MeshedRegion.plot() with PropertyField

[PProfizi]

Add the possibility to plot a PropertyField on a MeshedRegion.
Will have to add PropertyField.plot(mesh) with mandatory mesh as a PropertyField currently most likely does not have a mesh asosciated.

The PyVista options for the scalar bar still need a little bit of tweaking. Some colors are even missing from the scalar bar:

[PProfizi]

Hi @AlejandroFernandezLuces, could you please take a look at this PR?
The goal is to plot a discrete and potentially discontinuous field of integers, each value colored.
Do you think the approach with scalar_bar and categories options is the right one?

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 89.18919% with 4 lines in your changes are missing coverage. Please review.

Project coverage is 84.20%. Comparing base (ae7e203) to head (aea653e).
Report is 6 commits behind head on master.

❗ Current head aea653e differs from pull request most recent head 3ef8f12. Consider uploading reports for the commit 3ef8f12 to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1526      +/-   ##
==========================================
+ Coverage   84.16%   84.20%   +0.03%     
==========================================
  Files          82       82              
  Lines        9655     9672      +17     
==========================================
+ Hits         8126     8144      +18     
+ Misses       1529     1528       -1     

[AlejandroFernandezLuces]

Yes, I think activating categories flag and using the scalars_bar is the best way to represent what you want.

A side comment, why are you filling a kwargs map instead of using the arguments in the function calls? I personally find it clearer to use the parameters directly.

[AlejandroFernandezLuces]

I would prefer to directly use the add_scalar_bar directly, I think something like this should work in this context

Suggested change

	kwargs = self._set_scalar_bar_title(kwargs)
	kwargs["scalar_bar_args"]["n_labels"] = len(set(field.data))
	kwargs["scalar_bar_args"]["fmt"] = "%.0f"
	self._plotter.add_scalar_bar(
	title='Your title',
	n_labels=len(set(field.data))
	fmt='%.0f',
	)

[PProfizi]

@AlejandroFernandezLuces I think the original idea of using kwargs is to use setdefault so that if the argument is given as a kwarg to the plot function by the user then it is not overriden by what we put here. I'll improve the code to do that correctly.

[PProfizi]

Thanks @AlejandroFernandezLuces!
Also, any idea why a color is missing in the scalar bar of the plot I shared above?

[AlejandroFernandezLuces]

Thanks @AlejandroFernandezLuces! Also, any idea why a color is missing in the scalar bar of the plot I shared above?

Not sure what is happening there, you can try to set n_colors=len(set(field.data)) in add_scalar_bar or in kwargs, so you make sure that you have a color per label on PyVista side. If not, maybe you have a bug on pyDPF that causes a misrepresentation between the fields and the meshes?

[PProfizi]

I managed to get this look with the latest commit

[PProfizi]

NB: the default cmap used is tab20 as it is discrete and ensures a clear difference between neighboring values, however it only has 20 values and will not "circle back" if more values are present. It will simply associate several consecutive values to one color.
Improvement should be made to create a dynamic color map based on semi-random colors, and to properly handle discontinuous discrete series of integers.
For example:
having mat_id = [1, 2, 29, 30] will result in 1 and 2 being both the bottom color and 29 and 30 being both the top color. It will use an identical spacing to associate numbers with colors.

[rafacanton]

This needs to be reverted before merging, right?

[rafacanton]

Suggested change

	if isinstance(field_or_fields_container, PropertyField):
	elif isinstance(field_or_fields_container, PropertyField):