don't remove a[name] tags when they contain text.

This commit is contained in:
2014-02-09 17:57:32 +01:00
parent f4190ecbcd
commit c0b4635865
2 changed files with 37 additions and 2 deletions
@@ -139,7 +139,9 @@ namespace ReadSharp.Ports.NReadability
{
{ new Regex("^https?://(www|mobile)\\.theverge.com", RegexOptions.IgnoreCase), ".entry-body" },
{ new Regex("^https?://(www|blog)\\.bufferapp.com", RegexOptions.IgnoreCase), ".post" },
{ new Regex("^https?://(www.)?polygon.com", RegexOptions.IgnoreCase), ".body" }
{ new Regex("^https?://(www.)?polygon.com", RegexOptions.IgnoreCase), ".body" },
{ new Regex("^https?://(www.)?gizmodo.com", RegexOptions.IgnoreCase), ".post-container" },
{ new Regex("^https?://(www.)?it-scoop.com", RegexOptions.IgnoreCase), ".entry-content" }
};
#endregion
@@ -790,7 +792,22 @@ namespace ReadSharp.Ports.NReadability
rootElement.GetElementsByTagName("a")
.Where(aElement => aElement.Attribute("name") != null && aElement.Attribute("href") == null);
elementsToRemove.AddRange(anchorElements);
var elementsToBeReplaced = new List<XElement>();
foreach (XElement element in anchorElements)
{
if (!String.IsNullOrWhiteSpace(element.Value))
{
elementsToBeReplaced.Add(element);
}
else
{
elementsToRemove.Add(element);
}
}
elementsToBeReplaced.ForEach(element => element.ReplaceWith(element.Value));
RemoveElements(elementsToRemove);
/* Turn all double br's into p's and all font's into span's. */
+18
View File
@@ -129,9 +129,14 @@ namespace ReadSharp.Tests
[Fact]
public async Task TestDifferentCharsets()
{
// chinese?
string expectedTitle = "优艺客-专注互联网品牌建设-原韩雪冬网页设计工作室(公司站)";
Article result = await reader.Read(new Uri("http://www.uelike.com"));
Assert.Equal(result.Title, expectedTitle);
// arabic
result = await reader.Read(new Uri("http://www.it-scoop.com/2014/01/internet-of-things-google-nest/"));
Assert.NotEmpty(result.Content);
}
[Fact]
@@ -140,6 +145,9 @@ namespace ReadSharp.Tests
Article result = await reader.Read(new Uri("http://wpcentral.com.feedsportal.com/c/33999/f/616880/s/35a02b5e/sc/15/l/0L0Swpcentral0N0Cgameloft0Ediscusses0Etheir0Enew0Egame0Ebrothers0Earms0E30Esons0Ewar0Eceslive/story01.htm"));
Assert.NotEmpty(result.Content);
result = await reader.Read(new Uri("http://www.fastcoexist.com/3016005/futurist-forum/10-creative-ideas-for-thriving-cities-of-the-future"));
Assert.Contains("1: 311", result.Content);
result = await reader.Read(new Uri("http://msdn.microsoft.com/en-us/library/windows/apps/hh464925.aspx"));
Assert.NotEmpty(result.Content);
@@ -166,5 +174,15 @@ namespace ReadSharp.Tests
result = await reader.Read(new Uri("http://blog.bufferapp.com/connections-in-the-brain-understanding-creativity-and-intelligenceconnections"));
Assert.Contains("The Tweet resulted in over 1,000 retweets", result.Content);
}
[Fact]
public async Task AreMultipageArticlesWorking()
{
Article result = await reader.Read(new Uri("http://www.pcwelt.de/ratgeber/Acht_Tipps_fuer_die_sichere_Cloud-Google_Drive__Microsoft_Skydrive__Teamdrive-8205869.html?redirect=1"));
Assert.NotNull(result.NextPage);
result = await reader.Read(new Uri("http://arstechnica.com/apple/2014/01/two-steps-forward-a-review-of-the-2013-mac-pro/"));
Assert.NotNull(result.NextPage);
}
}
}