feat: implement query selectors, optimize performance, and fix benchmarks

- Implement select, selectAll, matches, closest, and isDefaultNamespace
- Optimize performance: enable LTO, use mimalloc, direct string parsing
- Fix clippy warnings and clean up dependencies
- Fix benchmarks and update README
- Add tests for new selector features

Author: xusd320

Cargo.lock

@@ -8,5 +8,8 @@ html5ever = "0.27.0"
 indexmap  = "2.2.6"
 
 [profile.release]
-lto   = true
-strip = "symbols"
+codegen-units = 1
+lto           = "fat"
+opt-level     = 3
+panic         = "abort"
+strip         = "symbols"

Cargo.toml

@@ -186,7 +186,6 @@ npm test
 ## Benchmark
 
 ```bash
-cargo benchmark
 npm run benchmark
 ```
 

README.md

@@ -14,7 +14,7 @@ export default function cheerio() {
 
     bench.on("result", function (stat) {
       console.log(
-        "niddle               :" +
+        "domparser-rs         :" +
           stat.mean().toPrecision(6) +
           " ms/file ± " +
           stat.sd().toPrecision(6),

benchmark/domparser.mjs

@@ -4,8 +4,8 @@ name    = "domparser"
 version = "0.0.0"
 
 [dependencies]
-html5ever         = { workspace = true }
-indexmap          = { workspace = true }
+html5ever         = "0.27.0"
+indexmap          = "2.2.6"
 markup5ever       = "0.11.0"
 markup5ever_rcdom = "0.3.0"
 tendril           = "0.4.3"

crates/domparser/Cargo.toml

@@ -1,8 +1,6 @@
+use html5ever::parse_document;
 use html5ever::tendril::TendrilSink;
-use html5ever::{local_name, namespace_url, ns, parse_document, QualName};
-use markup5ever_rcdom::{Node, NodeData, RcDom};
-use std::cell::RefCell;
-use std::rc::Rc;
+use markup5ever_rcdom::RcDom;
 
 pub mod node;
 pub mod serializer;
@@ -11,39 +9,9 @@ pub use markup5ever_rcdom;
 pub use node::DomNode;
 
 pub fn parse(html: String) -> DomNode {
-  let dom = parse_document(RcDom::default(), Default::default())
-    .from_utf8()
-    .read_from(&mut html.as_bytes())
-    .unwrap();
-
-  {
-    let mut children = dom.document.children.borrow_mut();
-    let has_html = children.iter().any(|c| {
-      if let NodeData::Element { name, .. } = &c.data {
-        name.local.as_ref() == "html"
-      } else {
-        false
-      }
-    });
-
-    if !has_html {
-      let html_name = QualName::new(None, ns!(html), local_name!("html"));
-      let html_node = Node::new(NodeData::Element {
-        name: html_name,
-        attrs: RefCell::new(vec![]),
-        template_contents: RefCell::new(None),
-        mathml_annotation_xml_integration_point: false,
-      });
-
-      for child in children.drain(..) {
-        html_node.children.borrow_mut().push(child.clone());
-        child.parent.set(Some(Rc::downgrade(&html_node)));
-      }
-
-      children.push(html_node.clone());
-      html_node.parent.set(Some(Rc::downgrade(&dom.document)));
-    }
-  }
+  let mut parser = parse_document(RcDom::default(), Default::default());
+  parser.process(html.into());
+  let dom = parser.finish();
 
   DomNode(dom.document)
 }

crates/domparser/src/lib.rs

@@ -70,11 +70,10 @@ fn clone_node_data(data: &NodeData) -> NodeData {
       mathml_annotation_xml_integration_point,
     } => {
       let new_attrs = attrs.borrow().clone();
-      let new_template_contents = if let Some(tc) = template_contents.borrow().as_ref() {
-        Some(clone_handle_recursive(tc))
-      } else {
-        None
-      };
+      let new_template_contents = template_contents
+        .borrow()
+        .as_ref()
+        .map(clone_handle_recursive);
       NodeData::Element {
         name: name.clone(),
         attrs: RefCell::new(new_attrs),

crates/domparser/src/node/mod.rs

@@ -222,11 +222,8 @@ impl DomNode {
 
   pub fn set_node_value(&self, value: Option<String>) {
     if let Some(val) = value {
-      match &self.0.data {
-        NodeData::Text { contents } => {
-          *contents.borrow_mut() = val.into();
-        }
-        _ => {}
+      if let NodeData::Text { contents } = &self.0.data {
+        *contents.borrow_mut() = val.into();
       }
     }
   }
@@ -286,12 +283,9 @@ impl DomNode {
     fn get_text(handle: &Handle) -> String {
       match &handle.data {
         NodeData::Text { contents } => contents.borrow().to_string(),
-        NodeData::Element { .. } | NodeData::Document => handle
-          .children
-          .borrow()
-          .iter()
-          .map(|child| get_text(child))
-          .collect(),
+        NodeData::Element { .. } | NodeData::Document => {
+          handle.children.borrow().iter().map(get_text).collect()
+        }
         _ => "".to_string(),
       }
     }
@@ -541,8 +535,8 @@ impl DomNode {
     if let NodeData::Element { attrs, .. } = &self.0.data {
       for attr in attrs.borrow().iter() {
         let name = attr.name.local.to_string();
-        if name.starts_with("data-") {
-          let key = kebab_to_camel(&name[5..]);
+        if let Some(stripped) = name.strip_prefix("data-") {
+          let key = kebab_to_camel(stripped);
           map.insert(key, attr.value.to_string());
         }
       }
@@ -769,7 +763,7 @@ impl DomNode {
         }
 
         let is_empty = if let NodeData::Text { contents } = &children[i].data {
-          contents.borrow().len() == 0
+          contents.borrow().is_empty()
         } else {
           false
         };
@@ -799,10 +793,8 @@ impl DomNode {
             {
               return Some(attr.value.to_string());
             }
-          } else {
-            if attr.name.local.as_ref() == "xmlns" && attr.name.prefix.is_none() {
-              return Some(attr.value.to_string());
-            }
+          } else if attr.name.local.as_ref() == "xmlns" && attr.name.prefix.is_none() {
+            return Some(attr.value.to_string());
           }
         }
       }
@@ -816,10 +808,10 @@ impl DomNode {
     while let Some(node) = current {
       if let NodeData::Element { attrs, .. } = &node.data {
         for attr in attrs.borrow().iter() {
-          if attr.value.as_ref() == namespace {
-            if attr.name.prefix.as_ref().map(|s| s.as_ref()) == Some("xmlns") {
-              return Some(attr.name.local.to_string());
-            }
+          if attr.value.as_ref() == namespace
+            && attr.name.prefix.as_ref().map(|s| s.as_ref()) == Some("xmlns")
+          {
+            return Some(attr.name.local.to_string());
           }
         }
       }
@@ -891,9 +883,9 @@ impl DomNode {
     let pos2 = children.iter().position(|x| Rc::ptr_eq(x, child2)).unwrap();
 
     if pos1 < pos2 {
-      return 4;
+      4
     } else {
-      return 2;
+      2
     }
   }
 }
@@ -904,13 +896,11 @@ fn kebab_to_camel(s: &str) -> String {
   for c in s.chars() {
     if c == '-' {
       next_upper = true;
+    } else if next_upper {
+      result.push(c.to_ascii_uppercase());
+      next_upper = false;
     } else {
-      if next_upper {
-        result.push(c.to_ascii_uppercase());
-        next_upper = false;
-      } else {
-        result.push(c);
-      }
+      result.push(c);
     }
   }
   result