Split shorthands into math and markup, add tilde shorthand (#4706)

6 files changed, 108 insertions, 82 deletions

diff --git a/crates/typst-syntax/src/ast.rs b/crates/typst-syntax/src/ast.rs
index 1cd9cd42..3d3a85ad 100644
--- a/crates/typst-syntax/src/ast.rs
+++ b/crates/typst-syntax/src/ast.rs
@@ -125,6 +125,8 @@ pub enum Expr<'a> {
     Math(Math<'a>),
     /// An identifier in math: `pi`.
     MathIdent(MathIdent<'a>),
+    /// A shorthand for a unicode codepoint in math: `a <= b`.
+    MathShorthand(MathShorthand<'a>),
     /// An alignment point in math: `&`.
     MathAlignPoint(MathAlignPoint<'a>),
     /// Matched delimiters in math: `[x + y]`.
@@ -232,6 +234,7 @@ impl<'a> AstNode<'a> for Expr<'a> {
             SyntaxKind::Equation => node.cast().map(Self::Equation),
             SyntaxKind::Math => node.cast().map(Self::Math),
             SyntaxKind::MathIdent => node.cast().map(Self::MathIdent),
+            SyntaxKind::MathShorthand => node.cast().map(Self::MathShorthand),
             SyntaxKind::MathAlignPoint => node.cast().map(Self::MathAlignPoint),
             SyntaxKind::MathDelimited => node.cast().map(Self::MathDelimited),
             SyntaxKind::MathAttach => node.cast().map(Self::MathAttach),
@@ -295,6 +298,7 @@ impl<'a> AstNode<'a> for Expr<'a> {
             Self::Equation(v) => v.to_untyped(),
             Self::Math(v) => v.to_untyped(),
             Self::MathIdent(v) => v.to_untyped(),
+            Self::MathShorthand(v) => v.to_untyped(),
             Self::MathAlignPoint(v) => v.to_untyped(),
             Self::MathDelimited(v) => v.to_untyped(),
             Self::MathAttach(v) => v.to_untyped(),
@@ -450,7 +454,7 @@ node! {
 
 impl Shorthand<'_> {
     /// A list of all shorthands in markup mode.
-    pub const MARKUP_LIST: &'static [(&'static str, char)] = &[
+    pub const LIST: &'static [(&'static str, char)] = &[
         ("...", '…'),
         ("~", '\u{00A0}'),
         ("-", '\u{2212}'), // Only before a digit
@@ -459,52 +463,11 @@ impl Shorthand<'_> {
         ("-?", '\u{00AD}'),
     ];
 
-    /// A list of all shorthands in math mode.
-    pub const MATH_LIST: &'static [(&'static str, char)] = &[
-        ("...", '…'),
-        ("-", '\u{2212}'),
-        ("'", '′'),
-        ("*", '∗'),
-        ("!=", '≠'),
-        (":=", '≔'),
-        ("::=", '⩴'),
-        ("=:", '≕'),
-        ("<<", '≪'),
-        ("<<<", '⋘'),
-        (">>", '≫'),
-        (">>>", '⋙'),
-        ("<=", '≤'),
-        (">=", '≥'),
-        ("->", '→'),
-        ("-->", '⟶'),
-        ("|->", '↦'),
-        (">->", '↣'),
-        ("->>", '↠'),
-        ("<-", '←'),
-        ("<--", '⟵'),
-        ("<-<", '↢'),
-        ("<<-", '↞'),
-        ("<->", '↔'),
-        ("<-->", '⟷'),
-        ("~>", '⇝'),
-        ("~~>", '⟿'),
-        ("<~", '⇜'),
-        ("<~~", '⬳'),
-        ("=>", '⇒'),
-        ("|=>", '⤇'),
-        ("==>", '⟹'),
-        ("<==", '⟸'),
-        ("<=>", '⇔'),
-        ("<==>", '⟺'),
-        ("[|", '⟦'),
-        ("|]", '⟧'),
-        ("||", '‖'),
-    ];
-
     /// Get the shorthanded character.
     pub fn get(self) -> char {
         let text = self.0.text();
-        (Self::MARKUP_LIST.iter().chain(Self::MATH_LIST))
+        Self::LIST
+            .iter()
             .find(|&&(s, _)| s == text)
             .map_or_else(char::default, |&(_, c)| c)
     }
@@ -771,6 +734,65 @@ impl Deref for MathIdent<'_> {
 }
 
