fix(ast): Improve UTF-8 handling

2 files changed, 48 insertions, 5 deletions

diff --git a/src/ast/container.rs b/src/ast/container.rs
index bf4166c..46c548b 100644
--- a/src/ast/container.rs
+++ b/src/ast/container.rs
@@ -224,10 +224,14 @@ impl Ast {
 
           nodes.push(Node::Heading {
             level,
-            // Here, we are `get`ing the `&str` starting at the `level`-th
-            // index, then trimming the start. These operations
-            // effectively off the line identifier.
-            text: line.get(level..).unwrap_or("").trim_start().to_string(),
+            // Here, the text after the heading markers is safely extracted.
+            // `chars().skip()` is used to safely handle UTF-8 boundaries.
+            text: line
+              .chars()
+              .skip(level)
+              .collect::<String>()
+              .trim_start()
+              .to_string(),
           });
 
           break;
diff --git a/tests/ast.rs b/tests/ast.rs
index 7a71f5c..5249931 100644
--- a/tests/ast.rs
+++ b/tests/ast.rs
@@ -95,7 +95,46 @@ mod test {
       assert_eq!(to, "");
       assert_eq!(text, &None);
     } else {
-      panic!("Expected Link node");
+      panic!("Expected link node");
+    }
+  }
+
+  #[test]
+  fn build_heading_with_unicode_and_edge_cases() {
+    // Unicode characters
+    let ast = Ast::from_string("# Hello, 世界!");
+
+    assert_eq!(ast.inner().len(), 1);
+
+    if let Node::Heading { level, text } = ast.inner().first().unwrap() {
+      assert_eq!(level, &1);
+      assert_eq!(text, "Hello, 世界!");
+    } else {
+      panic!("Expected heading node");
+    }
+
+    // Only hashes
+    let ast = Ast::from_string("###");
+
+    assert_eq!(ast.inner().len(), 1);
+
+    if let Node::Heading { level, text } = ast.inner().first().unwrap() {
+      assert_eq!(level, &3);
+      assert_eq!(text, "");
+    } else {
+      panic!("Expected heading node");
+    }
+
+    // Many hashes
+    let ast = Ast::from_string("########## Very Deep Heading");
+
+    assert_eq!(ast.inner().len(), 1);
+
+    if let Node::Heading { level, text } = ast.inner().first().unwrap() {
+      assert_eq!(level, &10);
+      assert_eq!(text, "Very Deep Heading");
+    } else {
+      panic!("Expected heading node");
     }
   }
 }