extend NReadability to allow output of body only

2013-10-15 22:21:17 +02:00
parent ef10ca23a4
commit 649ded199d
5 changed files with 30 additions and 3 deletions
@@ -16,7 +16,7 @@ namespace PocketSharp.Tests
    {
      PocketReader reader = new PocketReader();

-      string result = await reader.Read(new Uri("https://github.com/ceee/PocketSharp"));
+      string result = await reader.Read(new Uri("http://frontendplay.com/story/4/http-caching-demystified-part-2-implementation"));

      Assert.True(result.Length > 0);
    }
@@ -72,6 +72,7 @@ namespace PocketSharp
        Url = uri.ToString(),
        DomSerializationParams = new DomSerializationParams()
        {
+          BodyOnly = true,
          PrettyPrint = true,
          DontIncludeContentTypeMetaElement = true,
          DontIncludeMobileSpecificMetaElements = true,
@@ -82,7 +83,7 @@ namespace PocketSharp

      // process/transcode HTML
      TranscodingResult transcodingResult = transcoder.Transcode(transcodingInput);
-
+      
      return transcodingResult.ExtractedContent;
    }

@@ -43,6 +43,11 @@ namespace NReadability
    /// </summary>
    public bool DontIncludeGeneratorMetaElement { get; set; }

+    /// <summary>
+    /// Render complete Website or only the Body
+    /// </summary>
+    public bool BodyOnly { get; set; }
+
    #endregion
  }
 }
@@ -36,7 +36,15 @@ namespace NReadability
    /// </summary>
    /// <param name="document">System.Xml.Linq.XDocument instance containing the DOM to be serialized.</param>
    /// <param name="domSerializationParams">Contains parameters that modify the behaviour of the output serialization.</param>
-    /// <returns>Serialized representation of the DOM.</returns>
+    /// <param name="bodyOnly">if set to <c>true</c> [returns body only].</param>
+    /// <returns>
+    /// Serialized representation of the DOM.
+    /// </returns>
+    /// <exception cref="System.ArgumentException">
+    /// The document must have a root.
+    /// or
+    /// The document's root must be an html element.
+    /// </exception>
    public string SerializeDocument(XDocument document, DomSerializationParams domSerializationParams)
    {
      if (!domSerializationParams.DontIncludeContentTypeMetaElement
@@ -74,6 +82,16 @@ namespace NReadability
        result = "<!DOCTYPE html>\r\n" + result;
      }

+      if (domSerializationParams.BodyOnly && document.Root != null)
+      {
+        var body = document.Root.GetChildrenByTagName("body").FirstOrDefault();
+
+        if (body != null)
+        {
+          result = body.GetInnerHtml();
+        }
+      }
+
      return result;
    }

@@ -1,3 +1,4 @@
+using System.Xml.Linq;
 namespace NReadability
 {
  public class TranscodingResult
@@ -17,5 +18,7 @@ namespace NReadability
    public string ExtractedTitle { get; set; }

    public string NextPageUrl { get; set; }
+
+    public XDocument RawDocument { get; set; }
  }
 }