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/pdf/src/lib.rs

@@ -498,9 +498,6 @@ impl Backdrops {
 /// Most pixels one slide raster may have; it is RGBA, and the backdrop and the content are two rasters.
 const MAX_RASTER_PIXELS: f64 = 64_000_000.0;
 
-/// Most ruled note lines one handout area gets (a letter page holds about 30).
-const MAX_NOTE_LINES: usize = 500;
-
 /// Render scale for a slide of `slide` units drawn `box_width` points wide at `dpi`, with the total pixel count capped.
 fn raster_scale(box_width: f64, slide: (f64, f64), dpi: f64) -> f64 {
     let scale = (box_width / 72.0 * dpi.clamp(36.0, 600.0)) / slide.0;
@@ -576,17 +573,20 @@ impl Exporter<'_> {
 
     /// Invisible glyphs exactly over the rasterised text.
     fn text_layer(&mut self, s: &mut Surface, pt: &PlacedText, view: Affine, visible: bool) -> Result<()> {
-        // Each paragraph's characters, gathered once: a long paragraph wraps into many runs.
-        let mut para_chars: std::collections::HashMap<usize, Vec<char>> = std::collections::HashMap::new();
         for run in &pt.layout.runs {
             if run.glyphs.is_empty() || run.size <= 0.0 {
                 continue;
             }
-            let chars = para_chars.entry(run.para).or_insert_with(|| {
-                pt.body.paragraphs.get(run.para).map(|p| p.runs.iter().flat_map(|r| r.text.chars()).collect()).unwrap_or_default()
-            });
-            let (a, b) = (run.chars.0.min(chars.len()), run.chars.1.min(chars.len()));
-            let text: String = chars.get(a..b.max(a)).unwrap_or_default().iter().collect();
+            let text: String = pt
+                .body
+                .paragraphs
+                .get(run.para)
+                .map(|p| p.runs.iter().map(|r| r.text.as_str()).collect::<String>())
+                .unwrap_or_default()
+                .chars()
+                .skip(run.chars.0)
+                .take(run.chars.1.saturating_sub(run.chars.0))
+                .collect();
             if text.trim().is_empty() {
                 continue;
             }
@@ -666,28 +666,19 @@ impl Exporter<'_> {
         let lh = size * 1.3;
         let mut y = area.y0 + size;
         s.set_fill(Some(grey(0x20, 1.0)));
-        let space = width(" ");
         'outer: for para in text.split('\n') {
-            // The line's width is kept as words are added, not measured again for each word.
             let mut line = String::new();
-            let mut line_w = 0.0;
             for word in para.split(' ') {
-                let word_w = width(word);
-                let cand_w = if line.is_empty() { word_w } else { line_w + space + word_w };
-                if cand_w > area.width() && !line.is_empty() {
+                let cand = if line.is_empty() { word.to_string() } else { format!("{line} {word}") };
+                if width(&cand) > area.width() && !line.is_empty() {
                     if y > area.y1 {
                         break 'outer;
                     }
                     s.draw_text(Point::from_xy(area.x0 as f32, y as f32), font.clone(), size as f32, &line, false, TextDirection::Auto);
                     y += lh;
                     line = word.to_string();
-                    line_w = word_w;
                 } else {
-                    if !line.is_empty() {
-                        line.push(' ');
-                    }
-                    line.push_str(word);
-                    line_w = cand_w;
+                    line = cand;
                 }
             }
             if y > area.y1 {
@@ -704,26 +695,20 @@ impl Exporter<'_> {
 
     /// Ruled lines for handwritten notes (3-per-page handouts).
     fn note_lines(&self, s: &mut Surface, r: Rect) {
-        // Counted, not stepped: at huge coordinates adding 22 leaves y unchanged, and a huge page
-        // would want millions of rules.
-        let first = r.y0 + 18.0;
-        let count = ((r.y1 - first) / 22.0).floor() + 1.0;
-        if count.is_nan() || count < 1.0 {
-            return;
-        }
+        let mut y = r.y0 + 18.0;
         s.set_stroke(Some(krilla::paint::Stroke {
             paint: krilla::color::rgb::Color::new(0xB0, 0xB0, 0xB0).into(),
             width: 0.5,
             ..Default::default()
         }));
-        for i in 0..(count.min(MAX_NOTE_LINES as f64) as usize) {
-            let y = first + 22.0 * i as f64;
+        while y <= r.y1 {
             let mut p = kurbo::BezPath::new();
             p.move_to((r.x0, y));
             p.line_to((r.x1, y));
             if let Some(path) = to_path(&p) {
                 s.draw_path(&path);
             }
+            y += 22.0;
         }
         s.set_stroke(None);
     }

crates/pdf/src/tests.rs

@@ -677,17 +677,3 @@ fn raster_scale_caps_total_pixels_not_each_side() {
     let s = raster_scale(720.0, (960.0, 540.0), 200.0);
     assert!((s - 720.0 / 72.0 * 200.0 / 960.0).abs() < 1e-12);
 }
-
-#[cfg_attr(target_arch = "wasm32", wasm_bindgen_test::wasm_bindgen_test)]
-#[cfg_attr(not(target_arch = "wasm32"), test)]
-fn huge_paper_and_long_notes_export_in_bounded_work() {
-    // At 1e19 pt adding a rule's 22 pt leaves y unchanged: the rules are counted, not stepped. A narrow page keeps the raster small.
-    let p = deck(1);
-    let bytes = export(&p, &PdfOptions { paper: (612.0, 1.0e19), ..opts(PageLayout::Handouts { per_page: 3 }) }).expect("export");
-    assert!(bytes.starts_with(b"%PDF-"));
-    // 20,000 short words of notes, wrapped with the line width kept as words are added.
-    let mut p = deck(1);
-    Arc::make_mut(&mut p.slides[0]).notes = TextBody::from_text(&"ab ".repeat(20_000));
-    let bytes = export(&p, &opts(PageLayout::Notes)).expect("export");
-    assert!(bytes.starts_with(b"%PDF-"));
-}