From 0ca1e58b1f1db9242a0408c87329c506b1cd4795 Mon Sep 17 00:00:00 2001
From: Hakan Seven <hakanseven8125@gmail.com>
Date: Wed, 10 Jun 2026 20:27:45 +0300
Subject: [PATCH] fix(io): purge zero-length point-collapsed splines that hang
 load

sample_layout_03/04.dwg would not open: a periodic 9-control-point spline
with every control point pinned at the origin (zero extent) reaches truck's
parameter_division, which never converges on a zero-length curve and hangs
the first-frame tessellation. The spline passed every existing guard
(finite points, no knots, degree 3). Reject splines whose control-point
extent is sub-precision (< 1e-6 on all axes).

Verified: all five sample_layout DWGs now load + tessellate (03/04 went
from a hang to ~356 ms; 01/02/05 unchanged, no valid spline dropped).

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
---
 src/io/mod.rs | 28 ++++++++++++++++++++++++++++
 1 file changed, 28 insertions(+)

diff --git a/src/io/mod.rs b/src/io/mod.rs
index 273571220..9afa5ec66 100644
--- a/src/io/mod.rs
+++ b/src/io/mod.rs
@@ -498,10 +498,28 @@ pub(crate) fn is_entity_corrupt(e: &EntityType) -> bool {
                 && (s.knots.iter().any(|k| !k.is_finite())
                     || s.knots.windows(2).any(|w| w[1] < w[0])
                     || s.knots.len() != n + deg + 1);
+            // Degenerate: every control point collapses onto (nearly) the same
+            // point, so the curve has zero length. truck's `circle_arc` /
+            // `parameter_division` never converges on it and the tessellation
+            // hangs — a periodic 9-point spline pinned at the origin is the seen
+            // case. Reject when the control-point extent is sub-precision.
+            let degenerate_extent = n >= 2 && {
+                let (mut mn, mut mx) = ([f64::MAX; 3], [f64::MIN; 3]);
+                for p in &s.control_points {
+                    mn[0] = mn[0].min(p.x);
+                    mx[0] = mx[0].max(p.x);
+                    mn[1] = mn[1].min(p.y);
+                    mx[1] = mx[1].max(p.y);
+                    mn[2] = mn[2].min(p.z);
+                    mx[2] = mx[2].max(p.z);
+                }
+                (mx[0] - mn[0]).max(mx[1] - mn[1]).max(mx[2] - mn[2]) < 1.0e-6
+            };
             n >= MAX_VERTS
                 || degree_bad
                 || s.control_points.iter().any(|p| !finite_vec3(p))
                 || knots_bad
+                || degenerate_extent
         }
         _ => false,
     }
@@ -651,6 +669,16 @@ mod corrupt_guard_tests {
         assert!(is_entity_corrupt(&EntityType::Spline(s)));
     }
 
+    // A periodic spline whose control points all collapse onto (nearly) one
+    // point has zero length; truck's parameter_division never converges and the
+    // tessellation hangs. The control-point extent floor must reject it.
+    #[test]
+    fn rejects_degenerate_point_spline() {
+        let pts = vec![Vector3::new(1e-12, -1e-12, 0.0); 9];
+        let s = Spline::from_control_points(3, pts);
+        assert!(is_entity_corrupt(&EntityType::Spline(s)));
+    }
+
     // A normal cubic spline (4 control points, valid clamped knots) survives.
     #[test]
     fn keeps_valid_spline() {