Merge branch 'cpetry/main'

MaxMelcher · MaxMelcher · commit e62b13b5c810 · 2022-01-20T12:45:57.000+01:00
diff --git a/AzureDevOps.WikiPDFExport.Test/AzureDevOps.WikiPDFExport.Test.csproj b/AzureDevOps.WikiPDFExport.Test/AzureDevOps.WikiPDFExport.Test.csproj
@@ -12,4 +12,8 @@
     <PackageReference Include="xunit.runner.visualstudio" Version="2.4.0" />
   </ItemGroup>
 
+  <ItemGroup>
+    <ProjectReference Include="..\AzureDevOps.WikiPDFExport\azuredevops-export-wiki.csproj" />
+  </ItemGroup>
+
 </Project>
diff --git a/AzureDevOps.WikiPDFExport.Test/Tests/TableOfContent/TableOfContent_uTest.cs b/AzureDevOps.WikiPDFExport.Test/Tests/TableOfContent/TableOfContent_uTest.cs
@@ -0,0 +1,101 @@
+﻿using azuredevops_export_wiki;
+using System.Collections.Generic;
+using System.Linq;
+using Xunit;
+
+namespace AzureDevOps.WikiPDFExport
+{
+    public class TableOfContent_uTest
+    {
+        [Fact]
+        public void CreateGlobalTableOfContent_ShouldReturnTOCandSingleHeaderLine()
+        {
+            // Arrange
+            var wikiPDFExporter = new WikiPDFExporter(new Options());
+            var mdContent1 = "\n# SomeHeader\n"
+                + "SomeText";
+
+            // Act
+            var result = wikiPDFExporter.CreateGlobalTableOfContent(new List<string> { mdContent1 });
+
+            Assert.Equal("[TOC]", result[0]);
+            Assert.Equal("# SomeHeader", result[1]);
+        }
+
+        [Fact]
+        public void CreateGlobalTableOfContent_ShouldNotReturnTOC_WhenNoHeaderFound()
+        {
+            // Arrange
+            var wikiPDFExporter = new WikiPDFExporter(new Options());
+            var mdContent1 = "\nOnly boring text\n"
+                + "No header here";
+
+            // Act
+            var result = wikiPDFExporter.CreateGlobalTableOfContent(new List<string> { mdContent1 });
+
+            Assert.False(result.Any());
+        }
+
+        [Fact]
+        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 #";
+
+            // Act
+            var result = wikiPDFExporter.CreateGlobalTableOfContent(new List<string> { mdContent1, mdContent2 });
+
+            Assert.Equal("[TOC]", result[0]);
+            Assert.Equal("# SomeHeader", result[1]);
+            Assert.Equal("## SomeOtherHeader", result[2]);
+        }
+
+        [Fact]
+        public void RemoveDuplicatedHeadersFromGlobalTOC()
+        {
+            // Arrange
+            var wikiPDFExporter = new WikiPDFExporter(new Options());
+            var htmlContent = "<h1>SomeHeader</h1>\n"
+                + "<h2>SomeOtherHeader</h2>\n";
+
+            // Act
+            var result = wikiPDFExporter.RemoveDuplicatedHeadersFromGlobalTOC(htmlContent);
+
+            Assert.Equal("", result);
+        }
+
+        [Fact]
+        public void RemoveDuplicatedHeadersFromGlobalTOC_WhenIdsDefined()
+        {
+            // Arrange
+            var wikiPDFExporter = new WikiPDFExporter(new Options());
+            var htmlContent = "<h1 id='interestingID'>SomeHeader</h1>\n"
+                + "<h2>SomeOtherHeader</h2>\n";
+
+            // Act
+            var result = wikiPDFExporter.RemoveDuplicatedHeadersFromGlobalTOC(htmlContent);
+
+            Assert.Equal("", result);
+        }
+
+        [Fact]
+        public void RemoveDuplicatedHeadersFromGlobalTOC_ExceptNavTag()
+        {
+            // Arrange
+            var wikiPDFExporter = new WikiPDFExporter(new Options());
+            var nav = "<nav>Some cool nav content</nav>\n";
+            var htmlContent = nav
+                + "<h1>SomeHeader</h1>\n"
+                + "<h2>SomeOtherHeader</h2>\n";
+
+            // Act
+            var result = wikiPDFExporter.RemoveDuplicatedHeadersFromGlobalTOC(htmlContent);
+
+            Assert.Equal(nav.Trim('\n'), result);
+        }
+    }
+}
diff --git a/AzureDevOps.WikiPDFExport/Program.cs b/AzureDevOps.WikiPDFExport/Program.cs
@@ -100,5 +100,8 @@ public class Options
 
