address review: size text cost by the Paragraph and Run struct sizes

crates/engine/src/lib.rs

@@ -811,10 +811,32 @@ fn clamp_selection(st: &mut DocState) {
 /// Default for [`History::byte_limit`].
 pub const UNDO_BYTE_BUDGET: usize = 256 << 20;
 
+/// A rough size of a text body in memory: its paragraphs and runs, at their inline struct sizes.
+fn text_cost(t: &deckcraft_model::text::TextBody) -> usize {
+    use deckcraft_model::text::{Paragraph, Run};
+    let (para, run) = (std::mem::size_of::<Paragraph>(), std::mem::size_of::<Run>());
+    t.paragraphs.iter().map(|p| para + p.runs.iter().map(|r| run + r.text.len()).sum::<usize>()).sum()
+}
+
+/// A rough size of comments in memory, replies included.
+fn comments_cost(cs: &[deckcraft_model::Comment], depth: usize) -> usize {
+    let kids = |c: &deckcraft_model::Comment| if depth < 64 { comments_cost(&c.replies, depth + 1) } else { 0 };
+    cs.iter().map(|c| 256 + c.author.len() + c.initials.len() + c.text.len() + c.date.len() + kids(c)).sum()
+}
+
+/// A rough size of a slide's own data besides its shapes: notes, comments, tags and kept XML,
+/// all copied with the slide when an edit changes it.
+fn slide_extra_cost(s: &deckcraft_model::Slide) -> usize {
+    text_cost(&s.notes)
+        + comments_cost(&s.comments, 0)
+        + s.name.len()
+        + s.tags.iter().map(|(k, v)| 48 + k.len() + v.len()).sum::<usize>()
+        + s.raw_ext.as_ref().map_or(0, String::len)
+}
+
 /// A rough size of a shape in memory: the struct, its text and kept XML, its children.
 fn shape_cost(sh: &deckcraft_model::Shape, depth: usize) -> usize {
-    let text: usize =
-        sh.text.as_ref().map_or(0, |t| t.paragraphs.iter().map(|p| 96 + p.runs.iter().map(|r| 128 + r.text.len()).sum::<usize>()).sum());
+    let text: usize = sh.text.as_ref().map_or(0, text_cost);
     let kids: usize = if depth < 64 { sh.children().iter().map(|c| shape_cost(c, depth + 1)).sum() } else { 0 };
     768 + text + sh.raw_ext.as_ref().map_or(0, String::len) + kids
 }
@@ -832,7 +854,7 @@ pub fn undo_cost(doc: &Presentation, newer: &Presentation) -> usize {
     let media: HashSet<*const Vec<u8>> = newer.media.iter().map(|m| Arc::as_ptr(&m.data)).collect();
     let mut n = 64 * doc.slides.len();
     for s in doc.slides.iter().filter(|s| !slides.contains(&Arc::as_ptr(s))) {
-        n += 512 + shapes_cost(&s.shapes);
+        n += 512 + shapes_cost(&s.shapes) + slide_extra_cost(s);
     }
     for m in doc.masters.iter().filter(|m| !masters.contains(&Arc::as_ptr(m))) {
         n += 4096 + shapes_cost(&m.shapes) + m.layouts.iter().map(|l| 512 + shapes_cost(&l.shapes)).sum::<usize>();

crates/engine/src/tests.rs

@@ -3338,6 +3338,25 @@ fn undo_history_is_capped_by_memory_as_well_as_steps() {
     assert_eq!(s.doc().unwrap().history.undo.len(), 120);
 }
 
+/// A slide's notes count toward the undo budget: small edits to a slide with large notes keep a
+/// copy of those notes per step, and must not slip past the budget as if the slide were empty.
+#[test]
+fn undo_history_counts_slide_notes() {
+    let mut s = Session::with_new();
+    let notes = "n".repeat(1 << 20);
+    s.edit(|doc, _| {
+        std::sync::Arc::make_mut(&mut doc.slides[0]).notes = deckcraft_model::text::TextBody::from_text(&notes);
+        Ok(())
+    })
+    .unwrap();
+    s.doc_mut().unwrap().history = History { limit: 500, byte_limit: 8 << 20, ..Default::default() };
+    for i in 0..40 {
+        s.execute("slide.rename", &json!({"name": format!("v{i}"), "index": 0})).unwrap();
+    }
+    let n = s.doc().unwrap().history.undo.len();
+    assert!((1..=8).contains(&n), "{n} steps of 1 MB notes kept under an 8 MB budget");
+}
+
 #[test]
 fn picture_styles_replace_each_other_in_one_step() {
     const PNG: &[u8] = &[

crates/metafile/src/cliprect.rs

@@ -11,10 +11,6 @@ use crate::geom::{Op, P};
 /// Line segments per Bézier when a crossing shape is flattened.
 const CURVE_STEPS: usize = 16;
 
