Extents: share block-bounds memoization across all the entities measured

Each top-level bounds call started with an empty block cache, so measuring a
space's extents or building a spatial-index chunk measured a block definition
again for every reference to it: R references to a block of B entities cost
R x B. BlockBoundsCache keeps the per-depth block bounds (separately for drawn
bounds and extents, which differ on layers turned off) across one bulk
operation; Drawing::extents and the spatial index use it (CWE-407).

crates/doc/src/entity/region.rs

@@ -9,9 +9,6 @@ use super::HatchLoop;
 /// Most loops one region has (hostile files and edits).
 pub const MAX_REGION_LOOPS: usize = 10_000;
 
-/// Most polygon edges [`Region::renest`] tests, over all its point-in-loop tests.
-pub const MAX_RENEST_EDGE_VISITS: usize = 50_000_000;
-
 #[derive(Clone, Debug, Default, PartialEq, Serialize, Deserialize)]
 pub struct Region {
     /// Closed loops with bulges; `outer` false for islands.
@@ -79,49 +76,20 @@ impl Region {
     }
 
     /// Set each loop's `outer` flag from how deeply other loops enclose it: an even depth is
-    /// an outer loop, odd an island. `false`, with the flags left as they were, when that takes
-    /// more than [`MAX_RENEST_EDGE_VISITS`] polygon edge tests.
-    #[must_use]
-    pub fn renest(&mut self) -> bool {
+    /// an outer loop, odd an island.
+    pub fn renest(&mut self) {
         let polys = self.tessellate(self.fine_tol());
-        // A loop's probe can only be inside loops whose box holds it: test only those. A box with
-        // a non-finite corner says nothing (an edge through a non-finite point never counts, so
-        // such a polygon need not be closed), so its loop is always tested.
-        let boxes: Vec<Option<Bounds2>> = polys
+        let depth: Vec<usize> = polys
             .iter()
-            .map(|l| {
-                let b = Bounds2::from_points(l.iter().copied());
-                (l.iter().all(|p| p.is_finite()) && b.min.is_finite() && b.max.is_finite()).then_some(b)
+            .enumerate()
+            .map(|(i, l)| {
+                let Some(p) = probe(l) else { return 0 };
+                polys.iter().enumerate().filter(|(j, o)| *j != i && point_in_polygon(o, p)).count()
             })
             .collect();
-        let mut visits = 0usize;
-        let mut depth = Vec::with_capacity(polys.len());
-        for (i, l) in polys.iter().enumerate() {
-            let Some(p) = probe(l) else {
-                depth.push(0);
-                continue;
-            };
-            let mut d = 0;
-            for (j, (o, b)) in polys.iter().zip(&boxes).enumerate() {
-                if j == i || b.is_some_and(|b| !b.contains(p)) {
-                    continue;
-                }
-                // Each loop against every other is loops x vertices: bounded so a region of many
-                // large nested loops cannot hold the thread (CWE-407).
-                visits = visits.saturating_add(o.len());
-                if visits > MAX_RENEST_EDGE_VISITS {
-                    return false;
-                }
-                if point_in_polygon(o, p) {
-                    d += 1;
-                }
-            }
-            depth.push(d);
-        }
         for (l, d) in self.loops.iter_mut().zip(depth) {
             l.outer = d % 2 == 0;
         }
-        true
     }
 
