Excluded html tags from indexed body content

--HG--
branch : dev

src/Orchard.Core.Tests/Indexing/DefaultIndexProviderTests.cs

@@ -155,7 +155,18 @@ namespace Orchard.Tests.Indexing {
             Assert.That(searchBuilder.Get(1).Id, Is.EqualTo(1));
             Assert.That(searchBuilder.Get(11).Id, Is.EqualTo(11));
             Assert.That(searchBuilder.Get(111).Id, Is.EqualTo(111));
+        }
+        
+        [Test]
+        public void TagsShouldBeRemoved() {
+            _provider.CreateIndex("default");
+            _provider.Store("default", _provider.New(1).Add("body", "<hr>some content</hr>"));
+            _provider.Store("default", _provider.New(2).Add("body", "<hr>some content</hr>", true));
 
+            var searchBuilder = _provider.CreateSearchBuilder("default");
+
+            Assert.That(searchBuilder.WithField("body", "hr").Search().Count(), Is.EqualTo(1));
+            Assert.That(searchBuilder.WithField("body", "hr").Search().First().Id, Is.EqualTo(1));
         }
     }
 }

src/Orchard.Web/Core/Common/Handlers/BodyAspectHandler.cs

@@ -10,7 +10,7 @@ namespace Orchard.Core.Common.Handlers {
             Filters.Add(StorageFilter.For(bodyRepository));
 
             OnIndexing<BodyAspect>((context, bodyAspect) => context.IndexDocument
-                                                                .Add("body", bodyAspect.Record.Text).Store(false)
+                                                                .Add("body", bodyAspect.Record.Text, true).Store(false)
                                                                 .Add("format", bodyAspect.Record.Format).Analyze(false));
         }
     }

src/Orchard.Web/Core/Indexing/Lucene/DefaultIndexDocument.cs

@@ -1,7 +1,10 @@
 using System;
 using System.Collections.Generic;
+using System.Web.Mvc;
 using Lucene.Net.Documents;
 using Orchard.Indexing;
+using Orchard.Mvc.Html;
+using Orchard.Utility.Extensions;
 
 namespace Orchard.Core.Indexing.Lucene {
 
@@ -17,7 +20,15 @@ namespace Orchard.Core.Indexing.Lucene {
         }
 
         public IIndexDocument Add(string name, string value) {
+            return Add(name, value, false);
+        }
+
+        public IIndexDocument Add(string name, string value, bool removeTags) {
             AppendPreviousField();
+            if(removeTags) {
+                value = value.RemoveTags();
+            }
+
             _previousField = new Field(name, value, Field.Store.YES, Field.Index.ANALYZED);
             return this;
         }
@@ -27,6 +38,7 @@ namespace Orchard.Core.Indexing.Lucene {
             _previousField = new Field(name, DateTools.DateToString(value, DateTools.Resolution.SECOND), Field.Store.YES, Field.Index.NOT_ANALYZED);
             return this;
         }
+
         public IIndexDocument Add(string name, int value) {
             AppendPreviousField();
             _previousField = new NumericField(name, Field.Store.YES, true).SetIntValue(value);

src/Orchard/Indexing/IIndexDocument.cs

@@ -8,20 +8,22 @@ namespace Orchard.Indexing {
         IIndexDocument SetContentItemId(int documentId);
 
         IIndexDocument Add(string name, string value);
+        IIndexDocument Add(string name, string value, bool removeTags);
         IIndexDocument Add(string name, DateTime value);
         IIndexDocument Add(string name, int value);
         IIndexDocument Add(string name, bool value);
         IIndexDocument Add(string name, float value);
 
         /// <summary>
-        /// Whether to store the original value to the index
+        /// Whether to store the original value to the index.
         /// </summary>
         IIndexDocument Store(bool store);
 
         /// <summary>
-        /// Whether the content should be tokenized or not. If not, value will be taken as a whole
+        /// Whether the content should be tokenized or not. If not, value will be taken as a whole.
         /// </summary>
         IIndexDocument Analyze(bool analyze);
 
+
     }
 }

src/Orchard/Mvc/Html/HtmlHelperExtensions.cs

@@ -94,10 +94,8 @@ namespace Orchard.Mvc.Html {
         #region Excerpt
 
         public static MvcHtmlString Excerpt(this HtmlHelper html, string markup, int length) {
-            var tagRegex = new Regex("<[^<>]*>", RegexOptions.Singleline);
-            var text = tagRegex.Replace(markup, "");
 
-            return MvcHtmlString.Create(text.Ellipsize(length));
+            return MvcHtmlString.Create(markup.RemoveTags().Ellipsize(length));
         }
 
         #endregion

src/Orchard/Utility/Extensions/StringExtensions.cs

@@ -29,5 +29,12 @@ namespace Orchard.Utility.Extensions {
                 ? defaultValue
                 : text;
         }
+
+        public static string RemoveTags(this string html) {
+            var tagRegex = new Regex("<[^<>]*>", RegexOptions.Singleline);
+            var text = tagRegex.Replace(html, "");
+
+            return text;
+        }
     }
 }