Merge branch 'main' into table-subheaders

Cargo.lock

@@ -1259,6 +1259,12 @@ dependencies = [
  "serde",
 ]
 
+[[package]]
+name = "infer"
+version = "0.19.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "a588916bfdfd92e71cacef98a63d9b1f0d74d6599980d11894290e7ddefffcf7"
+
 [[package]]
 name = "inotify"
 version = "0.11.0"
@@ -3127,6 +3133,7 @@ dependencies = [
  "comemo",
  "ecow",
  "image",
+ "infer",
  "krilla",
  "krilla-svg",
  "serde",

Cargo.toml

@@ -71,6 +71,7 @@ icu_segmenter = { version = "1.4", features = ["serde"] }
 if_chain = "1"
 image = { version = "0.25.5", default-features = false, features = ["png", "jpeg", "gif"] }
 indexmap = { version = "2", features = ["serde"] }
+infer = { version = "0.19.0", default-features = false }
 kamadak-exif = "0.6"
 krilla = { version = "0.4.0", default-features = false, features = ["raster-images", "comemo", "rayon"] }
 krilla-svg = "0.1.0"

crates/typst-layout/src/inline/linebreak.rs

@@ -690,13 +690,34 @@ fn breakpoints(p: &Preparation, mut f: impl FnMut(usize, Breakpoint)) {
         let breakpoint = if point == text.len() {
             Breakpoint::Mandatory
         } else {
+            const OBJ_REPLACE: char = '\u{FFFC}';
             match lb.get(c) {
-                // Fix for: https://github.com/unicode-org/icu4x/issues/4146
-                LineBreak::Glue | LineBreak::WordJoiner | LineBreak::ZWJ => continue,
                 LineBreak::MandatoryBreak
                 | LineBreak::CarriageReturn
                 | LineBreak::LineFeed
                 | LineBreak::NextLine => Breakpoint::Mandatory,
+
+                // https://github.com/typst/typst/issues/5489
+                //
+                // OBJECT-REPLACEMENT-CHARACTERs provide Contingent Break
+                // opportunities before and after by default. This behaviour
+                // is however tailorable, see:
+                // https://www.unicode.org/reports/tr14/#CB
+                // https://www.unicode.org/reports/tr14/#TailorableBreakingRules
+                // https://www.unicode.org/reports/tr14/#LB20
+                //
+                // Don't provide a line breaking opportunity between a LTR-
+                // ISOLATE (or any other Combining Mark) and an OBJECT-
+                // REPLACEMENT-CHARACTER representing an inline item, if the
+                // LTR-ISOLATE could end up as the only character on the
+                // previous line.
+                LineBreak::CombiningMark
+                    if text[point..].starts_with(OBJ_REPLACE)
+                        && last + c.len_utf8() == point =>
+                {
+                    continue;
+                }
+
                 _ => Breakpoint::Normal,
             }
         };

crates/typst-layout/src/inline/shaping.rs

@@ -20,7 +20,7 @@ use unicode_bidi::{BidiInfo, Level as BidiLevel};
 use unicode_script::{Script, UnicodeScript};
 
 use super::{decorate, Item, Range, SpanMapper};
-use crate::modifiers::{FrameModifiers, FrameModify};
+use crate::modifiers::FrameModifyText;
 
 /// The result of shaping text.
 ///
@@ -327,7 +327,7 @@ impl<'a> ShapedText<'a> {
             offset += width;
         }
 
-        frame.modify(&FrameModifiers::get_in(self.styles));
+        frame.modify_text(self.styles);
         frame
     }
 

crates/typst-layout/src/modifiers.rs

@@ -1,6 +1,6 @@
 use typst_library::foundations::StyleChain;
-use typst_library::layout::{Fragment, Frame, FrameItem, HideElem, Point};
-use typst_library::model::{Destination, LinkElem};
+use typst_library::layout::{Abs, Fragment, Frame, FrameItem, HideElem, Point, Sides};
+use typst_library::model::{Destination, LinkElem, ParElem};
 
 /// Frame-level modifications resulting from styles that do not impose any
 /// layout structure.
@@ -52,14 +52,7 @@ pub trait FrameModify {
 
 impl FrameModify for Frame {
     fn modify(&mut self, modifiers: &FrameModifiers) {
-        if let Some(dest) = &modifiers.dest {
-            let size = self.size();
-            self.push(Point::zero(), FrameItem::Link(dest.clone(), size));
-        }
-
-        if modifiers.hidden {
-            self.hide();
-        }
+        modify_frame(self, modifiers, None);
     }
 }
 
@@ -82,6 +75,41 @@ where
     }
 }
 
+pub trait FrameModifyText {
+    /// Resolve and apply [`FrameModifiers`] for this text frame.
+    fn modify_text(&mut self, styles: StyleChain);
+}
+
+impl FrameModifyText for Frame {
+    fn modify_text(&mut self, styles: StyleChain) {
+        let modifiers = FrameModifiers::get_in(styles);
+        let expand_y = 0.5 * ParElem::leading_in(styles);
+        let outset = Sides::new(Abs::zero(), expand_y, Abs::zero(), expand_y);
+        modify_frame(self, &modifiers, Some(outset));
+    }
+}
+
+fn modify_frame(
+    frame: &mut Frame,
+    modifiers: &FrameModifiers,
+    link_box_outset: Option<Sides<Abs>>,
+) {
+    if let Some(dest) = &modifiers.dest {
+        let mut pos = Point::zero();
+        let mut size = frame.size();
+        if let Some(outset) = link_box_outset {
+            pos.y -= outset.top;
+            pos.x -= outset.left;
+            size += outset.sum_by_axis();
+        }
+        frame.push(pos, FrameItem::Link(dest.clone(), size));
+    }
+
+    if modifiers.hidden {
+        frame.hide();
+    }
+}
+
 /// Performs layout and modification in one step.
 ///
 /// This just runs `layout(styles).modified(&FrameModifiers::get_in(styles))`,

