@twoertwein Does ReadCsvBuffer suport next in addition to iter?

As far as I understand: As long as .__iter__() returns an object that has __next__, I don't think the file object itself has to have __next__.

Hm weird, mypy complains about ReadCsvBuffer[str]

I'm sorry, you are right! I think only read_fwf calls next explicitly, or? read_csv does not explicitly call next. The following works with read_csv

from io import StringIO

import pandas as pd


class Test:
    def __init__(self):
        self.buffer = StringIO("a,b,c\n0,1,2")

    @property
    def mode(self):
        return self.buffer.mode

    def fileno(self):
        return self.buffer.fileno()

    def seek(self, __offset, __whence=-1):
        if __whence != -1:
            return self.seek(__offset)
        return self.seek(__offset, __whence=__whence)

    def seekable(self):
        return self.buffer.seekable()

    def tell(self):
        return self.buffer.tell()

    def read(self, __n=None):
        self.buffer.read(__n)

    def __iter__(self):
        return self.buffer.__iter__()

    def readline(self):
        return self.buffer.readline()

    @property
    def closed(self):
        return self.buffer.closed()


print(pd.read_csv(Test(), engine="python"))

Yep I think this is only called from read_fwf.

Can we add next to ReadCsvBuffer without breaking anything (from a correctness standpoint)

Would be fine with adding it or (probably better) creating ReadFwfBuffer which inherits from ReadCsvBuffer (not sure whether it needs everything that is in ReadCsvBuffer).

Also happy to have the ignores for now and address this in a later PR.

Lets go with the ignore for now then until we have typed more. Can probably make a better decision then