Merge pull request #66 from niklak/feature/minor-adjustments

Code minor adjustments

Cargo.toml

@@ -21,6 +21,7 @@ tendril = "0.4.3"
 foldhash = "0.1.4"
 hashbrown = {version = "0.15.2", default-features = false, features = ["allocator-api2", "inline-more", "default-hasher"], optional = true}
 precomputed-hash = "0.1.1"
+bit-set = "0.8.0"
 
 [dev-dependencies]
 wasm-bindgen-test = "0.3"

src/document.rs

@@ -311,7 +311,7 @@ impl TreeSink for Document {
         };
 
         self.tree.create_node(NodeData::Element(Element::new(
-            name.clone(),
+            name,
             attrs,
             template_contents,
             flags.mathml_annotation_xml_integration_point,
@@ -459,7 +459,7 @@ impl TreeSink for Document {
     }
 }
 
-fn append_to_existing_text(prev: &mut TreeNode, text: &str) -> bool {
+fn append_to_existing_text(prev: &mut TreeNode, text: &StrTendril) -> bool {
     match prev.data {
         NodeData::Text { ref mut contents } => {
             contents.push_slice(text);

src/matcher.rs

@@ -1,15 +1,14 @@
 use std::{fmt, iter};
 
+use bit_set::BitSet;
 use cssparser::{CowRcStr, ParseError, SourceLocation, ToCss};
 use html5ever::Namespace;
 use selectors::context::SelectorCaches;
 use selectors::parser::{self, SelectorList, SelectorParseErrorKind};
 use selectors::{context, matching, visitor, Element};
 
 use crate::css::{CssLocalName, CssString};
-use crate::entities::InnerHashSet;
 use crate::node::NodeRef;
-
 /// CSS selector.
 #[derive(Clone, Debug)]
 pub struct Matcher {
@@ -52,7 +51,7 @@ impl Matcher {
 pub struct Matches<'a, 'b> {
     nodes: Vec<NodeRef<'a>>,
     matcher: &'b Matcher,
-    set: InnerHashSet<usize>,
+    seen: BitSet,
     caches: SelectorCaches,
 }
 
@@ -77,10 +76,11 @@ impl<'a, 'b> Matches<'a, 'b> {
     }
     pub fn from_one(root_node: NodeRef<'a>, matcher: &'b Matcher, match_scope: MatchScope) -> Self {
         let nodes = Self::nodes_from_root(iter::once(root_node), match_scope);
+        let set = BitSet::new();
         Self {
             nodes,
             matcher,
-            set: Default::default(),
+            seen: set,
             caches: Default::default(),
         }
     }
@@ -92,10 +92,11 @@ impl<'a, 'b> Matches<'a, 'b> {
     ) -> Self {
         let nodes = Self::nodes_from_root(root_nodes, match_scope);
 
+        let set = BitSet::new();
         Self {
             nodes,
             matcher,
-            set: Default::default(),
+            seen: set,
             caches: Default::default(),
         }
     }
@@ -106,7 +107,7 @@ impl<'a> Iterator for Matches<'a, '_> {
 
     fn next(&mut self) -> Option<Self::Item> {
         while let Some(node) = self.nodes.pop() {
-            if self.set.contains(&node.id.value) {
+            if self.seen.contains(node.id.value) {
                 continue;
             }
             self.nodes
@@ -116,7 +117,7 @@ impl<'a> Iterator for Matches<'a, '_> {
                 .matcher
                 .match_element_with_caches(&node, &mut self.caches)
             {
-                self.set.insert(node.id.value);
+                self.seen.insert(node.id.value);
                 return Some(node);
             }
         }

src/node/inner.rs

@@ -147,6 +147,15 @@ impl TreeNode {
             element.remove_class(class);
         }
     }
+
+    /// If element is link.
+    pub fn is_link(&self) -> bool {
+        if let Some(element) = self.as_element() {
+            element.is_link()
+        } else {
+            false
+        }
+    }
 }
 
