>>> dateutil.parser.parse('13NOV2017', fuzzy=True)
datetime.datetime(2017, 4, 13, 0, 0)  # --- today's date

>>> dateutil.parser.parse('13NOV2017', fuzzy=True, dayfirst=True)
datetime.datetime(2017, 4, 13, 0, 0)  # --- today's date

Without fuzzy, it raises an error.

These should parse to the proper dates. I am not sure how the tokenization is done, but a possible solution could be to use an alphabet character (in addition to what I think is used - any non-alphanum character)