 node! {
+    /// A shorthand for a unicode codepoint in math: `a <= b`.
+    MathShorthand
+}
+
+impl MathShorthand<'_> {
+    /// A list of all shorthands in math mode.
+    pub const LIST: &'static [(&'static str, char)] = &[
+        ("...", '…'),
+        ("-", '−'),
+        ("'", '′'),
+        ("*", '∗'),
+        ("~", '∼'),
+        ("!=", '≠'),
+        (":=", '≔'),
+        ("::=", '⩴'),
+        ("=:", '≕'),
+        ("<<", '≪'),
+        ("<<<", '⋘'),
+        (">>", '≫'),
+        (">>>", '⋙'),
+        ("<=", '≤'),
+        (">=", '≥'),
+        ("->", '→'),
+        ("-->", '⟶'),
+        ("|->", '↦'),
+        (">->", '↣'),
+        ("->>", '↠'),
+        ("<-", '←'),
+        ("<--", '⟵'),
+        ("<-<", '↢'),
+        ("<<-", '↞'),
+        ("<->", '↔'),
+        ("<-->", '⟷'),
+        ("~>", '⇝'),
+        ("~~>", '⟿'),
+        ("<~", '⇜'),
+        ("<~~", '⬳'),
+        ("=>", '⇒'),
+        ("|=>", '⤇'),
+        ("==>", '⟹'),
+        ("<==", '⟸'),
+        ("<=>", '⇔'),
+        ("<==>", '⟺'),
+        ("[|", '⟦'),
+        ("|]", '⟧'),
+        ("||", '‖'),
+    ];
+
+    /// Get the shorthanded character.
+    pub fn get(self) -> char {
+        let text = self.0.text();
+        Self::LIST
+            .iter()
+            .find(|&&(s, _)| s == text)
+            .map_or_else(char::default, |&(_, c)| c)
+    }
+}
+
+node! {
     /// An alignment point in math: `&`.
     MathAlignPoint
 }
diff --git a/crates/typst-syntax/src/highlight.rs b/crates/typst-syntax/src/highlight.rs
index 0c1f3d5f..ddd29326 100644
--- a/crates/typst-syntax/src/highlight.rs
+++ b/crates/typst-syntax/src/highlight.rs
@@ -172,6 +172,7 @@ pub fn highlight(node: &LinkedNode) -> Option<Tag> {
 
         SyntaxKind::Math => None,
         SyntaxKind::MathIdent => highlight_ident(node),
+        SyntaxKind::MathShorthand => Some(Tag::Escape),
         SyntaxKind::MathAlignPoint => Some(Tag::MathOperator),
         SyntaxKind::MathDelimited => None,
         SyntaxKind::MathAttach => None,
diff --git a/crates/typst-syntax/src/kind.rs b/crates/typst-syntax/src/kind.rs
index 7505dbc6..a4456b9f 100644
--- a/crates/typst-syntax/src/kind.rs
+++ b/crates/typst-syntax/src/kind.rs
@@ -75,6 +75,8 @@ pub enum SyntaxKind {
     Math,
     /// An identifier in math: `pi`.
     MathIdent,
+    /// A shorthand for a unicode codepoint in math: `a <= b`.
+    MathShorthand,
     /// An alignment point in math: `&`.
     MathAlignPoint,
     /// Matched delimiters in math: `[x + y]`.
@@ -400,6 +402,7 @@ impl SyntaxKind {
             Self::Equation => "equation",
             Self::Math => "math",
             Self::MathIdent => "math identifier",
+            Self::MathShorthand => "math shorthand",
             Self::MathAlignPoint => "math alignment point",
             Self::MathDelimited => "delimited math",
             Self::MathAttach => "math attachments",
diff --git a/crates/typst-syntax/src/lexer.rs b/crates/typst-syntax/src/lexer.rs
index 993af080..92e78b2d 100644
--- a/crates/typst-syntax/src/lexer.rs
+++ b/crates/typst-syntax/src/lexer.rs
@@ -514,42 +514,42 @@ impl Lexer<'_> {
             '\\' => self.backslash(),
             '"' => self.string(),
 
