Prioritize smaller distance gizmos in selection

When two GizmoDimension hit regions overlap (a short dimension
nested inside a longer one along the same axis), the larger one
used to win because hit boxes are scaled by world-space length —
the long box fully contains the short one, leaving the short
gizmo unreachable. The larger gizmo stays clickable at its
exposed ends, so smaller-wins is the right UX default.

Sets self.select_bias = -self._dimension_length inside
GizmoDimension.set_dimension_length. The smaller gizmo writes a
less-negative depth value in the GPU select buffer and wins the
tie-break. select_bias is unused elsewhere in the codebase, so
icon and arrow gizmos keep bias=0 and are unaffected (icons
correctly still win against dimensions, since 0 > -length).

Adds test/bim/module/drawing/test_dimension_gizmo_priority.py
with 5 cases: direct ordering, monotonicity across length ranges,
abs() handling for signed dimensions, and NaN/Inf safety.

Generated with the assistance of an AI coding tool.
This commit is contained in:
Gorgious56
2026-05-21 10:30:32 +02:00
committed by Thomas Krijnen
parent ddd9b4fa23
commit 74906ac9fe
2 changed files with 105 additions and 0 deletions
@@ -3864,6 +3864,11 @@ class GizmoDimension(GizmoMovable):
self._display_value = max(-10000.0, min(length, 10000.0))
# Clamp to valid range (0 to 10000 meters is reasonable for BIM) for drawing
self._dimension_length = max(0.0, min(abs(length), 10000.0))
# Smaller dimensions win selection when hit regions overlap: a long gizmo's
# hit box fully contains a nested short one's, so without a bias the long
# one wins and the short one is unreachable. The long one stays clickable
# at its exposed ends regardless of bias.
self.select_bias = -self._dimension_length
def invoke(self, context: bpy.types.Context, event: bpy.types.Event) -> set:
"""Initialize dimension gizmo interaction with click-position tracking.
@@ -0,0 +1,100 @@
# Bonsai - OpenBIM Blender Add-on
# Copyright (C) 2026
#
# This file is part of Bonsai.
#
# Bonsai is free software: you can redistribute it and/or modify
# it under the terms of the GNU General Public License as published by
# the Free Software Foundation, either version 3 of the License, or
# (at your option) any later version.
#
# Bonsai is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU General Public License
# along with Bonsai. If not, see <http://www.gnu.org/licenses/>.
#
# This file was generated with the assistance of an AI coding tool.
"""Regression guard: overlapping distance gizmos must let the smaller one win.
When two ``GizmoDimension`` instances overlap on screen (e.g. a short dimension
nested inside a longer one along the same axis), the longer one's hit box fully
contains the shorter one's. Without a depth bias the longer one wins the GPU
select tie-break and the shorter one becomes unreachable.
``GizmoDimension.set_dimension_length`` writes ``select_bias = -dimension_length``
so the smaller one writes a higher (less-negative) bias and wins. The longer one
stays clickable at its exposed ends regardless of bias.
We call ``set_dimension_length`` as an unbound method on a ``SimpleNamespace``
fake ``self``. Its body only *writes* attributes (``_display_value``,
``_dimension_length``, ``select_bias``), so it doesn't need a real
``bpy.types.Gizmo`` instance those only exist inside a registered
``GizmoGroup`` and aren't constructible in a headless test."""
import types
from types import SimpleNamespace
import bpy
import pytest
from bonsai.bim.module.drawing.gizmos import GizmoDimension
pytestmark = pytest.mark.drawing
@pytest.fixture(autouse=True)
def _require_real_bpy():
if not isinstance(bpy, types.ModuleType) or hasattr(bpy, "_mock_name"):
pytest.skip("requires real Blender (bpy is mocked or absent)")
def test_smaller_dimension_wins_select_bias():
small = SimpleNamespace()
large = SimpleNamespace()
GizmoDimension.set_dimension_length(small, 0.077)
GizmoDimension.set_dimension_length(large, 0.109)
assert small.select_bias > large.select_bias
@pytest.mark.parametrize(
"lengths",
[
[0.0, 0.05, 0.077, 0.109, 1.0, 5.0, 10.0],
[0.001, 0.5, 2.5, 100.0, 9999.0],
],
)
def test_select_bias_is_non_increasing_in_length(lengths):
"""A monotonic mapping is all Blender's GPU select needs to break the tie."""
biases = []
for length in lengths:
gizmo = SimpleNamespace()
GizmoDimension.set_dimension_length(gizmo, length)
biases.append(gizmo.select_bias)
for prev, curr in zip(biases, biases[1:]):
assert prev >= curr, f"select_bias must be non-increasing in length, got {biases}"
def test_negative_length_uses_absolute_value_for_bias():
"""Negative dimension values (e.g. inverted angles) clamp to abs() for hit-box scaling;
select_bias follows the same clamped magnitude so signed-direction gizmos still
obey the smaller-wins rule against their positive-sided peers."""
positive = SimpleNamespace()
negative = SimpleNamespace()
GizmoDimension.set_dimension_length(positive, 0.5)
GizmoDimension.set_dimension_length(negative, -0.5)
assert positive.select_bias == negative.select_bias
def test_nan_and_inf_length_falls_back_to_zero_bias():
"""Invalid inputs are coerced to 0.0 before the bias is written, so a malformed
update can't push a gizmo arbitrarily far forward or backward in the select buffer."""
import math
for bad in (math.nan, math.inf, -math.inf, "not a number"):
gizmo = SimpleNamespace()
GizmoDimension.set_dimension_length(gizmo, bad)
assert gizmo.select_bias == 0.0