extract description, image and favicon

PortablePorts/NReadability/MetaExtractor.cs

@@ -0,0 +1,137 @@
+using System;
+using System.Collections.Generic;
+using System.Linq;
+using System.Xml.Linq;
+
+namespace ReadSharp.Ports.NReadability
+{
+  public class MetaExtractor
+  {
+    /// <summary>
+    /// Gets or sets a value indicating whether [has value].
+    /// </summary>
+    /// <value>
+    ///   <c>true</c> if [has value]; otherwise, <c>false</c>.
+    /// </value>
+    public bool HasValue { get; set; }
+
+    /// <summary>
+    /// Gets or sets the tags.
+    /// </summary>
+    /// <value>
+    /// The tags.
+    /// </value>
+    public IEnumerable<XElement> Tags { get; private set; }
+
+
+
+    /// <summary>
+    /// Initializes a new instance of the <see cref="MetaExtractor"/> class.
+    /// </summary>
+    /// <param name="document">The document.</param>
+    public MetaExtractor(XDocument document)
+    {
+      var documentRoot = document.Root;
+
+      if (documentRoot == null || documentRoot.Name == null || !"html".Equals(documentRoot.Name.LocalName, StringComparison.OrdinalIgnoreCase))
+      {
+        HasValue = false;
+        return;
+      }
+
+      var headElement = documentRoot.GetChildrenByTagName("head").FirstOrDefault();
+
+      if (headElement == null)
+      {
+        HasValue = false;
+        return;
+      }
+
+      IEnumerable<XElement> meta = headElement.GetChildrenByTagName("meta");
+      IEnumerable<XElement> link = headElement.GetChildrenByTagName("link");
+
+      Tags = meta != null ? meta.Concat(link) : link;
+      HasValue = Tags != null && Tags.Count() > 0;
+    }
+
+
+    /// <summary>
+    /// Gets the meta description.
+    /// </summary>
+    /// <returns></returns>
+    public string GetMetaDescription()
+    {
+      return SearchCandidates(new Dictionary<string, string>()
+      {
+        { "property|og:description", "content" },
+        { "name|description", "content" }
+      });
+    }
+
+
+    /// <summary>
+    /// Gets the meta image.
+    /// </summary>
+    /// <returns></returns>
+    public string GetMetaImage()
+    {
+      return SearchCandidates(new Dictionary<string, string>()
+      {
+        { "property|og:image", "content" },
+        { "rel|apple-touch-icon", "href" },
+        { "rel|apple-touch-icon-precomposed", "href"},
+        { "name|msapplication-square310x310logo", "content" },
+        { "name|msapplication-square150x150logo", "content" },
+        { "name|msapplication-square70x70logo", "content" },
+        { "name|msapplication-TileImage", "content" },
+        { "rel|image_src", "href" }
+      });
+    }
+
+
+    /// <summary>
+    /// Gets the meta favicon.
+    /// </summary>
+    /// <returns></returns>
+    public string GetMetaFavicon()
+    {
+      return SearchCandidates(new Dictionary<string, string>()
+      {
+        { "rel|icon", "href" },
+        { "rel|shortcut icon", "href" }
+      });
+    }
+
+
+    /// <summary>
+    /// Searches the candidates.
+    /// </summary>
+    /// <param name="candidates">The candidates.</param>
+    /// <returns></returns>
+    private string SearchCandidates(Dictionary<string, string> candidates)
+    {
+      string result = null;
+
+      foreach (var candidate in candidates)
+      {
+        string[] type = candidate.Key.Split('|');
+
+        XElement element = Tags
+          .Where(item => String.Equals(item.GetAttributeValue(type[0], null), type[1], StringComparison.OrdinalIgnoreCase))
+          .FirstOrDefault();
+
+        if (element != null)
+        {
+          result = element.GetAttributeValue(candidate.Value, "");
+        }
+
+        if (result != null && result.Length > 1)
+        {
+          break;
+        }
+      }
+
+      return result;
+    }
+  }
+}

PortablePorts/NReadability/NReadability.csproj

@@ -49,6 +49,7 @@
     <Compile Include="Enums.cs" />
     <Compile Include="HtmlUtils.cs" />
     <Compile Include="InternalErrorException.cs" />
+    <Compile Include="MetaExtractor.cs" />
     <Compile Include="NReadabilityTranscoder.cs" />
     <Compile Include="Properties\AssemblyInfo.cs" />
     <Compile Include="SgmlDomBuilder.cs" />

PortablePorts/NReadability/NReadabilityTranscoder.cs

@@ -159,6 +159,7 @@ namespace ReadSharp.Ports.NReadability
 
     private Func<AttributeTransformationInput, AttributeTransformationResult> _imageSourceTranformer;
     private Func<AttributeTransformationInput, AttributeTransformationResult> _anchorHrefTransformer;
+    private Func<AttributeTransformationInput, AttributeTransformationResult> _metaTransformer;
 
