Use "subs" and "sups" font features for typographic scripts (#5777)

6 files changed, 155 insertions, 47 deletions

diff --git a/crates/typst-layout/src/inline/collect.rs b/crates/typst-layout/src/inline/collect.rs
index 5a1b7b4f..829f64b6 100644
--- a/crates/typst-layout/src/inline/collect.rs
+++ b/crates/typst-layout/src/inline/collect.rs
@@ -206,7 +206,7 @@ pub fn collect<'a>(
                     }
                     InlineItem::Frame(mut frame) => {
                         frame.modify(&FrameModifiers::get_in(styles));
-                        apply_baseline_shift(&mut frame, styles);
+                        apply_shift(&engine.world, &mut frame, styles);
                         collector.push_item(Item::Frame(frame));
                     }
                 }
@@ -221,7 +221,7 @@ pub fn collect<'a>(
                 let mut frame = layout_and_modify(styles, |styles| {
                     layout_box(elem, engine, loc, styles, region)
                 })?;
-                apply_baseline_shift(&mut frame, styles);
+                apply_shift(&engine.world, &mut frame, styles);
                 collector.push_item(Item::Frame(frame));
             }
         } else if let Some(elem) = child.to_packed::<TagElem>() {
diff --git a/crates/typst-layout/src/inline/line.rs b/crates/typst-layout/src/inline/line.rs
index 7bf4d4c7..b850e50e 100644
--- a/crates/typst-layout/src/inline/line.rs
+++ b/crates/typst-layout/src/inline/line.rs
@@ -5,7 +5,7 @@ use typst_library::engine::Engine;
 use typst_library::introspection::{SplitLocator, Tag};
 use typst_library::layout::{Abs, Dir, Em, Fr, Frame, FrameItem, Point};
 use typst_library::model::ParLineMarker;
-use typst_library::text::{Lang, TextElem};
+use typst_library::text::{variant, Lang, TextElem};
 use typst_utils::Numeric;
 
 use super::*;
@@ -330,7 +330,7 @@ fn adjust_cj_at_line_start(p: &Preparation, items: &mut Items) {
         let glyph = shaped.glyphs.to_mut().first_mut().unwrap();
         let shrink = glyph.shrinkability().0;
         glyph.shrink_left(shrink);
-        shaped.width -= shrink.at(shaped.size);
+        shaped.width -= shrink.at(glyph.size);
     } else if p.config.cjk_latin_spacing
         && glyph.is_cj_script()
         && glyph.x_offset > Em::zero()
@@ -342,7 +342,7 @@ fn adjust_cj_at_line_start(p: &Preparation, items: &mut Items) {
         glyph.x_advance -= shrink;
         glyph.x_offset = Em::zero();
         glyph.adjustability.shrinkability.0 = Em::zero();
-        shaped.width -= shrink.at(shaped.size);
+        shaped.width -= shrink.at(glyph.size);
     }
 }
 
@@ -360,7 +360,7 @@ fn adjust_cj_at_line_end(p: &Preparation, items: &mut Items) {
         let shrink = glyph.shrinkability().1;
         let punct = shaped.glyphs.to_mut().last_mut().unwrap();
         punct.shrink_right(shrink);
-        shaped.width -= shrink.at(shaped.size);
+        shaped.width -= shrink.at(punct.size);
     } else if p.config.cjk_latin_spacing
         && glyph.is_cj_script()
         && (glyph.x_advance - glyph.x_offset) > Em::one()
@@ -371,7 +371,7 @@ fn adjust_cj_at_line_end(p: &Preparation, items: &mut Items) {
         let glyph = shaped.glyphs.to_mut().last_mut().unwrap();
         glyph.x_advance -= shrink;
         glyph.adjustability.shrinkability.1 = Em::zero();
-        shaped.width -= shrink.at(shaped.size);
+        shaped.width -= shrink.at(glyph.size);
     }
 }
 
