Pandas stub fixes

[gramster]

Add a few missing files.
Update .gitignore.
Make some fixes based on the tests from the pandas-stubs package (more to come).

[bschnurr]

DataFrameGroupBy removed on purpose?

[yanyongyu]

pandas.io.parsers.TextFileReader now support context manager with __enter__ and __exit__ function but not include in stub file

python-type-stubs/pandas/io/parsers.pyi

Lines 553 to 564 in e7def7e

	class TextFileReader(abc.Iterator):
	f = ...
	orig_options = ...
	engine = ...
	chunksize = ...
	nrows = ...
	squeeze = ...
	def __init__(self, f, engine = ..., **kwds) -> None: ...
	def close(self) -> None: ...
	def __next__(self): ...
	def read(self, nrows = ...): ...
	def get_chunk(self, size = ...): ...

[Dr-Irv]

@gramster this change broke stuff in read_csv that I fixed in a previous PR. See the test here:
#109 (comment)

read_csv() was OK in pylance 2022.1.3, but now broken in pylance 2022.1.5

[gramster]

I don't think it's this change; there was a reversion to IO that happened in the copy in pylance that has since been fixed but must have made it out in the release. This change already includes that reversion.

If you edit your ~/.vscode/extensions/ms-python.vscode-pylance-2022.1.5/dist/bundled/stubs/pandas/io/parsers.pyi file and change the two lines that say:

    reader: IO,

to

    filepath: FilePathOrBuffer,

I think you should be good.

[Dr-Irv]

@gramster even if I do the change above, then the following test fails:

from io import StringIO
import pandas as pd
from pandas.io.parsers import TextFileReader

dio = StringIO("a,b\n 1,2\n 3,4\n")

df: pd.DataFrame = pd.read_csv(dio)

print(df)

dio.seek(0)
tr: pd.DataFrame = pd.read_csv(dio)

dio.seek(0)
tr2: TextFileReader = pd.read_csv(dio, iterator=True)

dio.seek(0)
tr3: pd.DataFrame = pd.read_csv(dio, iterator=False)

dio.seek(0)
tr5: TextFileReader = pd.read_csv(dio, chunksize=10)

dio.seek(0)
tr6: pd.DataFrame = pd.read_csv(dio, chunksize=None)

dio.seek(0)
tr7: pd.DataFrame = pd.read_csv(dio, header="infer")

dio.seek(0)
tr8: TextFileReader = pd.read_csv(dio, header="infer", iterator=True)


ftr1: TextFileReader = pd.read_csv("hey.csv")
ftr2: pd.DataFrame = pd.read_csv("hey.csv", iterator=True)
ftr3: TextFileReader = pd.read_csv("hey.csv", iterator=False)
ftr5: pd.DataFrame = pd.read_csv("hey.csv", chunksize=10)
ftr6: TextFileReader = pd.read_csv("he.csv", chunksize=None)
ftr7: TextFileReader = pd.read_csv("hey.csv", header="infer")
ftr8: pd.DataFrame = pd.read_csv("hey.csv", header="infer", iterator=True)

All the assignments to variables starting with tr should be legal, while the assignments to the ones using ftr should fail.

The deletion of the * argument at the beginning is what I believe to be the cause of those failures.