How does one use doctest with a pandas dataframe?

Question

I have a simple function (log_return) in a file called test.py (see below) that I'm trying to implement with a doctest.

import pandas as pd

def log_return(df):
    '''Return the log return based on closing prices
    
    >>> df = pd.DataFrame({'Close': [100, 101, 102, 99]}, index = pd.date_range('2022-01-01', periods=4, freq='D'))
    >>> log_return(df)

            Close         r
2022-01-01    100       NaN
2022-01-02    101  0.009950
2022-01-03    102  0.009852
2022-01-04     99 -0.029853
    '''
    df['r'] = np.log(df['Close']).diff()

However I'm getting the below error related to whitespace when I try to execute the doctest from the command line (e.g. $ python test.py). How can I fix this error?

ValueError: line 5 of the docstring for __main__.log_return has inconsistent leading whitespace: '2022-01-01    100       NaN'

should add `return df` after indenting so the test passes. :) — tlgs, Jan 17 '22 at 04:12
@tlgs Python convention is that [mutators don't return the object they mutate](/a/18829224/4518341), so instead I'd probably recommend `return np.log...` and assign it to a df column outside the function. — wjandrea, Jan 17 '22 at 04:15

score 1 · Accepted Answer · answered Jan 17 '22 at 04:10

You need to indent it like this:

    '''
    ...

    >>> log_return(df)
                Close         r
    2022-01-01    100       NaN
    2022-01-02    101  0.009950
    2022-01-03    102  0.009852
    2022-01-04     99 -0.029853
    '''

Keep in mind doctests are meant to look like interactive snippets, so in this case, that means things should be aligned like they are in an interactive session (prompt and output).

Now, once you fix this, the test will fail, but that's a separate issue.

How does one use doctest with a pandas dataframe?

1 Answers1