Replace is_mathml_annotation_xml_integration_point with attr lookup.

Author: SimonSapin

html5ever/examples/noop-tree-builder.rs

@@ -61,6 +61,11 @@ impl TreeSink for Sink {
         self.names.get(target).expect("not an element").expanded()
     }
 
+    fn elem_any_attr<P>(&self, _target: &usize, _predicate: P) -> bool
+    where P: FnMut(ExpandedName, &str) -> bool {
+        false
+    }
+
     fn create_element(&mut self, name: QualName, _attrs: Vec<Attribute>) -> usize {
         let id = self.get_id();
         self.names.insert(id, name);

html5ever/examples/print-tree-actions.rs

@@ -65,6 +65,10 @@ impl TreeSink for Sink {
         self.names.get(target).expect("not an element").expanded()
     }
 
+    fn elem_any_attr<P>(&self, _target: &usize, _predicate: P) -> bool
+    where P: FnMut(ExpandedName, &str) -> bool {
+        false
+    }
 
     fn create_element(&mut self, name: QualName, _attrs: Vec<Attribute>) -> usize {
         let id = self.get_id();

html5ever/src/tree_builder/actions.rs

@@ -953,7 +953,16 @@ impl<Handle, Sink> TreeBuilderActions<Handle>
                 TagToken(Tag { kind: StartTag, name: local_name!("svg"), .. }) => return false,
                 CharacterTokens(..) | NullCharacterToken |
                 TagToken(Tag { kind: StartTag, .. }) => {
-                    return !self.sink.is_mathml_annotation_xml_integration_point(self.adjusted_current_node())
+                    let integration_point = self.sink.elem_any_attr(
+                        self.adjusted_current_node(),
+                        |attr_name, attr_value| {
+                            attr_name == expanded_name!("", "encoding") && (
+                                attr_value.eq_ignore_ascii_case("text/html") ||
+                                attr_value.eq_ignore_ascii_case("application/xhtml+xml")
+                            )
+                        }
+                    );
+                    return !integration_point
                 }
                 _ => {}
             };

html5ever/src/tree_builder/mod.rs

@@ -569,6 +569,11 @@ mod test {
             self.rcdom.elem_name(target)
         }
 
+        fn elem_any_attr<P>(&self, target: &Self::Handle, predicate: P) -> bool
+        where P: FnMut(ExpandedName, &str) -> bool {
+            self.rcdom.elem_any_attr(target, predicate)
+        }
+
         fn create_element(&mut self, name: QualName, attrs: Vec<Attribute>) -> Handle {
             self.line_vec.push((name.clone(), self.current_line));
             self.rcdom.create_element(name, attrs)
@@ -619,10 +624,6 @@ mod test {
             self.rcdom.mark_script_already_started(target);
         }
 
-        fn is_mathml_annotation_xml_integration_point(&self, handle: &Handle) -> bool {
-            self.rcdom.is_mathml_annotation_xml_integration_point(handle)
-        }
-
         fn set_current_line(&mut self, line_number: u64) {
             self.current_line = line_number;
         }

markup5ever/interface/tree_builder.rs

@@ -79,6 +79,13 @@ pub trait TreeSink {
     /// feel free to `panic!`.
     fn elem_name<'a>(&'a self, target: &'a Self::Handle) -> ExpandedName<'a>;
 
+    /// Does this element have an attribute matching the given predicate?
+    ///
+    /// Should never be called on a non-element node;
+    /// feel free to `panic!`.
+    fn elem_any_attr<P>(&self, target: &Self::Handle, predicate: P) -> bool
+    where P: FnMut(ExpandedName, &str) -> bool;
+
     /// Create an element.
     ///
     /// When creating a template element (`name.ns.expanded() == expanded_name!(html "template")`),
@@ -108,7 +115,7 @@ pub trait TreeSink {
                                   system_id: StrTendril);
 
