Relax <select> parser (WHATWG proposal) (#560)

* Relax selector parsing

Neither the proposal nor the test changes are merged yet. Also, the
tests are still failing.

html5lib/html5lib-tests#178
whatwg/html#10557

* fmt

* whoops

* bump html5lib-tests again

* renamed methods

* fixup

* fix select state

* fmt

* remove xxx

* fix input case

* wip

* select

* select

* fmt

* clippy

* ignore failing selectedcontent tests

* lint fix

Author: untitaker

html5ever/examples/arena.rs

@@ -55,6 +55,7 @@ pub struct Node<'arena> {
 }
 
 /// HTML node data which can be an element, a comment, a string, a DOCTYPE, etc...
+#[derive(Clone)]
 pub enum NodeData<'arena> {
     Document,
     Doctype {
@@ -338,6 +339,21 @@ impl<'arena> TreeSink for Sink<'arena> {
             new_parent.append(child)
         }
     }
+
+    fn clone_subtree(&self, node: &Self::Handle) -> Self::Handle {
+        // Allocate the new node in the arena using Clone
+        let cloned_node = self.arena.alloc(Node::new(node.data.clone()));
+
+        // Clone all children and append them
+        let mut child = node.first_child.get();
+        while let Some(current_child) = child {
+            let cloned_child = self.clone_subtree(&current_child);
+            cloned_node.append(cloned_child);
+            child = current_child.next_sibling.get();
+        }
+
+        cloned_node
+    }
 }
 
 /// In this example an "arena" is created and filled with the DOM nodes.
@@ -352,5 +368,33 @@ fn main() {
     io::stdin().read_to_end(&mut bytes).unwrap();
 
     let arena = typed_arena::Arena::new();
-    html5ever_parse_slice_into_arena(&bytes, &arena);
+    let dom = html5ever_parse_slice_into_arena(&bytes, &arena);
+
+    // Print the DOM structure
+    print_node(dom, 0);
+}
+
+fn print_node<'arena>(node: &Node<'arena>, depth: usize) {
+    let indent = "  ".repeat(depth);
+
+    match &node.data {
+        NodeData::Document => println!("{}Document", indent),
+        NodeData::Doctype { name, .. } => println!("{}<!DOCTYPE {}>", indent, name),
+        NodeData::Text { contents } => {
+            let text = contents.borrow();
+            if !text.trim().is_empty() {
+                println!("{}\"{}\"", indent, text.trim());
+            }
+        },
+        NodeData::Comment { contents } => println!("{}<!-- {} -->", indent, contents),
+        NodeData::Element { name, .. } => println!("{}<{}>", indent, name.local),
+        NodeData::ProcessingInstruction { target, .. } => println!("{}<?{}>", indent, target),
+    }
+
+    // Print all children
+    let mut child = node.first_child.get();
+    while let Some(current_child) = child {
+        print_node(current_child, depth + 1);
+        child = current_child.next_sibling.get();
+    }
 }

html5ever/examples/noop-tree-builder.rs

@@ -113,6 +113,11 @@ impl TreeSink for Sink {
     fn remove_from_parent(&self, _target: &usize) {}
     fn reparent_children(&self, _node: &usize, _new_parent: &usize) {}
     fn mark_script_already_started(&self, _node: &usize) {}
+
+    fn clone_subtree(&self, _node: &Self::Handle) -> Self::Handle {
+        // For this noop example, just return a new placeholder ID
+        self.get_id()
+    }
 }
 
 /// In this example we implement the TreeSink trait which takes each parsed elements and insert

html5ever/examples/print-tree-actions.rs

@@ -167,6 +167,12 @@ impl TreeSink for Sink {
     fn pop(&self, elem: &usize) {
         println!("Popped element {elem}");
     }
+
+    fn clone_subtree(&self, node: &Self::Handle) -> Self::Handle {
+        println!("Clone subtree for node {node}");
+        // For this example, just return a new placeholder ID
+        self.get_id()
+    }
 }
 
