fix!: make start/end tag name spans encoding-independent

3 files changed, 40 insertions, 22 deletions

diff --git a/src/emitter.rs b/src/emitter.rs
index ff6e863..aa84215 100644
--- a/src/emitter.rs
+++ b/src/emitter.rs
@@ -49,10 +49,10 @@ pub trait Emitter<O> {
     fn emit_string(&mut self, c: &str);
 
     /// Set the _current token_ to a start tag.
-    fn init_start_tag(&mut self, offset: O);
+    fn init_start_tag(&mut self, tag_offset: O, name_offset: O);
 
     /// Set the _current token_ to an end tag.
-    fn init_end_tag(&mut self, offset: O);
+    fn init_end_tag(&mut self, tag_offset: O, name_offset: O);
 
     /// Set the _current token_ to a comment.
     fn init_comment(&mut self, data_offset: O);
@@ -78,6 +78,11 @@ pub trait Emitter<O> {
     /// If the current token is not a doctype, this method may panic.
     fn emit_current_doctype(&mut self, offset: O);
 
+    /// Called after the last [`push_tag_name`] call for a tag name.
+    ///
+    /// [`push_tag_name`]: Self::push_tag_name
+    fn terminate_tag_name(&mut self, offset: O) {}
+
     /// Called after the last [`push_attribute_value`] call for an attribute value.
     ///
     /// [`push_attribute_value`]: Self::push_attribute_value
@@ -273,18 +278,20 @@ impl<O: Offset> Emitter<O> for DefaultEmitter<O> {
         self.current_characters.push_str(s);
     }
 
-    fn init_start_tag(&mut self, offset: O) {
+    fn init_start_tag(&mut self, tag_offset: O, name_offset: O) {
         self.current_token = Some(Token::StartTag(StartTag {
-            span: offset..O::default(),
+            span: tag_offset..O::default(),
             self_closing: false,
             name: String::new(),
             attributes: Default::default(),
+            name_span: name_offset..O::default(),
         }));
     }
-    fn init_end_tag(&mut self, offset: O) {
+    fn init_end_tag(&mut self, tag_offset: O, name_offset: O) {
         self.current_token = Some(Token::EndTag(EndTag {
-            span: offset..O::default(),
+            span: tag_offset..O::default(),
             name: String::new(),
+            name_span: name_offset..O::default(),
         }));
         self.seen_attributes.clear();
     }
@@ -367,6 +374,22 @@ impl<O: Offset> Emitter<O> for DefaultEmitter<O> {
         }
     }
 
+    fn terminate_tag_name(&mut self, offset: O) {
+        match self.current_token {
+            Some(Token::StartTag(StartTag {
+                ref mut name_span, ..
+            })) => {
+                name_span.end = offset;
+            }
+            Some(Token::EndTag(EndTag {
+                ref mut name_span, ..
+            })) => {
+                name_span.end = offset;
+            }
+            _ => debug_assert!(false),
+        }
+    }
+
     fn push_comment(&mut self, s: &str) {
         match self.current_token {
             Some(Token::Comment(Comment { ref mut data, .. })) => data.push_str(s),
@@ -483,14 +506,9 @@ pub struct StartTag<O> {
 
     /// The source code span of the tag.
     pub span: Range<O>,
-}
 
-impl<O: Offset> StartTag<O> {
-    /// Calculates the span for the tag name and returns it.
-    pub fn name_span(&self) -> Range<O> {
-        let start = self.span.start + b"<".len();
-        start..start + self.name.len()
-    }
+    /// The span of the tag name.
+    pub name_span: Range<O>,
 }
 
 /// An HTML end/close tag, such as `</p>` or `</a>`.
@@ -502,14 +520,9 @@ pub struct EndTag<O> {
 
     /// The source code span of the tag.
     pub span: Range<O>,
-}
 
-impl<O: Offset> EndTag<O> {
-    /// Calculates the span for the tag name and returns it.
-    pub fn name_span(&self) -> Range<O> {
-        let start = self.span.start + b"</".len();
-        start..start + self.name.len()
-    }
+    /// The span of the tag name.
+    pub name_span: Range<O>,
 }
 
 /// An HTML comment.
diff --git a/src/machine.rs b/src/machine.rs
index 5b36eee..c27708d 100644
--- a/src/machine.rs
+++ b/src/machine.rs
@@ -190,14 +190,17 @@ where
         },
         State::TagName => match slf.read_char()? {
             Some(whitespace_pat!()) => {
+                slf.emitter.terminate_tag_name(slf.position_before_match);
                 slf.state = State::BeforeAttributeName;
                 Ok(ControlToken::Continue)
             }
             Some('/') => {
+                slf.emitter.terminate_tag_name(slf.position_before_match);
                 slf.state = State::SelfClosingStartTag;
                 Ok(ControlToken::Continue)
             }
             Some('>') => {
+                slf.emitter.terminate_tag_name(slf.position_before_match);
                 slf.state = State::Data;
                 slf.emit_current_tag();
                 Ok(ControlToken::Continue)
diff --git a/src/tokenizer.rs b/src/tokenizer.rs
index 58f7b80..e0402b9 100644
--- a/src/tokenizer.rs
+++ b/src/tokenizer.rs
@@ -221,14 +221,16 @@ impl<R: Reader + Position<O>, O: Offset, E: Emitter<O>> Tokenizer<R, O, E> {
 
     #[inline]
     pub(crate) fn init_start_tag(&mut self) {
-        self.emitter.init_start_tag(self.some_offset);
+        self.emitter
+            .init_start_tag(self.some_offset, self.position_before_match);
         self.current_tag_name.clear();
         self.is_start_tag = true;
     }
 
     #[inline]
     pub(crate) fn init_end_tag(&mut self) {
-        self.emitter.init_end_tag(self.some_offset);
+        self.emitter
+            .init_end_tag(self.some_offset, self.position_before_match);
         self.current_tag_name.clear();
         self.is_start_tag = false;
     }