crates/typst-pdf/Cargo.toml

@@ -23,6 +23,7 @@ bytemuck = { workspace = true }
 comemo = { workspace = true }
 ecow = { workspace = true }
 image = { workspace = true }
+infer = { workspace = true }
 krilla = { workspace = true }
 krilla-svg = { workspace = true }
 serde = { workspace = true }

crates/typst-pdf/src/embed.rs

@@ -34,6 +34,8 @@ pub(crate) fn embed_files(
             },
         };
         let data: Arc<dyn AsRef<[u8]> + Send + Sync> = Arc::new(embed.data.clone());
+        // TODO: update when new krilla version lands (https://github.com/LaurenzV/krilla/pull/203)
+        let compress = should_compress(&embed.data).unwrap_or(true);
 
         let file = EmbeddedFile {
             path,
@@ -41,7 +43,7 @@ pub(crate) fn embed_files(
             description,
             association_kind,
             data: data.into(),
-            compress: true,
+            compress,
             location: Some(span.into_raw().get()),
         };
 
@@ -52,3 +54,69 @@ pub(crate) fn embed_files(
 
     Ok(())
 }
+
+fn should_compress(data: &[u8]) -> Option<bool> {
+    let ty = infer::get(data)?;
+    match ty.matcher_type() {
+        infer::MatcherType::App => None,
+        infer::MatcherType::Archive => match ty.mime_type() {
+            #[rustfmt::skip]
+            "application/zip"
+            | "application/vnd.rar"
+            | "application/gzip"
+            | "application/x-bzip2"
+            | "application/vnd.bzip3"
+            | "application/x-7z-compressed"
+            | "application/x-xz"
+            | "application/vnd.ms-cab-compressed"
+            | "application/vnd.debian.binary-package"
+            | "application/x-compress"
+            | "application/x-lzip"
+            | "application/x-rpm"
+            | "application/zstd"
+            | "application/x-lz4"
+            | "application/x-ole-storage" => Some(false),
+            _ => None,
+        },
+        infer::MatcherType::Audio => match ty.mime_type() {
+            #[rustfmt::skip]
+            "audio/mpeg"
+            | "audio/m4a"
+            | "audio/opus"
+            | "audio/ogg"
+            | "audio/x-flac"
+            | "audio/amr"
+            | "audio/aac"
+            | "audio/x-ape" => Some(false),
+            _ => None,
+        },
+        infer::MatcherType::Book => None,
+        infer::MatcherType::Doc => None,
+        infer::MatcherType::Font => None,
+        infer::MatcherType::Image => match ty.mime_type() {
+            #[rustfmt::skip]
+            "image/jpeg"
+            | "image/jp2"
+            | "image/png"
+            | "image/webp"
+            | "image/vnd.ms-photo"
+            | "image/heif"
+            | "image/avif"
+            | "image/jxl"
+            | "image/vnd.djvu" => None,
+            _ => None,
+        },
+        infer::MatcherType::Text => None,
+        infer::MatcherType::Video => match ty.mime_type() {
+            #[rustfmt::skip]
+            "video/mp4"
+            | "video/x-m4v"
+            | "video/x-matroska"
+            | "video/webm"
+            | "video/quicktime"
+            | "video/x-flv" => Some(false),
+            _ => None,
+        },
+        infer::MatcherType::Custom => None,
+    }
+}

crates/typst-realize/src/lib.rs

@@ -655,6 +655,7 @@ fn visit_grouping_rules<'a>(
     let matching = s.rules.iter().find(|&rule| (rule.trigger)(content, &s.kind));
 
     // Try to continue or finish an existing grouping.
+    let mut i = 0;
     while let Some(active) = s.groupings.last() {
         // Start a nested group if a rule with higher priority matches.
         if matching.is_some_and(|rule| rule.priority > active.rule.priority) {
@@ -670,6 +671,16 @@ fn visit_grouping_rules<'a>(
         }
 
         finish_innermost_grouping(s)?;
+        i += 1;
+        if i > 512 {
+            // It seems like this case is only hit when there is a cycle between
+            // a show rule and a grouping rule. The show rule produces content
+            // that is matched by a grouping rule, which is then again processed
+            // by the show rule, and so on. The two must be at an equilibrium,
+            // otherwise either the "maximum show rule depth" or "maximum
+            // grouping depth" errors are triggered.
+            bail!(content.span(), "maximum grouping depth exceeded");
+        }
     }
 
     // Start a new grouping.

tests/suite/layout/inline/linebreak.typ

@@ -139,3 +139,11 @@ Some texts feature many longer
 words. Those are often exceedingly
 challenging to break in a visually
 pleasing way.
+
+--- issue-5489-matrix-stray-linebreak ---
+#table(
+  columns: (70pt,) * 1,
+  align: horizon + center,
+  stroke: 0.6pt,
+  [$mat(2241/2210,-71/1105;-71/1105,147/1105)$],
+)

tests/suite/styling/show.typ

@@ -258,3 +258,11 @@ I am *strong*, I am _emphasized_, and I am #[special<special>].
 
 = Hello
 *strong*
+
+--- issue-5690-oom-par-box ---
+// Error: 3:6-5:1 maximum grouping depth exceeded
+#show par: box
+
+Hello
+
+World