[FIXED] Doctest: How can I identify differences between expected and actual output?

Issue

In the below doctest, the expected and actual output look identical. I'm pretty sure there is some white space error since when I enable NORMALIZE_WHITESPACE, the test is successful. Is there another flag that can help a user triangulate on the location of such a whitespace error?

"""
>>> df  # doctest:-NORMALIZE_WHITESPACE
              Open
Date
2021-05-17  126.82
"""

import pandas as pd

df = pd.DataFrame({'Date': ['2021-05-17'], 'Open': [126.82]}).set_index('Date')

if __name__ == "__main__":
    import doctest
    doctest.testmod()

Failed example:
    df  # doctest:-NORMALIZE_WHITESPACE
Expected:
                  Open
    Date
    2021-05-17  126.82
Got:
                  Open
    Date
    2021-05-17  126.82

Solution

Try using the flag REPORT_NDIFF. The docs say, "This is the only method that marks differences within lines as well as across lines", which will help identify any whitespace problems, for example:

"""
>>> print('Hello goodbye')
Hello goodbye\r
"""

Failed example:
    print('Hello goodbye')
Differences (ndiff with -expected +actual):
    - Hello goodbye
    ?              -
    + Hello goodbye

Note that the carriage return doesn't actually appear here.

---

In this particular case, the problem is that dataframes with an index name have trailing whitespace, so it's failing on Date vs Date [...]. The whitespace in the docstring was probably stripped by your editor (which is good, and NORMALIZE_WHITESPACE is the right tool to account for that), but I'm not sure why it was stripped from the "Got" output too — maybe your terminal did that (which is not good) or maybe you pasted it into your editor then copied that to Stack Overflow.

Answered By - wjandrea