Fix repetition of Thai characters (#4977)

Co-authored-by: Martin Haug <mhaug@live.de>

4 files changed, 18 insertions, 11 deletions

diff --git a/crates/typst/src/layout/inline/shaping.rs b/crates/typst/src/layout/inline/shaping.rs
index 43dc351a..0b4a17d7 100644
--- a/crates/typst/src/layout/inline/shaping.rs
+++ b/crates/typst/src/layout/inline/shaping.rs
@@ -515,14 +515,14 @@ impl<'a> ShapedText<'a> {
             std::mem::swap(&mut start, &mut end);
         }
 
-        let left = self.find_safe_to_break(start, Side::Left)?;
-        let right = self.find_safe_to_break(end, Side::Right)?;
+        let left = self.find_safe_to_break(start)?;
+        let right = self.find_safe_to_break(end)?;
         Some(&self.glyphs[left..right])
     }
 
     /// Find the glyph offset matching the text index that is most towards the
-    /// given side and safe-to-break.
-    fn find_safe_to_break(&self, text_index: usize, towards: Side) -> Option<usize> {
+    /// start of the text and safe-to-break.
+    fn find_safe_to_break(&self, text_index: usize) -> Option<usize> {
         let ltr = self.dir.is_positive();
 
         // Handle edge cases.
@@ -542,6 +542,7 @@ impl<'a> ShapedText<'a> {
                 ordering.reverse()
             }
         });
+
         let mut idx = match found {
             Ok(idx) => idx,
             Err(idx) => {
@@ -565,13 +566,11 @@ impl<'a> ShapedText<'a> {
             }
         };
 
-        let next = match towards {
-            Side::Left => usize::checked_sub,
-            Side::Right => usize::checked_add,
-        };
-
-        // Search for the outermost glyph with the text index.
-        while let Some(next) = next(idx, 1) {
+        // Search for the start-most glyph with the text index. This means
+        // we take empty range glyphs at the start and leave those at the end
+        // for the next line.
+        let dec = if ltr { usize::checked_sub } else { usize::checked_add };
+        while let Some(next) = dec(idx, 1) {
             if self.glyphs.get(next).map_or(true, |g| g.range.start != text_index) {
                 break;
             }
diff --git a/tests/ref/issue-4468-linebreak-thai.png b/tests/ref/issue-4468-linebreak-thai.png
new file mode 100644
index 00000000..6257c560
--- /dev/null
+++ b/tests/ref/issue-4468-linebreak-thai.png
diff --git a/tests/ref/linebreak-thai.png b/tests/ref/linebreak-thai.png
index 8053a212..c31a61bd 100644
--- a/tests/ref/linebreak-thai.png
+++ b/tests/ref/linebreak-thai.png
diff --git a/tests/suite/layout/inline/linebreak.typ b/tests/suite/layout/inline/linebreak.typ
index 7e959352..8371d76e 100644
--- a/tests/suite/layout/inline/linebreak.typ
+++ b/tests/suite/layout/inline/linebreak.typ
@@ -115,3 +115,11 @@ For info see #link("https://myhost.tld").
 // for links because it now splits on word boundaries. We avoid the link markup
 // syntax because it's show rule interferes.
 #"http://creativecommons.org/licenses/by-nc-sa/4.0/"
+
+--- issue-4468-linebreak-thai ---
+// In this bug, empty-range glyphs at line break boundaries could be duplicated.
+// This happens for Thai specifically because it has both
+// - line break opportunities
+// - shaping that results in multiple glyphs in the same cluster
+#set text(font: "Noto Sans Thai")
+#h(85pt) งบิก