use lexer subtrees to parse decorators

5 files changed, 44 insertions, 59 deletions

diff --git a/crates/typst-syntax/src/highlight.rs b/crates/typst-syntax/src/highlight.rs
index 59cad25e..83c15b14 100644
--- a/crates/typst-syntax/src/highlight.rs
+++ b/crates/typst-syntax/src/highlight.rs
@@ -286,7 +286,6 @@ pub fn highlight(node: &LinkedNode) -> Option<Tag> {
         SyntaxKind::Destructuring => None,
         SyntaxKind::DestructAssignment => None,
 
-        SyntaxKind::DecoratorMarker => Some(Tag::Comment),
         SyntaxKind::Decorator => None,
 
         SyntaxKind::LineComment => Some(Tag::Comment),
diff --git a/crates/typst-syntax/src/kind.rs b/crates/typst-syntax/src/kind.rs
index b9ccf45c..50d40960 100644
--- a/crates/typst-syntax/src/kind.rs
+++ b/crates/typst-syntax/src/kind.rs
@@ -13,8 +13,8 @@ pub enum SyntaxKind {
     LineComment,
     /// A block comment: `/* ... */`.
     BlockComment,
-    /// A decorator's marker: `/!`
-    DecoratorMarker,
+    /// A decorator: `/! allow("warning")`
+    Decorator,
 
     /// The contents of a file or content block.
     Markup,
@@ -280,9 +280,6 @@ pub enum SyntaxKind {
     Destructuring,
     /// A destructuring assignment expression: `(x, y) = (1, 2)`.
     DestructAssignment,
-
-    /// A decorator: `/! allow("amogus")`
-    Decorator,
 }
 
 impl SyntaxKind {
@@ -360,7 +357,11 @@ impl SyntaxKind {
     pub fn is_trivia(self) -> bool {
         matches!(
             self,
-            Self::LineComment | Self::BlockComment | Self::Space | Self::Parbreak
+            Self::LineComment
+                | Self::BlockComment
+                | Self::Space
+                | Self::Parbreak
+                | Self::Decorator
         )
     }
 
@@ -376,7 +377,7 @@ impl SyntaxKind {
             Self::Error => "syntax error",
             Self::LineComment => "line comment",
             Self::BlockComment => "block comment",
-            Self::DecoratorMarker => "decorator marker",
+            Self::Decorator => "decorator",
             Self::Markup => "markup",
             Self::Text => "text",
             Self::Space => "space",
@@ -504,7 +505,6 @@ impl SyntaxKind {
             Self::FuncReturn => "`return` expression",
             Self::Destructuring => "destructuring pattern",
             Self::DestructAssignment => "destructuring assignment expression",
-            Self::Decorator => "decorator",
         }
     }
 }
diff --git a/crates/typst-syntax/src/lexer.rs b/crates/typst-syntax/src/lexer.rs
index 05b12570..341484c7 100644
--- a/crates/typst-syntax/src/lexer.rs
+++ b/crates/typst-syntax/src/lexer.rs
@@ -18,8 +18,10 @@ pub(super) struct Lexer<'s> {
     newline: bool,
     /// The state held by raw line lexing.
     raw: Vec<(SyntaxKind, usize)>,
-    /// The state held by decorator lexing.
-    decorator: Vec<(SyntaxKind, usize)>,
+    /// The subtree of tokens associated with this token.
+    /// The parser is responsible for converting this subtree into syntax nodes
+    /// matching this structure.
+    subtree: Vec<(SyntaxKind, usize)>,
     /// An error for the last token.
     error: Option<SyntaxError>,
 }
@@ -35,8 +37,6 @@ pub(super) enum LexMode {
     Code,
     /// The contents of a raw block.
     Raw,
-    /// The contents of a decorator.
-    Decorator,
 }
 
 impl<'s> Lexer<'s> {
@@ -49,7 +49,7 @@ impl<'s> Lexer<'s> {
             newline: false,
             error: None,
             raw: Vec::new(),
-            decorator: Vec::new(),
+            subtree: Vec::new(),
         }
     }
 
@@ -113,16 +113,9 @@ impl Lexer<'_> {
             return kind;
         }
 
-        if self.mode == LexMode::Decorator {
-            let Some((kind, end)) = self.decorator.pop() else {
-                return SyntaxKind::End;
-            };
-            self.s.jump(end);
-            return kind;
-        }
-
         self.newline = false;
         self.error = None;
+        self.subtree.clear();
         let start = self.s.cursor();
         match self.s.eat() {
             Some(c) if is_space(c, self.mode) => self.whitespace(start, c),
@@ -142,13 +135,17 @@ impl Lexer<'_> {
                 LexMode::Math => self.math(start, c),
                 LexMode::Code => self.code(start, c),
                 LexMode::Raw => unreachable!(),
-                LexMode::Decorator => unreachable!(),
             },
 