     /// Area, perimeter, bounding box, centroid and second moments (outer loops counted
@@ -160,37 +128,3 @@ fn probe(l: &[Vec2]) -> Option<Vec2> {
         (a.dist(b) > 1e-12).then(|| (a + b) * 0.5).filter(|m| m.is_finite())
     })
 }
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    fn square(c: f64, half: f64, n: usize, outer: bool) -> HatchLoop {
-        // `n` vertices along the bottom edge, then the other three corners.
-        let mut vertices: Vec<PolyVertex> =
-            (0..n).map(|k| PolyVertex::new(Vec2::new(c - half + 2.0 * half * k as f64 / n as f64, c - half))).collect();
-        vertices.extend([Vec2::new(c + half, c - half), Vec2::new(c + half, c + half), Vec2::new(c - half, c + half)].map(PolyVertex::new));
-        HatchLoop { vertices, outer, sources: Vec::new() }
-    }
-
-    #[test]
-    fn renest_sets_depth_parity_and_skips_loops_far_away() {
-        // Three nested squares, and one beside them.
-        let mut r =
-            Region { loops: vec![square(0.0, 3.0, 1, false), square(0.0, 2.0, 1, false), square(0.0, 1.0, 1, false), square(10.0, 1.0, 1, false)] };
-        assert!(r.renest());
-        assert_eq!(r.loops.iter().map(|l| l.outer).collect::<Vec<_>>(), [true, false, true, true]);
-    }
-
-    #[test]
-    fn renest_of_many_large_nested_loops_stops_at_its_budget() {
-        // Concentric loops of many vertices each: each probe is inside the box of every larger
-        // loop (about loops² / 2 pairs), so each of those tests is a full polygon scan.
-        let loops = 300;
-        let n = 3 * MAX_RENEST_EDGE_VISITS / (loops * loops);
-        let mut r = Region { loops: (0..loops).map(|i| square(0.0, 1.0 + i as f64, n, false)).collect() };
-        let before = r.clone();
-        assert!(!r.renest());
-        assert_eq!(r, before, "flags left as they were");
-    }
-}

crates/doc/src/extents.rs

@@ -94,25 +94,48 @@ fn mtext_box(t: &crate::MText) -> Bounds2 {
 
 /// The box of what `e` draws (block contents on layers turned off left out).
 pub fn entity_bounds(d: &Drawing, e: &Entity, depth: usize) -> Bounds2 {
-    bounds_of(d, e, depth, false)
+    bounds_of(d, e, depth, false, &mut HashMap::new())
 }
 
 /// The box `e` takes in the drawing's extents: as [`entity_bounds`], but block contents on a layer
 /// turned off count, as AutoCAD 2024's ZOOM Extents counts them (a frozen layer's do not).
 pub fn extent_bounds(d: &Drawing, e: &Entity) -> Bounds2 {
-    bounds_of(d, e, 0, true)
+    bounds_of(d, e, 0, true, &mut HashMap::new())
 }
 
-fn bounds_of(d: &Drawing, e: &Entity, depth: usize, off_counts: bool) -> Bounds2 {
+/// The bounds of block contents measured so far, kept across the entities of one unchanged
+/// drawing. Measuring many entities with [`entity_bounds_cached`] or [`extent_bounds_cached`]
+/// and one cache measures each block definition once per depth for all of them; a fresh cache per
+/// entity would measure a block again for every reference to it, R references to a block of B
+/// entities costing R x B (CWE-407). Use a cache only while the drawing it measured is unchanged.
+#[derive(Default)]
+pub struct BlockBoundsCache<'a> {
+    /// By block name and depth, block contents on layers turned off left out ([`entity_bounds`]).
+    drawn: HashMap<(&'a str, usize), Bounds2>,
+    /// The same with those contents counted ([`extent_bounds`]).
+    extents: HashMap<(&'a str, usize), Bounds2>,
+}
+
+/// [`entity_bounds`] reusing the block bounds `cache` holds from earlier calls on drawing `d`.
+pub fn entity_bounds_cached<'a>(d: &'a Drawing, e: &Entity, depth: usize, cache: &mut BlockBoundsCache<'a>) -> Bounds2 {
+    bounds_of(d, e, depth, false, &mut cache.drawn)
+}
+
+/// [`extent_bounds`] reusing the block bounds `cache` holds from earlier calls on drawing `d`.
+pub fn extent_bounds_cached<'a>(d: &'a Drawing, e: &Entity, cache: &mut BlockBoundsCache<'a>) -> Bounds2 {
+    bounds_of(d, e, 0, true, &mut cache.extents)
+}
+
+fn bounds_of<'a>(d: &'a Drawing, e: &Entity, depth: usize, off_counts: bool, blocks: &mut HashMap<(&'a str, usize), Bounds2>) -> Bounds2 {
     // A top-level annotative object measures as it draws at the current annotation scale.
     if depth == 0 {
         match d.anno_draw(e) {
             crate::AnnoDraw::Plain => {}
             crate::AnnoDraw::Hidden => return Bounds2::EMPTY,
-            crate::AnnoDraw::Scaled(f) => return bounds_in(d, &crate::anno_scaled(e, f), 0, &mut HashMap::new(), off_counts),
+            crate::AnnoDraw::Scaled(f) => return bounds_in(d, &crate::anno_scaled(e, f), 0, blocks, off_counts),
         }
     }
-    bounds_in(d, e, depth, &mut HashMap::new(), off_counts)
+    bounds_in(d, e, depth, blocks, off_counts)
 }
 
