[CleanRepo] Fix bugs with multiple docsets (#655)

Author: gewarren

cleanrepo/Program.cs

@@ -1,6 +1,5 @@
 using System.Data;
 using System.Diagnostics;
-using System.IO;
 using System.Text;
 using System.Text.Json;
 using System.Text.RegularExpressions;
@@ -55,26 +54,26 @@ static async Task Main(string[] args)
 
     static async Task RunOptions(Options options)
     {
-        if (String.IsNullOrEmpty(options.Function))
+        if (string.IsNullOrEmpty(options.Function))
         {
             Console.WriteLine($"\nYou didn't specify which function to perform, " +
                 $"such as {s_functions[0]}, {s_functions[1]}, {s_functions[2]}, or {s_functions[3]}.");
             return;
         }
 
-        if (String.IsNullOrEmpty(options.DocFxDirectory))
+        if (string.IsNullOrEmpty(options.DocFxDirectory))
         {
             Console.WriteLine("\nYou didn't specify the directory that contains the docfx.json file.");
             return;
         }
 
-        if (String.IsNullOrEmpty(options.TargetDirectory))
+        if (string.IsNullOrEmpty(options.TargetDirectory))
         {
             Console.WriteLine("\nYou didn't specify the directory to search/clean.");
             return;
         }
 
-        if (String.IsNullOrEmpty(options.UrlBasePath))
+        if (string.IsNullOrEmpty(options.UrlBasePath))
         {
             Console.WriteLine("\nYou didn't specify the URL base path, such as /dotnet or /windows/uwp.");
             return;
@@ -232,8 +231,7 @@ static async Task RunOptions(Options options)
                     docFxRepo._imageLinkRegExes.Add($"social_image_url: ?\"?(?<path>{docFxRepo.UrlBasePath}.*?(\\.(png|jpg|gif|svg))+)");
 
                     // Gather media file names.
-                    if (docFxRepo._imageRefs is null)
-                        docFxRepo._imageRefs = HelperMethods.GetMediaFiles(options.TargetDirectory);
+                    docFxRepo._imageRefs ??= HelperMethods.GetMediaFiles(options.TargetDirectory);
 
                     Console.WriteLine($"\nCataloging '{docFxRepo._imageRefs.Count}' images (recursively) " +
                         $"in the '{options.TargetDirectory}' directory...\n");
@@ -1668,7 +1666,11 @@ public static bool IsFileLinkedFromFile(FileInfo linkedFile, FileInfo linkingFil
     public static List<FileInfo> GetRedirectionFiles(string directoryPath)
     {
         DirectoryInfo dir = new(directoryPath);
-        return dir.EnumerateFiles(".openpublishing.redirection.*.json", SearchOption.AllDirectories).ToList();
+        return
+        [
+            .. dir.EnumerateFiles(".openpublishing.redirection.json", SearchOption.AllDirectories),
+            .. dir.EnumerateFiles(".openpublishing.redirection.*.json", SearchOption.AllDirectories)
+        ];
     }
 
     /// <summary>

cleanrepo/Repo.cs

@@ -443,7 +443,7 @@ void TryAddLinkingFile(string key, string linkingFile)
     }
 
     /// <summary>
-    /// Pulls docset information, including URL base path, from the OPS config and docfx.json files.
+    /// Pulls docset information from the OPS config and docfx.json files.
     /// </summary>
     internal Dictionary<string, string>? GetDocsetInfo()
     {
@@ -464,7 +464,14 @@ void TryAddLinkingFile(string key, string linkingFile)
                 if (sourceFolder.build_source_folder is null)
                     continue;
 
-                string docfxFilePath = Path.Combine(OpsConfigFile.DirectoryName!, sourceFolder.build_source_folder, "docfx.json");
+                string docfxFilePath = Path.GetFullPath(Path.Combine(OpsConfigFile.DirectoryName!, sourceFolder.build_source_folder, "docfx.json"));
+
+                if (!string.Equals(docfxFilePath, Path.Combine(DocFxDirectory!.FullName, "docfx.json"), StringComparison.InvariantCultureIgnoreCase))
+                {
+                    // This is a different docset in the same repo. Ignore it.
+                    continue;
+                }
+
                 DocFx? docfx = LoadDocfxFile(docfxFilePath);
                 if (docfx == null)
                     continue;
@@ -497,10 +504,10 @@ void TryAddLinkingFile(string key, string linkingFile)
                                         docsetFilePath = string.Concat(docsetFilePath, "/", item.src);
                                 }
                             }
-
-                            if (!mappingInfo.ContainsKey(docsetFilePath))
-                                mappingInfo.Add(docsetFilePath, UrlBasePath);
                         }
+
+                        if (!mappingInfo.ContainsKey(docsetFilePath))
+                            mappingInfo.Add(docsetFilePath, UrlBasePath);
                     }
                 }
             }
@@ -641,18 +648,28 @@ private static void RemoveRedirectHopsFromFile(FileInfo redirectsFile, Dictionar
             if (redirect.source_path != null)
             {
                 // Construct the full path to the redirected file
-                fullPath = Path.Combine(redirectsFile.DirectoryName!, redirect.source_path);
+                fullPath = Path.GetFullPath(Path.Combine(redirectsFile.DirectoryName!, redirect.source_path));
             }
             else if (redirect.source_path_from_root != null)
             {
                 // Construct the full path to the redirected file
-                fullPath = Path.Combine(rootPath, redirect.source_path_from_root.Substring(1));
+                fullPath = Path.GetFullPath(Path.Combine(rootPath, redirect.source_path_from_root.Substring(1)));
             }
 
             // Path.GetFullPath doesn't require the file or directory to exist,
             // so this works on case-sensitive file systems too.
             if (redirect.redirect_url is not null)
-                redirectsLookup.Add(Path.GetFullPath(fullPath!), redirect.redirect_url);
+            {
+                try
+                {
+                    redirectsLookup.Add(fullPath!, redirect.redirect_url);
+                }
+                catch (ArgumentException)
+                {
+                    Console.WriteLine($"WARNING: The source path '{fullPath}' appears more than once in the redirection file. " +
+                        $"Please remove duplicates to ensure that all hops are removed.\n");
+                }
+            }
         }
 
         foreach (KeyValuePair<string, string> redirectPair in redirectsLookup)