
AssertionError: Noooh don't do it.

Giggle

Why use format_stack(limit=25)? – Felting

@Semnodime. It limits the number of traces to 25, just like native python does it. – Giggle

Please reference the source of limit at 25. The following code disproves the limit and has been evaluated at least for python 3.8 online-python.com/HZv5EnpRfQ – Felting

Answer 10

my 2-cents:

import sys, traceback
try: 
  ...
except Exception, e:
  T, V, TB = sys.exc_info()
  print ''.join(traceback.format_exception(T,V,TB))

Triazine

Answer 11

If you would like to convert your traceback to a list of dict (for python > 3.5):

from traceback import TracebackException


def list_traceback(exc_value: BaseException):
    result = list()

    # get previous fails, so errors are appended by order of execution
    if exc_value.__context__:
        result += list_traceback(exc_value.__context__)

    # convert Exception into TracebackException
    tbe = TracebackException.from_exception(exc_value)

    # get stacktrace (cascade methods calls)
    error_lines = list()
    for frame_summary in tbe.stack:
        summary_details = {
            'filename': frame_summary.filename,
            'method'  : frame_summary.name,
            'lineno'  : frame_summary.lineno,
            'code'    : frame_summary.line
        }
        error_lines.append(summary_details)

    # append error, by order of execution
    result.append({"error_lines": error_lines,
                   "type"       : tbe.exc_type.__name__,
                   "message"    : str(tbe)})

    return result

This will be (an example of) the result:

[
   {
      "error_lines": [
         {
            "filename": "/home/demo/file2.py",
            "method": "do_error_2",
            "lineno": 18,
            "code": "a=1/0"
         }
      ],
      "type": "ZeroDivisionError",
      "message": "division by zero"
   },
   {
      "error_lines": [
         {
            "filename": "/home/demo/file_main.py",
            "method": "demo2",
            "lineno": 50,
            "code": "file2.DEMO().do_error_2()"
         },
         {
            "filename": "/home/demo/file2.py",
            "method": "do_error_2",
            "lineno": 20,
            "code": "raise AssertionError(\"Raised inside the except, after division by zero\")"
         }
      ],
      "type": "AssertionError",
      "message": "Raised inside the except, after division by zero"
   }
]

Cysticercoid

Answer 12

I defined following helper class:

import traceback
class TracedExeptions(object):
    def __init__(self):
        pass
    def __enter__(self):
        pass

    def __exit__(self, etype, value, tb):
      if value :
        if not hasattr(value, 'traceString'):
          value.traceString = "\n".join(traceback.format_exception(etype, value, tb))
        return False
      return True

Which I can later use like this:

with TracedExeptions():
  #some-code-which-might-throw-any-exception

And later can consume it like this:

def log_err(ex):
  if hasattr(ex, 'traceString'):
    print("ERROR:{}".format(ex.traceString));
  else:
    print("ERROR:{}".format(ex));

(Background: I was frustraded because of using Promises together with Exceptions, which unfortunately passes exceptions raised in one place to a on_rejected handler in another place, and thus it is difficult to get the traceback from original location)

Monson