Handling NameError is not working

[shazarivf]

We recently upgraded to a new version and we have a several scripts that relied on handling NameError as we couldn't ensure all user_symbols to be present always.
The following is a small example snippet doesn't work as expected anymore.

from asteval import Interpreter

aeval = Interpreter(user_symbols={"one": 1, "two": 2, "fallback_var": 0})

code = """
try:
    output = some_var
except NameError:
    output = None
output or fallback_var
"""

a = aeval(code)

This results in:

NameError
   
try:
    output = some_var
except NameError:
    output = None
output or fallback_var
                 ^^^
name 'some_var' is not defined

NameError doesn't get handled and the code just breaks and returns None
It looks like this behavior changed with version 0.9.17 and this commit.

[newville]

@shazarivf Thasnks - I agree that this looks wrong. I'll need to investigate, and I can't look at this in depth this week and can't promise this will be fixed quickly.

Does undoing that check of if len(self.error) == 0 fix this?

[shazarivf]

@newville yes, removing the check fixes it. I can raise a PR if you think the check is safe to remove

[newville]

@shazarivf Yes please -- I'd be interested to see what tests fail!