 impl Clone for TreeNode {

src/node/node_data.rs

@@ -120,7 +120,7 @@ impl Element {
         let attr = self
             .attrs
             .iter_mut()
-            .find(|attr| &attr.name.local == "class");
+            .find(|attr| attr.name.local == local_name!("class"));
 
         match attr {
             Some(attr) => {
@@ -154,7 +154,7 @@ impl Element {
         if let Some(attr) = self
             .attrs
             .iter_mut()
-            .find(|attr| &attr.name.local == "class")
+            .find(|attr| attr.name.local == local_name!("class"))
         {
             let mut set: InnerHashSet<&str> = attr
                 .value
@@ -230,7 +230,7 @@ impl Element {
             .attrs
             .iter()
             .map(|e| e.name.clone())
-            .collect::<InnerHashSet<_>>();
+            .collect::<Vec<_>>();
 
         self.attrs.extend(
             attrs
@@ -244,4 +244,15 @@ impl Element {
         let new_name = QualName::new(None, ns!(), LocalName::from(name));
         self.name = new_name;
     }
+
+    /// If element is a link.
+    pub fn is_link(&self) -> bool {
+        matches!(
+            self.name.local,
+            local_name!("a") | local_name!("area") | local_name!("link")
+        ) && self
+            .attrs
+            .iter()
+            .any(|a| a.name.local == local_name!("href"))
+    }
 }

src/node/selector.rs

@@ -141,12 +141,7 @@ impl selectors::Element for NodeRef<'_> {
             Active | Focus | Hover | Enabled | Disabled | Checked | Indeterminate | Visited => {
                 false
             }
-            AnyLink | Link => match self.node_name() {
-                Some(node_name) => {
-                    matches!(node_name.deref(), "a" | "area" | "link") && self.has_attr("href")
-                }
-                None => false,
-            },
+            AnyLink | Link => self.query_or(false, |n| n.is_link()),
             OnlyText => self.has_only_text(),
             HasText(s) => self.has_text(s.as_str()),
             Contains(s) => self.text().contains(s.as_str()),
@@ -166,15 +161,7 @@ impl selectors::Element for NodeRef<'_> {
         // TODO: This function adds some overhead.
         // Its purpose in dom_query is unclear.
         // Returning `false` works just fine.
-        self.query_or(false, |node| {
-            if let NodeData::Element(ref e) = node.data {
-                return matches!(
-                    e.name.local,
-                    local_name!("a") | local_name!("area") | local_name!("link")
-                ) && e.attrs.iter().any(|a| a.name.local == local_name!("href"));
-            }
-            false
-        })
+        self.query_or(false, |node| node.is_link())
     }
 
     /// Whether the element is an HTML element.

src/node/serializing.rs

@@ -196,7 +196,7 @@ fn push_normalized_text(text: &mut StrTendril, new_text: &str) {
     if result.is_empty() && follows_newline {
         return;
     }
-    
+
     text.push_tendril(&result);
 
     if push_end_whitespace && !text.ends_with(char::is_whitespace) {

tests/data.rs

@@ -130,7 +130,6 @@ pub static DMC_CONTENTS: &str = r#"<!DOCTYPE html>
     </body>
 </html>"#;
 
-
 pub static MINI_TABLE_CONTENTS: &str = r#"<!DOCTYPE html>
 <html>
     <head></head>
@@ -148,4 +147,4 @@ pub static MINI_TABLE_CONTENTS: &str = r#"<!DOCTYPE html>
             </tr>
         </table>
     </body>
-</html>"#;
+</html>"#;

tests/node-traversal.rs

@@ -352,12 +352,11 @@ That's how we keep our code development!
     assert_eq!(text.as_ref(), expected);
 }
 
-
 #[cfg_attr(not(target_arch = "wasm32"), test)]
 #[cfg_attr(target_arch = "wasm32", wasm_bindgen_test)]
 fn test_doc_table_formatted_text() {
     let doc = Document::from(MINI_TABLE_CONTENTS);
     let text = doc.formatted_text();
     let expected = "1 2 3\n4 5 6";
     assert_eq!(text.as_ref(), expected);
-}
+}