Separate items for hyphens, fixing style of repeated hyphen (#6798)

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

@@ -18,6 +18,25 @@ const EN_DASH: char = '–';
 const EM_DASH: char = '—';
 const LINE_SEPARATOR: char = '\u{2028}'; // We use LS to distinguish justified breaks.
 
+// We use indices to remember the logical (as opposed to visual) order of items.
+// During line building, the items are stored in visual (BiDi-reordered) order.
+// When committing to a line and building its frame, we sort by logical index.
+//
+// - Special layout-generated items have custom indices that ensure correct
+//   ordering w.r.t. to each other and normal elements, listed below.
+// - Normal items have their position in `p.items` plus the number of special
+//   reserved prefix indices.
+//
+// Logical indices must be unique within a line because we use an unstable sort.
+const START_HYPHEN_IDX: usize = 0;
+const fn logical_item_idx(i: usize) -> usize {
+    // This won't overflow because the `idx` comes from a vector which is
+    // limited to `isize::MAX` elements.
+    i + 1
+}
+const FALLBACK_TEXT_IDX: usize = usize::MAX - 1;
+const END_HYPHEN_IDX: usize = usize::MAX;
+
 /// A layouted line, consisting of a sequence of layouted inline items that are
 /// mostly borrowed from the preparation phase. This type enables you to measure
 /// the size of a line in a range before committing to building the line's
@@ -128,7 +147,7 @@ pub fn line<'a>(
     p: &'a Preparation,
     range: Range,
     breakpoint: Breakpoint,
-    pred: Option<&Line>,
+    pred: Option<&Line<'a>>,
 ) -> Line<'a> {
     // The line's full text.
     let full = &p.text[range.clone()];
@@ -152,20 +171,26 @@ pub fn line<'a>(
     let trim = range.start + breakpoint.trim(full).len();
 
     // Collect the items for the line.
-    let mut items = collect_items(engine, p, range, trim);
+    let mut items = Items::new();
 
     // Add a hyphen at the line start, if a previous dash should be repeated.
-    if pred.is_some_and(|pred| should_repeat_hyphen(pred, full))
+    if let Some(pred) = pred
-        && let Some(shaped) = items.first_text_mut()
+        && pred.dash == Some(Dash::Hard)
+        && let Some(base) = pred.items.last_text()
+        && should_repeat_hyphen(base.lang, full)
+        && let Some(hyphen) = ShapedText::hyphen(engine, p.config.fallback, base, trim)
     {
-        shaped.prepend_hyphen(engine, p.config.fallback);
+        items.push(Item::Text(hyphen), START_HYPHEN_IDX);
     }
 
+    collect_items(&mut items, engine, p, range, trim);
+
     // Add a hyphen at the line end, if we ended on a soft hyphen.
     if dash == Some(Dash::Soft)
-        && let Some(shaped) = items.last_text_mut()
+        && let Some(base) = items.last_text()
+        && let Some(hyphen) = ShapedText::hyphen(engine, p.config.fallback, base, trim)
     {
-        shaped.push_hyphen(engine, p.config.fallback);
+        items.push(Item::Text(hyphen), END_HYPHEN_IDX);
     }
 
     // Ensure that there is no weak spacing at the start and end of the line.
@@ -189,18 +214,18 @@ pub fn line<'a>(
 /// We do not factor the `trim` directly into the `range` because we still want
 /// to keep non-text items after the trim (e.g. tags).
 fn collect_items<'a>(
+    items: &mut Items<'a>,
     engine: &Engine,
     p: &'a Preparation,
     range: Range,
     trim: usize,
-) -> Items<'a> {
+) {
-    let mut items = Items::new();
     let mut fallback = None;
 
     // Collect the items for each consecutively ordered run.
     reorder(p, range.clone(), |subrange, rtl| {
         let from = items.len();
-        collect_range(engine, p, subrange, trim, &mut items, &mut fallback);
+        collect_range(engine, p, subrange, trim, items, &mut fallback);
         if rtl {
             items.reorder(from);
         }
@@ -210,10 +235,8 @@ fn collect_items<'a>(
     if !items.iter().any(|item| matches!(item, Item::Text(_)))
         && let Some(fallback) = fallback
     {
-        items.push(fallback, usize::MAX);
+        items.push(fallback, FALLBACK_TEXT_IDX);
     }
-
-    items
 }
 