             None => SyntaxKind::End,
         }
     }
 
+    /// Takes the subtree associated with the latest token.
+    pub fn take_subtree(&mut self) -> Vec<(SyntaxKind, usize)> {
+        std::mem::take(&mut self.subtree)
+    }
+
     /// Eat whitespace characters greedily.
     fn whitespace(&mut self, start: usize, c: char) -> SyntaxKind {
         let more = self.s.eat_while(|c| is_space(c, self.mode));
@@ -194,15 +191,8 @@ impl Lexer<'_> {
 
         SyntaxKind::BlockComment
     }
-}
 
-/// Decorators.
-impl Lexer<'_> {
     fn decorator(&mut self) -> SyntaxKind {
-        let start = self.s.cursor() - 2;
-
-        self.decorator.clear();
-
         while !self.s.eat_newline() {
             let start = self.s.cursor();
             let token = match self.s.eat() {
@@ -215,25 +205,20 @@ impl Lexer<'_> {
                 Some(c @ '0'..='9') => self.number(start, c),
                 Some(',') => SyntaxKind::Comma,
                 Some(c) if is_id_start(c) => self.ident(start),
-                Some(c) => {
-                    return self.error(eco_format!(
-                        "the character {c} is not valid in a decorator"
-                    ))
-                }
+                Some(c) => self
+                    .error(eco_format!("the character {c} is not valid in a decorator")),
                 None => break,
             };
 
+            if token.is_error() {
+                return token;
+            }
+
             let end = self.s.cursor();
-            self.decorator.push((token, end));
+            self.subtree.push((token, end));
         }
 
-        // The saved tokens will be removed in reverse.
-        self.decorator.reverse();
-
-        // Already collected all we need from the decorator.
-        self.s.jump(start + 2);
-
-        SyntaxKind::DecoratorMarker
+        SyntaxKind::Decorator
     }
 }
 
diff --git a/crates/typst-syntax/src/parser.rs b/crates/typst-syntax/src/parser.rs
index 54417e01..e1d4bb95 100644
--- a/crates/typst-syntax/src/parser.rs
+++ b/crates/typst-syntax/src/parser.rs
@@ -108,7 +108,8 @@ fn markup_expr(p: &mut Parser, at_start: &mut bool) {
         SyntaxKind::Space
         | SyntaxKind::Parbreak
         | SyntaxKind::LineComment
-        | SyntaxKind::BlockComment => {
+        | SyntaxKind::BlockComment
+        | SyntaxKind::Decorator => {
             p.eat();
             return;
         }
@@ -121,7 +122,6 @@ fn markup_expr(p: &mut Parser, at_start: &mut bool) {
         | SyntaxKind::Link
         | SyntaxKind::Label => p.eat(),
 
-        SyntaxKind::DecoratorMarker => decorator(p),
         SyntaxKind::Hash => embedded_code_expr(p),
         SyntaxKind::Star => strong(p),
         SyntaxKind::Underscore => emph(p),
@@ -147,19 +147,6 @@ fn markup_expr(p: &mut Parser, at_start: &mut bool) {
     *at_start = false;
 }
 
-fn decorator(p: &mut Parser) {
-    let m = p.marker();
-    p.enter(LexMode::Decorator);
-    p.assert(SyntaxKind::DecoratorMarker);
-
-    while !p.end() {
-        p.eat();
-    }
-
-    p.exit();
-    p.wrap(m, SyntaxKind::Decorator);
-}
-
 /// Parses strong content: `*Strong*`.
 fn strong(p: &mut Parser) {
     const END: SyntaxSet = SyntaxSet::new()
@@ -1775,9 +1762,23 @@ impl<'s> Parser<'s> {
 
     fn save(&mut self) {
         let text = self.current_text();
+        let subtree = self.lexer.take_subtree();
         if self.at(SyntaxKind::Error) {
             let error = self.lexer.take_error().unwrap();
             self.nodes.push(SyntaxNode::error(error, text));
+        } else if !subtree.is_empty() {
+            let mut text_cursor = self.current_start;
+            let mut children = Vec::with_capacity(subtree.len());
+
+            for (kind, end) in subtree {
+                // Ensure no errors in the subtree
+                assert!(!kind.is_error());
+
+                children.push(SyntaxNode::leaf(kind, &self.text[text_cursor..end]));
+                text_cursor = end;
+            }
+
+            self.nodes.push(SyntaxNode::inner(self.current, children));
         } else {
             self.nodes.push(SyntaxNode::leaf(self.current, text));
         }
diff --git a/crates/typst-syntax/src/set.rs b/crates/typst-syntax/src/set.rs
index b5b19b46..3f5a2606 100644
--- a/crates/typst-syntax/src/set.rs
+++ b/crates/typst-syntax/src/set.rs
@@ -54,7 +54,7 @@ pub const MARKUP_EXPR: SyntaxSet = SyntaxSet::new()
     .add(SyntaxKind::Parbreak)
     .add(SyntaxKind::LineComment)
     .add(SyntaxKind::BlockComment)
-    .add(SyntaxKind::DecoratorMarker)
+    .add(SyntaxKind::Decorator)
     .add(SyntaxKind::Text)
     .add(SyntaxKind::Linebreak)
     .add(SyntaxKind::Escape)