Linebreaking for links

2 files changed, 41 insertions, 31 deletions

diff --git a/crates/typst-syntax/src/lexer.rs b/crates/typst-syntax/src/lexer.rs
index 18622154..a909dfa0 100644
--- a/crates/typst-syntax/src/lexer.rs
+++ b/crates/typst-syntax/src/lexer.rs
@@ -253,43 +253,16 @@ impl Lexer<'_> {
     }
 
     fn link(&mut self) -> SyntaxKind {
-        let mut brackets = Vec::new();
-
-        #[rustfmt::skip]
-        self.s.eat_while(|c: char| {
-            match c {
-                | '0' ..= '9'
-                | 'a' ..= 'z'
-                | 'A' ..= 'Z'
-                | '!' | '#' | '$' | '%' | '&' | '*' | '+'
-                | ',' | '-' | '.' | '/' | ':' | ';' | '='
-                | '?' | '@' | '_' | '~' | '\'' => true,
-                '[' => {
-                    brackets.push(SyntaxKind::LeftBracket);
-                    true
-                }
-                '(' => {
-                    brackets.push(SyntaxKind::LeftParen);
-                    true
-                }
-                ']' => brackets.pop() == Some(SyntaxKind::LeftBracket),
-                ')' => brackets.pop() == Some(SyntaxKind::LeftParen),
-                _ => false,
-            }
-        });
+        let (link, balanced) = link_prefix(self.s.after());
+        self.s.jump(self.s.cursor() + link.len());
 
-        if !brackets.is_empty() {
+        if !balanced {
             return self.error(
                 "automatic links cannot contain unbalanced brackets, \
                  use the `link` function instead",
             );
         }
 
-        // Don't include the trailing characters likely to be part of text.
-        while matches!(self.s.scout(-1), Some('!' | ',' | '.' | ':' | ';' | '?' | '\'')) {
-            self.s.uneat();
-        }
-
         SyntaxKind::Link
     }
 
@@ -662,6 +635,43 @@ pub fn is_newline(character: char) -> bool {
     )
 }
 
+/// Extracts a prefix of the text that is a link and also returns whether the
+/// parentheses and brackets in the link were balanced.
+pub fn link_prefix(text: &str) -> (&str, bool) {
+    let mut s = unscanny::Scanner::new(text);
+    let mut brackets = Vec::new();
+
+    #[rustfmt::skip]
+    s.eat_while(|c: char| {
+        match c {
+            | '0' ..= '9'
+            | 'a' ..= 'z'
+            | 'A' ..= 'Z'
+            | '!' | '#' | '$' | '%' | '&' | '*' | '+'
+            | ',' | '-' | '.' | '/' | ':' | ';' | '='
+            | '?' | '@' | '_' | '~' | '\'' => true,
+            '[' => {
+                brackets.push(b'[');
+                true
+            }
+            '(' => {
+                brackets.push(b'(');
+                true
+            }
+            ']' => brackets.pop() == Some(b'['),
+            ')' => brackets.pop() == Some(b'('),
+            _ => false,
+        }
+    });
+
+    // Don't include the trailing characters likely to be part of text.
+    while matches!(s.scout(-1), Some('!' | ',' | '.' | ':' | ';' | '?' | '\'')) {
+        s.uneat();
+    }
+
+    (s.before(), brackets.is_empty())
+}
+
 /// Split text at newlines.
 pub(super) fn split_newlines(text: &str) -> Vec<&str> {
     let mut s = Scanner::new(text);
diff --git a/crates/typst-syntax/src/lib.rs b/crates/typst-syntax/src/lib.rs
index 4ee37096..5cf740e7 100644
--- a/crates/typst-syntax/src/lib.rs
+++ b/crates/typst-syntax/src/lib.rs
@@ -15,7 +15,7 @@ mod span;
 pub use self::file::{FileId, PackageSpec, PackageVersion, VirtualPath};
 pub use self::highlight::{highlight, highlight_html, Tag};
 pub use self::kind::SyntaxKind;
-pub use self::lexer::{is_id_continue, is_id_start, is_ident, is_newline};
+pub use self::lexer::{is_id_continue, is_id_start, is_ident, is_newline, link_prefix};
 pub use self::node::{LinkedChildren, LinkedNode, SyntaxError, SyntaxNode};
 pub use self::parser::{parse, parse_code, parse_math};
 pub use self::source::Source;