Add removal of lists in posts so they won't be taken part into analytics

ktos · Feb 11, 2020 · 5f43567 · 5f43567
1 parent 5bfc4f5
commit 5f43567
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 5 deletions.
diff --git a/src/HtmlCleaner.cs b/src/HtmlCleaner.cs
@@ -51,13 +51,14 @@ public static string RemoveProperCode(string postText)
         }
 
         /// <summary>
-        /// Removes quotations, which should be ignored
+        /// Removes quotations, lists and other HTML tags which content should be ignored
         /// </summary>
         /// <param name="postText">Post content to be cleaned (in HTML)</param>
         /// <returns>Post content with Blockquote elements removed</returns>
-        public static string RemoveQuotes(string postText)
+        public static string RemoveHTMLContent(string postText)
         {
             postText = Regex.Replace(postText, "<blockquote>(.|\n)*?</blockquote>", "", RegexOptions.Multiline);
+            postText = Regex.Replace(postText, "<ul>(.|\n)*?</ul>", "", RegexOptions.Multiline);
             return postText;
         }
 

diff --git a/src/PostAnalyzer.cs b/src/PostAnalyzer.cs
@@ -78,7 +78,7 @@ private NotFormattedCodeFound CheckForUnformattedCode(CoyoteApi.Post post)
         {
             var text = HtmlCleaner.RemoveProperCode(post.html);
             text = HtmlCleaner.RemoveDownloadLinks(text);
-            text = HtmlCleaner.RemoveQuotes(text);
+            text = HtmlCleaner.RemoveHTMLContent(text);
 
             foreach (var para in text.Split("</p>").Select(CleanParagraph))
             {

diff --git a/test/HtmlCleanerTests.cs b/test/HtmlCleanerTests.cs
@@ -27,15 +27,25 @@ public void RemoveDownloadLinks_PostWithDownloadLinks_LinksRemoved()
         }
 
         [Fact]
-        public void RemoveQuotes_PostWithQuotes_QuotesRemoved()
+        public void RemoveHtmlContent_PostWithQuotes_QuotesRemoved()
         {
             string postText = "<blockquote>cytat</blockquote>nie cytat<blockquote>cytat</blockquote>";
 
-            var result = HtmlCleaner.RemoveQuotes(postText);
+            var result = HtmlCleaner.RemoveHTMLContent(postText);
 
             Assert.Equal("nie cytat", result);
         }
 
+        [Fact]
+        public void RemoveHtmlContent_PostWithList_ListRemoved()
+        {
+            string postText = "<ul>\n<li>\n<p>libGDX</p>\n</li>\n<li>\n<p>jMonkeyEngine</p>\n</li>\n<li>\n<p>lwjgl</p>\n</li>\n</ul>";
+
+            var result = HtmlCleaner.RemoveHTMLContent(postText);
+
+            Assert.Equal(string.Empty, result);
+        }
+
         [Fact]
         public void StripTags_TextWithTags_TagsRemoved()
         {