BUG: fix matplotlib warning on CN color #36981

ivanovmg · 2020-10-08T12:42:03Z

closes BUG: warning when using colors 'CN' #36972
tests added / passed
passes black pandas
passes git diff upstream/master -u -- "*.py" | flake8 --diff
whatsnew entry

Refer to commit aaa69d1, where the way of treating CN colors was originally developed. This is not relevant anymore as more readable alternative is provided.

ivanovmg · 2020-10-08T13:48:55Z

CI checks failure due to pyx files, not because of my change.

jreback · 2020-10-08T21:19:30Z

pandas/plotting/_matplotlib/style.py

- # check whether the string can be convertible to single color
- maybe_single_color = _maybe_valid_colors([colors])
- # check whether each character can be convertible to colors
- maybe_color_cycle = _maybe_valid_colors(list(colors))


are these not relevant / texted? (e.g. the axes.prop_cycle) or is this handled by conv.to_rbga now?

I guess that this was the way of handling compatibility with matplotlib <= 2.0.
If the color 'CN' is passed, then it would pick N-th color from "axes.prop_cycle".
Before matplotlib 2.0 there were no CN colors, so this is probably why this construction is here.

jreback · 2020-10-08T21:19:48Z

cc @charlesdong1991

mwaskom · 2020-10-08T22:49:54Z

pandas/plotting/_matplotlib/style.py

+
+def _is_cn_color(color: str) -> bool:
+ """Check if color string is CN color, like 'C0', 'C1', etc."""
+ cn_colors = ["C" + str(x) for x in range(10)]


The default matplotlib color cycle has 10 colors, but if set differently this test won't be valid.

mwaskom · 2020-10-08T22:50:37Z

pandas/plotting/_matplotlib/style.py

+ False otherwise.
+ """
+ conv = matplotlib.colors.ColorConverter()
+ if _is_cn_color(color):


to_rgba handles the "Cn" style, so why is this separate check needed?

Correct! It turns out that we can make it even cleaner.

conv.to_rgba handles CN colors itself.

charlesdong1991

interesting! so now matplotlib handles all color validations? one minor coment

charlesdong1991 · 2020-10-09T08:28:07Z

pandas/plotting/_matplotlib/style.py

- hex_color = [c["color"] for c in list(plt.rcParams["axes.prop_cycle"])]
- colors = [hex_color[int(colors[1])]]
- elif maybe_single_color:
+ if _is_single_color(colors):


maybe put it in one line?

`if isinstance(colors, str) and _is_single_color(colors)`

also nice to put an issue number below this line for future reference!

ivanovmg · 2020-10-09T08:40:21Z

Just confirming that the concerned warning is not raised anymore.

(base) root@f31dd4516194:/workspaces/pandas# pytest pandas/tests/plotting/test_frame.py ======================================================================================================================== test session starts ========================================================================================================================= platform linux -- Python 3.7.7, pytest-6.0.1, py-1.9.0, pluggy-0.13.1 rootdir: /workspaces/pandas, configfile: setup.cfg plugins: cov-2.10.0, forked-1.2.0, xdist-1.34.0, hypothesis-5.23.11, asyncio-0.12.0 collected 186 items pandas/tests/plotting/test_frame.py ...........................x.................................................................................................x............................................................ [100%] ========================================================================================================================== warnings summary ========================================================================================================================== pandas/tests/plotting/test_frame.py::TestDataFramePlots::test_errorbar_plot pandas/tests/plotting/test_frame.py::TestDataFramePlots::test_errorbar_plot pandas/tests/plotting/test_frame.py::TestDataFramePlots::test_errorbar_plot pandas/tests/plotting/test_frame.py::TestDataFramePlots::test_errorbar_timeseries pandas/tests/plotting/test_frame.py::TestDataFramePlots::test_errorbar_timeseries pandas/tests/plotting/test_frame.py::TestDataFramePlots::test_errorbar_timeseries /workspaces/pandas/pandas/plotting/_matplotlib/__init__.py:61: UserWarning: To output multiple subplots, the figure containing the passed axes is being cleared plot_obj.generate() -- Docs: https://docs.pytest.org/en/stable/warnings.html ======================================================================================================= 184 passed, 2 xfailed, 6 warnings in 62.41s (0:01:02) ======================================================================================================== (base) root@f31dd4516194:/workspaces/pandas#

The other warnings are handled here #36982

ivanovmg · 2020-10-09T10:07:49Z

For some reason the test fails in Travis CI.

=================================== FAILURES =================================== _________________ TestDataFramePlots.test_mpl2_color_cycle_str _________________ [gw0] linux -- Python 3.7.8 /home/travis/miniconda3/envs/pandas-dev/bin/python self = <pandas.tests.plotting.test_frame.TestDataFramePlots object at 0x7f022d75ca90> def test_mpl2_color_cycle_str(self): # GH 15516 df = DataFrame(randn(10, 3), columns=["a", "b", "c"]) colors = ["C0", "C1", "C2", "C3", "C4", "C5", "C6", "C7", "C8", "C9"] with warnings.catch_warnings(record=True) as w: # GH 36972 warnings.simplefilter("always", "MatplotlibDeprecationWarning") for color in colors: _check_plot_works(df.plot, color=color) no_warning_raised = len(w) == 0 > assert no_warning_raised, "MatplotlibDeprecationWarning was raised" E AssertionError: MatplotlibDeprecationWarning was raised E assert False pandas/tests/plotting/test_frame.py:183: AssertionError

ivanovmg · 2020-10-09T10:35:38Z

I would only think that the error happened because of some other deprecation warning for that particular environment.
I thus changed assertion to match the actual deprecation warning message. We'll see how it goes.

