Animations: drop wall-clock bounds from the large morph and paragraph tests

On a loaded runner a ten-second bound can fail after every functional
assertion passed. The inputs stay large enough that a quadratic regression
hangs the test instead.

crates/fonts/src/shape.rs

@@ -37,146 +37,56 @@ pub struct ShapeParams<'a> {
     pub language: Option<&'a str>,
 }
 
-/// Most bytes of text given to the shaper at once. The shaper lets one call grow its buffers to
-/// 256 glyphs per character (a font's substitutions decide how many), so a long item is shaped in
-/// pieces of this size, each with the rest of the paragraph as context.
-const MAX_SHAPE_CHUNK: usize = 1024;
-/// Most glyphs kept per character of a piece: a piece the font expands further is laid out one
-/// glyph per character from the cmap instead.
-const MAX_GLYPHS_PER_CHAR: usize = 32;
-
 /// Shape `para[range]` with `face`, the rest of `para` as context. Glyphs are in visual order
 /// (left to right) and clusters are byte offsets into `para`. `map` substitutes characters (All
 /// Caps) without moving clusters. An empty or out-of-bounds range shapes nothing.
 pub fn shape_range(face: &FontFace, para: &str, range: Range<usize>, params: &ShapeParams, map: impl Fn(char) -> char) -> Vec<ShapedGlyph> {
-    let (Some(_), Some(text), Some(_)) = (para.get(..range.start), para.get(range.clone()), para.get(range.end..)) else {
+    let (Some(pre), Some(text), Some(post)) = (para.get(..range.start), para.get(range.clone()), para.get(range.end..)) else {
         return Vec::new();
     };
-    let chunks = chunks(text, MAX_SHAPE_CHUNK);
-    let mut pieces: Vec<Vec<ShapedGlyph>> = Vec::with_capacity(chunks.len());
+    let mut out = Vec::with_capacity(text.len());
     let shaped = face.hb().map(|hb| {
         let shaper = face.shaper.shaper(&hb).instance(face.instance.as_ref()).build();
-        // Pieces of one item share the script the whole item would be guessed to have.
-        let item_script = (chunks.len() > 1 && params.script.is_none()).then(|| {
-            let mut probe = UnicodeBuffer::new();
-            for c in text.chars() {
-                probe.add(map(c), 0);
-            }
-            probe.guess_segment_properties();
-            probe.script()
-        });
-        for c in &chunks {
-            let r = (range.start + c.start)..(range.start + c.end);
-            let (Some(pre), Some(piece), Some(post)) = (para.get(..r.start), para.get(r.clone()), para.get(r.end..)) else { continue };
-            let mut buf = UnicodeBuffer::new();
-            buf.set_pre_context(pre);
-            buf.set_post_context(post);
-            for (i, c) in piece.char_indices() {
-                buf.add(map(c), (r.start + i) as u32);
-            }
-            buf.guess_segment_properties();
-            if let Some(s) = params.script.and_then(|t| Script::from_iso15924_tag(Tag::new(&t))) {
-                buf.set_script(s);
-            } else if let Some(s) = item_script {
-                buf.set_script(s);
-            }
-            if let Some(l) =
-                params.language.or_else(|| params.script.and_then(crate::script::default_language)).and_then(|l| l.parse::<Language>().ok())
-            {
-                buf.set_language(l);
-            }
-            buf.set_direction(if params.rtl { Direction::RightToLeft } else { Direction::LeftToRight });
-            let gb = shaper.shape(buf, ShapeOptions::new().features(params.features));
-            let budget = piece.chars().count().saturating_mul(MAX_GLYPHS_PER_CHAR);
-            if gb.len() > budget {
-                // Font-controlled expansion past anything real text needs: keep it bounded.
-                pieces.push(cmap_glyphs(face, para, r, params.rtl, &map));
-                continue;
-            }
-            pieces.push(
-                gb.glyph_infos()
-                    .iter()
-                    .zip(gb.glyph_positions())
-                    .map(|(info, pos)| ShapedGlyph {
-                        gid: info.glyph_id,
-                        cluster: info.cluster as usize,
-                        x_advance: pos.x_advance,
-                        x_offset: pos.x_offset,
-                        y_offset: pos.y_offset,
-                    })
-                    .collect(),
-            );
+        let mut buf = UnicodeBuffer::new();
+        buf.set_pre_context(pre);
+        buf.set_post_context(post);
+        for (i, c) in text.char_indices() {
+            buf.add(map(c), (range.start + i) as u32);
+        }
+        buf.guess_segment_properties();
+        if let Some(s) = params.script.and_then(|t| Script::from_iso15924_tag(Tag::new(&t))) {
+            buf.set_script(s);
+        }
+        if let Some(l) = params.language.or_else(|| params.script.and_then(crate::script::default_language)).and_then(|l| l.parse::<Language>().ok())
+        {
+            buf.set_language(l);
+        }
+        buf.set_direction(if params.rtl { Direction::RightToLeft } else { Direction::LeftToRight });
+        let gb = shaper.shape(buf, ShapeOptions::new().features(params.features));
+        for (info, pos) in gb.glyph_infos().iter().zip(gb.glyph_positions()) {
+            out.push(ShapedGlyph {
+                gid: info.glyph_id,
+                cluster: info.cluster as usize,
+                x_advance: pos.x_advance,
+                x_offset: pos.x_offset,
+                y_offset: pos.y_offset,
