BUG: Pandas 1.1.0 apply function with axis=1 seems to mishandle the rows

[arshad171]

The functionality of the apply(func, axis=1) in the newly released pandas==1.1.0 is not working as expected.

Pandas seems to be overwriting all the rows in the data frame with the 1st row present. This is happening specifically when introducing a new column in the data frame when running func method on each of the rows.

This working in pandas==1.0.5, but seems to be a bug in pandas=1.10.

I am attaching a sample script and the logs captured for pandas==1.0.5 and pandas==1.10.

Attachments:

1. sample script to reproduce the issue (rename to .py before running) --> script.txt

2. output 1 (pandas==1.0.5) - working as expected --> out_pandas_1.0.5.log

3. output 2 (pandas==1.1.0) - buggy --> out_pandas_1.1.0.log

As you can see in the out_pandas_1.1.0.log log, after preprocessing the data frame using df = df.apply(process_text, axis=1) all the rows in the data frame have been overwritten with the 1st row.

This was not the case with pandas==1.0.5, check the out_pandas_1.0.5.log log.

Environment

• OS: Ubuntu 20.04
• Python: 3.7.7 (anaconda env)

[simonjayhawkins]

Thanks @arshad171 for the report. This appears to be a duplicate of #35462 so closing.