DOC: ExcelWriter constructor types

[alexhtn]

Location of the documentation

https://pandas.pydata.org/docs/dev/reference/api/pandas.ExcelWriter.html?highlight=excelwriter
or

pandas/pandas/io/excel/_base.py

Line 544 in d9fff27

path : str

Documentation problem

Parameter path has type str, but based on this feature https://github.com/pandas-dev/pandas/issues/7074 I can pass io.BytesIO.

Suggested fix for documentation

Change type description to str | io.BytesIO.

[rhshadrach]

The links in the issue are broken, you need to use

[description](url)

[MJafarMashhadi]

It's better to use bytes-like object instead of io.BytesIO specifically.

[alexhtn]

The links in the issue are broken, you need to use

I fixed links

[alexhtn]

It's better to use bytes-like object instead of io.BytesIO specifically.

I didn't understand how can I use bytes-like object as output for ExcelWriter.

[MJafarMashhadi]

I mean it doesn't strictly require io.BytesIO instance, it can be a subclass or anything else that behaves like a bytes buffer (a zip file, a network socket, etc). So I'd suggest to use a more generic term in your patch.

with zipfile.ZipFile('samplezip.zip', 'w') as zf:
    with zf.open('XLfile.xlsx', 'w') as inf:
        with ExcelWriter(inf) as ew:
            df.to_excel(ew)

[alexhtn]

Now I got your idea.
I think correct type is io.BufferedIOBase. ZipFile and io.Bytes inherits from it, and function open returns io.BufferedWriter.
Also I check socket.makefile, for binary format in writing mode it returns io.BufferedWriter.
io.BufferedWriter inherits from io.BufferedIOBase.
But I'm not sure that it covers all cases. Maybe correct type would be io.IOBase.

[MJafarMashhadi]

I believe bytes like object means the same thing. It's more about having the expected interface (duck typing), rather than being a subclass of a certain class (which is the case in Java).

[alexhtn]

I understand this. But now possibility of using bytes-likes objects isn't described in docs at all and also pycharm shows warnings.
I think it would be enough add io.BufferedIOBase to path type and some examples to show how it works.

[MJafarMashhadi]

Sounds good, Would you like to submit a PR?

[alexhtn]

take

[alexhtn]

After I made changes to docstring I ran python scripts/validate_docstrings.py pandas.ExcelWriter and it showed errors for new examples and also for old examples: NameError: name 'ExcelWriter' is not defined and NameError: name 'df' is not defined.
Should I ignore these errors?

[MJafarMashhadi]

The current version has that warning anyway, so I guess you can ignore that. Worst case, you'll get a comment in code review on how to fix it.