Added handling of axes to regular plotting functions #1942

philippjfr · 2016-02-17T00:08:41Z

This PR makes it possible to pass an existing matplotlib Axes object to the various plotting functions in the plotting module. I realize that in PR #1112 drawing directly onto the axes was disabled, instead always drawing to the axis that is currently open using the plt function calls. While that is a good fix for users using the plt interface, which is useful for simple usage, when building more complex plots the object-oriented approach is the way to go. This is crucial for some of the functionality we're building as part of the cube-explorer project integrating Iris functionality within HoloViews.

It's a fairly small set of changes but not being familiar with the code I may have missed something. I think only providing this functionality in plot but not quickplot makes sense since the latter is suited towards usage in an interactive interpreter anyway. I'd be happy to add examples, documentation and tests once we agree these are the correct changes.

marqh · 2016-02-17T13:47:26Z

lib/iris/plot.py

        u, v = plot_arrays
        u, v = _broadcast_2d(u, v)
-        draw_method = getattr(plt, draw_method_name)
+        axes = kwargs.pop('axes')


there will need to be a None returned if axes doesn't exist

Thanks for catching that, will make any fixes once you're done reviewing.

marqh · 2016-02-17T13:53:05Z

you also need to change the copyright statement in the first line of plot.py

it needs to read

# (C) British Crown Copyright 2010 - 2016, Met Office

as you've changed it in 2016

philippjfr · 2016-02-17T13:54:41Z

Thanks @marqh, will do that and will hopefully get the contributor agreement sorted out quickly. Any thoughts/recommendations on examples, documentation and tests?

marqh · 2016-02-17T14:00:05Z

i recommend
https://github.com/SciTools/iris/tree/master/lib/iris/tests/unit/plot
as the start point for testing

the unit tests (lib/iris/tests/unit/.../.py) are gradually replacing the older tests (lib/iris/tests/.py)

marqh · 2016-02-17T14:01:37Z

I think a lot of the plot types in
http://scitools.org.uk/iris/docs/latest/iris/iris/plot.html

will want this behaviour

I think that we should explicitly mention 'axes' on any plot type that we have implemented the change for
e.g.
https://github.com/philippjfr/iris/blob/master/lib/iris/plot.py#L615

philippjfr · 2016-02-17T14:05:14Z

I think that we should explicitly mention 'axes' on any plot type that we have implemented the change for.

Agreed, I think they should all support it now, so I'll just make sure to document the kwarg everywhere.

marqh · 2016-02-18T08:37:45Z

CLA received, with thanks
SciTools/scitools.org.uk#120

marqh · 2016-02-19T16:41:17Z

Hi @philippjfr

this looks grand. please add a stub to
https://github.com/philippjfr/iris/tree/master/docs/iris/src/whatsnew/contributions_1.10
and squish your commits

then this should be good to merge
thank you

philippjfr · 2016-02-19T16:44:04Z

@marqh Thanks for the review, I'll make sure to do that. I've also looked into adding some unit tests based on the existing unit tests but as far as I can tell only the contour and contourf calls are actually setting the x- and y-ticks correctly, so give me some time to investigate that.

philippjfr · 2016-02-19T21:34:23Z

@marqh Made some final fixes, added unit tests, added an entry under contributions and squashed the commits. I think it's ready to merge, but let me know if you want me to refactor those unit tests a little bit.

pelson · 2016-02-22T08:47:42Z

lib/iris/plot.py

    # Ensure the current axes are a cartopy.mpl.GeoAxes instance.
-    _replace_axes_with_cartopy_axes(cartopy_proj)
+    if not kwargs.get('axes'):
+        _replace_axes_with_cartopy_axes(cartopy_proj)


I'm not sure we can make this conditional - what if you've made an axes which isn't a cartopy GeoAxes and passed that through... it may be that we need to raise an exception in that situation.

Good point, raising an exception makes sense to me.