 /// Trims weak spacing from the start and end of the line.
@@ -277,7 +300,9 @@ fn collect_range<'a>(
     items: &mut Items<'a>,
     fallback: &mut Option<ItemEntry<'a>>,
 ) {
-    for (idx, (subrange, item)) in p.slice(range.clone()) {
+    for (i, (subrange, item)) in p.slice(range.clone()) {
+        let idx = logical_item_idx(i);
+
         // All non-text items are just kept, they can't be split.
         let Item::Text(shaped) = item else {
             items.push(item, idx);
@@ -382,19 +407,10 @@ fn adjust_cj_at_line_end(p: &Preparation, items: &mut Items) {
     }
 }
 
-/// Whether a hyphen should be inserted at the start of the next line.
+/// Whether a hyphen should be repeated at the start of the line in the given
-fn should_repeat_hyphen(pred_line: &Line, text: &str) -> bool {
+/// language, when the following text is the given one.
-    // If the predecessor line does not end with a `Dash::Hard`, we shall
+fn should_repeat_hyphen(lang: Lang, following_text: &str) -> bool {
-    // not place a hyphen at the start of the next line.
+    match lang {
-    if pred_line.dash != Some(Dash::Hard) {
-        return false;
-    }
-
-    // The hyphen should repeat only in the languages that require that feature.
-    // For more information see the discussion at https://github.com/typst/typst/issues/3235
-    let Some(Item::Text(shaped)) = pred_line.items.last() else { return false };
-
-    match shaped.lang {
         // - Lower Sorbian: see https://dolnoserbski.de/ortografija/psawidla/K3
         // - Czech: see https://prirucka.ujc.cas.cz/?id=164
         // - Croatian: see http://pravopis.hr/pravilo/spojnica/68/
@@ -413,7 +429,7 @@ fn should_repeat_hyphen(pred_line: &Line, text: &str) -> bool {
         //
         // See § 4.1.1.1.2.e on the "Ortografía de la lengua española"
         // https://www.rae.es/ortografía/como-signo-de-división-de-palabras-a-final-de-línea
-        Lang::SPANISH => text.chars().next().is_some_and(|c| !c.is_uppercase()),
+        Lang::SPANISH => following_text.chars().next().is_some_and(|c| !c.is_uppercase()),
 
         _ => false,
     }
@@ -675,7 +691,11 @@ impl<'a> Items<'a> {
         self.0.iter().map(|(_, item)| &**item)
     }
 
-    /// Iterate over the items with indices
+    /// Iterate over the items with the indices that define their logical order.
+    /// See the docs above `logical_item_idx` for more details.
+    ///
+    /// Note that this is different from `.iter().enumerate()` which would
+    /// provide the indices in visual order!
     pub fn indexed_iter(&self) -> impl Iterator<Item = &(usize, ItemEntry<'a>)> {
         self.0.iter()
     }
@@ -690,6 +710,11 @@ impl<'a> Items<'a> {
         self.0.last().map(|(_, item)| &**item)
     }
 
+    /// Access the last item, if it is text.
+    pub fn last_text(&self) -> Option<&ShapedText<'a>> {
+        self.0.last()?.1.text()
+    }
+
     /// Access the first item mutably, if it is text.
     pub fn first_text_mut(&mut self) -> Option<&mut ShapedText<'a>> {
         self.0.first_mut()?.1.text_mut()
@@ -706,12 +731,6 @@ impl<'a> Items<'a> {
     }
 }
 
-impl<'a> FromIterator<ItemEntry<'a>> for Items<'a> {
-    fn from_iter<I: IntoIterator<Item = ItemEntry<'a>>>(iter: I) -> Self {
-        Self(iter.into_iter().enumerate().collect())
-    }
-}
-
 impl<'a> Deref for Items<'a> {
     type Target = Vec<(usize, ItemEntry<'a>)>;
 

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

@@ -196,14 +196,6 @@ impl ShapedGlyph {
     }
 }
 
-/// A side you can go toward.
-enum Side {
-    /// To the left-hand side.
-    Left,
-    /// To the right-hand side.
-    Right,
-}
-
 impl<'a> ShapedText<'a> {
     /// Build the shaped text's frame.
     ///
@@ -448,27 +440,23 @@ impl<'a> ShapedText<'a> {
         }
     }
 