@@ -412,9 +412,30 @@ fn should_repeat_hyphen(pred_line: &Line, text: &str) -> bool {
     }
 }
 
-/// Apply the current baseline shift to a frame.
-pub fn apply_baseline_shift(frame: &mut Frame, styles: StyleChain) {
-    frame.translate(Point::with_y(TextElem::baseline_in(styles)));
+/// Apply the current baseline shift and italic compensation to a frame.
+pub fn apply_shift<'a>(
+    world: &Tracked<'a, dyn World + 'a>,
+    frame: &mut Frame,
+    styles: StyleChain,
+) {
+    let mut baseline = TextElem::baseline_in(styles);
+    let mut compensation = Abs::zero();
+    if let Some(scripts) = TextElem::shift_settings_in(styles) {
+        let font_metrics = TextElem::font_in(styles)
+            .into_iter()
+            .find_map(|family| {
+                world
+                    .book()
+                    .select(family.as_str(), variant(styles))
+                    .and_then(|id| world.font(id))
+            })
+            .map_or(*scripts.kind.default_metrics(), |f| {
+                *scripts.kind.read_metrics(f.metrics())
+            });
+        baseline -= scripts.shift.unwrap_or(font_metrics.vertical_offset).resolve(styles);
+        compensation += font_metrics.horizontal_offset.resolve(styles);
+    }
+    frame.translate(Point::new(compensation, baseline));
 }
 
 /// Commit to a line and build its frame.
@@ -444,7 +465,7 @@ pub fn commit(
                 && TextElem::overhang_in(text.styles)
                 && (line.items.len() > 1 || text.glyphs.len() > 1)
             {
-                let amount = overhang(glyph.c) * glyph.x_advance.at(text.size);
+                let amount = overhang(glyph.c) * glyph.x_advance.at(glyph.size);
                 offset -= amount;
                 remaining += amount;
             }
@@ -458,7 +479,7 @@ pub fn commit(
                 && TextElem::overhang_in(text.styles)
                 && (line.items.len() > 1 || text.glyphs.len() > 1)
             {
-                let amount = overhang(glyph.c) * glyph.x_advance.at(text.size);
+                let amount = overhang(glyph.c) * glyph.x_advance.at(glyph.size);
                 remaining += amount;
             }
         }
@@ -519,7 +540,7 @@ pub fn commit(
                     let mut frame = layout_and_modify(*styles, |styles| {
                         layout_box(elem, engine, loc.relayout(), styles, region)
                     })?;
-                    apply_baseline_shift(&mut frame, *styles);
+                    apply_shift(&engine.world, &mut frame, *styles);
                     push(&mut offset, frame, idx);
                 } else {
                     offset += amount;
diff --git a/crates/typst-layout/src/inline/linebreak.rs b/crates/typst-layout/src/inline/linebreak.rs
index ada048c7..709745ed 100644
--- a/crates/typst-layout/src/inline/linebreak.rs
+++ b/crates/typst-layout/src/inline/linebreak.rs
@@ -927,9 +927,9 @@ impl Estimates {
                     let byte_len = g.range.len();
                     let stretch = g.stretchability().0 + g.stretchability().1;
                     let shrink = g.shrinkability().0 + g.shrinkability().1;
-                    widths.push(byte_len, g.x_advance.at(shaped.size));
-                    stretchability.push(byte_len, stretch.at(shaped.size));
-                    shrinkability.push(byte_len, shrink.at(shaped.size));
+                    widths.push(byte_len, g.x_advance.at(g.size));
+                    stretchability.push(byte_len, stretch.at(g.size));
+                    shrinkability.push(byte_len, shrink.at(g.size));
                     justifiables.push(byte_len, g.is_justifiable() as usize);
                 }
             } else {
diff --git a/crates/typst-layout/src/inline/mod.rs b/crates/typst-layout/src/inline/mod.rs
index 6cafb9b0..506fa5ea 100644
--- a/crates/typst-layout/src/inline/mod.rs
+++ b/crates/typst-layout/src/inline/mod.rs
@@ -29,7 +29,7 @@ use typst_utils::{Numeric, SliceExt};
 use self::collect::{collect, Item, Segment, SpanMapper};
 use self::deco::decorate;
 use self::finalize::finalize;
-use self::line::{apply_baseline_shift, commit, line, Line};
+use self::line::{apply_shift, commit, line, Line};
 use self::linebreak::{linebreak, Breakpoint};
 use self::prepare::{prepare, Preparation};
 use self::shaping::{
diff --git a/crates/typst-layout/src/inline/prepare.rs b/crates/typst-layout/src/inline/prepare.rs
index 5d7fcd7c..ab39bdb1 100644
--- a/crates/typst-layout/src/inline/prepare.rs
+++ b/crates/typst-layout/src/inline/prepare.rs
@@ -144,7 +144,7 @@ fn add_cjk_latin_spacing(items: &mut [(Range, Item)]) {
                 // The spacing is default to 1/4 em, and can be shrunk to 1/8 em.
                 glyph.x_advance += Em::new(0.25);
                 glyph.adjustability.shrinkability.1 += Em::new(0.125);
-                text.width += Em::new(0.25).at(text.size);
+                text.width += Em::new(0.25).at(glyph.size);
             }
 