ivanovmg · 2020-10-09T16:18:13Z

Looks like now everything works. The new test fails on master branch, but succeeds with this PR changes.

charlesdong1991 · 2020-10-09T18:49:26Z

emm, just make sure it's not a flaky error, maybe try to rebase one more time to see if the failure comes back?

btw, does this need a whatsnew? probably no, right?

ivanovmg · 2020-10-10T04:07:37Z

All checks pass. Neither the warning was raised (I checked all Travis logs) nor the test fail (like it should if this warning is raised).
Regarding the whatsnew - I don't know. What would others think?

jreback · 2020-10-10T16:43:49Z

lgtm. @charlesdong1991 merge if good here.

jreback · 2020-10-10T22:33:07Z

this is fine

ivanovmg added 7 commits October 8, 2020 19:35

TST: add failing test

e163302

FIX: eliminate warning regarding uppercase color

ff57346

CLN: simplify if statements

29ea95e

REF: drop old way of handling CN colors

a9d125d

Refer to commit aaa69d1, where the way of treating CN colors was originally developed. This is not relevant anymore as more readable alternative is provided.

REF: move _is_*_color funcs to module level

d489a10

TYP/DOC: fix typing, add docs

aa14f33

CLN: remove resetwarnings

749d63a

ivanovmg requested a review from jreback October 8, 2020 13:48

jreback added the Visualization plotting label Oct 8, 2020

jreback added this to the 1.2 milestone Oct 8, 2020

jreback added the Clean label Oct 8, 2020

jreback requested changes Oct 8, 2020

View reviewed changes

mwaskom reviewed Oct 8, 2020

View reviewed changes

REF: remove function _is_cn_color

25c0df0

conv.to_rgba handles CN colors itself.

ivanovmg requested a review from jreback October 9, 2020 03:50

ivanovmg mentioned this pull request Oct 9, 2020

FIX: fix cleanup warnings for errorbar timeseries #36982

Merged

5 tasks

charlesdong1991 suggested changes Oct 9, 2020

View reviewed changes

ivanovmg added 2 commits October 9, 2020 15:34

CLN: make if one-liner, reference GH issue

58cebb2

Merge branch 'master' into warnings/matplotlib

b4937b6

FIX: match exact warning message

f003217

Merge branch 'master' into warnings/matplotlib

4f1178d

jreback approved these changes Oct 10, 2020

View reviewed changes

jreback merged commit 0435d91 into pandas-dev:master Oct 10, 2020

ivanovmg deleted the warnings/matplotlib branch October 20, 2020 11:37

kesmit13 pushed a commit to kesmit13/pandas that referenced this pull request Nov 2, 2020

BUG: fix matplotlib warning on CN color (pandas-dev#36981)

c442745

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Uh oh!

BUG: fix matplotlib warning on CN color #36981

BUG: fix matplotlib warning on CN color #36981

Uh oh!

ivanovmg commented Oct 8, 2020

ivanovmg commented Oct 8, 2020

jreback Oct 8, 2020

ivanovmg Oct 9, 2020

jreback commented Oct 8, 2020

mwaskom Oct 8, 2020

mwaskom Oct 8, 2020

ivanovmg Oct 9, 2020

charlesdong1991 left a comment

charlesdong1991 Oct 9, 2020

ivanovmg Oct 9, 2020

ivanovmg commented Oct 9, 2020

ivanovmg commented Oct 9, 2020 •

edited

Loading

ivanovmg commented Oct 9, 2020

ivanovmg commented Oct 9, 2020

charlesdong1991 commented Oct 9, 2020

ivanovmg commented Oct 10, 2020

jreback commented Oct 10, 2020

jreback commented Oct 10, 2020

Labels

4 participants

Uh oh!

BUG: fix matplotlib warning on CN color #36981

BUG: fix matplotlib warning on CN color #36981

Uh oh!

Conversation

ivanovmg commented Oct 8, 2020

ivanovmg commented Oct 8, 2020

jreback Oct 8, 2020

Choose a reason for hiding this comment

ivanovmg Oct 9, 2020

Choose a reason for hiding this comment

jreback commented Oct 8, 2020

mwaskom Oct 8, 2020

Choose a reason for hiding this comment

mwaskom Oct 8, 2020

Choose a reason for hiding this comment

ivanovmg Oct 9, 2020

Choose a reason for hiding this comment

charlesdong1991 left a comment

Choose a reason for hiding this comment

charlesdong1991 Oct 9, 2020

Choose a reason for hiding this comment

ivanovmg Oct 9, 2020

Choose a reason for hiding this comment

ivanovmg commented Oct 9, 2020

ivanovmg commented Oct 9, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

ivanovmg commented Oct 9, 2020

ivanovmg commented Oct 9, 2020

charlesdong1991 commented Oct 9, 2020

ivanovmg commented Oct 10, 2020

jreback commented Oct 10, 2020

jreback commented Oct 10, 2020

Labels

4 participants

ivanovmg commented Oct 9, 2020 •

edited

Loading