Merge pull request #1134 from lahma/issue/#584

Improve QuerySelectorAll performance

src/AngleSharp.Benchmarks/AngleSharp.Benchmarks.csproj

@@ -1,7 +1,7 @@
 <Project Sdk="Microsoft.NET.Sdk">
   <PropertyGroup>
     <OutputType>Exe</OutputType>
-    <TargetFramework>net472</TargetFramework>
+    <TargetFrameworks>net472;net6.0</TargetFrameworks>
     <SignAssembly>false</SignAssembly>
   </PropertyGroup>
 
@@ -10,16 +10,14 @@
   </ItemGroup>
 
   <ItemGroup>
-    <PackageReference Include="BenchmarkDotNet" Version="0.13.5" />
-    <PackageReference Include="CsQuery" Version="1.3.5-beta5" />
+    <PackageReference Include="BenchmarkDotNet" Version="0.13.7" />
+    <PackageReference Include="CsQuery" Version="1.3.5-beta5" Condition=" '$(TargetFramework)' == 'net472' " />
     <PackageReference Include="HtmlAgilityPack" Version="1.11.46" />
     <PackageReference Include="System.Net.Http" Version="4.3.4" />
   </ItemGroup>
 
   <ItemGroup>
-    <None Update="page.html">
-      <CopyToOutputDirectory>PreserveNewest</CopyToOutputDirectory>
-    </None>
+    <None Update="page.html" CopyToOutputDirectory="PreserveNewest" />
   </ItemGroup>
 
 </Project>

src/AngleSharp.Benchmarks/ParserBenchmark.cs

@@ -1,11 +1,14 @@
 using System.Collections.Generic;
-using System.Text;
 using AngleSharp.Html.Parser;
 using BenchmarkDotNet.Attributes;
 using BenchmarkDotNet.Configs;
+
+#if NETFRAMEWORK
 using CsQuery;
 using CsQuery.ExtensionMethods.Internal;
 using CsQuery.HtmlParser;
+#endif
+
 using HtmlAgilityPack;
 
 namespace AngleSharp.Benchmarks
@@ -20,7 +23,7 @@ public IEnumerable<UrlTest> GetSources()
             var websites = new UrlTests(
                 ".html",
                 true);
-            
+
             websites.Include(
                 "http://www.amazon.com",
                 "http://www.blogspot.com",
@@ -78,14 +81,16 @@ public IEnumerable<UrlTest> GetSources()
 
         [ParamsSource(nameof(GetSources))] public UrlTest UrlTest { get; set; }
 
+#if NETFRAMEWORK
         [Benchmark]
         public void CsQuery()
         {
             var factory = new ElementFactory(DomIndexProviders.Simple);
 
             using var stream = UrlTest.Source.ToStream();
-            factory.Parse(stream, Encoding.UTF8);
+            factory.Parse(stream, System.Text.Encoding.UTF8);
         }
+#endif
 
         [Benchmark]
         public void HTMLAgilityPack()

src/AngleSharp.Benchmarks/SelectorBenchmark.cs

@@ -5,7 +5,10 @@
 using AngleSharp.Html.Parser;
 using BenchmarkDotNet.Attributes;
 using BenchmarkDotNet.Configs;
+
+#if NETFRAMEWORK
 using CsQuery;
+#endif
 
 namespace AngleSharp.Benchmarks
 {
@@ -14,14 +17,19 @@ public class SelectorBenchmark
     {
         private static readonly HtmlParser angleSharpParser = new HtmlParser();
         private IDocument angleSharpDocument;
+
+#if NETFRAMEWORK
         private CQ cqDocument;
+#endif
 
         [GlobalSetup]
         public void GlobalSetup()
         {
             var pageContent = File.ReadAllText("page.html");
             angleSharpDocument = angleSharpParser.ParseDocument(pageContent);
+#if NETFRAMEWORK
             cqDocument = CQ.CreateDocument(pageContent);
+#endif
         }
 
         [ParamsSource(nameof(GetSelectors))]
@@ -73,11 +81,13 @@ public void GlobalSetup()
             "div > p > a"
         };
 
+#if NETFRAMEWORK
         [Benchmark]
         public void CsQuery()
         {
             cqDocument.Select(Selector);
         }
+#endif
 
         [Benchmark]
         public void AngleSharp()

src/AngleSharp/Css/Dom/Internal/ComplexSelector.cs

@@ -133,11 +133,12 @@ public void AppendSelector(ISelector selector, CssCombinator combinator)
 
         private Boolean MatchCascade(Int32 pos, IElement element, IElement? scope)
         {
-            var newElements = _combinators[pos].Transform!(element);
+            var combinatorSelector = _combinators[pos];
+            var newElements = combinatorSelector.Transform!(element);
 
             foreach (var newElement in newElements)
             {
-                if (_combinators[pos].Selector.Match(newElement, scope) && (pos == 0 || MatchCascade(pos - 1, newElement, scope)))
+                if (combinatorSelector.Selector.Match(newElement, scope) && (pos == 0 || MatchCascade(pos - 1, newElement, scope)))
                 {
                     return true;
                 }