             // Case 2: Latin followed by a CJ character
@@ -152,7 +152,7 @@ fn add_cjk_latin_spacing(items: &mut [(Range, Item)]) {
                 glyph.x_advance += Em::new(0.25);
                 glyph.x_offset += Em::new(0.25);
                 glyph.adjustability.shrinkability.0 += Em::new(0.125);
-                text.width += Em::new(0.25).at(text.size);
+                text.width += Em::new(0.25).at(glyph.size);
             }
 
             prev = Some(glyph);
diff --git a/crates/typst-layout/src/inline/shaping.rs b/crates/typst-layout/src/inline/shaping.rs
index 935a86b3..48747cd5 100644
--- a/crates/typst-layout/src/inline/shaping.rs
+++ b/crates/typst-layout/src/inline/shaping.rs
@@ -3,14 +3,15 @@ use std::fmt::{self, Debug, Formatter};
 use std::sync::Arc;
 
 use az::SaturatingAs;
-use rustybuzz::{BufferFlags, ShapePlan, UnicodeBuffer};
+use rustybuzz::{BufferFlags, Feature, ShapePlan, UnicodeBuffer};
+use ttf_parser::gsub::SubstitutionSubtable;
 use ttf_parser::Tag;
 use typst_library::engine::Engine;
 use typst_library::foundations::{Smart, StyleChain};
 use typst_library::layout::{Abs, Dir, Em, Frame, FrameItem, Point, Size};
 use typst_library::text::{
     families, features, is_default_ignorable, language, variant, Font, FontFamily,
-    FontVariant, Glyph, Lang, Region, TextEdgeBounds, TextElem, TextItem,
+    FontVariant, Glyph, Lang, Region, ShiftSettings, TextEdgeBounds, TextElem, TextItem,
 };
 use typst_library::World;
 use typst_utils::SliceExt;
@@ -41,8 +42,6 @@ pub struct ShapedText<'a> {
     pub styles: StyleChain<'a>,
     /// The font variant.
     pub variant: FontVariant,
-    /// The font size.
-    pub size: Abs,
     /// The width of the text's bounding box.
     pub width: Abs,
     /// The shaped glyphs.
@@ -62,6 +61,8 @@ pub struct ShapedGlyph {
     pub x_offset: Em,
     /// The vertical offset of the glyph.
     pub y_offset: Em,
+    /// The font size for the glyph.
+    pub size: Abs,
     /// The adjustability of the glyph.
     pub adjustability: Adjustability,
     /// The byte range of this glyph's cluster in the full inline layout. A
@@ -222,14 +223,17 @@ impl<'a> ShapedText<'a> {
         let mut frame = Frame::soft(size);
         frame.set_baseline(top);
 
