address review: use a narrow tall page in the bounded-work test

crates/pdf/src/lib.rs

@@ -498,6 +498,9 @@ 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;
@@ -573,20 +576,17 @@ 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 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();
+            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();
             if text.trim().is_empty() {
                 continue;
             }
@@ -666,19 +666,28 @@ 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 cand = if line.is_empty() { word.to_string() } else { format!("{line} {word}") };
-                if width(&cand) > area.width() && !line.is_empty() {
+                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() {
                     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 {
-                    line = cand;
+                    if !line.is_empty() {
+                        line.push(' ');
+                    }
+                    line.push_str(word);
+                    line_w = cand_w;
                 }
             }
             if y > area.y1 {
@@ -695,20 +704,26 @@ impl Exporter<'_> {
 
     /// Ruled lines for handwritten notes (3-per-page handouts).
     fn note_lines(&self, s: &mut Surface, r: Rect) {
-        let mut y = r.y0 + 18.0;
+        // 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;
+        }
         s.set_stroke(Some(krilla::paint::Stroke {
             paint: krilla::color::rgb::Color::new(0xB0, 0xB0, 0xB0).into(),
             width: 0.5,
             ..Default::default()
         }));
-        while y <= r.y1 {
+        for i in 0..(count.min(MAX_NOTE_LINES as f64) as usize) {
+            let y = first + 22.0 * i as f64;
             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

@@ -677,3 +677,16 @@ 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 _ = export(&p, &PdfOptions { paper: (612.0, 1.0e19), ..opts(PageLayout::Handouts { per_page: 3 }) });
+    // 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-"));
+}

crates/pptx/src/read.rs

@@ -46,9 +46,6 @@ pub struct Imp<'a> {
     pub tag_bytes: usize,
     /// The tag budget ran out and the warning was given.
     pub tags_capped: bool,
-    /// Whether a part has relationships of its own, by normalized part name: looked up once
-    /// however many embedded objects point at the part.
-    pub has_rels: HashMap<String, bool>,
 }
 
 impl Imp<'_> {
@@ -449,7 +446,6 @@ pub fn import_theme(bytes: &[u8]) -> Result<(deckcraft_model::theme::Theme, Vec<
         tags_by_part: HashMap::new(),
         tag_bytes: 0,
         tags_capped: false,
-        has_rels: HashMap::new(),
     };
     let part = Part::new(&pkg, &theme_path, &doc);
     let theme = theme::read_theme(&mut imp, &part, &doc.root);
@@ -491,7 +487,6 @@ pub fn import_checked(bytes: &[u8]) -> Result<(Presentation, Vec<String>), PptxE
         tags_by_part: HashMap::new(),
         tag_bytes: 0,
         tags_capped: false,
-        has_rels: HashMap::new(),
     };
     // Slide ids first so hyperlinks anywhere can point at slides.
     let slide_ids = register_slides(&mut imp, &docs, &slide_entries, &mut lost_slides);

crates/pptx/src/read/dml.rs

@@ -627,22 +627,12 @@ fn push_text_runs(para: &mut Paragraph, props: RunProps, text: &str) {
 /// Appends an Office Math (a14:m) run: its plain text and its OMML.
 fn push_math_run(para: &mut Paragraph, part: &Part, e: &El) {
     if para.runs.len() < MAX_PARAGRAPH_RUNS {
-        // The text of the outermost `t` elements: a `t` nested in another is already part of
-        // the outer one's text, and taking it again would copy it once per nesting level.
-        fn outer_t(e: &El, depth: usize, text: &mut String) {
-            if depth > crate::xml::MAX_DEPTH {
-                return;
-            }
-            for c in e.elements() {
-                if c.is("t") {
-                    text.push_str(&c.text());
-                } else {
-                    outer_t(c, depth + 1, text);
-                }
-            }
-        }
         let mut text = String::new();
-        outer_t(e, 0, &mut text);
+        let mut ts = vec![];
+        e.find_all("t", &mut ts);
+        for t in ts {
+            text.push_str(&t.text());
+        }
         let omml = e.elements().map(|c| part.keep_inner(c)).collect::<String>();
         para.runs.push(Run { text, props: RunProps::default(), kind: RunKind::Math { omml } });
     }
@@ -687,17 +677,6 @@ mod tests {
     use super::*;
     use crate::xml::parse;
 
-    #[test]
-    fn nested_math_text_is_taken_once() {
-        // 200 `m:t` nested in each other around one word: the word, not 200 copies of it.
-        let x = format!(r#"<a14:m xmlns:a14="a" xmlns:m="m"><m:oMath>{}word{}</m:oMath></a14:m>"#, "<m:t>".repeat(200), "</m:t>".repeat(200));
-        let d = parse(x.as_bytes()).unwrap();
-        let part = Part { name: "ppt/slides/slide1.xml".into(), rels: Default::default(), ns: vec![], inner_ns: vec![] };
-        let mut para = Paragraph::default();
-        push_math_run(&mut para, &part, &d.root);
-        assert_eq!(para.runs[0].text, "word");
-    }
-
     #[test]
     fn colors_and_transforms() {
         let d = parse(

crates/pptx/src/read/shapes.rs

@@ -475,8 +475,7 @@ fn frame(imp: &mut Imp, part: &Part, ctx: &mut IdCtx, e: &El, depth: usize) -> S
         .into_iter()
         .map(|id| {
             let rel = part.rels.get(&id).filter(|r| !r.external)?;
-            let pkg = imp.pkg;
-            if *imp.has_rels.entry(crate::opc::norm(&rel.target)).or_insert_with(|| !pkg.rels(&rel.target).items.is_empty()) {
+            if !imp.pkg.rels(&rel.target).items.is_empty() {
                 return None;
             }
             let (kind, target) = (rel.ty.clone(), rel.target.clone());