 /// Same example as the "noop-tree-builder", but this time every function implemented in our

html5ever/src/tree_builder/mod.rs

@@ -118,6 +118,9 @@ pub struct TreeBuilder<Handle, Sink> {
 
     /// Form element pointer.
     form_elem: RefCell<Option<Handle>>,
+
+    /// selectedcontent element pointer.
+    selectedcontent_elem: RefCell<Option<Handle>>,
     //§ END
     /// Frameset-ok flag.
     frameset_ok: Cell<bool>,
@@ -163,6 +166,7 @@ where
             active_formatting: Default::default(),
             head_elem: Default::default(),
             form_elem: Default::default(),
+            selectedcontent_elem: Default::default(),
             frameset_ok: Cell::new(true),
             ignore_lf: Default::default(),
             foster_parenting: Default::default(),
@@ -203,6 +207,7 @@ where
             active_formatting: Default::default(),
             head_elem: Default::default(),
             form_elem: RefCell::new(form_elem),
+            selectedcontent_elem: Default::default(),
             frameset_ok: Cell::new(true),
             ignore_lf: Default::default(),
             foster_parenting: Default::default(),
@@ -285,6 +290,10 @@ where
             tracer.trace_handle(form_elem);
         }
 
+        if let Some(selectedcontent_elem) = self.selectedcontent_elem.borrow().as_ref() {
+            tracer.trace_handle(selectedcontent_elem);
+        }
+
         if let Some(context_elem) = self.context_elem.borrow().as_ref() {
             tracer.trace_handle(context_elem);
         }
@@ -923,6 +932,7 @@ where
             .borrow_mut()
             .pop()
             .expect("no current element");
+
         self.sink.pop(&elem);
         elem
     }
@@ -1183,6 +1193,7 @@ where
         n
     }
 
+    /// Pop element until an element with the given name has been popped.
     fn pop_until_named(&self, name: LocalName) -> usize {
         self.pop_until(|p| *p.ns == ns!(html) && *p.local == name)
     }
@@ -1269,16 +1280,6 @@ where
                 _ => continue,
             };
             match *name {
-                local_name!("select") => {
-                    for ancestor in self.open_elems.borrow()[0..i].iter().rev() {
-                        if self.html_elem_named(ancestor, local_name!("template")) {
-                            return InsertionMode::InSelect;
-                        } else if self.html_elem_named(ancestor, local_name!("table")) {
-                            return InsertionMode::InSelectInTable;
-                        }
-                    }
-                    return InsertionMode::InSelect;
-                },
                 local_name!("td") | local_name!("th") => {
                     if !last {
                         return InsertionMode::InCell;
@@ -1401,6 +1402,12 @@ where
 
         self.insert_at(insertion_point, AppendNode(elem.clone()));
 
+        if qname.local == local_name!("selectedcontent")
+            && self.selectedcontent_elem.borrow().is_none()
+        {
+            *self.selectedcontent_elem.borrow_mut() = Some(elem.clone());
+        }
+
         match push {
             PushFlag::Push => self.push(&elem),
             PushFlag::NoPush => (),
@@ -1585,6 +1592,19 @@ where
         self.remove_from_stack(&node);
     }
 
+    fn maybe_clone_option_into_selectedcontent(&self, option: &Handle) {
+        if let Some(selectedcontent) = self.selectedcontent_elem.borrow().as_ref().cloned() {
+            self.clone_option_into_selectedcontent(option, &selectedcontent);
+        }
+    }
+
+    fn clone_option_into_selectedcontent(&self, option: &Handle, selectedcontent: &Handle) {
+        self.sink
+            .reparent_children(selectedcontent, &self.sink.get_document());
+        let cloned_option = self.sink.clone_subtree(option);
+        self.sink.reparent_children(&cloned_option, selectedcontent);
+    }
+
     //§ tree-construction
     fn is_foreign(&self, token: &Token) -> bool {
         if let Token::Eof = *token {