+        let size = TextElem::size_in(self.styles);
         let shift = TextElem::baseline_in(self.styles);
         let decos = TextElem::deco_in(self.styles);
         let fill = TextElem::fill_in(self.styles);
         let stroke = TextElem::stroke_in(self.styles);
         let span_offset = TextElem::span_offset_in(self.styles);
 
-        for ((font, y_offset), group) in
-            self.glyphs.as_ref().group_by_key(|g| (g.font.clone(), g.y_offset))
+        for ((font, y_offset, glyph_size), group) in self
+            .glyphs
+            .as_ref()
+            .group_by_key(|g| (g.font.clone(), g.y_offset, g.size))
         {
             let mut range = group[0].range.clone();
             for glyph in group {
@@ -237,7 +241,7 @@ impl<'a> ShapedText<'a> {
                 range.end = range.end.max(glyph.range.end);
             }
 
-            let pos = Point::new(offset, top + shift - y_offset.at(self.size));
+            let pos = Point::new(offset, top + shift - y_offset.at(size));
             let glyphs: Vec<Glyph> = group
                 .iter()
                 .map(|shaped: &ShapedGlyph| {
@@ -257,11 +261,11 @@ impl<'a> ShapedText<'a> {
                         adjustability_right * justification_ratio;
                     if shaped.is_justifiable() {
                         justification_right +=
-                            Em::from_length(extra_justification, self.size)
+                            Em::from_abs(extra_justification, glyph_size)
                     }
 
-                    frame.size_mut().x += justification_left.at(self.size)
-                        + justification_right.at(self.size);
+                    frame.size_mut().x += justification_left.at(glyph_size)
+                        + justification_right.at(glyph_size);
 
                     // We may not be able to reach the offset completely if
                     // it exceeds u16, but better to have a roughly correct
@@ -304,7 +308,7 @@ impl<'a> ShapedText<'a> {
 
             let item = TextItem {
                 font,
-                size: self.size,
+                size: glyph_size,
                 lang: self.lang,
                 region: self.region,
                 fill: fill.clone(),
@@ -336,12 +340,13 @@ impl<'a> ShapedText<'a> {
         let mut top = Abs::zero();
         let mut bottom = Abs::zero();
 
+        let size = TextElem::size_in(self.styles);
         let top_edge = TextElem::top_edge_in(self.styles);
         let bottom_edge = TextElem::bottom_edge_in(self.styles);
 
         // Expand top and bottom by reading the font's vertical metrics.
         let mut expand = |font: &Font, bounds: TextEdgeBounds| {
-            let (t, b) = font.edges(top_edge, bottom_edge, self.size, bounds);
+            let (t, b) = font.edges(top_edge, bottom_edge, size, bounds);
             top.set_max(t);
             bottom.set_max(b);
         };
@@ -388,18 +393,16 @@ impl<'a> ShapedText<'a> {
     pub fn stretchability(&self) -> Abs {
         self.glyphs
             .iter()
-            .map(|g| g.stretchability().0 + g.stretchability().1)
-            .sum::<Em>()
-            .at(self.size)
+            .map(|g| (g.stretchability().0 + g.stretchability().1).at(g.size))
+            .sum()
     }
 
     /// The shrinkability of the text
     pub fn shrinkability(&self) -> Abs {
         self.glyphs
             .iter()
-            .map(|g| g.shrinkability().0 + g.shrinkability().1)
-            .sum::<Em>()
-            .at(self.size)
+            .map(|g| (g.shrinkability().0 + g.shrinkability().1).at(g.size))
+            .sum()
     }
 
