I've tried to interpret this code change several times, and it seems to be the logic of the pre-fix state and post-fix state is functionally identical.

1. If this term is not any of the characters in this string, vs
2. None of the characters in this string are my term.

Clearly it has fixed the issue, the tests are passing, but I'm not clear on how the change works.

The issue is that the terms are not always strings. The __eq__ method on CompValue successfully returns False in this case, but the __contains__ method on str apparently raises a type error, although I can’t find any documentation of this behaviour. Compare with the similar test on line 72.

Edit: The test we're running afoul of is here.

Edit 2: This is likely because __contains__ on str is defined in terms of substrings, rather then the usual container membership. The functionality we want here is, however, container membership. A simpler fix might have been to make the string into a simpler container:

if i + 1 < l_ and terms[i + 1] not in list(".,;"):

or simply avoid representing lists as strings in the first place:

if i + 1 < l_ and terms[i + 1] not in [".", ",", ";"]:

Edit 3: I changed to the last version with a brief comment. (Sadly it is four lines longer after black formatting.)