-            '-' if self.s.eat_if(">>") => SyntaxKind::Shorthand,
-            '-' if self.s.eat_if('>') => SyntaxKind::Shorthand,
-            '-' if self.s.eat_if("->") => SyntaxKind::Shorthand,
-            ':' if self.s.eat_if('=') => SyntaxKind::Shorthand,
-            ':' if self.s.eat_if(":=") => SyntaxKind::Shorthand,
-            '!' if self.s.eat_if('=') => SyntaxKind::Shorthand,
-            '.' if self.s.eat_if("..") => SyntaxKind::Shorthand,
-            '[' if self.s.eat_if('|') => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if("==>") => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if("-->") => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if("--") => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if("-<") => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if("->") => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if("<-") => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if("<<") => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if("=>") => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if("==") => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if("~~") => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if('=') => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if('<') => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if('-') => SyntaxKind::Shorthand,
-            '<' if self.s.eat_if('~') => SyntaxKind::Shorthand,
-            '>' if self.s.eat_if("->") => SyntaxKind::Shorthand,
-            '>' if self.s.eat_if(">>") => SyntaxKind::Shorthand,
-            '=' if self.s.eat_if("=>") => SyntaxKind::Shorthand,
-            '=' if self.s.eat_if('>') => SyntaxKind::Shorthand,
-            '=' if self.s.eat_if(':') => SyntaxKind::Shorthand,
-            '>' if self.s.eat_if('=') => SyntaxKind::Shorthand,
-            '>' if self.s.eat_if('>') => SyntaxKind::Shorthand,
-            '|' if self.s.eat_if("->") => SyntaxKind::Shorthand,
-            '|' if self.s.eat_if("=>") => SyntaxKind::Shorthand,
-            '|' if self.s.eat_if(']') => SyntaxKind::Shorthand,
-            '|' if self.s.eat_if('|') => SyntaxKind::Shorthand,
-            '~' if self.s.eat_if("~>") => SyntaxKind::Shorthand,
-            '~' if self.s.eat_if('>') => SyntaxKind::Shorthand,
-            '*' | '-' => SyntaxKind::Shorthand,
+            '-' if self.s.eat_if(">>") => SyntaxKind::MathShorthand,
+            '-' if self.s.eat_if('>') => SyntaxKind::MathShorthand,
+            '-' if self.s.eat_if("->") => SyntaxKind::MathShorthand,
+            ':' if self.s.eat_if('=') => SyntaxKind::MathShorthand,
+            ':' if self.s.eat_if(":=") => SyntaxKind::MathShorthand,
+            '!' if self.s.eat_if('=') => SyntaxKind::MathShorthand,
+            '.' if self.s.eat_if("..") => SyntaxKind::MathShorthand,
+            '[' if self.s.eat_if('|') => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if("==>") => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if("-->") => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if("--") => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if("-<") => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if("->") => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if("<-") => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if("<<") => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if("=>") => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if("==") => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if("~~") => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if('=') => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if('<') => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if('-') => SyntaxKind::MathShorthand,
+            '<' if self.s.eat_if('~') => SyntaxKind::MathShorthand,
+            '>' if self.s.eat_if("->") => SyntaxKind::MathShorthand,
+            '>' if self.s.eat_if(">>") => SyntaxKind::MathShorthand,
+            '=' if self.s.eat_if("=>") => SyntaxKind::MathShorthand,
+            '=' if self.s.eat_if('>') => SyntaxKind::MathShorthand,
+            '=' if self.s.eat_if(':') => SyntaxKind::MathShorthand,
+            '>' if self.s.eat_if('=') => SyntaxKind::MathShorthand,
+            '>' if self.s.eat_if('>') => SyntaxKind::MathShorthand,
+            '|' if self.s.eat_if("->") => SyntaxKind::MathShorthand,
+            '|' if self.s.eat_if("=>") => SyntaxKind::MathShorthand,
+            '|' if self.s.eat_if(']') => SyntaxKind::MathShorthand,
+            '|' if self.s.eat_if('|') => SyntaxKind::MathShorthand,
+            '~' if self.s.eat_if("~>") => SyntaxKind::MathShorthand,
+            '~' if self.s.eat_if('>') => SyntaxKind::MathShorthand,
+            '*' | '-' | '~' => SyntaxKind::MathShorthand,
 
             '#' => SyntaxKind::Hash,
             '_' => SyntaxKind::Underscore,
diff --git a/crates/typst-syntax/src/parser.rs b/crates/typst-syntax/src/parser.rs
index 2cbce1d7..4ce3917e 100644
--- a/crates/typst-syntax/src/parser.rs
+++ b/crates/typst-syntax/src/parser.rs
@@ -319,7 +319,7 @@ fn math_expr_prec(p: &mut Parser, min_prec: usize, stop: SyntaxKind) {
             }
         }
 
-        SyntaxKind::Text | SyntaxKind::Shorthand => {
+        SyntaxKind::Text | SyntaxKind::MathShorthand => {
             continuable = matches!(
                 math_class(p.current_text()),
                 None | Some(MathClass::Alphabetic)
diff --git a/crates/typst-syntax/src/set.rs b/crates/typst-syntax/src/set.rs
index 39e64651..2297cacb 100644
--- a/crates/typst-syntax/src/set.rs
+++ b/crates/typst-syntax/src/set.rs
@@ -80,7 +80,7 @@ pub const MATH_EXPR: SyntaxSet = SyntaxSet::new()
     .add(SyntaxKind::Hash)
     .add(SyntaxKind::MathIdent)
     .add(SyntaxKind::Text)
-    .add(SyntaxKind::Shorthand)
+    .add(SyntaxKind::MathShorthand)
     .add(SyntaxKind::Linebreak)
     .add(SyntaxKind::MathAlignPoint)
     .add(SyntaxKind::Escape)