fix!: add adjusted_current_node_present_and_not_in_html_namespace to Emitter

1 files changed, 24 insertions, 0 deletions

diff --git a/src/emitter.rs b/src/emitter.rs
index ed8d9e9..30e1d17 100644
--- a/src/emitter.rs
+++ b/src/emitter.rs
@@ -156,6 +156,13 @@ pub trait Emitter<O> {
     ///
     /// If the current token is not a doctype, this method may panic.
     fn push_doctype_system_id(&mut self, s: &str);
+
+    /// Returns true if there is an _adjusted current node_ and it is not an element in the HTML namespace.
+    ///
+    /// See the third list item under [Markup declaration open state].
+    ///
+    /// [Markup declaration open state]: https://html.spec.whatwg.org/multipage/parsing.html#markup-declaration-open-state
+    fn adjusted_current_node_present_and_not_in_html_namespace(&mut self) -> bool;
 }
 
 /// The default implementation of [`Emitter`], used to produce tokens.
@@ -174,6 +181,19 @@ pub trait Emitter<O> {
 ///   assert!(matches!(tokens.next().unwrap(), Token::StartTag(tag) if tag.name == "b"));
 ///   ```
 ///
+/// * The DefaultEmitter implements [`Emitter::adjusted_current_node_present_and_not_in_html_namespace`]
+///   by returning false, which results in all CDATA sections being tokenized as bogus comments.
+///
+///   ```
+///   # use html5tokenizer::{DefaultEmitter, Tokenizer, Token};
+///   let emitter = DefaultEmitter::default();
+///   let html = "<svg><![CDATA[I love SVG]]>";
+///   let mut tokens = Tokenizer::new(html, emitter).flatten();
+///   assert!(matches!(tokens.next().unwrap(), Token::StartTag(tag) if tag.name == "svg"));
+///   assert!(matches!(tokens.next().unwrap(), Token::Error {..}));
+///   assert!(matches!(tokens.next().unwrap(), Token::Comment(_bogus_comment)));
+///   ```
+///
 /// [`Tokenizer::set_state`]: crate::Tokenizer::set_state
 pub struct DefaultEmitter<O = NoopOffset> {
     current_characters: String,
@@ -447,6 +467,10 @@ impl<O: Offset> Emitter<O> for DefaultEmitter<O> {
             debug_assert!(false);
         }
     }
+
+    fn adjusted_current_node_present_and_not_in_html_namespace(&mut self) -> bool {
+        false
+    }
 }
 
 /// An HTML start tag, such as `<p>` or `<a>`.