address review: skip the tab-stop lookup for a NaN position

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

@@ -677,16 +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 _ = 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/text/src/breaking.rs

@@ -260,12 +260,15 @@ pub(crate) struct Tab {
     align: TabAlign,
 }
 
+/// The paragraph's tab stops in position order (a stop at no finite position is left out), so
+/// [`tab_advance`] finds the next one by binary search instead of a scan per tab.
 pub(crate) fn tabs_of(props: &ParaProps) -> Vec<Tab> {
-    props
+    let mut tabs: Vec<Tab> = props
         .tabs
         .as_ref()
         .map(|t| {
             t.iter()
+                .filter(|s| s.pos.is_finite())
                 .map(|s| Tab {
                     pos: s.pos,
                     align: match s.align.as_str() {
@@ -277,7 +280,9 @@ pub(crate) fn tabs_of(props: &ParaProps) -> Vec<Tab> {
                 })
                 .collect()
         })
-        .unwrap_or_default()
+        .unwrap_or_default();
+    tabs.sort_by(|a, b| a.pos.total_cmp(&b.pos));
+    tabs
 }
 
 /// Width of the text after the tab at cell `i` up to the next tab or line break, and the width of
@@ -301,18 +306,30 @@ pub(crate) fn tab_segment(cells: &[Cell], i: usize) -> (f64, f64) {
 /// (`seg`: its width, and its width before the decimal point) as the stop aligns it but never
 /// moving back, as PowerPoint does; past the last stop, to the next default stop.
 pub(crate) fn tab_advance(x: f64, stops: &[Tab], default_tab: f64, seg: (f64, f64)) -> f64 {
-    for s in stops {
-        if s.pos > x + 0.5 {
-            let to = s.pos - x;
-            return match s.align {
-                TabAlign::Left => to,
-                TabAlign::Right => (to - seg.0).max(0.0),
-                TabAlign::Center => (to - seg.0 / 2.0).max(0.0),
-                TabAlign::Decimal => (to - seg.1).max(0.0),
-            };
-        }
+    // `stops` is in position order (`tabs_of`): the next stop is the first past x.
+    // A NaN `x` (a NaN spacing upstream) skips the lookup and takes the finite default-stop path.
+    if let Some(s) = stops.get(stops.partition_point(|s| s.pos <= x + 0.5)).filter(|_| !x.is_nan()) {
+        let to = s.pos - x;
+        return match s.align {
+            TabAlign::Left => to,
+            TabAlign::Right => (to - seg.0).max(0.0),
+            TabAlign::Center => (to - seg.0 / 2.0).max(0.0),
+            TabAlign::Decimal => (to - seg.1).max(0.0),
+        };
     }
     let d = if default_tab > 1.0 { default_tab } else { 72.0 };
     let next = ((x / d).floor() + 1.0) * d;
     (next - x).max(1.0)
 }
+
+#[cfg(test)]
+mod tab_tests {
+    use super::*;
+
+    #[test]
+    fn a_nan_position_takes_the_finite_default_stop() {
+        let stops = [Tab { pos: 100.0, align: TabAlign::Left }];
+        assert!(tab_advance(f64::NAN, &stops, 72.0, (0.0, 0.0)).is_finite());
+        assert_eq!(tab_advance(10.0, &stops, 72.0, (0.0, 0.0)), 90.0);
+    }
+}

crates/text/src/tests.rs

@@ -1416,6 +1416,16 @@ fn tab_stops_align_the_text_after_them() {
     assert!((l.lines[0].edges[4].0 - x0(&l) - 72.0).abs() < 0.01);
 }
 
+#[test]
+fn many_tabs_over_many_stops_lay_out_by_position() {
+    // 20,000 stops given out of order, and 2,000 tabs: each tab finds the next stop by position.
+    let mut stops: Vec<(f64, &str)> = (1..=20_000).rev().map(|k| (k as f64 * 0.05 + 30.0, "l")).collect();
+    stops.push((f64::NAN, "l"));
+    let text = "a\t".repeat(2_000);
+    let l = tabbed(&text, &stops, 1.0e6);
+    assert!((l.lines[0].edges[2].0 - l.inner.x0 - 30.05).abs() < 0.01, "{:?}", &l.lines[0].edges[..3]);
+}
+
 #[test]
 fn a_tab_past_the_end_moves_to_the_next_line() {
     // PowerPoint: "abcdefghij\tk\tl\tm…" in a narrow box puts the tab that doesn't fit on the next