src/AngleSharp/Css/Dom/SelectorExtensions.cs

@@ -19,14 +19,19 @@ public static class SelectorExtensions
         /// <returns>The resulting element or null.</returns>
         public static IElement? MatchAny(this ISelector selector, IEnumerable<IElement> elements, IElement? scope)
         {
+            var stack = new Stack<INode>();
             foreach (var element in elements)
             {
-                foreach (var descendentAndSelf in element.DescendentsAndSelf<IElement>())
+                stack.Clear();
+                var nodes = element.GetDescendantsAndSelf(
+                    stack,
+                    filter: static (node, state) => node is IElement e && state.Selector.Match(e, state.Scope),
+                    state: new SelectorState(selector, scope));
+
+                var enumerator = nodes.GetEnumerator();
+                if (enumerator.MoveNext())
                 {
-                    if (selector.Match(descendentAndSelf, scope))
-                    {
-                        return descendentAndSelf;
-                    }
+                    return (IElement?) enumerator.Current;
                 }
             }
 
@@ -59,16 +64,32 @@ public static IHtmlCollection<IElement> MatchAll(this ISelector selector, IEnume
 
         private static void MatchAll(this ISelector selector, IEnumerable<IElement> elements, IElement? scope, List<IElement> result)
         {
+            var stack = new Stack<INode>();
             foreach (var element in elements)
             {
-                foreach (var descendentAndSelf in element.DescendentsAndSelf<IElement>())
+                stack.Clear();
+                var nodes = element.GetDescendantsAndSelf(
+                    stack,
+                    filter: static (node, state) => node is IElement e && state.Selector.Match(e, state.Scope),
+                    state: new SelectorState(selector, scope));
+
+                foreach (var descendentAndSelf in nodes)
                 {
-                    if (selector.Match(descendentAndSelf, scope))
-                    {
-                        result.Add(descendentAndSelf);
-                    }
+                    result.Add((IElement) descendentAndSelf);
                 }
             }
         }
+
+        private readonly struct SelectorState
+        {
+            public readonly ISelector Selector;
+            public readonly IElement? Scope;
+
+            public SelectorState(ISelector selector, IElement? scope)
+            {
+                Selector = selector;
+                Scope = scope;
+            }
+        }
     }
 }

src/AngleSharp/Dom/Internal/NodeList.cs

@@ -4,6 +4,7 @@ namespace AngleSharp.Dom
     using System.Collections;
     using System.Collections.Generic;
     using System.IO;
+    using System.Runtime.CompilerServices;
 
     /// <summary>
     /// Represents a list of Node instances or nodes.
@@ -34,17 +35,27 @@ internal NodeList()
 
         public Node this[Int32 index]
         {
+            [MethodImpl(MethodImplOptions.AggressiveInlining)]
             get => _entries[index];
+            [MethodImpl(MethodImplOptions.AggressiveInlining)]
             set => _entries[index] = value;
         }
 
-        INode INodeList.this[Int32 index] => this[index];
+        INode INodeList.this[Int32 index]
+        {
+            [MethodImpl(MethodImplOptions.AggressiveInlining)]
+            get => this[index];
+        }
 
         #endregion
 
         #region Properties
 
-        public Int32 Length => _entries.Count;
+        public Int32 Length
+        {
+            [MethodImpl(MethodImplOptions.AggressiveInlining)]
+            get => _entries.Count;
+        }
 
         #endregion
 

src/AngleSharp/Dom/NodeExtensions.cs

@@ -103,19 +103,49 @@ public static Boolean IsDescendantOf(this INode node, INode parent)
         /// <returns>An iterator over all descendants and itself.</returns>
         public static IEnumerable<INode> GetDescendantsAndSelf(this INode parent)
         {
-            var stack = new Stack<INode>();
+            return GetDescendantsAndSelf<Object?>(parent, new Stack<INode>(), null, null);
+        }
+
+        /// <summary>
+        /// Gets the descendant nodes and itself of the provided parent, in tree order.
+        /// </summary>
+        /// <param name="parent">The parent of the descendants.</param>
+        /// <param name="stack">Stack instance to be used (allows reuse).</param>
+        /// <param name="filter">Optional filter to run against items.</param>
+        /// <param name="state">Optional state to help with filtering.</param>
+        /// <returns>An iterator over all descendants and itself.</returns>
+        internal static IEnumerable<INode> GetDescendantsAndSelf<TState>(this INode parent, Stack<INode> stack, Func<INode, TState?, Boolean>? filter = null, TState? state = default)
+        {
             stack.Push(parent);
 
             while (stack.Count > 0)
             {
                 var next = stack.Pop();
-                yield return next;
 
-                var length = next.ChildNodes.Length;
+                if (filter == null || filter(next, state))
+                {
+                    yield return next;
+                }
+
+                var childNodes = next.ChildNodes;
 
-                while (length > 0)
+                // we only have one implementation
+                if (childNodes is NodeList nodeList)
+                {
+                    var length = nodeList.Length;
+                    while (length > 0)
+                    {
+                        stack.Push(nodeList[--length]);
+                    }
+                }
+                else
                 {
-                    stack.Push(next.ChildNodes[--length]);
+                    // unlikely virtual dispatch
+                    var length = childNodes.Length;
+                    while (length > 0)
+                    {
+                        stack.Push(childNodes[--length]);
+                    }
                 }
             }
         }