[ Python 常見問題 ] How to properly assert that an exception gets raised in pytest?

Source From Here 
Question 
Check testing code below: 

view plaincopy to clipboardprint?

1. # coding=utf-8  
2. import pytest  
3.   
4.   
5. def whatever():  
6.     return 9/0  
7.   
8. def test_whatever():  
9.     try:  
10.         whatever()  
11.     except ZeroDivisionError as exc:  
12.         pytest.fail(exc, pytrace=True)  

The output will be: 

view plaincopy to clipboardprint?

1. ================================ test session starts =================================  
2. platform linux2 -- Python 2.7.3 -- py-1.4.20 -- pytest-2.5.2  
3. plugins: django, cov  
4. collected 1 items   
5.   
6. pytest_test.py F  
7.   
8. ====================================== FAILURES ======================================  
9. ___________________________________ test_whatever ____________________________________  
10.   
11.     def test_whatever():  
12.         try:  
13.             whatever()  
14.         except ZeroDivisionError as exc:  
15. >           pytest.fail(exc, pytrace=True)  
16. E           Failed: integer division or modulo by zero  
17.   
18. pytest_test.py:12: Failed  
19. ============================== 1 failed in 1.16 seconds ==============================  

How to make pytest print traceback, so I would see where in the whatever function an exception was raised? 

How-To 
There are two ways to handle these kind of cases in pytest by catching expected exception (Assertions about expected exceptions): 

* Using pytest.raises function
* Using pytest.mark.xfail decorator

Usage of pytest.raises: 

view plaincopy to clipboardprint?

1. def whatever():  
2.     return 9/0  
3. def test_whatever():  
4.     with pytest.raises(ZeroDivisionError):  
5.         whatever()  

Output of pytest.raises: 

view plaincopy to clipboardprint?

1. ============================= test session starts ============================  
2. platform linux2 -- Python 2.7.10, pytest-3.2.3, py-1.4.34, pluggy-0.4.0 --   
3. /usr/local/python_2.7_10/bin/python  
4. cachedir: .cache  
5. rootdir: /home/ukpdl, inifile:  
6. collected 1 item  
7.   
8. test_fun.py::test_whatever PASSED  
9.   
10.   
11. ======================== 1 passed in 0.01 seconds =============================  

Usage of pytest.mark.xfail: 

view plaincopy to clipboardprint?

1. @pytest.mark.xfail(raises=ZeroDivisionError)  
2. def test_whatever():  
3.     whatever()  

Output of pytest.xfail marker: 

view plaincopy to clipboardprint?

1. ============================= test session starts ============================  
2. platform linux2 -- Python 2.7.10, pytest-3.2.3, py-1.4.34, pluggy-0.4.0 --   
3. /usr/local/python_2.7_10/bin/python  
4. cachedir: .cache  
5. rootdir: /home/ukpdl, inifile:  
6. collected 1 item  
7.   
8. test_fun.py::test_whatever xfail  
9.   
10. ======================== 1 xfailed in 0.03 seconds=============================  

As the document says: 

Using pytest.raises is likely to be better for cases where you are testing exceptions your own code is deliberately raising, whereas using @pytest.mark.xfail with a check function is probably better for something like documenting unfixed bugs(where the test describes what “should” happen) or bugs in dependencies.

Regarding to print the trackback from pytest, pytest.raises(Exception) as e_info is what you need. Check sample code as below: 
- /tmp/test.py 

view plaincopy to clipboardprint?

1. import pytest  
2. import logging  
3.   
4. def test_passes():  
5.     with pytest.raises(Exception) as e_info:  
6.         x = 1 / 0  
7.     logging.error('Expected exception: {}'.format(e_info))  
8.   
9. def test_passes_without_info():  
10.     with pytest.raises(Exception):  
11.         x = 1 / 0  
12.   
13. def test_fails():  
14.     with pytest.raises(Exception) as e_info:  
15.         x = 1 / 1  
16.   
17. def test_fails_without_info():  
18.     with pytest.raises(Exception):  
19.         x = 1 / 1  
20.   
21. # Don't do this. Assertions are caught as exceptions.  
22. def test_passes_but_should_not():  
23.     try:  
24.         x = 1 / 1  
25.         assert False  
26.     except Exception:  
27.         assert True  
28.   
29. # Even if the appropriate exception is caught, it is bad style,  
30. # because the test result is less informative  
31. # than it would be with pytest.raises(e)  
32. # (it just says pass or fail.)  
33.   
34. def test_passes_but_bad_style():  
35.     try:  
36.         x = 1 / 0  
37.         assert False  
38.     except ZeroDivisionError:  
39.         assert True  
40.   
41. def test_fails_but_bad_style():  
42.     try:  
43.         x = 1 / 1  
44.         assert False  
45.     except ZeroDivisionError:  
46.         assert True  

Output (By executing pytest -s /tmp/test.py): 

view plaincopy to clipboardprint?

1. ...  
2. ../../tmp/test.py ERROR:root:Expected exception: /tmp/test.py:6: ZeroDivisionError: division by zero  
3. ...  
4. === test session starts ===  
5. platform linux2 -- Python 2.7.6 -- py-1.4.26 -- pytest-2.6.4  
6. collected 7 items   
7.   
8. test.py ..FF..F  
9.   
10. === FAILURES ===  
11. ___ test_fails ____  
12.   
13.     def test_fails():  
14.         with pytest.raises(Exception) as e_info:  
15. >           x = 1 / 1  
16. E           Failed: DID NOT RAISE  
17.   
18. test.py:13: Failed  
19. ___ test_fails_without_info ____  
20.   
21.     def test_fails_without_info():  
22.         with pytest.raises(Exception):  
23. >           x = 1 / 1  
24. E           Failed: DID NOT RAISE  
25.   
26. test.py:17: Failed  
27. ___ test_fails_but_bad_style ___  
28.   
29.     def test_fails_but_bad_style():  
30.         try:  
31.             x = 1 / 1  
32. >           assert False  
33. E           assert False  
34.   
35. test.py:43: AssertionError  
36. === 3 failed, 4 passed in 0.02 seconds ===  

Note that e_info saves the exception object so you can extract details from it. For example, if you want to check the exception call stack or another nested exception inside.