bodyOnly and noHeadline options for PocketReader.Read

2013-11-15 14:20:56 +01:00
parent 6f6d054044
commit b2ce585966
4 changed files with 45 additions and 8 deletions
@@ -59,15 +59,17 @@ namespace PocketSharp
    /// The PocketReader is based on a custom PCL port of NReadability and SgmlReader.
    /// </summary>
    /// <param name="uri">An URI.</param>
+    /// <param name="bodyOnly">if set to <c>true</c> [only body is returned].</param>
+    /// <param name="noHeadline">if set to <c>true</c> [no headline (h1) is included].</param>
    /// <returns>A Pocket article with extracted content and title.</returns>
    /// <exception cref="PocketRequestException"></exception>
-    public async Task<PocketArticle> Read(Uri uri)
+    public async Task<PocketArticle> Read(Uri uri, bool bodyOnly = true, bool noHeadline = false)
    {
      return await Read(new PocketItem()
      {
        ID = 0,
        Uri = uri
-      });
+      }, bodyOnly, noHeadline);
    }


@@ -78,9 +80,13 @@ namespace PocketSharp
    /// The PocketReader is based on a custom PCL port of NReadability and SgmlReader.
    /// </summary>
    /// <param name="item">The pocket item.</param>
-    /// <returns>A Pocket article with extracted content and title.</returns>
+    /// <param name="bodyOnly">if set to <c>true</c> [only body is returned].</param>
+    /// <param name="noHeadline">if set to <c>true</c> [no headline (h1) is included].</param>
+    /// <returns>
+    /// A Pocket article with extracted content and title.
+    /// </returns>
    /// <exception cref="PocketRequestException"></exception>
-    public async Task<PocketArticle> Read(PocketItem item)
+    public async Task<PocketArticle> Read(PocketItem item, bool bodyOnly = true, bool noHeadline = false)
    {
      // initialize transcoder
      NReadabilityTranscoder transcoder = new NReadabilityTranscoder(
@@ -101,11 +107,12 @@ namespace PocketSharp
        Url = item.Uri.ToString(),
        DomSerializationParams = new DomSerializationParams()
        {
-          BodyOnly = true,
+          BodyOnly = bodyOnly,
+          NoHeadline = noHeadline,
          PrettyPrint = true,
          DontIncludeContentTypeMetaElement = true,
          DontIncludeMobileSpecificMetaElements = true,
-          DontIncludeDocTypeMetaElement = true,
+          DontIncludeDocTypeMetaElement = false,
          DontIncludeGeneratorMetaElement = true
        }
      };
@@ -26,6 +26,23 @@ namespace PocketSharp.Tests
        Uri = new Uri("http://frontendplay.com/story/4/http-caching-demystified-part-2-implementation")
      });

+      Assert.DoesNotContain("<!DOCTYPE html>", result.Content);
+      Assert.Contains("<h1>", result.Content);
+      Assert.True(result.Content.Length > 15000);
+    }
+
+
+    [Fact]
+    public async Task ReadArticleWithContainerNoHeadlineTest()
+    {
+      PocketArticle result = await reader.Read(new PocketItem()
+      {
+        ID = 99,
+        Uri = new Uri("http://frontendplay.com/story/4/http-caching-demystified-part-2-implementation")
+      }, false, true);
+
+      Assert.Contains("<!DOCTYPE html>", result.Content);
+      Assert.DoesNotContain("<h1>", result.Content);
      Assert.True(result.Content.Length > 15000);
    }

@@ -1,5 +1,4 @@
-using System;
-
+
 namespace PocketSharp.Ports.NReadability
 {
  public class DomSerializationParams
@@ -48,6 +47,10 @@ namespace PocketSharp.Ports.NReadability
    /// </summary>
    public bool BodyOnly { get; set; }

+    /// <summary>
+    /// Remove headline of website
+    /// </summary>
+    public bool NoHeadline { get; set; }
    #endregion
  }
 }
@@ -92,6 +92,16 @@ namespace PocketSharp.Ports.NReadability
        }
      }

+      if (domSerializationParams.NoHeadline)
+      {
+        var h1 = document.Root.GetElementsByTagName("h1").FirstOrDefault();
+
+        if (h1 != null)
+        {
+          result = result.Replace(h1.ToString(), "");
+        }
+      }
+
      return result;
    }