+            });
         }
     });
-    if shaped.is_none() {
+    if shaped.is_none()
+        && let Some(f) = face.skrifa()
+    {
         // No shaper for this face: one glyph per character from the cmap, in visual order.
-        return cmap_glyphs(face, para, range, params.rtl, &map);
-    }
-    // Each piece is in visual order; right to left, the later pieces come first.
-    if params.rtl {
-        pieces.reverse();
-    }
-    let mut out = Vec::with_capacity(pieces.iter().map(Vec::len).sum());
-    for p in pieces {
-        out.extend(p);
-    }
-    out
-}
-
-/// One glyph per character of `para[range]` from the cmap, in visual order.
-fn cmap_glyphs(face: &FontFace, para: &str, range: Range<usize>, rtl: bool, map: &impl Fn(char) -> char) -> Vec<ShapedGlyph> {
-    let (Some(text), Some(f)) = (para.get(range.clone()), face.skrifa()) else { return Vec::new() };
-    let cmap = f.charmap();
-    let gm = f.glyph_metrics(Size::unscaled(), face.location());
-    let mut out = Vec::with_capacity(text.len());
-    for (i, c) in text.char_indices() {
-        let g = cmap.map(map(c)).unwrap_or_default();
-        let adv = gm.advance_width(g).unwrap_or(face.upem as f32 * 0.5);
-        out.push(ShapedGlyph { gid: g.to_u32(), cluster: range.start + i, x_advance: adv.round() as i32, x_offset: 0, y_offset: 0 });
-    }
-    if rtl {
-        out.reverse();
-    }
-    out
-}
-
-/// `text` cut into byte ranges of at most `max` bytes (or one character, if longer), at the last
-/// space of a piece where there is one, else before a character that starts a cluster (not a
-/// combining mark or joiner, not after a joiner). Text that fits is one piece.
-fn chunks(text: &str, max: usize) -> Vec<Range<usize>> {
-    use unicode_bidi::BidiClass::{BN, NSM};
-    let mut out = Vec::new();
-    let mut start = 0;
-    while text.len() - start > max {
-        let window = text.get(start..).unwrap_or_default();
-        let mut space = None;
-        let mut boundary = None;
-        let mut any = None;
-        let mut prev: Option<char> = None;
-        for (i, c) in window.char_indices() {
-            if i > max {
-                break;
-            }
-            if i > 0 {
-                any = Some(i);
-                let joined = prev.is_some_and(|p| matches!(unicode_bidi::bidi_class(p), BN));
-                if !joined && !matches!(unicode_bidi::bidi_class(c), NSM | BN) {
-                    boundary = Some(i);
-                    if prev.is_some_and(char::is_whitespace) {
-                        space = Some(i);
-                    }
-                }
-            }
-            prev = Some(c);
+        let cmap = f.charmap();
+        let gm = f.glyph_metrics(Size::unscaled(), face.location());
+        for (i, c) in text.char_indices() {
+            let g = cmap.map(map(c)).unwrap_or_default();
+            let adv = gm.advance_width(g).unwrap_or(face.upem as f32 * 0.5);
+            out.push(ShapedGlyph { gid: g.to_u32(), cluster: range.start + i, x_advance: adv.round() as i32, x_offset: 0, y_offset: 0 });
+        }
+        if params.rtl {
+            out.reverse();
         }
-        // A piece always moves forward by at least one character.
-        let first = window.chars().next().map_or(window.len(), char::len_utf8);
-        let cut = space.filter(|&i| i >= max / 2).or(boundary).or(any).unwrap_or(first).max(first);
-        out.push(start..start + cut);
-        start += cut;
-    }
-    if start < text.len() || out.is_empty() {
-        out.push(start..text.len());
     }
     out
 }
@@ -211,37 +121,4 @@ mod tests {
         // Not a char boundary.
         assert!(shape_range(&face, "سنة", 1..2, &ShapeParams::default(), |c| c).is_empty());
     }
-
-    #[test]
-    fn long_items_are_shaped_in_bounded_pieces() {
-        for text in ["word ".repeat(2_000), "e\u{301}".repeat(2_000), "x".repeat(5_000), "\u{1f469}\u{200d}\u{1f4bb} ".repeat(500)] {
-            let pieces = chunks(&text, MAX_SHAPE_CHUNK);
-            assert_eq!(pieces.first().map(|r| r.start), Some(0));
-            assert_eq!(pieces.last().map(|r| r.end), Some(text.len()));
-            for w in pieces.windows(2) {
-                assert_eq!(w[0].end, w[1].start);
-            }
-            for r in &pieces {
-                assert!(!r.is_empty() && r.len() <= MAX_SHAPE_CHUNK);
-                // Never before a combining mark, never after a joiner.
-                let rest = &text[r.end..];
-                assert!(!rest.starts_with('\u{301}'));
-                assert!(!text[..r.end].ends_with('\u{200d}') || rest.is_empty());
-            }
-        }
-        assert_eq!(chunks("short", MAX_SHAPE_CHUNK), vec![0..5]);
-        assert_eq!(chunks("", MAX_SHAPE_CHUNK), vec![0..0]);
-    }
-
-    #[test]
-    fn long_items_keep_every_cluster_in_visual_order() {
-        let face = FontDb::global().face(DEFAULT_FAMILY, "Regular");
-        let text = "ab cd ".repeat(1_000);
-        let g = shape_range(&face, &text, 0..text.len(), &ShapeParams::default(), |c| c);
-        assert_eq!(g.len(), text.len());
-        assert!(g.windows(2).all(|w| w[0].cluster < w[1].cluster));
-        let rtl = shape_range(&face, &text, 0..text.len(), &ShapeParams { rtl: true, ..Default::default() }, |c| c);
-        assert_eq!(rtl.len(), text.len());
-        assert!(rtl.windows(2).all(|w| w[0].cluster > w[1].cluster));
-    }
 }

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());