     #endregion
 
@@ -249,11 +250,38 @@ namespace ReadSharp.Ports.NReadability
 
       bool titleExtracted = !string.IsNullOrEmpty(extractedTitle);
 
+      MetaExtractor metaExtractor = new MetaExtractor(transcodedXmlDocument);
+
+      string description = null;
+      Uri image = null;
+      Uri favicon = null;
+
+      if (metaExtractor.HasValue)
+      {
+        description = metaExtractor.GetMetaDescription();
+        string imageString = metaExtractor.GetMetaImage();
+        string faviconString = metaExtractor.GetMetaFavicon();
+
+        if (imageString != null)
+        {
+          imageString = ResolveElementUrl(imageString, transcodingInput.Url);
+          Uri.TryCreate(imageString, UriKind.Absolute, out image);
+        }
+        if (faviconString != null)
+        {
+          faviconString = ResolveElementUrl(faviconString, transcodingInput.Url);
+          Uri.TryCreate(faviconString, UriKind.Absolute, out favicon);
+        }
+      }
+
       return
         new TranscodingResult(contentExtracted, titleExtracted)
           {
             ExtractedContent = transcodedContent,
             ExtractedTitle = extractedTitle,
+            ExtractedDescription = description,
+            ExtractedFavicon = favicon,
+            ExtractedImage = image,
             NextPageUrl = nextPageUrl,
             Images = images
           };
@@ -1835,6 +1863,13 @@ namespace ReadSharp.Ports.NReadability
       set { _anchorHrefTransformer = value; }
     }
 
+
+    public Func<AttributeTransformationInput, AttributeTransformationResult> MetaTranformer
+    {
+      get { return _metaTransformer; }
+      set { _metaTransformer = value; }
+    }
+
     #endregion
   }
 }

PortablePorts/NReadability/TranscodingResult.cs

@@ -1,3 +1,4 @@
+using System;
 using System.Collections.Generic;
 using System.Xml.Linq;
 
@@ -21,6 +22,12 @@ namespace ReadSharp.Ports.NReadability
 
     public string NextPageUrl { get; set; }
 
+    public string ExtractedDescription { get; set; }
+
+    public Uri ExtractedImage { get; set; }
+
+    public Uri ExtractedFavicon { get; set; }
+
     public XDocument RawDocument { get; set; }
 
     public IEnumerable<XElement> Images { get; set; }

ReadSharp/Models/Article.cs

@@ -8,6 +8,22 @@ namespace ReadSharp
   /// </summary>
   public class Article
   {
+    /// <summary>
+    /// Gets or sets the title.
+    /// </summary>
+    /// <value>
+    /// The title.
+    /// </value>
+    public string Title { get; set; }
+
+    /// <summary>
+    /// Gets or sets the description.
+    /// </summary>
+    /// <value>
+    /// The description.
+    /// </value>
+    public string Description { get; set; }
+
     /// <summary>
     /// Gets or sets the content.
     /// </summary>
@@ -16,6 +32,22 @@ namespace ReadSharp
     /// </value>
     public string Content { get; set; }
 
+    /// <summary>
+    /// Gets or sets the front image.
+    /// </summary>
+    /// <value>
+    /// The front image.
+    /// </value>
+    public Uri FrontImage { get; set; }
+
+    /// <summary>
+    /// Gets or sets the favicon.
+    /// </summary>
+    /// <value>
+    /// The favicon.
+    /// </value>
+    public Uri Favicon { get; set; }
+
     /// <summary>
     /// Gets or sets the images.
     /// </summary>
@@ -24,14 +56,6 @@ namespace ReadSharp
     /// </value>
     public List<ArticleImage> Images { get; set; }
 
-    /// <summary>
-    /// Gets or sets the title.
-    /// </summary>
-    /// <value>
-    /// The title.
-    /// </value>
-    public string Title { get; set; }
-
     /// <summary>
     /// Gets or sets the next page URL.
     /// </summary>

ReadSharp/Reader.cs

@@ -68,7 +68,7 @@ namespace ReadSharp
     /// </summary>
     /// <param name="uri">An URI to extract the content from.</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>
+    /// <param name="noHeadline">if set to <c>true</c> [no headline (h1) is included in generated HTML].</param>
     /// <returns>
     /// An article with extracted content and meta information.
     /// </returns>
@@ -124,9 +124,12 @@ namespace ReadSharp
       // create article
       return new Article()
       {
-        Content = transcodingResult.ExtractedContent,
-        Images = images,
         Title = transcodingResult.ExtractedTitle,
+        Description = transcodingResult.ExtractedDescription,
+        Content = transcodingResult.ExtractedContent,
+        FrontImage = transcodingResult.ExtractedImage,
+        Images = images,
+        Favicon = transcodingResult.ExtractedFavicon,
         NextPage = transcodingResult.NextPageUrl != null ? new Uri(transcodingResult.NextPageUrl, UriKind.Absolute) : null
       };
     }