Raising an exception now.

philippjfr · 2016-02-29T13:06:45Z

@pelson I've finally made the revisions you asked for, except regarding your comment to attach the anchored text to the figure. I'd be happy to revert adding an axes argument there for for now, just let me know what you think is best.

pelson · 2016-03-01T10:54:59Z

lib/iris/plot.py

-    _replace_axes_with_cartopy_axes(cartopy_proj)
+    if 'axes' not in kwargs:
+        _replace_axes_with_cartopy_axes(cartopy_proj)
+    elif not not isinstance(kwargs['axes'], cartopy.mpl.geoaxes.GeoAxes):


double not. I'm not not sure what happens here...

This highlights the need for a test of this error I think.

Agreed, will add one now.

Looks like some of the tests are failing because of this too:

====================================================================== ERROR: test_outline (iris.tests.test_mapping.TestLimitedAreaCube) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/Iris-1.10.0.dev0-py2.7-linux-x86_64.egg/iris/tests/test_mapping.py", line 240, in test_outline iplt.outline(self.cube) File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/Iris-1.10.0.dev0-py2.7-linux-x86_64.egg/iris/plot.py", line 845, in outline antialiased=True, coords=coords, axes=axes) File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/Iris-1.10.0.dev0-py2.7-linux-x86_64.egg/iris/plot.py", line 244, in _draw_2d_from_bounds cube, plot_defn, *args, **kwargs) File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/Iris-1.10.0.dev0-py2.7-linux-x86_64.egg/iris/plot.py", line 614, in _map_common return plotfn(*new_args, **kwargs) File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/matplotlib/pyplot.py", line 2946, in pcolormesh ret = ax.pcolormesh(*args, **kwargs) File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/matplotlib/axes.py", line 7766, in pcolormesh t = t._as_mpl_transform(self.axes) File "lib/cartopy/_crs.pyx", line 213, in cartopy._crs.CRS._as_mpl_transform (lib/cartopy/_crs.c:4071) ValueError: Axes should be an instance of GeoAxes, got <class 'matplotlib.axes.AxesSubplot'> ====================================================================== ERROR: test_grid (iris.tests.test_mapping.TestBoundedCube) ---------------------------------------------------------------------- Traceback (most recent call last): File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/Iris-1.10.0.dev0-py2.7-linux-x86_64.egg/iris/tests/test_mapping.py", line 205, in test_grid iplt.outline(self.cube) File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/Iris-1.10.0.dev0-py2.7-linux-x86_64.egg/iris/plot.py", line 845, in outline antialiased=True, coords=coords, axes=axes) File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/Iris-1.10.0.dev0-py2.7-linux-x86_64.egg/iris/plot.py", line 244, in _draw_2d_from_bounds cube, plot_defn, *args, **kwargs) File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/Iris-1.10.0.dev0-py2.7-linux-x86_64.egg/iris/plot.py", line 614, in _map_common return plotfn(*new_args, **kwargs) File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/matplotlib/pyplot.py", line 2946, in pcolormesh ret = ax.pcolormesh(*args, **kwargs) File "/home/travis/miniconda/envs/test-environment/lib/python2.7/site-packages/matplotlib/axes.py", line 7766, in pcolormesh t = t._as_mpl_transform(self.axes) File "lib/cartopy/_crs.pyx", line 213, in cartopy._crs.CRS._as_mpl_transform (lib/cartopy/_crs.c:4071) ValueError: Axes should be an instance of GeoAxes, got <class 'matplotlib.axes.AxesSubplot'>

Likely due to the double not above. Fixing that and adding new unit tests now.

philippjfr · 2016-03-01T11:33:38Z

Fixed the double not bug above, which will hopefully fix the tests and added some further unit tests to ensure that plots actually raise the GeoAxes exception when passed a regular axes.

philippjfr · 2016-03-01T11:58:07Z

Seems there are further tests failing due the new GeoAxes exception. I don't quite understand how that's happening unless those tests have been passing axes in all along and they were simply ignored until now. Will investigate.

philippjfr · 2016-03-01T13:26:44Z

Resolved the failing tests, the logic for raising an exception was off. I think it's ready to merge now.

