fix!: make attribute spans encoding-independent

3 files changed, 28 insertions, 21 deletions

diff --git a/src/attr.rs b/src/attr.rs
index 72dfcd9..096235e 100644
--- a/src/attr.rs
+++ b/src/attr.rs
@@ -30,8 +30,8 @@ pub struct AttributeMap<O> {
 #[derive(Debug, Eq, PartialEq)]
 pub(crate) struct AttrInternal<O> {
     pub value: String,
-    /// The start offset of the attribute name.
-    pub name_offset: O,
+    /// The span of the attribute name.
+    pub name_span: Range<O>,
     /// The span of the attribute value.
     /// For the empty attribute syntax this is just `O::default()..O::default()`.
     /// We intentionally don't use `Option<Range<O>>` here to spare us a byte (and padding) per attribute.
@@ -65,8 +65,8 @@ pub struct AttributeOwned<O> {
     pub name: String,
     /// The attribute value. Character references have been resolved.
     pub value: String,
-    /// The start offset of the attribute name.
-    pub name_offset: O,
+    /// The span of the attribute name.
+    pub name_span: Range<O>,
     /// The span of the attribute value.
     /// `None` in case of the empty attribute syntax (e.g. `disabled` in `<input disabled>`).
     pub value_span: Option<Range<O>>,
@@ -99,9 +99,9 @@ impl<'a, O: Offset> Attribute<'a, O> {
         &self.map_val.value
     }
 
-    /// Calculates the span of the attribute name and returns it.
+    /// Returns the span of the attribute name.
     pub fn name_span(&self) -> Range<O> {
-        self.map_val.name_offset..self.map_val.name_offset + self.name.len()
+        self.map_val.name_span.clone()
     }
 
     /// For explicitly defined values returns the span of the attribute value.
@@ -157,7 +157,7 @@ impl<O> Iterator for AttrIntoIter<O> {
         Some(AttributeOwned {
             name,
             value: map_val.value,
-            name_offset: map_val.name_offset,
+            name_span: map_val.name_span,
             value_span: map_val.value_syntax.is_some().then_some(map_val.value_span),
             value_syntax: map_val.value_syntax,
         })
@@ -196,7 +196,7 @@ impl<O: Default> FromIterator<(String, String)> for AttributeMap<O> {
                         name,
                         AttrInternal {
                             value,
-                            name_offset: O::default(),
+                            name_span: O::default()..O::default(),
                             value_span: O::default()..O::default(),
                             value_syntax: Some(AttrValueSyntax::DoubleQuoted),
                         },
@@ -206,9 +206,3 @@ impl<O: Default> FromIterator<(String, String)> for AttributeMap<O> {
         }
     }
 }
-
-impl<O: Offset> AttrInternal<O> {
-    pub(crate) fn name_span(&self, name_len: usize) -> Range<O> {
-        self.name_offset..self.name_offset + name_len
-    }
-}
diff --git a/src/emitter.rs b/src/emitter.rs
index aa84215..bb12ca4 100644
--- a/src/emitter.rs
+++ b/src/emitter.rs
@@ -83,6 +83,11 @@ pub trait Emitter<O> {
     /// [`push_tag_name`]: Self::push_tag_name
     fn terminate_tag_name(&mut self, offset: O) {}
 
+    /// Called after the last [`push_attribute_name`] call for an attribute name.
+    ///
+    /// [`push_attribute_name`]: Self::push_attribute_name
+    fn terminate_attribute_name(&mut self, offset: O) {}
+
     /// Called after the last [`push_attribute_value`] call for an attribute value.
     ///
     /// [`push_attribute_value`]: Self::push_attribute_value
@@ -224,16 +229,14 @@ impl<O> DefaultEmitter<O> {
                     Entry::Vacant(vacant) => {
                         vacant.insert(map_val);
                     }
-                    Entry::Occupied(entry) => {
-                        let name_len = entry.key().len();
-                        self.push_error(Error::DuplicateAttribute, map_val.name_span(name_len));
+                    Entry::Occupied(_) => {
+                        self.push_error(Error::DuplicateAttribute, map_val.name_span);
                     }
                 },
                 Some(Token::EndTag(_)) => {
-                    let name_span = map_val.name_span(name.len());
-                    self.attr_in_end_tag_span = Some(name_span.clone());
+                    self.attr_in_end_tag_span = Some(map_val.name_span.clone());
                     if !self.seen_attributes.insert(name) {
-                        self.push_error(Error::DuplicateAttribute, name_span);
+                        self.push_error(Error::DuplicateAttribute, map_val.name_span);
                     }
                 }
                 _ => {
@@ -420,7 +423,7 @@ impl<O: Offset> Emitter<O> for DefaultEmitter<O> {
         self.current_attribute = Some((
             String::new(),
             crate::attr::AttrInternal {
-                name_offset: offset,
+                name_span: offset..O::default(),
                 value: String::new(),
                 value_span: O::default()..O::default(),
                 value_syntax: None,
@@ -437,6 +440,12 @@ impl<O: Offset> Emitter<O> for DefaultEmitter<O> {
         let current_attr = self.current_attribute.as_mut().unwrap();
         current_attr.0.push_str(s);
     }
+
+    fn terminate_attribute_name(&mut self, offset: O) {
+        let current_attr = self.current_attribute.as_mut().unwrap();
+        current_attr.1.name_span.end = offset;
+    }
+
     fn push_attribute_value(&mut self, s: &str) {
         let current_attr = self.current_attribute.as_mut().unwrap();
         current_attr.1.value.push_str(s);
diff --git a/src/machine.rs b/src/machine.rs
index c27708d..4f2d129 100644
--- a/src/machine.rs
+++ b/src/machine.rs
@@ -715,11 +715,15 @@ where
         },
         State::AttributeName => match slf.read_char()? {
             c @ Some(whitespace_pat!() | '/' | '>') | c @ None => {
+                slf.emitter
+                    .terminate_attribute_name(slf.position_before_match);
                 slf.state = State::AfterAttributeName;
                 slf.unread_char(c);
                 Ok(ControlToken::Continue)
             }
             Some('=') => {
+                slf.emitter
+                    .terminate_attribute_name(slf.position_before_match);
                 slf.state = State::BeforeAttributeValue;
                 Ok(ControlToken::Continue)
             }