     /// Reshape a range of the shaped text, reusing information from this
@@ -418,9 +421,8 @@ impl<'a> ShapedText<'a> {
                 lang: self.lang,
                 region: self.region,
                 styles: self.styles,
-                size: self.size,
                 variant: self.variant,
-                width: glyphs.iter().map(|g| g.x_advance).sum::<Em>().at(self.size),
+                width: glyphs_width(glyphs),
                 glyphs: Cow::Borrowed(glyphs),
             }
         } else {
@@ -484,13 +486,15 @@ impl<'a> ShapedText<'a> {
             // that subtracting either of the endpoints by self.base doesn't
             // underflow. See <https://github.com/typst/typst/issues/2283>.
             .unwrap_or_else(|| self.base..self.base);
-            self.width += x_advance.at(self.size);
+            let size = TextElem::size_in(self.styles);
+            self.width += x_advance.at(size);
             let glyph = ShapedGlyph {
                 font,
                 glyph_id: glyph_id.0,
                 x_advance,
                 x_offset: Em::zero(),
                 y_offset: Em::zero(),
+                size,
                 adjustability: Adjustability::default(),
                 range,
                 safe_to_break: true,
@@ -666,6 +670,7 @@ fn shape<'a>(
     region: Option<Region>,
 ) -> ShapedText<'a> {
     let size = TextElem::size_in(styles);
+    let shift_settings = TextElem::shift_settings_in(styles);
     let mut ctx = ShapingContext {
         engine,
         size,
@@ -676,6 +681,7 @@ fn shape<'a>(
         features: features(styles),
         fallback: TextElem::fallback_in(styles),
         dir,
+        shift_settings,
     };
 
     if !text.is_empty() {
@@ -698,12 +704,17 @@ fn shape<'a>(
         region,
         styles,
         variant: ctx.variant,
-        size,
-        width: ctx.glyphs.iter().map(|g| g.x_advance).sum::<Em>().at(size),
+        width: glyphs_width(&ctx.glyphs),
         glyphs: Cow::Owned(ctx.glyphs),
     }
 }
 
+/// Computes the width of a run of glyphs relative to the font size, accounting
+/// for their individual scaling factors and other font metrics.
+fn glyphs_width(glyphs: &[ShapedGlyph]) -> Abs {
+    glyphs.iter().map(|g| g.x_advance.at(g.size)).sum()
+}
+
 /// Holds shaping results and metadata common to all shaped segments.
 struct ShapingContext<'a, 'v> {
     engine: &'a Engine<'v>,
@@ -715,6 +726,7 @@ struct ShapingContext<'a, 'v> {
     features: Vec<rustybuzz::Feature>,
     fallback: bool,
     dir: Dir,
+    shift_settings: Option<ShiftSettings>,
 }
 
 /// Shape text with font fallback using the `families` iterator.
@@ -789,6 +801,18 @@ fn shape_segment<'a>(
     // text extraction.
     buffer.set_flags(BufferFlags::REMOVE_DEFAULT_IGNORABLES);
 
+    let (script_shift, script_compensation, scale, shift_feature) = ctx
+        .shift_settings
+        .map_or((Em::zero(), Em::zero(), Em::one(), None), |settings| {
+            determine_shift(text, &font, settings)
+        });
+
+    let has_shift_feature = shift_feature.is_some();
+    if let Some(feat) = shift_feature {
+        // Temporarily push the feature.
+        ctx.features.push(feat)
+    }
+
     // Prepare the shape plan. This plan depends on direction, script, language,
     // and features, but is independent from the text and can thus be memoized.
     let plan = create_shape_plan(
@@ -799,6 +823,10 @@ fn shape_segment<'a>(
         &ctx.features,
     );
 
