From ea378e89b4f2267bb85ec56c905111a6c73d4721 Mon Sep 17 00:00:00 2001
From: Laurenz <laurmaedje@gmail.com>
Date: Sun, 22 Jan 2023 13:27:49 +0100
Subject: [PATCH] Better math atoms

---
 library/src/math/atom.rs | 47 ++++++++++++++++++++++++----------------
 src/syntax/lexer.rs      | 29 +++++++++++++++++--------
 2 files changed, 48 insertions(+), 28 deletions(-)

diff --git a/library/src/math/atom.rs b/library/src/math/atom.rs
index 142461474..5b35d289a 100644
--- a/library/src/math/atom.rs
+++ b/library/src/math/atom.rs
@@ -10,7 +10,7 @@ use super::*;
 /// ## Category
 /// math
 #[func]
-#[capable(Texify)]
+#[capable(LayoutMath)]
 #[derive(Debug, Hash)]
 pub struct AtomNode(pub EcoString);
 
@@ -21,26 +21,35 @@ impl AtomNode {
     }
 }
 
-impl Texify for AtomNode {
-    fn texify(&self, t: &mut Texifier) -> SourceResult<()> {
-        let multi = self.0.graphemes(true).count() > 1;
-        if multi {
-            t.push_str("\\mathrm{");
-        }
-
-        for c in self.0.chars() {
-            let supportive = c == '|';
-            if supportive {
-                t.support();
+impl LayoutMath for AtomNode {
+    fn layout_math(&self, ctx: &mut MathContext) -> SourceResult<()> {
+        let mut chars = self.0.chars();
+        if let Some(glyph) = chars
+            .next()
+            .filter(|_| chars.next().is_none())
+            .and_then(|c| GlyphFragment::try_new(ctx, c))
+        {
+            // A single letter that is available in the math font.
+            if ctx.style.size == MathSize::Display
+                && glyph.class() == Some(MathClass::Large)
+            {
+                let height = scaled!(ctx, display_operator_min_height);
+                ctx.push(glyph.stretch_vertical(ctx, height, Abs::zero()));
+            } else {
+                ctx.push(glyph);
             }
-            t.push_escaped(c);
-            if supportive {
-                t.support();
+        } else if self.0.chars().all(|c| c.is_ascii_digit()) {
+            // A number that should respect math styling and can therefore
+            // not fall back to the normal text layout.
+            let mut vec = vec![];
+            for c in self.0.chars() {
+                vec.push(GlyphFragment::new(ctx, c).into());
             }
-        }
-
-        if multi {
-            t.push_str("}");
+            let frame = MathRow(vec).to_frame(ctx);
+            ctx.push(frame);
+        } else {
+            // Anything else is handled by Typst's standard text layout.
+            TextNode(self.0.clone()).pack().layout_math(ctx)?;
         }
 
         Ok(())
diff --git a/src/syntax/lexer.rs b/src/syntax/lexer.rs
index e3c291509..433d0defc 100644
--- a/src/syntax/lexer.rs
+++ b/src/syntax/lexer.rs
@@ -1,3 +1,4 @@
+use unicode_segmentation::UnicodeSegmentation;
 use unicode_xid::UnicodeXID;
 use unscanny::Scanner;
 
@@ -103,7 +104,7 @@ impl Lexer<'_> {
 
             Some(c) => match self.mode {
                 LexMode::Markup => self.markup(start, c),
-                LexMode::Math => self.math(c),
+                LexMode::Math => self.math(start, c),
                 LexMode::Code => self.code(start, c),
             },
 
@@ -404,7 +405,7 @@ impl Lexer<'_> {
 
 /// Math.
 impl Lexer<'_> {
-    fn math(&mut self, c: char) -> SyntaxKind {
+    fn math(&mut self, start: usize, c: char) -> SyntaxKind {
         match c {
             '\\' => self.backslash(),
             ':' if self.s.at(is_id_start) => self.maybe_symbol(),
@@ -441,13 +442,7 @@ impl Lexer<'_> {
             }
 
             // Other math atoms.
-            _ => {
-                // Keep numbers together.
-                if c.is_numeric() {
-                    self.s.eat_while(char::is_numeric);
-                }
-                SyntaxKind::Atom
-            }
+            _ => self.atom(start, c),
         }
     }
 
@@ -469,6 +464,22 @@ impl Lexer<'_> {
 
         SyntaxKind::Ident
     }
+
+    fn atom(&mut self, start: usize, c: char) -> SyntaxKind {
+        // Keep numbers and grapheme clusters together.
+        if c.is_numeric() {
+            self.s.eat_while(char::is_numeric);
+        } else {
+            let len = self
+                .s
+                .get(start..self.s.string().len())
+                .graphemes(true)
+                .next()
+                .map_or(0, str::len);
+            self.s.jump(start + len);
+        }
+        SyntaxKind::Atom
+    }
 }
 
 /// Code.