         [Option("organization", Required = false, HelpText = "Azure Devops organization URL used to convert work item references to work item links. Ex: https://dev.azure.com/MyOrganizationName/")]
         public string AzureDevopsOrganization { get; set; }
+
+        [Option("globaltoc", Required = false, HelpText = "Title for a global table of content for all markdown files. When not specified each markdown creates its own toc if defined")]
+        public string GlobalTOC { get; set; }
     }
 }
diff --git a/AzureDevOps.WikiPDFExport/WikiPDFExporter.cs b/AzureDevOps.WikiPDFExport/WikiPDFExporter.cs
@@ -30,6 +30,9 @@
 
 using Microsoft.VisualStudio.Services.WebApi;
 using Process = System.Diagnostics.Process;
+using System.Runtime.CompilerServices;
+
+[assembly:InternalsVisibleTo("AzureDevOps.WikiPDFExport.Test")]
 
 namespace azuredevops_export_wiki
 {
@@ -100,7 +103,6 @@ public WikiPDFExporter(Options options)
                 {"icon_pull_request", "bowtie-tfvc-pull-request"},
                 {"icon_github_issue", "bowtie-status-error-outline"},
             };
-
         }
 
         public async Task Export()
@@ -145,7 +147,6 @@ public async Task Export()
                             }
                         };
                     }
-                    else
                     {
                         files = ReadOrderFiles(_path, 0); // root level
                     }
@@ -399,7 +400,8 @@ private string ConvertMarkdownToHTML(List<MarkdownFile> files)
                 .UsePipeTables()
                 .UseEmojiAndSmiley()
                 .UseAdvancedExtensions()
-                .UseYamlFrontMatter();
+                .UseYamlFrontMatter()
+                .UseTableOfContent();
 
             //must be handled by us to have linking across files
             pipelineBuilder.Extensions.RemoveAll(x => x is Markdig.Extensions.AutoIdentifiers.AutoIdentifierExtension);
@@ -430,11 +432,38 @@ private string ConvertMarkdownToHTML(List<MarkdownFile> files)
                     continue;
                 }
 
-                var md = File.ReadAllText(file.FullName);
+                var markdownContent = File.ReadAllText(file.FullName);
+                files[i].Content = markdownContent;
+            }
+
+            if (!string.IsNullOrEmpty(_options.GlobalTOC))
+            {
+                var firstMDFileInfo = new FileInfo(files[0].AbsolutePath);
+                var directoryName = firstMDFileInfo.Directory.Name;
+                var tocName = _options.GlobalTOC == "" ? directoryName : _options.GlobalTOC;
+                var relativePath = "/" + tocName + ".md";
+                var tocMDFilePath = new FileInfo(files[0].AbsolutePath).DirectoryName + relativePath;
+
+                var contents = files.Select(x => x.Content).ToList();
+                var tocContent = CreateGlobalTableOfContent(contents);
+                var tocString = string.Join("\n", tocContent);
+
+                var tocMarkdownFile = new MarkdownFile { AbsolutePath = tocMDFilePath, Level = 0, RelativePath = relativePath, Content = tocString };
+                files.Insert(0, tocMarkdownFile);
+            }
+
+            for (var i = 0; i < files.Count; i++)
+            {
+                var mf = files[i];
+                var file = new FileInfo(files[i].AbsolutePath);
+
+                Log($"{file.Name}", LogLevel.Information, 1);
 
-                //replace Table of Content
-                md = RemoveTableOfContent(md);
+                var md = mf.Content;
 
+                //rename TOC tags to fit to MarkdigToc or delete them from each markdown document
+                var newTOCString = _options.GlobalTOC != null ? "" : "[TOC]"; 
+                md = md.Replace("[[_TOC_]]", newTOCString);
 
                 // remove scalings from image links, width & height: file.png =600x500
                 var regexImageScalings = @"\((.[^\)]*?[png|jpg|jpeg]) =(\d+)x(\d+)\)";
@@ -457,7 +486,7 @@ private string ConvertMarkdownToHTML(List<MarkdownFile> files)
                 var pipeline = pipelineBuilder.Build();
 
                 //parse the markdown document so we can alter it later
-                var document = (MarkdownDocument)Markdown.Parse(md, pipeline);
+                var document = Markdown.Parse(md, pipeline);
 
                 if (_options.NoFrontmatter)
                 {
@@ -477,6 +506,7 @@ private string ConvertMarkdownToHTML(List<MarkdownFile> files)
                     }
                 }
 
+
                 //adjust the links
                 CorrectLinksAndImages(document, file, mf);
 
@@ -491,6 +521,12 @@ private string ConvertMarkdownToHTML(List<MarkdownFile> files)
                 }
                 html = builder.ToString();
 
+                if (!string.IsNullOrEmpty(_options.GlobalTOC) && i == 0)
+                {
+                    html = RemoveDuplicatedHeadersFromGlobalTOC(html);
+                    Log($"Removed duplicated headers from toc html", LogLevel.Information, 1);
+                }
+
                 //add html anchor
                 var anchorPath = file.FullName.Substring(_path.Length);
                 anchorPath = anchorPath.Replace("\\", "");
@@ -513,6 +549,13 @@ private string ConvertMarkdownToHTML(List<MarkdownFile> files)
                     html = heading + html;
                 }
 