Added handling of axes to regular plotting functions

pelson · 2016-03-02T09:44:12Z

Thanks @philippjfr - this is an excellent addition. 👍

philippjfr · 2016-03-02T10:55:15Z

Yay! Thanks for merging @pelson.

Added a what's new for #1942.

philippjfr changed the title ~~Added handling of axes handling to regular plotting functions~~ Added handling of axes to regular plotting functions Feb 17, 2016

marqh reviewed Feb 17, 2016
View reviewed changes

philippjfr force-pushed the master branch 2 times, most recently from c03e6ac to ffb30aa Compare February 19, 2016 21:32

pelson reviewed Feb 22, 2016
View reviewed changes

philippjfr mentioned this pull request Feb 23, 2016

Added initial prototype of holocube package and notebook CubeBrowser/cube-explorer#5

Merged

5 tasks

philippjfr force-pushed the master branch 2 times, most recently from c95232f to 2888005 Compare February 29, 2016 13:03

pelson reviewed Mar 1, 2016
View reviewed changes

philippjfr force-pushed the master branch from 2888005 to ccd5028 Compare March 1, 2016 11:30

philippjfr force-pushed the master branch from ccd5028 to 90efc58 Compare March 1, 2016 11:37

philippjfr force-pushed the master branch 2 times, most recently from 57ed2ed to ab2808a Compare March 1, 2016 12:23

Allowed passing explicit axes argument to plotting functions

b0513b6

philippjfr force-pushed the master branch from ab2808a to b0513b6 Compare March 1, 2016 12:41

pelson added a commit that referenced this pull request Mar 2, 2016

Merge pull request #1942 from philippjfr/master

808d208

Added handling of axes to regular plotting functions

pelson merged commit 808d208 into SciTools:master Mar 2, 2016

QuLogic added this to the v1.10 milestone Mar 2, 2016

pelson added a commit to pelson/iris that referenced this pull request Mar 2, 2016

Added a what's new for SciTools#1942.

3a6920c

pelson mentioned this pull request Mar 2, 2016

Added a what's new for #1942. #1951

Merged

rhattersley mentioned this pull request May 5, 2016

Complete v1.10 deprecations to prepare for v2.0 #1948

Closed

9 tasks

pp-mo added a commit that referenced this pull request May 6, 2016

Merge pull request #1951 from pelson/whats_new_axes

fc0a38d

Added a what's new for #1942.

ajdawson mentioned this pull request Feb 7, 2018

quickplot.contourf features not attached to supplied axes #2946

Closed

rcomer mentioned this pull request Apr 8, 2021

iris.plot arguments #4085

Closed

Added handling of axes to regular plotting functions #1942

Added handling of axes to regular plotting functions #1942

Uh oh!

Conversation

philippjfr commented Feb 17, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

marqh commented Feb 17, 2016

Uh oh!

philippjfr commented Feb 17, 2016

Uh oh!

marqh commented Feb 17, 2016

Uh oh!

marqh commented Feb 17, 2016

Uh oh!

philippjfr commented Feb 17, 2016

Uh oh!

marqh commented Feb 18, 2016

Uh oh!

marqh commented Feb 19, 2016

Uh oh!

philippjfr commented Feb 19, 2016

Uh oh!

philippjfr commented Feb 19, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

philippjfr commented Feb 29, 2016

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

philippjfr commented Mar 1, 2016

Uh oh!

philippjfr commented Mar 1, 2016

Uh oh!

philippjfr commented Mar 1, 2016

Uh oh!

pelson commented Mar 2, 2016

Uh oh!

philippjfr commented Mar 2, 2016

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants