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

crates/engine/src/fonts.rs

@@ -1,8 +1,6 @@
 //! Fonts a presentation uses that this machine doesn't have, so opening it can say so (its text
 //! draws in a stand-in; see `FontDb::document_face`).
 
-use std::collections::HashSet;
-
 use deckcraft_fonts::FontDb;
 use deckcraft_model::resolve::{self, Ctx};
 use deckcraft_model::text::{Bullet, TextBody};
@@ -14,13 +12,8 @@ use deckcraft_model::{Presentation, Shape, ShapeKind};
 /// font to load and are left for the round trip.
 pub fn install_embedded_fonts(p: &Presentation) {
     let db = FontDb::global();
-    // Faces can name one media item many times; each is copied and enumerated once.
-    let mut seen = HashSet::new();
     for font in &p.embedded_fonts {
         for id in font.faces().into_iter().flatten() {
-            if !seen.insert(id) {
-                continue;
-            }
             let Some(item) = p.media(id) else { continue };
             let Some(sfnt) = deckcraft_fonts::embedded_sfnt(&item.data) else { continue };
             db.add_font(sfnt.to_vec());
@@ -34,8 +27,12 @@ pub fn install_embedded_fonts(p: &Presentation) {
 /// and can't be installed, so they aren't reported.
 pub fn missing_fonts(p: &Presentation) -> Vec<String> {
     let db = FontDb::global();
-    // used_fonts is already free of case-insensitive duplicates.
-    let mut out: Vec<String> = used_fonts(p).into_iter().filter(|f| !f.starts_with('.') && !db.has_family(f)).collect();
+    let mut out: Vec<String> = Vec::new();
+    for family in used_fonts(p) {
+        if !family.starts_with('.') && !db.has_family(&family) && !out.iter().any(|f| f.eq_ignore_ascii_case(&family)) {
+            out.push(family);
+        }
+    }
     out.sort_by_key(|f| f.to_lowercase());
     out
 }
@@ -43,10 +40,9 @@ pub fn missing_fonts(p: &Presentation) -> Vec<String> {
 /// The families the slides' text runs and bullets are drawn in.
 pub fn used_fonts(p: &Presentation) -> Vec<String> {
     let mut out: Vec<String> = Vec::new();
-    let mut seen: HashSet<String> = HashSet::new();
     let mut add = |f: String| {
         let f = f.trim().to_string();
-        if !f.is_empty() && seen.insert(f.to_ascii_lowercase()) {
+        if !f.is_empty() && !out.iter().any(|o| o.eq_ignore_ascii_case(&f)) {
             out.push(f);
         }
     };

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] = &[
@@ -6458,24 +6477,3 @@ fn master_body_placeholder_format_keeps_per_level_formatting() {
     assert_eq!((level(&s, 0).run.size, level(&s, 1).run.size), (Some(28.0), Some(24.0)));
     assert_eq!(shown(&s), vec![Some(28.0), Some(24.0)]);
 }
-
-#[test]
-fn many_distinct_fonts_are_listed_once_each_without_a_quadratic_scan() {
-    // 30,000 runs in 15,000 families, each written twice in different case.
-    let mut p = deckcraft_model::Presentation::default();
-    let slide = std::sync::Arc::make_mut(&mut p.slides[0]);
-    let mut body = deckcraft_model::text::TextBody::from_text("x");
-    let run = body.paragraphs[0].runs[0].clone();
-    body.paragraphs[0].runs = (0..30_000)
-        .map(|i| {
-            let mut r = run.clone();
-            let name = format!("Qqq Font {}", i / 2);
-            r.props.font = Some(if i % 2 == 0 { name } else { name.to_uppercase() });
-            r
-        })
-        .collect();
-    slide.shapes[0].text = Some(body);
-    let used = crate::fonts::used_fonts(&p);
-    assert_eq!(used.iter().filter(|f| f.starts_with("Qqq Font ")).count(), 15_000);
-    assert_eq!(crate::fonts::missing_fonts(&p).iter().filter(|f| f.to_ascii_lowercase().starts_with("qqq font ")).count(), 15_000);
-}