fix block comment parsing with proper nesting support

core/src/doc.rs

@@ -88,36 +88,40 @@ fn is_ident(s: &str) -> bool {
 
 pub fn classify_document(text: &str) -> Vec<ClassifiedLine> {
     let mut result = Vec::new();
-    let mut in_block_comment = false;
+    let mut depth: usize = 0;
 
     for (i, line) in text.lines().enumerate() {
-        if in_block_comment {
+        let was_in_comment = depth > 0;
-            let cl = ClassifiedLine { index: i, kind: LineKind::Comment, content: line.to_string() };
+        depth = scan_comment_depth(line, depth);
-            if line.contains("*/") {
-                in_block_comment = false;
-            }
-            result.push(cl);
-            continue;
-        }
 
-        let trimmed = line.trim();
+        if was_in_comment || line.trim().starts_with("/*") {
-
-        if trimmed.starts_with("/*") {
-            if trimmed.contains("*/") && trimmed.find("*/").unwrap() > trimmed.find("/*").unwrap() {
-                // single-line block comment
-            } else {
-                in_block_comment = true;
-            }
             result.push(ClassifiedLine { index: i, kind: LineKind::Comment, content: line.to_string() });
-            continue;
+        } else {
+            result.push(classify_line(i, line));
         }
-
-        result.push(classify_line(i, line));
     }
 
     result
 }
 
+fn scan_comment_depth(line: &str, mut depth: usize) -> usize {
+    let bytes = line.as_bytes();
+    let len = bytes.len();
+    let mut i = 0;
+    while i < len.saturating_sub(1) {
+        if bytes[i] == b'/' && bytes[i + 1] == b'*' {
+            depth += 1;
+            i += 2;
+        } else if bytes[i] == b'*' && bytes[i + 1] == b'/' {
+            depth = depth.saturating_sub(1);
+            i += 2;
+        } else {
+            i += 1;
+        }
+    }
+    depth
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
@@ -199,4 +203,22 @@ mod tests {
         assert_eq!(lines[0].kind, LineKind::Comment);
         assert_eq!(lines[1].kind, LineKind::Eval);
     }
+
+    #[test]
+    fn nested_block_comments() {
+        let lines = classify_document("/* outer /* inner */ still comment */\nlet x = 5");
+        assert_eq!(lines[0].kind, LineKind::Comment);
+        assert_eq!(lines[1].kind, LineKind::Cordial);
+    }
+
+    #[test]
+    fn nested_multiline_block_comments() {
+        let doc = "/* outer\n/* inner */\nstill in outer\n*/\nlet x = 5";
+        let lines = classify_document(doc);
+        assert_eq!(lines[0].kind, LineKind::Comment);
+        assert_eq!(lines[1].kind, LineKind::Comment);
+        assert_eq!(lines[2].kind, LineKind::Comment);
+        assert_eq!(lines[3].kind, LineKind::Comment);
+        assert_eq!(lines[4].kind, LineKind::Cordial);
+    }
 }