Combine wrapped collection into GeoMesh data when get_array is called #1656
Conversation
When calling get_array with a QuadMesh object (from pcolormesh), the returned array only includes the unmasked data. This change fills in the masked data when present.
|
Looks like a good way to do it to me, @sdeastham! Can you add a test for this too? As simple as just putting Also, it looks like you've got one failing test currently. That is actually one that I just put in to make sure the cells were being wrapped properly. A good fix for that test would be to assert that the |
|
ping @sdeastham. I think all that is needed is updated tests here. |
|
@greglucas Got it. Sorry about the delay - I've never used pytest before and am having terrible trouble getting it to do anything except throw |
|
No worries at all. I think that error can be fixed by adding the pyargs argument to pytest: |
A new test has been added which verifies that the data returned by a pcolormesh get_array() command matches the data which was first given to the pcolormesh command, even if wrapping has occurred. One of the other tests (pcolormesh_diagonal_wrap) was also failing due to a faulty check, which has been removed.
| @@ -321,6 +321,48 @@ def test_pcolormesh_global_with_wrap1(): | |||
| ax.coastlines() | |||
| ax.set_global() # make sure everything is visible | |||
|
|
|||
| #@pytest.mark.natural_earth | |||
| #@ImageTesting(['pcolormesh_global_wrap1'], tolerance=1.27) | |||
There was a problem hiding this comment.
E265 block comment should start with '# '
| @@ -321,6 +321,48 @@ def test_pcolormesh_global_with_wrap1(): | |||
| ax.coastlines() | |||
| ax.set_global() # make sure everything is visible | |||
|
|
|||
| #@pytest.mark.natural_earth | |||
| #@ImageTesting(['pcolormesh_global_wrap1'], tolerance=1.27) | |||
| def test_pcolormesh_get_array_with_mask(): | |||
There was a problem hiding this comment.
E302 expected 2 blank lines, found 1
| assert c._wrapped_collection_fix is not None, \ | ||
| 'No pcolormesh wrapping was done when it should have been.' | ||
|
|
||
| assert np.array_equal(data.ravel(),c.get_array()), \ |
| ax.coastlines() | ||
| ax.set_global() # make sure everything is visible | ||
|
|
||
| assert getattr(c,"_wrapped_collection_fix",None) is None, \ |
| assert getattr(c,"_wrapped_collection_fix",None) is None, \ | ||
| 'pcolormesh wrapping was done when it should not have been.' | ||
|
|
||
| assert np.array_equal(data2.ravel(),c.get_array()), \ |
Stickler Bot identified six issues with the previous commit. Two were with commented-out code which was no longer necessary, one was because of a missing newline, and three were because of missing whitespace after commas. All have now been resolved.
|
@greglucas Thanks again for your patience! I've attempted to build some tests which test that |
There was a problem hiding this comment.
@sdeastham, glad you stuck with it and got all the tests working. This looks good to me! Thanks for contributing and hope to see you back again!
When calling get_array with a QuadMesh object (from pcolormesh), the returned array only included the unmasked data. This change fills in the masked data when present. This follows from the discussion in #1654.
Rationale
The
pcolormeshset_arraymethod now correctly handles the wrapped, masked collections which can be generated when polygons cross the antimeridian. However, theget_arraymethod still only returns the data for the standard cells (those which did not cross the antimeridian). This caused unexpected behavior when running, for example:This would result in the same plot being shown, but now with the cells at 180E missing. This fix modifies the
get_arraycall to compensate.Implications
Calling
get_arraynow returns the complete array, with the masked elements filled in.