-/// Point-against-edge tests one clipped item may cost (all its subpaths together); an item over
-/// it is dropped, as a single polygon over it already was.
-const MAX_CLIP_WORK: usize = 1_000_000;
-
 /// The bounding box `[l, t, r, b]` of the points in `ops` (control points included).
 pub(crate) fn ops_bbox(ops: &[Op]) -> Option<[f64; 4]> {
     let mut b: Option<[f64; 4]> = None;
@@ -104,20 +100,14 @@ fn is_rect(clip: &[P]) -> bool {
 
 /// Sutherland–Hodgman: `poly` cut by convex `clip`.
 pub(crate) fn clip_polygon(poly: &[P], clip: &[P]) -> Vec<P> {
-    clip_polygon_within(poly, clip, &mut { MAX_CLIP_WORK }).unwrap_or_default()
-}
-
-/// [`clip_polygon`] charging each point visited against each clip edge to `budget`: the polygon
-/// can gain a vertex at every edge, so its cost isn't known from its input size alone. `None`
-/// once the budget is spent.
-fn clip_polygon_within(poly: &[P], clip: &[P], budget: &mut usize) -> Option<Vec<P>> {
+    if poly.len().saturating_mul(clip.len()) > 1_000_000 {
+        return Vec::new();
+    }
     let mut poly = poly.to_vec();
     for i in 0..clip.len() {
-        // The work actually done: this edge visits the polygon as it now stands.
-        if poly.len() > *budget {
-            return None;
+        if poly.len().saturating_mul(clip.len() - i) > 1_000_000 {
+            return Vec::new();
         }
-        *budget -= poly.len();
         let (a, b) = (clip[i], clip[(i + 1) % clip.len()]);
         let keep = |p: P| cross(a, b, p) >= 0.0;
         let cut = |p: P, q: P| {
@@ -143,7 +133,7 @@ fn clip_polygon_within(poly: &[P], clip: &[P], budget: &mut usize) -> Option<Vec
             break;
         }
     }
-    Some(poly)
+    poly
 }
 
 fn disjoint(b: [f64; 4], c: [f64; 4]) -> bool {
@@ -168,12 +158,9 @@ pub(crate) fn clip_raw(raw: Raw, clip: &[P]) -> Vec<Raw> {
                 return Vec::new();
             }
             let subs = flatten(&ops);
-            if too_much_work(&subs, clip, usize::from(fill.is_some()) + usize::from(stroke.is_some())) {
-                return Vec::new();
-            }
             let mut out = Vec::new();
             if fill.is_some() {
-                let Some(f) = clip_fill(&subs, clip) else { return Vec::new() };
+                let f = clip_fill(&subs, clip);
                 if !f.is_empty() {
                     out.push(Raw::Path { ops: f, fill, stroke: None, even_odd, region: region.clone() });
                 }
@@ -194,11 +181,7 @@ pub(crate) fn clip_raw(raw: Raw, clip: &[P]) -> Vec<Raw> {
             if disjoint(b, cb) {
                 return Vec::new();
             }
-            let subs = flatten(&ops);
-            if too_much_work(&subs, clip, 1) {
-                return Vec::new();
-            }
-            let f = clip_fill(&subs, clip).unwrap_or_default();
+            let f = clip_fill(&flatten(&ops), clip);
             if f.is_empty() { Vec::new() } else { vec![Raw::Pattern { ops: f, even_odd, tile, region }] }
         }
         Raw::Bitmap { rect, img, region } => crop(rect, img, clip).map(|(rect, img)| vec![Raw::Bitmap { rect, img, region }]).unwrap_or_default(),
@@ -247,28 +230,17 @@ pub(crate) fn flatten(ops: &[Op]) -> Vec<(Vec<P>, bool)> {
     subs
 }
 
-/// Whether cutting every subpath of one item by `clip` would exceed [`MAX_CLIP_WORK`]; the
-/// per-polygon check alone lets many small subpaths each spend the whole allowance. `passes` is
-/// how many clipping passes the item gets (fill and stroke each cut every subpath), so one item
-/// shares a single allowance between them.
-fn too_much_work(subs: &[(Vec<P>, bool)], clip: &[P], passes: usize) -> bool {
-    subs.iter().map(|s| s.0.len() + 1).sum::<usize>().saturating_mul(clip.len()).saturating_mul(passes) > MAX_CLIP_WORK
-}
-
-/// The clipped fill of every subpath, its clipping work shared by all of them: `None` (the whole
-/// item goes) once [`MAX_CLIP_WORK`] is spent.
-fn clip_fill(subs: &[(Vec<P>, bool)], clip: &[P]) -> Option<Vec<Op>> {
+fn clip_fill(subs: &[(Vec<P>, bool)], clip: &[P]) -> Vec<Op> {
     let mut ops = Vec::new();
-    let mut budget = MAX_CLIP_WORK;
     for (pts, _) in subs {
-        let poly = clip_polygon_within(pts, clip, &mut budget)?;
+        let poly = clip_polygon(pts, clip);
         if poly.len() >= 3 {
             ops.push(Op::M(poly[0]));
             ops.extend(poly[1..].iter().map(|p| Op::L(*p)));
             ops.push(Op::Z);
         }
     }
-    Some(ops)
+    ops
 }
 
