Fixes #94 - Removing code sections when searching for table of content headers

Author: chrPetry

AzureDevOps.WikiPDFExport.Test/Tests/TableOfContent/TableOfContent_uTest.cs

@@ -41,10 +41,11 @@ public void CreateGlobalTableOfContent_ShouldReturnTOCandMultipleHeaderLines()
         {
             // Arrange
             var wikiPDFExporter = new WikiPDFExporter(new Options());
-            var mdContent1 = "\n# SomeHeader\n"
-                + "SomeText";
-            var mdContent2 = "    ## SomeOtherHeader   \n"
-                + " []() #Some very interesting text in wrong header format #";
+            var mdContent1 = @"
+                # SomeHeader
+                SomeText";
+            var mdContent2 = @"    ## SomeOtherHeader   
+                []() #Some very interesting text in wrong header format #";
 
             // Act
             var result = wikiPDFExporter.CreateGlobalTableOfContent(new List<string> { mdContent1, mdContent2 });
@@ -54,6 +55,81 @@ public void CreateGlobalTableOfContent_ShouldReturnTOCandMultipleHeaderLines()
             Assert.Equal("## SomeOtherHeader", result[2]);
         }
 
+        [Fact]
+        public void CreateGlobalTableOfContent_ShouldIgnoreCodeSectionsSingle()
+        {
+            // Arrange
+            var wikiPDFExporter = new WikiPDFExporter(new Options());
+            var mdContent1 = @"
+                ``` code section
+                # SomeHeader
+                ```";
+            
+            // Act
+            var result = wikiPDFExporter.CreateGlobalTableOfContent(new List<string> { mdContent1 });
+
+            Assert.False(result.Any());
+        }
+
+        [Fact]
+        public void CreateGlobalTableOfContent_ShouldIgnoreCodeSectionsMultiple()
+        {
+            // Arrange
+            var wikiPDFExporter = new WikiPDFExporter(new Options());
+            var mdContent1 = @"
+                ``` code section
+                ## SomeHeader
+                Some other text
+                ```
+                ```
+                ## Another header ```
+                # A valid header";
+
+            // Act
+            var result = wikiPDFExporter.CreateGlobalTableOfContent(new List<string> { mdContent1 });
+
+            Assert.Equal(2, result.Count);
+            Assert.Equal("[TOC]", result[0]);
+            Assert.Equal("# A valid header", result[1]);
+        }
+
+        [Fact]
+        public void CreateGlobalTableOfContent_ShouldNotIgnoreInvalidCodeSections()
+        {
+            // Arrange
+            var wikiPDFExporter = new WikiPDFExporter(new Options());
+            var mdContent1 = @"
+                Not at the beginning ```
+                # A valid header
+                ```
+                ";
+
+            // Act
+            var result = wikiPDFExporter.CreateGlobalTableOfContent(new List<string> { mdContent1 });
+
+            Assert.Equal(2, result.Count);
+            Assert.Equal("[TOC]", result[0]);
+            Assert.Equal("# A valid header", result[1]);
+        }
+
+        [Fact]
+        public void CreateGlobalTableOfContent_ShouldNotIgnoreUnclosedCodeSections()
+        {
+            // Arrange
+            var wikiPDFExporter = new WikiPDFExporter(new Options());
+            var mdContent1 = @"
+                ``` unclosed comment
+                # A valid header
+                ";
+
+            // Act
+            var result = wikiPDFExporter.CreateGlobalTableOfContent(new List<string> { mdContent1 });
+
+            Assert.Equal(2, result.Count);
+            Assert.Equal("[TOC]", result[0]);
+            Assert.Equal("# A valid header", result[1]);
+        }
+
         [Fact]
         public void RemoveDuplicatedHeadersFromGlobalTOC()
         {

AzureDevOps.WikiPDFExport/WikiPDFExporter.cs

@@ -634,7 +634,9 @@ internal string RemoveDuplicatedHeadersFromGlobalTOC(string html)
         internal List<string> CreateGlobalTableOfContent(List<string> contents)
         {
             var headers = new List<string>();
-            foreach (var content in contents) 
+            var filteredContentList = RemoveCodeSections(contents);
+
+            foreach (var content in filteredContentList) 
             {
                 var headerMatches = Regex.Matches(content, "^ *#+ ?.*$", RegexOptions.Multiline);
                 headers.AddRange(headerMatches.Select(x => x.Value.Trim()));
@@ -648,6 +650,17 @@ internal List<string> CreateGlobalTableOfContent(List<string> contents)
             return tocContent;
         }
 
+        private List<string> RemoveCodeSections(List<string> contents)
+        {
+            var contentWithoutCode = new List<string>();
+            for(var i=0; i < contents.Count; i++)
+            {
+                var contentWithoutCodeSection = Regex.Replace(contents[i], "^[ \t]*```[^`]*```", "", RegexOptions.Multiline);
+                contentWithoutCode.Add(contentWithoutCodeSection);
+            }
+            return contentWithoutCode;
+        }
+
         private MarkdownDocument RemoveFrontmatter(MarkdownDocument document)
         {
             var frontmatter = document.Descendants<YamlFrontMatterBlock>().FirstOrDefault();