to_excel doesn't support dateutil.relativedelta

[s-celles]

Hello,

to_excel doesn't support dateutil.relativedelta

import pandas as pd
from dateutil.relativedelta import relativedelta
df = pd.DataFrame([relativedelta(years=2)])
df.to_excel('out.xls')

raises Exception: Unexpected data type <type 'instance'>

df.to_excel('out.xlsx')

raises AttributeError: relativedelta instance has no attribute '__float__'

Not sure if it's a Pandas issue, an XLS writer issue or a dateutil issue.

[jorisvandenbossche]

What do you expect as the result in excel, as I think excel does not have such a datatype? A string representation?

[s-celles]

to_csv returns a string so it could be a string also

[jorisvandenbossche]

Related to #8188 (same question, but for lists). There is a PR for that (#9181, although that will not solve this case)

[s-celles]

Maybe by default to_excel should raise an error when type is not "known" but an additional flag could be add to save to string (using __str__ or __repr__ ?).

[WillAyd]

I can't reproduce this. issue on master; if you are still seeing feel free to ping and can reopen

[s-celles]

Thanks. I haven't try this feature recently.

Here is Excel content (both out.xls and out.xlsx):

	0
0	relativedelta(years=+2)

It's better than raising an exception.