Merge pull request #2927 from ehuss/fix-unbalanced-html-in-header

Handle unclosed HTML tags inside a markdown element

Author: ehuss

crates/mdbook-html/src/html/tree.rs

@@ -306,25 +306,7 @@ where
             trace!("event={event:?}");
             match event {
                 Event::Start(tag) => self.start_tag(tag),
-                Event::End(tag) => {
-                    // TODO: This should validate that the event stack is
-                    // properly synchronized with the tag stack.
-                    self.pop();
-                    match tag {
-                        TagEnd::TableHead => {
-                            self.table_state = TableState::Body;
-                            self.push(Node::Element(Element::new("tbody")));
-                        }
-                        TagEnd::TableCell => {
-                            self.table_cell_index += 1;
-                        }
-                        TagEnd::Table => {
-                            // Pop tbody or thead
-                            self.pop();
-                        }
-                        _ => {}
-                    }
-                }
+                Event::End(tag) => self.end_tag(tag),
                 Event::Text(text) => {
                     self.append_text(text.into_tendril());
                 }
@@ -600,6 +582,46 @@ where
         self.push(Node::Element(element));
     }
 
+    fn end_tag(&mut self, tag: TagEnd) {
+        // TODO: This should validate that the event stack is properly
+        // synchronized with the tag stack. That, would likely require keeping
+        // a parallel "expected end tag" with the tag stack, since mapping a
+        // pulldown-cmark event tag to an HTML tag isn't always clear.
+        //
+        // Check for unclosed HTML tags when exiting a markdown event.
+        while let Some(node_id) = self.tag_stack.last() {
+            let node = self.tree.get(*node_id).unwrap().value();
+            let Node::Element(el) = node else {
+                break;
+            };
+            if !el.was_raw {
+                break;
+            }
+            warn!(
+                "unclosed HTML tag `<{}>` found in `{}` while exiting {tag:?}\n\
+                HTML tags must be closed before exiting a markdown element.",
+                el.name.local,
+                self.options.path.display(),
+            );
+            self.pop();
+        }
+        self.pop();
+        match tag {
+            TagEnd::TableHead => {
+                self.table_state = TableState::Body;
+                self.push(Node::Element(Element::new("tbody")));
+            }
+            TagEnd::TableCell => {
+                self.table_cell_index += 1;
+            }
+            TagEnd::Table => {
+                // Pop tbody or thead
+                self.pop();
+            }
+            _ => {}
+        }
+    }
+
     /// Given some HTML, parse it into [`Node`] elements and append them to
     /// the current node.
     fn append_html(&mut self, html: &str) {

tests/testsuite/rendering.rs

@@ -283,3 +283,24 @@ Check that the HTML tags are properly balanced.
         })
         .check_main_file("book/chapter_1.html", str!["<div>x<span>foo</span></div>"]);
 }
+
+// Test for bug with unbalanced HTML handling in the heading.
+#[test]
+fn heading_with_unbalanced_html() {
+    BookTest::init(|_| {})
+        .change_file("src/chapter_1.md", "### Option<T>")
+        .run("build", |cmd| {
+            cmd.expect_stderr(str![[r#"
+ INFO Book building has started
+ INFO Running the html backend
+ WARN unclosed HTML tag `<t>` found in `chapter_1.md` while exiting Heading(H3)
+HTML tags must be closed before exiting a markdown element.
+ INFO HTML book written to `[ROOT]/book`
+
+"#]]);
+        })
+        .check_main_file(
+            "book/chapter_1.html",
+            str![[r##"<h3 id="option"><a class="header" href="#option">Option<t></t></a></h3>"##]],
+        );
+}