diff --git a/src/BirdsiteLive.Common/Regexes/UrlRegexes.cs b/src/BirdsiteLive.Common/Regexes/UrlRegexes.cs
index ea3e5c2..1f2b279 100644
--- a/src/BirdsiteLive.Common/Regexes/UrlRegexes.cs
+++ b/src/BirdsiteLive.Common/Regexes/UrlRegexes.cs
@@ -4,6 +4,6 @@ namespace BirdsiteLive.Common.Regexes
{
public class UrlRegexes
{
- public static readonly Regex Url = new Regex(@"((http|ftp|https):\/\/[\w\-_]+(\.[\w\-_]+)+([\w\-\.,@?^=%&:/~\+#]*[\w\-\@?^=%&/~\+#])?)");
+ public static readonly Regex Url = new Regex(@"(.?)(((http|ftp|https):\/\/)[\w\-_]+(\.[\w\-_]+)+([\w\-\.,@?^=%&:/~\+#]*[\w\-\@?^=%&/~\+#])?)");
}
}
\ No newline at end of file
diff --git a/src/BirdsiteLive.Domain/Tools/StatusExtractor.cs b/src/BirdsiteLive.Domain/Tools/StatusExtractor.cs
index a43ccb0..7df4c0e 100644
--- a/src/BirdsiteLive.Domain/Tools/StatusExtractor.cs
+++ b/src/BirdsiteLive.Domain/Tools/StatusExtractor.cs
@@ -44,11 +44,8 @@ namespace BirdsiteLive.Domain.Tools
var urlMatch = UrlRegexes.Url.Matches(messageContent);
foreach (Match m in urlMatch)
{
- var url = m.ToString().Replace("\n", string.Empty).Trim();
-
- var protocol = "https://";
- if (url.StartsWith("http://")) protocol = "http://";
- else if (url.StartsWith("ftp://")) protocol = "ftp://";
+ var url = m.Groups[2].ToString();
+ var protocol = m.Groups[3].ToString();
var truncatedUrl = url.Replace(protocol, string.Empty);
@@ -68,7 +65,7 @@ namespace BirdsiteLive.Domain.Tools
}
messageContent = Regex.Replace(messageContent, m.ToString(),
- $@" {protocol}{firstPart}{secondPart}");
+ $@"{m.Groups[1]}{protocol}{firstPart}{secondPart}");
}
// Extract Hashtags
diff --git a/src/BirdsiteLive/BirdsiteLive.csproj b/src/BirdsiteLive/BirdsiteLive.csproj
index 346a936..cb2c708 100644
--- a/src/BirdsiteLive/BirdsiteLive.csproj
+++ b/src/BirdsiteLive/BirdsiteLive.csproj
@@ -4,7 +4,7 @@
netcoreapp3.1
d21486de-a812-47eb-a419-05682bb68856
Linux
- 0.14.0
+ 0.14.1
diff --git a/src/Tests/BirdsiteLive.Domain.Tests/Tools/StatusExtractorTests.cs b/src/Tests/BirdsiteLive.Domain.Tests/Tools/StatusExtractorTests.cs
index 924461f..06d8f9a 100644
--- a/src/Tests/BirdsiteLive.Domain.Tests/Tools/StatusExtractorTests.cs
+++ b/src/Tests/BirdsiteLive.Domain.Tests/Tools/StatusExtractorTests.cs
@@ -136,7 +136,7 @@ namespace BirdsiteLive.Domain.Tests.Tools
}
[TestMethod]
- public void Extract_MultiUrls__Test()
+ public void Extract_MultiUrls_Test()
{
#region Stubs
var message = $"https://t.co/L8BpyHgg25 Bla!{Environment.NewLine}https://www.eff.org/deeplinks/2020/07/pact-act-not-solution-problem-harmful-online-content";
@@ -160,6 +160,63 @@ namespace BirdsiteLive.Domain.Tests.Tools
#endregion
}
+ [TestMethod]
+ public void Extract_SmallUrl_Test()
+ {
+ #region Stubs
+ var message = @"🚀 test http://GOV.UK date 🎉 data http://GOV.UK woopsi.";
+ #endregion
+
+ #region Mocks
+ var logger = new Mock>();
+ #endregion
+
+ var service = new StatusExtractor(_settings, logger.Object);
+ var result = service.Extract(message);
+
+ #region Validations
+ Assert.AreEqual(@"🚀 test http://GOV.UK date 🎉 data http://GOV.UK woopsi.", result.content);
+ #endregion
+ }
+
+ [TestMethod]
+ public void Extract_SmallUrl_2_Test()
+ {
+ #region Stubs
+ var message = @"🚀http://GOV.UK";
+ #endregion
+
+ #region Mocks
+ var logger = new Mock>();
+ #endregion
+
+ var service = new StatusExtractor(_settings, logger.Object);
+ var result = service.Extract(message);
+
+ #region Validations
+ Assert.AreEqual(@"🚀http://GOV.UK", result.content);
+ #endregion
+ }
+
+ [TestMethod]
+ public void Extract_SmallUrl_3_Test()
+ {
+ #region Stubs
+ var message = @"🚀http://GOV.UK.";
+ #endregion
+
+ #region Mocks
+ var logger = new Mock>();
+ #endregion
+
+ var service = new StatusExtractor(_settings, logger.Object);
+ var result = service.Extract(message);
+
+ #region Validations
+ Assert.AreEqual(@"🚀http://GOV.UK.", result.content);
+ #endregion
+ }
+
[TestMethod]
public void Extract_SingleHashTag_Test()
{