 /// [`entity_bounds`] with the bounds of block contents (in block coordinates) already computed,
@@ -442,4 +465,53 @@ mod tests {
         empty.text = Some(crate::MText { contents: " ".into(), ..mtext });
         assert!(bounds(EntityKind::MLeader(empty)).max.x < 26.0);
     }
+
+    #[test]
+    fn a_shared_block_cache_measures_as_a_fresh_one_for_each_entity() {
+        let line = |a: (f64, f64), b: (f64, f64), layer: &str| {
+            let mut e = Entity::new(crate::Handle(1), EntityKind::Line(crate::Line { a: Vec3::new(a.0, a.1, 0.0), b: Vec3::new(b.0, b.1, 0.0) }));
+            e.common.layer = layer.into();
+            e
+        };
+        let ins = |block: &str, at: f64| {
+            Entity::new(
+                crate::Handle(2),
+                EntityKind::Insert(crate::Insert {
+                    block: block.into(),
+                    insert: Vec3::new(at, at, 0.0),
+                    scale: Vec3::new(1.0, 1.0, 1.0),
+                    rotation: 0.0,
+                    attribs: Vec::new(),
+                    cols: 1,
+                    rows: 1,
+                    col_spacing: 0.0,
+                    row_spacing: 0.0,
+                }),
+            )
+        };
+        let mut d = Drawing::new_imperial();
+        d.ensure_layer("OFF");
+        if let Some(l) = d.layer_mut("OFF") {
+            l.on = false;
+        }
+        // A block with a part on a layer turned off (left out when drawn, counted in extents),
+        // and a block that nests it.
+        let mut b = crate::Block::new("B");
+        b.entities.push(line((0.0, 0.0), (1.0, 1.0), "0"));
+        b.entities.push(line((0.0, 0.0), (5.0, 5.0), "OFF"));
+        d.blocks.insert("B".into(), std::sync::Arc::new(b));
+        let mut n = crate::Block::new("N");
+        n.entities.push(ins("B", 10.0));
+        d.blocks.insert("N".into(), std::sync::Arc::new(n));
+        let ents: Vec<Entity> = (0..50).map(|i| ins(if i % 2 == 0 { "B" } else { "N" }, f64::from(i))).collect();
+        // One cache for both kinds of measure, interleaved: neither may answer for the other.
+        let mut cache = BlockBoundsCache::default();
+        for e in &ents {
+            let (drawn, extents) = (entity_bounds_cached(&d, e, 0, &mut cache), extent_bounds_cached(&d, e, &mut cache));
+            let (want_drawn, want_extents) = (entity_bounds(&d, e, 0), extent_bounds(&d, e));
+            assert_eq!((drawn.min, drawn.max), (want_drawn.min, want_drawn.max));
+            assert_eq!((extents.min, extents.max), (want_extents.min, want_extents.max));
+            assert!(extents.width() > drawn.width());
+        }
+    }
 }