+    if has_shift_feature {
+        ctx.features.pop();
+    }
+
     // Shape!
     let buffer = rustybuzz::shape_with_plan(font.rusty(), &plan, buffer);
     let infos = buffer.glyph_infos();
@@ -869,8 +897,9 @@ fn shape_segment<'a>(
                 glyph_id: info.glyph_id as u16,
                 // TODO: Don't ignore y_advance.
                 x_advance,
-                x_offset: font.to_em(pos[i].x_offset),
-                y_offset: font.to_em(pos[i].y_offset),
+                x_offset: font.to_em(pos[i].x_offset) + script_compensation,
+                y_offset: font.to_em(pos[i].y_offset) + script_shift,
+                size: scale.at(ctx.size),
                 adjustability: Adjustability::default(),
                 range: start..end,
                 safe_to_break: !info.unsafe_to_break(),
@@ -932,6 +961,64 @@ fn shape_segment<'a>(
     ctx.used.pop();
 }
 
+/// Returns a `(script_shift, script_compensation, scale, feature)` quadruplet
+/// describing how to produce scripts.
+///
+/// Those values determine how the rendered text should be transformed to
+/// display sub-/super-scripts. If the OpenType feature can be used, the
+/// rendered text should not be transformed in any way, and so those values are
+/// neutral (`(0, 0, 1, None)`). If scripts should be synthesized, those values
+/// determine how to transform the rendered text to display scripts as expected.
+fn determine_shift(
+    text: &str,
+    font: &Font,
+    settings: ShiftSettings,
+) -> (Em, Em, Em, Option<Feature>) {
+    settings
+        .typographic
+        .then(|| {
+            // If typographic scripts are enabled (i.e., we want to use the
+            // OpenType feature instead of synthesizing if possible), we add
+            // "subs"/"sups" to the feature list if supported by the font.
+            // In case of a problem, we just early exit
+            let gsub = font.rusty().tables().gsub?;
+            let subtable_index =
+                gsub.features.find(settings.kind.feature())?.lookup_indices.get(0)?;
+            let coverage = gsub
+                .lookups
+                .get(subtable_index)?
+                .subtables
+                .get::<SubstitutionSubtable>(0)?
+                .coverage();
+            text.chars()
+                .all(|c| {
+                    font.rusty().glyph_index(c).is_some_and(|i| coverage.contains(i))
+                })
+                .then(|| {
+                    // If we can use the OpenType feature, we can keep the text
+                    // as is.
+                    (
+                        Em::zero(),
+                        Em::zero(),
+                        Em::one(),
+                        Some(Feature::new(settings.kind.feature(), 1, ..)),
+                    )
+                })
+        })
+        // Reunite the cases where `typographic` is `false` or where using the
+        // OpenType feature would not work.
+        .flatten()
+        .unwrap_or_else(|| {
+            let script_metrics = settings.kind.read_metrics(font.metrics());
+            (
+                settings.shift.unwrap_or(script_metrics.vertical_offset),
+                script_metrics.horizontal_offset,
+                settings.size.unwrap_or(script_metrics.height),
+                None,
+            )
+        })
+}
+
 /// Create a shape plan.
 #[comemo::memoize]
 pub fn create_shape_plan(
@@ -963,6 +1050,7 @@ fn shape_tofus(ctx: &mut ShapingContext, base: usize, text: &str, font: Font) {
             x_advance,
             x_offset: Em::zero(),
             y_offset: Em::zero(),
+            size: ctx.size,
             adjustability: Adjustability::default(),
             range: start..end,
             safe_to_break: true,
@@ -985,9 +1073,8 @@ fn shape_tofus(ctx: &mut ShapingContext, base: usize, text: &str, font: Font) {
 
 /// Apply tracking and spacing to the shaped glyphs.
 fn track_and_space(ctx: &mut ShapingContext) {
-    let tracking = Em::from_length(TextElem::tracking_in(ctx.styles), ctx.size);
-    let spacing =
-        TextElem::spacing_in(ctx.styles).map(|abs| Em::from_length(abs, ctx.size));
+    let tracking = Em::from_abs(TextElem::tracking_in(ctx.styles), ctx.size);
+    let spacing = TextElem::spacing_in(ctx.styles).map(|abs| Em::from_abs(abs, ctx.size));
 
     let mut glyphs = ctx.glyphs.iter_mut().peekable();
     while let Some(glyph) = glyphs.next() {