Added enhanced xpath capability in selector

FlorianRappl · FlorianRappl · commit 4c6bc3c28c1e · 2019-01-06T00:35:39.000+01:00
diff --git a/README.md b/README.md
@@ -21,9 +21,24 @@ var contentNode = document.Body.SelectSingleNode("//div[@id='content']");
 
 Besides `SelectSingleNode` we can also use `SelectNodes`. Both are extension methods defined in the `AngleSharp.XPath` namespace.
 
+If wanted we can also use XPath directly in CSS selectors such as in `QuerySelector` or `QuerySelectorAll` calls. For this we only need to apply the following configuration:
+
+```cs
+var config = Configuration.Default.WithXPath();
+```
+
+Now we can write queries such as
+
+```cs
+var secondLi = document.QuerySelector("*[xpath>'//li[2]']");
+```
+
+It is important that the original selector has all elements (`*`) as the intersection of the ordinary CSS selector and the XPath attribute is considered. The XPath attribute consists of a head (`xpath>`) and a value - provided as a string, e.g., `//li[2]`.
+
 ## Features
 
 - Uses `XPathNavigator` from `System.Xml.XPath`
+- Includes XPath capabilities to CSS query selectors if wanted
 
 ## Participating
 
diff --git a/src/AngleSharp.XPath.Tests/XPathConfigurationTests.cs b/src/AngleSharp.XPath.Tests/XPathConfigurationTests.cs
@@ -0,0 +1,45 @@
+namespace AngleSharp.XPath.Tests
+{
+    using NUnit.Framework;
+    using System.Threading.Tasks;
+
+    [TestFixture]
+    public class XPathConfigurationTests
+    {
+        [Test]
+        public async Task RunXPathQueryFromDocumentWithSelectAll()
+        {
+            var source = @"<body><ul><li><li><li></ul>";
+            var config = Configuration.Default.WithXPath();
+            var context = BrowsingContext.New(config);
+            var document = await context.OpenAsync(res => res.Content(source));
+            var elements = document.QuerySelectorAll("*[xpath>'//li']");
+            Assert.AreEqual(3, elements.Length);
+        }
+
+        [Test]
+        public async Task RunXPathQueryFromDocumentWithSelectSingle()
+        {
+            var source = @"<body><ul><li><li><li></ul>";
+            var config = Configuration.Default.WithXPath();
+            var context = BrowsingContext.New(config);
+            var document = await context.OpenAsync(res => res.Content(source));
+            var element = document.QuerySelector("*[xpath>'//li']");
+            Assert.IsNotNull(element);
+            Assert.AreEqual("LI", element.TagName);
+        }
+
+        [Test]
+        public async Task RunXPathQueryFromDocumentWithSelectSingleSpecialN()
+        {
+            var source = @"<body><ul><li><li class=two><li></ul>";
+            var config = Configuration.Default.WithXPath();
+            var context = BrowsingContext.New(config);
+            var document = await context.OpenAsync(res => res.Content(source));
+            var element = document.QuerySelector("*[xpath>'//li[2]']");
+            Assert.IsNotNull(element);
+            Assert.AreEqual("LI", element.TagName);
+            Assert.AreEqual("two", element.ClassName);
+        }
+    }
+}
diff --git a/src/AngleSharp.XPath/XPathAttrSelector.cs b/src/AngleSharp.XPath/XPathAttrSelector.cs
@@ -0,0 +1,36 @@
+namespace AngleSharp.XPath
+{
+    using AngleSharp.Css;
+    using AngleSharp.Css.Dom;
+    using AngleSharp.Dom;
+    using System;
+    using System.Collections.Generic;
+
+    sealed class XPathAttrSelector : ISelector
+    {
+        private readonly String _value;
+        private IElement _scope;
+        private List<INode> _result;
+
+        public XPathAttrSelector(String value) => _value = value;
+
+        public Priority Specificity => Priority.OneClass;
+
+        public String Text => $"[xpath>'${_value}']";
+
+        public void Accept(ISelectorVisitor visitor)
+        {
+        }
+
+        public Boolean Match(IElement element, IElement scope)
+        {
+            if (_scope != scope)
+            {
+                _scope = scope;
+                _result = scope.SelectNodes(_value);
+            }
+
+            return _result.Contains(element);
+        }
+    }
+}
diff --git a/src/AngleSharp.XPath/XPathConfigurationExtensions.cs b/src/AngleSharp.XPath/XPathConfigurationExtensions.cs
@@ -0,0 +1,30 @@
+namespace AngleSharp
+{
+    using AngleSharp.Css;
+    using AngleSharp.XPath;
+    using System.Linq;
+
+    /// <summary>
+    /// Additional extensions for integrating XPath.
+    /// </summary>
+    public static class XPathConfigurationExtensions
+    {
+        /// <summary>
+        /// Adds XPath to standard queries.
+        /// </summary>
+        /// <param name="configuration">The configuration to use.</param>
+        /// <returns>The new configuration.</returns>
+        public static IConfiguration WithXPath(this IConfiguration configuration)
+        {
+            var selectorFactory = configuration.Services.OfType<DefaultAttributeSelectorFactory>().FirstOrDefault();
+
+            if (selectorFactory != null)
+            {
+                selectorFactory.Unregister(">");
+                selectorFactory.Register(">", (name, value, prefix, mode) => new XPathAttrSelector(value));
+            }
+
+            return configuration;
+        }
+    }
+}
