ERR: Improve error message and doc for invalid labels in cut/qcut

[ryankarlos]

• closes ERR: cut/qcut need better error message when passing invalid input #13318
• tests added / passed
• passes black pandas
• passes git diff upstream/master -u -- "*.py" | flake8 --diff
• add whats new note

[pep8speaks]

Hello @ryankarlos! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2020-01-07 17:48:37 UTC

[alimcmaster1]

Looks like you already made most of these changes in this PR: https://github.com/pandas-dev/pandas/pull/30615/files ?

[jreback]

looks like this pickup up your other changes.

[ryankarlos]

sorry my mistake - just rebased with master now

[alimcmaster1]

hmm mypy seems to be giving:

Performing static analysis using mypy
pandas/core/generic.py:464: error: Non-overlapping identity check (left operand type: "str", right operand type: "Type[int]")
Found 1 error in 1 file (checked 842 source files)
Performing static analysis using mypy DONE
##[error]Process completed with exit code 1.

Need to wait for this to be merged: #30698

[ryankarlos]

Looks like im getting an unrelated compiler error in one of the builds - can see its throwing errors in other PRs as well

gcc -pthread -shared -B /home/vsts/miniconda3/envs/pandas-dev/compiler_compat -L/home/vsts/miniconda3/envs/pandas-dev/lib -Wl,-rpath=/home/vsts/miniconda3/envs/pandas-dev/lib -Wl,--no-as-needed -Wl,--sysroot=/ build/temp.linux-x86_64-3.7/pandas/io/sas/sas.o -o /home/vsts/work/1/s/pandas/io/sas/_sas.cpython-37m-x86_64-linux-gnu.so
error: command 'gcc' failed with exit status 1
##[error]Bash exited with code '1'.
Finishing: Setup environment and build pandas

[alimcmaster1]

Yeah the numpy dev build will fail if any warnings are raised.
Due to this flag https://github.com/pandas-dev/pandas/blob/master/ci/azure/posix.yml#L59

Hopefully someone will have a look soon..

[jreback]

ok, can you add an else that handles the non-list-like case (e.g. 'foo')

[jschendel]

Maybe a little bit cleaner to validate labels immediately instead of embedding the logic here. Right after the duplicates validation at the beginning of _bins_to_cuts you could add something like:

if labels is not None and labels is not False and not is_list_like(labels):
    raise ValueError(...)

Or alternatively could factor out the not:

if not (labels is None or labels is False or is_list_like(labels)):
    raise ValueError(...)

This would allow the rest of the logic in the _bins_to_cut to proceed under the assumption that labels is valid. The advantage being then that this block of code can remain unchanged as the else block here must be list-like labels since the other two options have already been eliminated by the preceding if statements.

[ryankarlos]

Thanks for the feedback. Ive now added a block for eliminating not None and not is_list_like (which allows me to get rid of the is True check and embed that in the previous check) but after doing a check for is not False as there is also an outer else block for cases where labels is False, so was a bit tricky doing them all in one go but hopefully it looks a lot cleaner now.

[jschendel]

Thanks. For checking all three in one go I meant moving the check further up in _bins_to_cuts to around line 369.

[jreback]

thanks @ryankarlos