 fn clip_stroke(subs: &[(Vec<P>, bool)], clip: &[P]) -> Vec<Op> {

crates/metafile/src/tests.rs

@@ -2074,63 +2074,3 @@ fn clip_polygon_work_bound() {
     let clipped = crate::cliprect::clip_polygon(&poly, &clip);
     assert!(clipped.is_empty(), "excessive polygon clip product rejected");
 }
-
-#[test]
-fn clip_polygon_charges_actual_work_not_projected_growth() {
-    // Two convex 1,000-gons: the product is exactly the budget, and a cut that adds a vertex must
-    // not make a later step look over budget.
-    let gon = |cx: f64| -> Vec<(f64, f64)> {
-        (0..1000).map(|i| (i as f64 / 1000.0 * std::f64::consts::TAU).sin_cos()).map(|(s, c)| (cx + 100.0 * c, 100.0 * s)).collect()
-    };
-    let clipped = crate::cliprect::clip_polygon(&gon(0.0), &gon(30.0));
-    assert!(clipped.len() > 100, "the intersection is kept: {}", clipped.len());
-}
-
-#[test]
-fn clip_work_is_shared_by_fill_and_stroke() {
-    use crate::canvas::Raw;
-    use crate::geom::Op;
-    // One zigzag subpath crossing a 512-gon's edge: about 510k point-edge tests per pass, so a
-    // fill alone fits the allowance but a fill plus a stroke would spend it twice.
-    let clip: Vec<(f64, f64)> =
-        (0..512).map(|i| (i as f64 / 512.0 * std::f64::consts::TAU).sin_cos()).map(|(s, c)| (100.0 + 100.0 * c, 100.0 + 100.0 * s)).collect();
-    let mut ops = vec![Op::M((0.0, 95.0))];
-    ops.extend((1..1000).map(|i| Op::L((i as f64 * 0.19, if i % 2 == 0 { 95.0 } else { 300.0 }))));
-    ops.push(Op::Z);
-    let raw = |fill, stroke| Raw::Path { ops: ops.clone(), fill, stroke, even_odd: false, region: None };
-    let black = [0, 0, 0, 255];
-    assert!(!crate::cliprect::clip_raw(raw(Some(black), None), &clip).is_empty(), "a fill alone is within the allowance");
-    assert!(crate::cliprect::clip_raw(raw(Some(black), Some((black, 1.0))), &clip).is_empty(), "fill and stroke share one allowance");
-}
-
-#[test]
-fn clip_work_is_bounded_across_subpaths() {
-    use crate::canvas::Raw;
-    use crate::geom::Op;
-    // 512-gon clip; 4,000 small triangles crossing its edge: each under the per-polygon cap, together
-    // far over it, for both the fill and the stroke.
-    let clip: Vec<(f64, f64)> =
-        (0..512).map(|i| (i as f64 / 512.0 * std::f64::consts::TAU).sin_cos()).map(|(s, c)| (100.0 + 100.0 * c, 100.0 + 100.0 * s)).collect();
-    let tri = |x: f64| [Op::M((x, 95.0)), Op::L((x + 1.0, 95.0)), Op::L((x, 300.0)), Op::Z];
-    let many: Vec<Op> = (0..4000).flat_map(|i| tri(i as f64 % 200.0)).collect();
-    let raw = |ops| Raw::Path { ops, fill: Some([0, 0, 0, 255]), stroke: Some(([0, 0, 0, 255], 1.0)), even_odd: false, region: None };
-    assert!(crate::cliprect::clip_raw(raw(many), &clip).is_empty(), "the whole item's clipping work is bounded");
-    assert_eq!(crate::cliprect::clip_raw(raw(tri(100.0).to_vec()), &clip).len(), 2, "a small crossing shape is still clipped");
-}
-
-/// Clipping a fill can add a vertex at every clip edge, so a polygon's work isn't its input size:
-/// shapes that pass the up-front estimate but grow while clipped are charged as they grow, against
-/// one budget for the whole item.
-#[test]
-fn growing_fill_clips_share_one_budget() {
-    use crate::canvas::Raw;
-    use crate::geom::Op;
-    let clip: Vec<(f64, f64)> =
-        (0..512).map(|i| (i as f64 / 512.0 * std::f64::consts::TAU).sin_cos()).map(|(s, c)| (100.0 + 100.0 * c, 100.0 + 100.0 * s)).collect();
-    // A triangle around the whole circle: clipped, it becomes the 512-gon, about 130,000 steps.
-    let tri = [Op::M((-1000.0, -1000.0)), Op::L((1000.0, -1000.0)), Op::L((100.0, 1000.0)), Op::Z];
-    let raw = |n: usize| Raw::Path { ops: tri.repeat(n), fill: Some([0, 0, 0, 255]), stroke: None, even_odd: false, region: None };
-    // 400 of them pass the estimate (400 x 4 x 512 is under a million) but cost about 52 million.
-    assert!(crate::cliprect::clip_raw(raw(400), &clip).is_empty(), "the growing work is charged");
-    assert_eq!(crate::cliprect::clip_raw(raw(2), &clip).len(), 1, "a couple of them are still clipped");
-}