-    /// Push a hyphen to end of the text.
+    /// Creates shaped text containing a hyphen.
-    pub fn push_hyphen(&mut self, engine: &Engine, fallback: bool) {
+    pub fn hyphen(
-        self.insert_hyphen(engine, fallback, Side::Right)
+        engine: &Engine,
-    }
+        fallback: bool,
-
+        base: &ShapedText<'a>,
-    /// Prepend a hyphen to start of the text.
+        pos: usize,
-    pub fn prepend_hyphen(&mut self, engine: &Engine, fallback: bool) {
+    ) -> Option<Self> {
-        self.insert_hyphen(engine, fallback, Side::Left)
-    }
-
-    fn insert_hyphen(&mut self, engine: &Engine, fallback: bool, side: Side) {
         let world = engine.world;
         let book = world.book();
         let fallback_func = if fallback {
-            Some(|| book.select_fallback(None, self.variant, "-"))
+            Some(|| book.select_fallback(None, base.variant, "-"))
         } else {
             None
         };
-        let mut chain = families(self.styles)
+        let mut chain = families(base.styles)
             .filter(|family| family.covers().is_none_or(|c| c.is_match("-")))
-            .map(|family| book.select(family.as_str(), self.variant))
+            .map(|family| book.select(family.as_str(), base.variant))
             .chain(fallback_func.iter().map(|f| f()))
             .flatten();
 
@@ -477,37 +465,33 @@ impl<'a> ShapedText<'a> {
             let ttf = font.ttf();
             let glyph_id = ttf.glyph_index('-')?;
             let x_advance = font.to_em(ttf.glyph_hor_advance(glyph_id)?);
-            let range = match side {
+            let size = base.styles.resolve(TextElem::size);
-                Side::Left => self.glyphs.first().map(|g| g.range.start..g.range.start),
+
-                Side::Right => self.glyphs.last().map(|g| g.range.end..g.range.end),
+            Some(ShapedText {
-            }
+                base: pos,
-            // In the unlikely chance that we hyphenate after an empty line,
+                text: "",
-            // ensure that the glyph range still falls after self.base so
+                dir: base.dir,
-            // that subtracting either of the endpoints by self.base doesn't
+                lang: base.lang,
-            // underflow. See <https://github.com/typst/typst/issues/2283>.
+                region: base.region,
-            .unwrap_or_else(|| self.base..self.base);
+                styles: base.styles,
-            let size = self.styles.resolve(TextElem::size);
+                variant: base.variant,
-            self.width += x_advance.at(size);
+                width: x_advance.at(size),
-            let glyph = ShapedGlyph {
+                glyphs: Cow::Owned(vec![ShapedGlyph {
-                font,
+                    font,
-                glyph_id: glyph_id.0,
+                    glyph_id: glyph_id.0,
-                x_advance,
+                    x_advance,
-                x_offset: Em::zero(),
+                    x_offset: Em::zero(),
-                y_offset: Em::zero(),
+                    y_offset: Em::zero(),
-                size,
+                    size,
-                adjustability: Adjustability::default(),
+                    adjustability: Adjustability::default(),
-                range,
+                    range: pos..pos,
-                safe_to_break: true,
+                    safe_to_break: true,
-                c: '-',
+                    c: '-',
-                is_justifiable: false,
+                    is_justifiable: false,
-                script: Script::Common,
+                    script: Script::Common,
-            };
+                }]),
-            match side {
+            })
-                Side::Left => self.glyphs.to_mut().insert(0, glyph),
+        })
-                Side::Right => self.glyphs.to_mut().push(glyph),
-            }
-            Some(())
-        });
     }
 
     /// Find the subslice of glyphs that represent the given text range if both

tests/suite/layout/inline/hyphenate.typ

@@ -112,6 +112,12 @@ Tras el estallido de la contienda Ruiz-Giménez fue detenido junto a sus
 dos hermanos y puesto bajo custodia por las autoridades republicanas, con
 el objetivo de protegerle de las patrullas de milicianos.
 
+--- hyphenate-repeat-style ---
+// Ensure that a repeated hard hyphen keeps its styles.
+#set page(width: 2cm)
+#set text(lang: "es")
+Hello-#text(red)[world]
+
 --- costs-widow-orphan ---
 #set page(height: 60pt)