+
+                if (!string.IsNullOrEmpty(_options.GlobalTOC) && i == 0 && !_options.Heading)
+                {
+                    var heading = $"<h1>{_options.GlobalTOC}</h1>";
+                    html = heading + html;
+                }
+
                 if (_options.Heading)
                 {
                     var filename = file.Name.Replace(".md", "");
@@ -564,6 +607,30 @@ private string ConvertMarkdownToHTML(List<MarkdownFile> files)
             return result;
         }
 
+        internal string RemoveDuplicatedHeadersFromGlobalTOC(string html)
+        {
+            var result = Regex.Replace(html, @"^ *<h[123456].*>.*<\/h[123456]> *\n?$", "", RegexOptions.Multiline);
+            result = result.Trim('\n');
+            return result;
+        }
+
+        internal List<string> CreateGlobalTableOfContent(List<string> contents)
+        {
+            var headers = new List<string>();
+            foreach (var content in contents) 
+            {
+                var headerMatches = Regex.Matches(content, "^ *#+ ?.*$", RegexOptions.Multiline);
+                headers.AddRange(headerMatches.Select(x => x.Value.Trim()));
+            }
+
+            if (!headers.Any())
+                return new List<string>(); // no header -> no toc
+
+            var tocContent = new List<string> { "[TOC]" }; // MarkdigToc style
+            tocContent.AddRange(headers);
+            return tocContent;
+        }
+
         private MarkdownDocument RemoveFrontmatter(MarkdownDocument document)
         {
             var frontmatter = document.Descendants<YamlFrontMatterBlock>().FirstOrDefault();
@@ -636,13 +703,10 @@ private bool PageMatchesFilter(MarkdownObject document)
             return true;
         }
 
-        private string RemoveTableOfContent(string document)
+        private string RenameTableOfContent(string document)
         {
             if (document.Contains("TOC"))
-            {
-                Log("Removing Table of contents [[_TOC_]] from pdf", LogLevel.Warning, 1);
-                document = document.Replace("[[_TOC_]]", "");
-            }
+                document = document.Replace("[[_TOC_]]", "[TOC]"); // MarkdigToc styled. See https://github.com/leisn/MarkdigToc
             return document;
         }
 
@@ -860,6 +924,7 @@ public class MarkdownFile
         public string AbsolutePath;
         public string RelativePath;
         public int Level;
+        public string Content;
 
         public override string ToString()
         {
diff --git a/AzureDevOps.WikiPDFExport/azuredevops-export-wiki.csproj b/AzureDevOps.WikiPDFExport/azuredevops-export-wiki.csproj
@@ -25,6 +25,7 @@
 
   <ItemGroup>
     <PackageReference Include="CommandLineParser" Version="2.8.0" />
+    <PackageReference Include="Leisn.MarkdigToc" Version="0.1.3" />
     <PackageReference Include="Markdig" Version="0.26.0" />
     <PackageReference Include="Microsoft.ApplicationInsights.AspNetCore" Version="2.19.0" />
     <PackageReference Include="Microsoft.Extensions.Logging" Version="6.0.0" />
diff --git a/azuredevops-export-wiki.sln b/azuredevops-export-wiki.sln
@@ -5,6 +5,8 @@ VisualStudioVersion = 16.0.30907.101
 MinimumVisualStudioVersion = 10.0.40219.1
 Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "azuredevops-export-wiki", "AzureDevOps.WikiPDFExport\azuredevops-export-wiki.csproj", "{9E8EFB6E-03E6-4C3A-B971-D31DFE483B59}"
 EndProject
+Project("{9A19103F-16F7-4668-BE54-9A1E7A4F7556}") = "AzureDevOps.WikiPDFExport.Test", "AzureDevOps.WikiPDFExport.Test\AzureDevOps.WikiPDFExport.Test.csproj", "{6DDA14AF-1CEA-4038-88C9-17BA6E8F3AC2}"
+EndProject
 Global
 	GlobalSection(SolutionConfigurationPlatforms) = preSolution
 		Debug|Any CPU = Debug|Any CPU

Original file line number	Diff line number	Diff line change
`@@ -100,5 +100,8 @@ public class Options`
`100`	`100`
`101`	`101`	`[Option("organization", Required = false, HelpText = "Azure Devops organization URL used to convert work item references to work item links. Ex: https://dev.azure.com/MyOrganizationName/")]`
`102`	`102`	`public string AzureDevopsOrganization { get; set; }`
	`103`	`+`
	`104`	`+ [Option("globaltoc", Required = false, HelpText = "Title for a global table of content for all markdown files. When not specified each markdown creates its own toc if defined")]`
	`105`	`+ public string GlobalTOC { get; set; }`
`103`	`106`	`}`
`104`	`107`	`}`