fix: labels visibility issue

[moe-ad]

This is a tricky one which I have been investigating since yesterday. Even trickier to describe but I will try my best.

Consider this code snippet

import pyvista as pv
from ansys.tools.visualization_interface import Plotter

plotter = Plotter()
sphere = pv.Sphere(radius=0.5, center=(0.5, 0.5, 0))
plotter.plot(sphere)

points = [[0, 0, 0], [1, 0, 0], [0, 1, 0], [1, 1, 0], [0.5, 0.5, 0.5]]
labels = ['A', 'B', 'C', 'D', 'E']

# Adding labels on at a time
for i, (pt, lbl) in enumerate(zip(points, labels)):
    # Pass font_size, point_size, and other kwargs
    plotter.add_labels([pt], [lbl], font_size=20, point_size=10, bold=True)

print(f"Total actors in scene: {len(plotter._backend.base_plotter.actors)}")
plotter.show()

Now one would expect 11 actors to be present on this scene (1 sphere, 5 points, 5 labels), but the running this code produces random results every single time as shown in the ffl examples (you can also trying running the snippet locally from the main branch):

Number of reported actors on the scene	Corresponding plot
3
5
3

As you can see, labels are being randomly dropped from the scene.

However, this bug doesn't occur when you go through pyvista directly. You can confirm that by running the ffl:

import pyvista as pv

pl = pv.Plotter()
pl.add_mesh(pv.Sphere(radius=0.5, center=(0.5, 0.5, 0)))

# Adding labels one at a time
points = [[0, 0, 0], [1, 0, 0], [0, 1, 0], [1, 1, 0], [0.5, 0.5, 0.5]]
labels = ['A', 'B', 'C', 'D', 'E']

for i, (pt, lbl) in enumerate(zip(points, labels)):
    # Create PolyData for each point
    point_cloud = pv.PolyData([pt])
    pl.add_point_labels(point_cloud, [lbl], font_size=20, point_size=10, bold=True)

print(f"Total actors in scene: {len(pl.actors)}")

pl.show()

In summary, the outcome of my debugging is that:

• (simple version) the labels are getting garbage collected at some point and pyvista loses this context.
• (more complicated version) this has to do with how PyVista stores the labels in memory.

The simple solution is just to hold a reference to the labels like I have done in this PR.

[AlejandroFernandezLuces]

Weird error.. thanks for fixing it 🙂