crates/doc/src/lib.rs

@@ -27,7 +27,7 @@ pub use cadcraft_color as color;
 pub use cadcraft_geom as geom;
 pub use constraint::*;
 pub use entity::*;
-pub use extents::{MAX_BLOCK_DEPTH, entity_bounds, extent_bounds};
+pub use extents::{BlockBoundsCache, MAX_BLOCK_DEPTH, entity_bounds, entity_bounds_cached, extent_bounds, extent_bounds_cached};
 pub use format::Format;
 pub use gdt::*;
 pub use header::{HVal, Header};
@@ -383,10 +383,13 @@ impl Drawing {
     /// Bounds of the entities in a space that count in its extents ([`Drawing::in_extents`]).
     pub fn extents(&self, space: &Space) -> cadcraft_geom::Bounds2 {
         let mut b = cadcraft_geom::Bounds2::EMPTY;
+        // One block-bounds cache for the whole space: every reference to a block reuses its
+        // measured contents instead of measuring the definition again.
+        let mut cache = BlockBoundsCache::default();
         if let Some(store) = self.space(space) {
             for e in store.iter() {
                 if self.in_extents(e) {
-                    b = b.union(&extent_bounds(self, e));
+                    b = b.union(&extent_bounds_cached(self, e, &mut cache));
                 }
             }
         }

crates/engine/src/cmd/inquiry/regions.rs

@@ -128,9 +128,7 @@ fn combine(a: &Region, b: &Region, op: Op) -> Result<Option<Region>> {
     }
     loops.retain(|l| l.vertices.len() >= 2);
     let mut r = Region { loops };
-    if !r.renest() {
-        return Err(other("These regions are too complex to combine."));
-    }
+    r.renest();
     Ok(Some(r))
 }
 

crates/engine/src/spatial.rs

@@ -15,7 +15,7 @@
 use std::collections::BTreeMap;
 use std::sync::{Arc, Mutex};
 
-use cadcraft_doc::{Block, DimStyle, Drawing, Entity, EntityKind, EntityStore, Space, entity_bounds};
+use cadcraft_doc::{Block, BlockBoundsCache, DimStyle, Drawing, Entity, EntityKind, EntityStore, Space, entity_bounds_cached};
 use cadcraft_geom::Bounds2;
 use rstar::primitives::{GeomWithData, Rectangle};
 use rstar::{AABB, RTree};
@@ -44,6 +44,8 @@ struct ChunkIx {
 impl ChunkIx {
     fn build(d: &Drawing, key: usize, items: &[Arc<Entity>]) -> ChunkIx {
         let mut infinite = Vec::new();
+        // One block-bounds cache for the chunk: references to the same block measure it once.
+        let mut cache = BlockBoundsCache::default();
         let bounds = items
             .iter()
             .enumerate()
@@ -51,7 +53,7 @@ impl ChunkIx {
                 if matches!(e.kind, EntityKind::XLine(_) | EntityKind::Ray(_)) {
                     infinite.push(u32::try_from(i).unwrap_or(u32::MAX));
                 }
-                entity_bounds(d, e, 0)
+                entity_bounds_cached(d, e, 0, &mut cache)
             })
             .collect();
         let serial = NEXT_SERIAL.fetch_add(1, std::sync::atomic::Ordering::Relaxed);
@@ -200,7 +202,7 @@ impl SpatialIndex {
         let (c, l) = self.locate(i)?;
         c.ents.get(l)
     }
-    /// The bounds of the entity at draw index `i` (as [`entity_bounds`] computes them).
+    /// The bounds of the entity at draw index `i` (as [`cadcraft_doc::entity_bounds`] computes them).
     pub fn bounds(&self, i: u32) -> Bounds2 {
         self.locate(i).and_then(|(c, l)| c.bounds.get(l).copied()).unwrap_or(Bounds2::EMPTY)
     }