InvalidNameError can't be copied or pickled, which crashes normal parsing:

import copy
from bibtexparser.middlewares.names import InvalidNameError
copy.deepcopy(InvalidNameError(name="Doe, John", reason="Too many commas"))
# TypeError: InvalidNameError.__init__() missing 1 required positional argument

Its __init__ takes name and reason but the base ValueError only keeps the formatted message in args, so deepcopy/pickle rebuild it with a single argument. This surfaces through the public API: a failed name is stored in a MiddlewareErrorBlock, and a later middleware like SortBlocksByTypeAndKeyMiddleware deepcopies the whole library, so any bad author name plus that middleware raises TypeError out of parse_string. I store name/reason and add __reduce__ so it round-trips. Added a test.