     /// Mark a HTML `<script>` as "already started".
-    fn mark_script_already_started(&mut self, node: &Self::Handle);
+    fn mark_script_already_started(&mut self, _node: &Self::Handle) {}
 
     /// Indicate that a node was popped off the stack of open elements.
     fn pop(&mut self, _node: &Self::Handle) {}
@@ -157,12 +164,6 @@ pub trait TreeSink {
     /// Remove all the children from node and append them to new_parent.
     fn reparent_children(&mut self, node: &Self::Handle, new_parent: &Self::Handle);
 
-    /// Returns true if the adjusted current node is an HTML integration point
-    /// and the token is a start tag.
-    fn is_mathml_annotation_xml_integration_point(&self, _handle: &Self::Handle) -> bool {
-        false
-    }
-
     /// Called whenever the line number changes.
     fn set_current_line(&mut self, _line_number: u64) {}
 

markup5ever/rcdom.rs

@@ -12,7 +12,6 @@
 //! This is sufficient as a static parse tree, but don't build a
 //! web browser using it. :)
 
-use std::ascii::AsciiExt;
 use std::cell::{RefCell, Cell};
 use std::collections::HashSet;
 use std::default::Default;
@@ -33,7 +32,7 @@ use serialize::{Serialize, Serializer};
 use serialize::TraversalScope;
 use serialize::TraversalScope::{IncludeNode, ChildrenOnly};
 
-pub use self::ElementEnum::{AnnotationXml, Normal, Script, Template};
+pub use self::ElementEnum::{Normal, Script, Template};
 pub use self::NodeEnum::{Document, Doctype, Text, Comment, Element, PI};
 
 /// The different kinds of elements in the DOM.
@@ -45,11 +44,6 @@ pub enum ElementEnum {
     /// A template element and its template contents.
     /// https://html.spec.whatwg.org/multipage/#template-contents
     Template(Handle),
-    /// An annotation-xml element in the MathML namespace whose start tag token had an attribute
-    /// with the name "encoding" whose value was an ASCII case-insensitive match for the string
-    /// "text/html" or "application/xhtml+xml"
-    /// https://html.spec.whatwg.org/multipage/embedded-content.html#math:annotation-xml
-    AnnotationXml(bool),
 }
 
 /// The different kinds of nodes in the DOM.
@@ -196,18 +190,20 @@ impl TreeSink for RcDom {
         };
     }
 
+    fn elem_any_attr<P>(&self, target: &Self::Handle, mut predicate: P) -> bool
+    where P: FnMut(ExpandedName, &str) -> bool {
+        return match target.node {
+            Element(_, _, ref attrs) => {
+                attrs.borrow().iter().any(|a| predicate(a.name.expanded(), &*a.value))
+            }
+            _ => panic!("not an element!"),
+        };
+    }
+
     fn create_element(&mut self, name: QualName, attrs: Vec<Attribute>) -> Handle {
         let info = match name.expanded() {
             expanded_name!(html "script") => Script(Cell::new(false)),
             expanded_name!(html "template") => Template(new_node(Document)),
-            expanded_name!(mathml "annotation-xml") => {
-                AnnotationXml(attrs.iter()
-                    .find(|attr| attr.name.expanded() == expanded_name!("", "encoding"))
-                    .map_or(false, |attr| {
-                        attr.value.eq_ignore_ascii_case("text/html") ||
-                        attr.value.eq_ignore_ascii_case("application/xhtml+xml")
-                    }))
-            },
             _ => Normal,
         };
         new_node(Element(name, info, RefCell::new(attrs)))
@@ -319,13 +315,6 @@ impl TreeSink for RcDom {
             panic!("not a script element!");
         }
     }
-
-    fn is_mathml_annotation_xml_integration_point(&self, handle: &Handle) -> bool {
-        match handle.node {
-            Element(_, AnnotationXml(ret), _) => ret,
-            _ => unreachable!(),
-        }
-    }
 }
 
 impl Default for RcDom {