wip

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 {
@@ -338,8 +339,25 @@ 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.
 /// "Arena" is a type of allocation in which a block of memory is allocated
 /// and later filled with data, DOM nodes in this case. When the arena is deallocated
@@ -352,5 +370,33 @@ 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 +113,12 @@ 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 +167,13 @@ 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
     }
@@ -1392,6 +1402,10 @@ 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 => (),
@@ -1576,6 +1590,18 @@ 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 {

html5ever/src/tree_builder/rules.rs

@@ -587,6 +587,7 @@ where
                     ProcessResult::Done
                 },
 
+
                 Token::Tag(
                     tag @
                     tag!(</address> | </article> | </aside> | </blockquote> | </button> | </center> |
@@ -638,6 +639,33 @@ where
                     ProcessResult::Done
                 },
 
+                Token::Tag(tag!(</select>)) => {
+                    if self.in_scope_named(default_scope, local_name!("select")) {
+                        self.generate_implied_end_tags(cursory_implied_end);
+                        self.expect_to_close(local_name!("select"));
+                    } else {
+                        self.unexpected(&token);
+                    }
+                    ProcessResult::Done
+                },
+
+                Token::Tag(tag @ tag!(</option>)) => {
+                    let option_in_stack = self.open_elems.borrow()
+                        .iter()
+                        .find(|elem| self.html_elem_named(elem, local_name!("option")))
+                        .cloned();
+
+                    self.process_end_tag_in_body(tag);
+
+                    if let Some(option) = option_in_stack {
+                        if !self.open_elems.borrow().iter().any(|elem| self.sink.same_node(elem, &option)) {
+                            self.maybe_clone_option_into_selectedcontent(&option);
+                        }
+                    }
+
+                    ProcessResult::Done
+                },
+
                 Token::Tag(tag!(</p>)) => {
                     if !self.in_scope_named(button_scope, local_name!("p")) {
                         self.sink.parse_error(Borrowed("No <p> tag to close"));

markup5ever/interface/tree_builder.rs

@@ -236,6 +236,9 @@ pub trait TreeSink {
     /// Remove all the children from node and append them to new_parent.
     fn reparent_children(&self, node: &Self::Handle, new_parent: &Self::Handle);
 
+    /// Clone a node and all its descendants, returning the cloned node.
+    fn clone_subtree(&self, node: &Self::Handle) -> 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 {

rcdom/lib.rs

@@ -60,7 +60,7 @@ use markup5ever::ExpandedName;
 use markup5ever::QualName;
 
 /// The different kinds of nodes in the DOM.
-#[derive(Debug)]
+#[derive(Debug, Clone)]
 pub enum NodeData {
     /// The `Document` itself - the root node of a HTML document.
     Document,
@@ -417,6 +417,19 @@ impl TreeSink for RcDom {
         new_children.extend(mem::take(&mut *children));
     }
 
+    fn clone_subtree(&self, node: &Handle) -> Handle {
+        let cloned_node = Node::new(node.data.clone());
+
+        // Clone all children recursively
+        for child in node.children.borrow().iter() {
+            let cloned_child = self.clone_subtree(child);
+            append(&cloned_node, cloned_child);
+        }
+
+        cloned_node
+    }
+
+
     fn is_mathml_annotation_xml_integration_point(&self, target: &Handle) -> bool {
         if let NodeData::Element {
             mathml_annotation_xml_integration_point,

rcdom/tests/html-tree-sink.rs

@@ -4,7 +4,7 @@ use html5ever::tendril::StrTendril;
 use html5ever::ExpandedName;
 use html5ever::QualName;
 use markup5ever::interface::{ElementFlags, NodeOrText, QuirksMode, TreeSink};
-use markup5ever::{local_name, ns, Attribute};
+use markup5ever::{local_name, ns, Attribute, LocalName};
 use markup5ever_rcdom::{Handle, RcDom};
 use std::borrow::Cow;
 use std::cell::{Cell, RefCell};
@@ -111,6 +111,11 @@ impl TreeSink for LineCountingDOM {
     fn set_current_line(&self, line_number: u64) {
         self.current_line.set(line_number);
     }
+
+    fn clone_subtree(&self, node: &Self::Handle) -> Self::Handle {
+        self.rcdom.clone_subtree(node)
+    }
+
 }
 
 #[test]

web_atoms/local_names.txt

@@ -879,6 +879,7 @@ section
 seed
 select
 selected
+selectedcontent
 selection
 selector
 semantics