From 2501da4bb62948e63c68c76cb45af3b7c8a57a0c Mon Sep 17 00:00:00 2001
From: Laurenz <laurmaedje@gmail.com>
Date: Thu, 21 Aug 2025 16:21:56 +0200
Subject: [PATCH] Fix Unicode mapping of hyphenation artifacts (#6799)

---
 crates/typst-layout/src/inline/line.rs    |  8 +++++---
 crates/typst-layout/src/inline/shaping.rs | 16 +++++++++++++---
 2 files changed, 18 insertions(+), 6 deletions(-)

diff --git a/crates/typst-layout/src/inline/line.rs b/crates/typst-layout/src/inline/line.rs
index 9b5b656e5..f89795de6 100644
--- a/crates/typst-layout/src/inline/line.rs
+++ b/crates/typst-layout/src/inline/line.rs
@@ -178,7 +178,8 @@ pub fn line<'a>(
         && 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)
+        && let Some(hyphen) =
+            ShapedText::hyphen(engine, p.config.fallback, base, trim, false)
     {
         items.push(Item::Text(hyphen), START_HYPHEN_IDX);
     }
@@ -188,7 +189,8 @@ pub fn line<'a>(
     // Add a hyphen at the line end, if we ended on a soft hyphen.
     if dash == Some(Dash::Soft)
         && let Some(base) = items.last_text()
-        && let Some(hyphen) = ShapedText::hyphen(engine, p.config.fallback, base, trim)
+        && let Some(hyphen) =
+            ShapedText::hyphen(engine, p.config.fallback, base, trim, true)
     {
         items.push(Item::Text(hyphen), END_HYPHEN_IDX);
     }
@@ -659,7 +661,7 @@ fn overhang(c: char) -> f64 {
     match c {
         // Dashes.
         '–' | '—' => 0.2,
-        '-' => 0.55,
+        '-' | '\u{ad}' => 0.55,
 
         // Punctuation.
         '.' | ',' => 0.8,
diff --git a/crates/typst-layout/src/inline/shaping.rs b/crates/typst-layout/src/inline/shaping.rs
index 287bca9a5..66987b2ad 100644
--- a/crates/typst-layout/src/inline/shaping.rs
+++ b/crates/typst-layout/src/inline/shaping.rs
@@ -21,6 +21,11 @@ use unicode_script::{Script, UnicodeScript};
 use super::{Item, Range, SpanMapper, decorate};
 use crate::modifiers::FrameModifyText;
 
+const SHY: char = '\u{ad}';
+const SHY_STR: &str = "\u{ad}";
+const HYPHEN: char = '-';
+const HYPHEN_STR: &str = "-";
+
 /// The result of shaping text.
 ///
 /// This type contains owned or borrowed shaped text runs, which can be
@@ -441,11 +446,15 @@ impl<'a> ShapedText<'a> {
     }
 
     /// Creates shaped text containing a hyphen.
+    ///
+    /// If `soft` is true, the item will map to plain text as a soft hyphen.
+    /// Otherwise, it will map to a normal hyphen.
     pub fn hyphen(
         engine: &Engine,
         fallback: bool,
         base: &ShapedText<'a>,
         pos: usize,
+        soft: bool,
     ) -> Option<Self> {
         let world = engine.world;
         let book = world.book();
@@ -466,10 +475,11 @@ impl<'a> ShapedText<'a> {
             let glyph_id = ttf.glyph_index('-')?;
             let x_advance = font.to_em(ttf.glyph_hor_advance(glyph_id)?);
             let size = base.styles.resolve(TextElem::size);
+            let (c, text) = if soft { (SHY, SHY_STR) } else { (HYPHEN, HYPHEN_STR) };
 
             Some(ShapedText {
                 base: pos,
-                text: "",
+                text,
                 dir: base.dir,
                 lang: base.lang,
                 region: base.region,
@@ -484,9 +494,9 @@ impl<'a> ShapedText<'a> {
                     y_offset: Em::zero(),
                     size,
                     adjustability: Adjustability::default(),
-                    range: pos..pos,
+                    range: pos..pos + text.len(),
                     safe_to_break: true,
-                    c: '-',
+                    c,
                     is_justifiable: false,
                     script: Script::Common,
                 }]),