Dataframe columns with lists in them do not work

[schwallie]

Got an error on what I think is a column with a list in it:

  File "get_paypal.py", line 789, in main
    donor, wkb='OTAT Donor Database', wks='DonorLevel')
  File "/Users/---/Desktop/repos/otat/config.py", line 246, in upload_to_google
    set_with_dataframe(worksheet, df, include_index=True)
  File "/usr/local/lib/python2.7/site-packages/gspread_dataframe.py", line 185, in set_with_dataframe
    _cellrepr(cell_value, allow_formulas))
  File "/usr/local/lib/python2.7/site-packages/gspread_dataframe.py", line 51, in _cellrepr
    if pd.isnull(value):
ValueError: The truth value of an array with more than one element is ambiguous. Use a.any() or a.all()```

[robin900]

Ok, I'll explore how to make the null check not raise an exception regardless of the value type. But I have a question: when the array value was written to a cell in the google spreadsheet, what were you expecting it to look like? A comma-delimited string? A space-delimited string? A Python array literal?

[schwallie]

It looks like a python array literal, which works for me

…

On Tue, Jul 25, 2017, 9:03 PM Robin Thomas ***@***.***> wrote: Ok, I'll explore how to make the null check not raise an exception regardless of the value type. But I have a question: when the array value was written to a cell in the google spreadsheet, what were you expecting it to look like? A comma-delimited string? A space-delimited string? A Python array literal? — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#6 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AD-vCsqHe1htcFWHz6GRc2QTWr_mpOaRks5sRp6BgaJpZM4OjSJ5> .

[robin900]

OK, my next question is: if you write a Python list literal to a Google spreadsheet cell, are you expecting to 1) read it back into a dataframe; and 2) if so, are you expecting it to be parsed back into a Python list object?

I can accept 1) being "yes", as long as 2) is "no", because I have to defer to pandas.read_csv's behavior, and it's always going to read a cell inputValue of [1,2,3] as the string "[1, 2, 3]".

[schwallie]

For me whatever behavior you think is best is fine by me. The list is not actually read back, and it's only generated for visual representation right before I put it on the Google sheet, I don't actually use lists in pandas df columns often

…

On Tue, Jul 25, 2017, 9:56 PM Robin Thomas ***@***.***> wrote: OK, my next question is: if you write a Python list literal to a Google spreadsheet cell, are you expecting to 1) read it back into a dataframe; and 2) if so, are you expecting it to be parsed back into a Python list object? I can accept 1) being "yes", as long as 2) is "no", because I have to defer to pandas.read_csv's behavior, and it's always going to read a cell inputValue of [1,2,3] as the string "[1, 2, 3]". — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub <#6 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AD-vCtlpCX5TdpbBDUIuz39BqrWCjlgzks5sRqrWgaJpZM4OjSJ5> .

[robin900]

Yes, I actually never use them. While I contemplate over the next 12 hours how best to code a fix for this, your best workaround is to do a str(list_value) and use the result as the value to assign to that cell in the dataframe.

[robin900]

@schwallie Would you like to checkout the branch referenced by the pull request (rbt-accept-list-value-for-cell) and test it with your own code? You can also run python setup.py test to run the included test suite. With your approval, I'll proceed with a new PyPI release.

[schwallie]

Awesome, thanks. lgtm 💯

[robin900]

@schwallie Release 2.1.0 is now on PyPI. (Because of the change in the code of how floats are represented in cell values submitted to Google sheet API, I bumped the minor version to 2.1 rather than make a bugfix version of 2.0.2.)