LINT: pandas/scripts

[jreback]

need to lint these & add a rule in ci/lint.sh to check.

[gfyoung]

Yeah, that directory could certainly benefit from a lot of clean-up.

[bhavybarca]

I would like to work on this as my first issue

[gfyoung]

@bhavybarca : Go for it!

[bhavybarca]

what all do i need to do, i am a beginner

[jreback]

you need to add a add pandas/scripts/ to the ci/lint.sh script (make it a separate step)
then run this export LINT=1; ci/lint.sh and see what turns up, which needs fixing. you can use autopep8 to fix most

[bhavybarca]

so first i need to make ci/lint.sh script and then copy the files to it ?

[jreback]

it’s aleady there just edit

[bhavybarca]

can you please simplify all this a bit am i just starting

[TomAugspurger]

This file: https://github.com/pandas-dev/pandas/blob/master/ci/lint.sh

You'll add another flake8 line like the others, but for scripts/ directory.

@bhavybarca are you on Windows or Mac or Linux? You'll set the LINT environment variable and then run ci/lint.sh, which will print out a list of errors that need to be fixed.

[TomAugspurger]

It'll be roughly like

$ flake8 scripts/
scripts/api_rst_coverage.py:24:1: E302 expected 2 blank lines, found 1
scripts/api_rst_coverage.py:64:80: E501 line too long (90 > 79 characters)
scripts/api_rst_coverage.py:70:80: E501 line too long (93 > 79 characters)
scripts/api_rst_coverage.py:78:80: E501 line too long (88 > 79 characters)
...

[gfyoung]

@TomAugspurger : I prefer to use flake8 scripts/. No need to lint the entire repository for this issue.

[TomAugspurger]

yeah, agreed.

[datapythonista]

I think it makes sense to include as part of this ticket, couple of other things related to the scripts:

• Fix PEP-8 issues, and add check to lint (as originally requested on the ticket)
• Add documentation to scripts that don't have it (like find_undoc_args.py)
• Make all the scripts executable (chmod a+x and #! header)
• Create a section in contributing.rst with the available script and what they are used for
• Make the scripts work regardless of the current directory (announce.py needs to be called from the pandas home, api_rst_coverage.py from the scripts directory...)
• Follow the same convention in script names (rename merge-pr.py to merge_pr.py)

What do you think?

[bhavybarca]

@TomAugspurger i m on linux 16.04, all this makes a lot of sense now thanks ,just last thing how do i setup lint environment to run lint.sh

[gfyoung]

@datapythonista : All of these are excellent ideas! I think linting is a pretty big step as it is though, so might be worthwhile to just open another issue.

[datapythonista]

@bhavybarca I'll take this, if that's all right

[bhavybarca]

@datapythonista yeah sure!!

[gfyoung]

As a follow-up, we should document all of these functions (@datapythonista : no need to this in your open PR, but wanted to add that just so that we don't forget 😄 )