read_excel and ExcelFile engine conflict

[WillAyd]

This is currently valid code:

>>> xl = pd.ExcelFile('foo.xlsx', engine='xlrd')
>>> pd.read_excel(xl, engine='openpyxl')
Empty DataFrame
Columns: []
Index: []

Though ambiguous as to which engine is actually being used. I think we should be raising if an engine is specified in read_excel when the target is a pd.ExcelFile, as the latter is a lower level construct and will already have an engine in place.

@simonjayhawkins

[WillAyd]

To clarify I suppose this isn’t ambiguous now since xlrd is our only reader but will be once that doesn’t hold true. May be a similar issue on the write side

[alimcmaster1]

Makes sense to me - happy to take a look

[sumanau7]

Isn't this a documentation issue, because only when input to read_excel is not an instance of ExcelFile, then only engine argument given to read_excel will be used, in all other cases, engine argument has no impact whatsoever.
https://github.com/pandas-dev/pandas/blob/master/pandas/io/excel/_base.py#L290
@WillAyd

[alimcmaster1]

I think Will is suggesting it should throw some kind of error - otherwise it’s ambiguous ( without reading the docs ) what the engine is . But a doc update would be welcome too - I can do that as part of the PR

…

Sent from my iPhone

On 6 Jun 2019, at 11:35, Sumanau Sareen ***@***.***> wrote: Isn't this a documentation issue, because only when input to read_excel is not an instance of ExcelFile, then only engine argument given to read_excel will be used, in all other cases, engine argument has no impact whatsoever. @WillAyd — You are receiving this because you commented. Reply to this email directly, view it on GitHub, or mute the thread.