Handling exceptions assert to NumPy testing in Python

Question

Understanding someone code and met this:

There is a function:

 def some_func():
       print("Hello")
       assert(len(y)==len(p))
       print("First assert")
       assert(p.all()<=0.99999)
       print("Second assert")
       return 1

Next, call assert_raises:

np.testing.assert_raises(AssertionError, some_func, np.asarray([1, 2, 3]), np.asarray([1, 2, 3, 4, 5]))

In the output, we simply get Hello without exception messages:

Hello

Next, call the function assert_array_less:

np.testing.assert_array_less(some_func(np.asarray([1, 2, 3]), np.asarray([1, 2, 3])), np.inf)

In the output, we get Hello First assert and then an error message and an AssertionError exception:

Hello
First assert
---------------------------------------------------------------------------
AssertionError  Traceback (most recent call last)
<ipython-input-26-df1a32b4f5a0> in <module>()
      9 np.testing.assert_raises(AssertionError, some_func, np.asarray([1, 2, 3]), np.asarray([1, 2, 3, 4, 5]))
     10 
---> 11 np.testing.assert_array_less(some_func(np.asarray([1, 2, 3]), np.asarray([1, 2, 3])), np.inf)

<ipython-input-26-df1a32b4f5a0> in some_func(a, b)
      3     assert(len(a)==len(b))
      4     print("First assert")
----> 5     assert(a.all()<=0.99999)
      6     print("Second assert")
      7     return 1

AssertionError:

Question:

Why in 1 case the code just stops and no exceptions are thrown, although it is called first assert in some_func ()?

And why in the second does not happen the same as in the first, and an exception is thrown?

tel · Accepted Answer · 2018-12-08 08:23:33Z

Based on the error message you show, I'm going to guess that the actual definition of your some_func function is:

def some_func(a, b):
    print("Hello")
    assert(len(a)==len(b))
    print("First assert")
    assert(a.all()<=0.99999)
    print("Second assert")
    return 1

Given that, here's the exact rundown on what happened during your assert calls:

You call
```
np.testing.assert_raises(AssertionError, some_func, np.asarray([1, 2, 3]), np.asarray([1, 2, 3, 4, 5]))
```
- The np.testing.assert_raises function in turn calls
```
some_func(np.asarray([1, 2, 3]), np.asarray([1, 2, 3, 4, 5]))
```
  - The first line of some_func runs, and prints Hello.
  - Next, some_func tries to assert that the lengths of a and b are the same. However, the length of a is 3 and the length of b is 5, so the assertion fails. This causes an AssertionError to be thrown. At this point, execution of some_func is terminated, and control returns to assert_raises.
- assert_raises was told to expect an AssertionError via the first argument passed to it. It sees that an AssertionError has indeed been thrown, so from its perspective all is well. It handles the AssertionError (preventing it from creating an error message that would be shown to you, the user), and execution of assert_raises ends normally.
Next you call
```
np.testing.assert_array_less(some_func(np.asarray([1, 2, 3]), np.asarray([1, 2, 3])), np.inf)
```
- Once again, the first line of some_func runs and prints Hello.
- This time, len(a)==len(b)==3, so the assert passes and execution of some_func continues normally.
- Next the third line of some_func runs and prints First assert.
- Every value in a is nonzero, so a.all() is True. The boolean value True has a numeric value of 1. Thus, a.all()<=0.99999 is False and so the assert fails. At this point, an AssertionError is raised and the execution of some_func is terminated.
This time, some_func was running in the top-level scope (if it had successfully completed, np.testing.assert_array_less would have been called and the return value of some_func would have been passed to it). Unlike the last time some_func was called there is no enclosing function that will handle the AssertionError that was raised. This means that the AssertionError will propagate and produce the visible error message that gets printed out beneath First assert.

And that's the whole story.

Collectives™ on Stack Overflow

Handling exceptions assert to NumPy testing in Python

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related