Merge pull request #98 from NicolasConstant/fix_tweet-parsing

Fix tweet parsing
NicolasConstant · Mar 30, 2021 · 07d09eb · 07d09eb
2 parents a4d62fa + c1f4aed
commit 07d09eb
Show file tree

Hide file tree

Showing 3 changed files with 53 additions and 17 deletions.
diff --git a/src/BirdsiteLive.Domain/Tools/StatusExtractor.cs b/src/BirdsiteLive.Domain/Tools/StatusExtractor.cs
@@ -81,12 +81,16 @@ public StatusExtractor(InstanceSettings instanceSettings, ILogger<StatusExtracto
                 }
 
                 var url = $"https://{_instanceSettings.Domain}/tags/{tag}";
-                tags.Add(new Tag
+
+                if (tags.All(x => x.href != url))
                 {
-                    name = $"#{tag}",
-                    href = url,
-                    type = "Hashtag"
-                });
+                    tags.Add(new Tag
+                    {
+                        name = $"#{tag}",
+                        href = url,
+                        type = "Hashtag"
+                    });
+                }
 
                 messageContent = Regex.Replace(messageContent, Regex.Escape(m.Groups[0].ToString()),
                     $@"{m.Groups[1]}<a href=""{url}"" class=""mention hashtag"" rel=""tag"">#<span>{tag}</span></a>{m.Groups[3]}");
@@ -96,7 +100,7 @@ public StatusExtractor(InstanceSettings instanceSettings, ILogger<StatusExtracto
             if (extractMentions)
             {
                 var mentionMatch = OrderByLength(UserRegexes.Mention.Matches(messageContent));
-                foreach (Match m in mentionMatch.OrderByDescending(x => x.Length))
+                foreach (Match m in mentionMatch)
                 {
                     var mention = m.Groups[2].ToString();
 
@@ -109,27 +113,34 @@ public StatusExtractor(InstanceSettings instanceSettings, ILogger<StatusExtracto
                     var url = $"https://{_instanceSettings.Domain}/users/{mention}";
                     var name = $"@{mention}@{_instanceSettings.Domain}";
 
-                    tags.Add(new Tag
+                    if (tags.All(x => x.href != url))
                     {
-                        name = name,
-                        href = url,
-                        type = "Mention"
-                    });
-
+                        tags.Add(new Tag
+                        {
+                            name = name,
+                            href = url,
+                            type = "Mention"
+                        });
+                    }
+
                     messageContent = Regex.Replace(messageContent, Regex.Escape(m.Groups[0].ToString()),
                         $@"{m.Groups[1]}<span class=""h-card""><a href=""https://{_instanceSettings.Domain}/@{mention}"" class=""u-url mention"">@<span>{mention}</span></a></span>{m.Groups[3]}");
                 }
             }
 
             return (messageContent.Trim(), tags.ToArray());
         }
-
+        
         private IEnumerable<Match> OrderByLength(MatchCollection matches)
         {
             var result = new List<Match>();
-
             foreach (Match m in matches) result.Add(m);
-            result = result.OrderByDescending(x => x.Length).ToList();
+
+            result = result
+                .OrderBy(x => x.Length)
+                .GroupBy(p => p.Value)
+                .Select(g => g.First())
+                .ToList();
 
             return result;
         }

diff --git a/src/BirdsiteLive/BirdsiteLive.csproj b/src/BirdsiteLive/BirdsiteLive.csproj
@@ -4,7 +4,7 @@
     <TargetFramework>netcoreapp3.1</TargetFramework>
     <UserSecretsId>d21486de-a812-47eb-a419-05682bb68856</UserSecretsId>
     <DockerDefaultTargetOS>Linux</DockerDefaultTargetOS>
-    <Version>0.16.0</Version>
+    <Version>0.16.1</Version>
   </PropertyGroup>
 
   <ItemGroup>

diff --git a/src/Tests/BirdsiteLive.Domain.Tests/Tools/StatusExtractorTests.cs b/src/Tests/BirdsiteLive.Domain.Tests/Tools/StatusExtractorTests.cs
@@ -364,7 +364,32 @@ public void Extract_SingleMentionTag_Test()
             Assert.IsTrue(result.content.Contains(@"<span class=""h-card""><a href=""https://domain.name/@mynickname"" class=""u-url mention"">@<span>mynickname</span></a></span>"));
             #endregion
         }
-
+
+        [TestMethod]
+        public void Extract_MultiMentionTag_MultiOccurrence_Test()
+        {
+            #region Stubs
+            var message = $"[RT @yamenbousrih]{Environment.NewLine}@KiwixOffline @photos_floues Bla. Cc @Pyb75 @photos_floues @KiwixOffline";
+            #endregion
+
+            #region Mocks
+            var logger = new Mock<ILogger<StatusExtractor>>();
+            #endregion
+
+            var service = new StatusExtractor(_settings, logger.Object);
+            var result = service.Extract(message);
+
+            #region Validations
+            logger.VerifyAll();
+            Assert.AreEqual(4, result.tags.Length);
+            Assert.AreEqual("Mention", result.tags.First().type);
+
+            Assert.IsTrue(result.content.Contains(@"<span class=""h-card""><a href=""https://domain.name/@photos_floues"" class=""u-url mention"">@<span>photos_floues</span></a></span>"));
+            Assert.IsTrue(result.content.Contains(@"<span class=""h-card""><a href=""https://domain.name/@KiwixOffline"" class=""u-url mention"">@<span>KiwixOffline</span></a></span> <span class=""h-card""><a href=""https://domain.name/@photos_floues"" class=""u-url mention"">@<span>photos_floues</span></a></span>"));
+            Assert.IsTrue(result.content.Contains(@"Cc <span class=""h-card""><a href=""https://domain.name/@Pyb75"" class=""u-url mention"">@<span>Pyb75</span></a></span> <span class=""h-card""><a href=""https://domain.name/@photos_floues"" class=""u-url mention"">@<span>photos_floues</span></a></span> <span class=""h-card""><a href=""https://domain.name/@KiwixOffline"" class=""u-url mention"">@<span>KiwixOffline</span></a></span>"));
+            #endregion
+        }
+
         [TestMethod]
         public void Extract_SingleMentionTag_RT_Test()
         {