Update shaping logic

Cargo.lock

@@ -1345,7 +1345,6 @@ dependencies = [
 [[package]]
 name = "krilla"
 version = "0.3.0"
-source = "git+https://github.com/LaurenzV/krilla?rev=399dc59#399dc596fdb27fde3ca1eef27b74625affed199a"
 dependencies = [
  "base64",
  "bumpalo",

Cargo.toml

@@ -70,7 +70,7 @@ if_chain = "1"
 image = { version = "0.25.5", default-features = false, features = ["png", "jpeg", "gif"] }
 indexmap = { version = "2", features = ["serde"] }
 kamadak-exif = "0.6"
-krilla = { git = "https://github.com/LaurenzV/krilla", rev = "399dc59", features = ["svg", "raster-images", "comemo", "rayon"] }
+krilla = { path = "../krilla/crates/krilla", features = ["svg", "raster-images", "comemo", "rayon"] }
 kurbo = "0.11"
 libfuzzer-sys = "0.4"
 lipsum = "0.9"

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

@@ -826,8 +826,47 @@ fn shape_segment<'a>(
         if info.glyph_id != 0 && is_covered(cluster) {
             // Determine the text range of the glyph.
             let start = base + cluster;
+            // Assume we have the following sequence of (glyph_id, cluster):
+            // [(120, 0), (80, 0), (3, 3), (755, 4), (69, 4), (424, 13), (63, 13), (193, 25), (80, 25), (3, 31)
+            // We then want the sequence of (glyph_id, text_range) to look as follows:
+            // [(120, 0..3), (80, 0..3), (3, 3..4), (755, 4..13), (69, 4..13), (424, 13..25),
+            // (63, 13..25), (193, 25..31), (80, 25..31), (3, 31..x)]
+            // i.e. each glyph in the same cluster should be assigned the full text range. This
+            // is necessary because only this way can krilla properly assign `ActualText` attributes
+            // in complex shaping scenarios.
+            let offset = if ltr {
+                let mut e = i.checked_add(1);
+                loop {
+                    if let Some(index) = e {
+                        if let Some(end_info) = infos.get(index) {
+                            if end_info.cluster == info.cluster {
+                                e = index.checked_add(1);
+                                continue;
+                            }
+                        }
+                    }
+
+                    break;
+                }
+
+                e
+            } else {
+                let mut e = i.checked_sub(1);
+                while let Some(index) = e {
+                    if let Some(end_info) = infos.get(index) {
+                        if end_info.cluster == info.cluster {
+                            e = index.checked_sub(1);
+                        } else {
+                            break;
+                        }
+                    }
+                }
+
+                e
+            };
+
             let end = base
-                + if ltr { i.checked_add(1) } else { i.checked_sub(1) }
+                + offset
                     .and_then(|last| infos.get(last))
                     .map_or(text.len(), |info| info.cluster as usize);