Add initial variable processing

This supports the work in whatwg/html#11392 with a new processor found in variables.rs, documented there.

This also updates the parser to store the line numbers for each element, for use in error messages.

For now this contains commented-out code to error on `<var>`s outside of `algorithm` and `var-scope` scopes, since CI will prevent us from merging a html-build change that cannot build the current main branch of whatwg/html. After the HTML PR is merged, we can uncomment that code.

Author: domenic

src/dom_utils.rs

@@ -28,6 +28,9 @@ pub trait NodeHandleExt {
     /// Sets an attribute on the element. Must be an element.
     fn set_attribute(&self, name: &QualName, value: StrTendril);
 
+    /// Removes an attribute from the element, if present. Must be an element.
+    fn remove_attribute(&self, name: &QualName);
+
     /// Returns true if the node is an element.
     fn is_element(&self) -> bool;
 
@@ -225,6 +228,16 @@ impl NodeHandleExt for Handle {
         }
     }
 
+    fn remove_attribute(&self, name: &QualName) {
+        let mut attrs = match self.data {
+            NodeData::Element { ref attrs, .. } => attrs.borrow_mut(),
+            _ => panic!("not an element"),
+        };
+        if let Some(i) = attrs.iter().position(|a| &a.name == name) {
+            attrs.remove(i);
+        }
+    }
+
     fn is_element(&self) -> bool {
         matches!(&self.data, NodeData::Element { .. })
     }

src/main.rs

@@ -19,6 +19,7 @@ mod rcdom_with_line_numbers;
 mod represents;
 mod self_link;
 mod tag_omission;
+mod variables;
 
 #[tokio::main]
 async fn main() -> io::Result<()> {
@@ -60,11 +61,13 @@ async fn run_preprocess() -> io::Result<()> {
     let mut tag_omission = tag_omission::Processor::new();
     let mut interface_index = interface_index::Processor::new();
     let mut self_link = self_link::Processor::new();
+    let mut variables = variables::Processor::new(&parsed);
 
     // We do exactly one pass to identify the changes that need to be made.
     dom_utils::scan_dom(&document, &mut |h| {
         boilerplate.visit(h);
         represents.visit(h);
+        variables.visit(h);
         annotate_attributes.visit(h);
         tag_omission.visit(h);
         interface_index.visit(h);
@@ -76,6 +79,7 @@ async fn run_preprocess() -> io::Result<()> {
     // conflicts between them.
     boilerplate.apply().await?;
     represents.apply()?;
+    variables.apply()?;
     annotate_attributes.apply().await?;
     tag_omission.apply()?;
     interface_index.apply()?;

src/rcdom_with_line_numbers.rs

@@ -7,14 +7,17 @@ use html5ever::{
     tendril::StrTendril,
     tree_builder::{ElementFlags, NodeOrText, QuirksMode},
 };
-use markup5ever_rcdom::{Handle, RcDom};
+use markup5ever_rcdom::{Handle, Node, RcDom};
 use std::borrow::Cow;
-use std::cell::Cell;
+use std::cell::{Cell, RefCell};
+use std::collections::HashMap;
 use std::io;
+use std::rc::Rc;
 
 pub struct RcDomWithLineNumbers {
     dom: RcDom,
     current_line: Cell<u64>,
+    node_line_map: RefCell<HashMap<*const Node, u64>>,
 }
 
 #[cfg(test)]
@@ -48,13 +51,21 @@ impl RcDomWithLineNumbers {
             Ok(())
         }
     }
+
+    /// Returns the 1-based line number where the element represented by `handle`
+    /// was created, if known.
+    pub fn line_number_for(&self, handle: &Handle) -> Option<u64> {
+        let key = Rc::as_ptr(handle);
+        self.node_line_map.borrow().get(&key).cloned()
+    }
 }
 
 impl Default for RcDomWithLineNumbers {
     fn default() -> Self {
         Self {
             dom: RcDom::default(),
             current_line: Cell::new(1),
+            node_line_map: RefCell::new(HashMap::new()),
         }
     }
 }
@@ -81,20 +92,27 @@ impl TreeSink for RcDomWithLineNumbers {
         self
     }
 
+    // Override to record the current line number for each created element handle.
+    fn create_element(
+        &self,
+        name: QualName,
+        attrs: Vec<Attribute>,
+        flags: ElementFlags,
+    ) -> Self::Handle {
+        let h = self.dom.create_element(name, attrs, flags);
+        let key = Rc::as_ptr(&h);
+        let line = self.current_line.get();
+        self.node_line_map.borrow_mut().insert(key, line);
+        h
+    }
+
     // Delegate all other methods to RcDom.
     delegate! {
         to self.dom {
             fn get_document(&self) -> Self::Handle;
 
             fn elem_name<'a>(&'a self, target: &'a Self::Handle) -> ExpandedName<'a>;
 
-            fn create_element(
-                &self,
-                name: QualName,
-                attrs: Vec<Attribute>,
-                flags: ElementFlags,
-            ) -> Self::Handle;
-
             fn create_comment(&self, text: StrTendril) -> Self::Handle;
 
             fn create_pi(&self, target: StrTendril, data: StrTendril) -> Self::Handle;