Use a custom cross-platform GetInvalidFileNameChars to ensure consistent behavior on Ubuntu and Windows

Author: Jericho

Source/HttpMultipartParser/FilePart.cs

@@ -93,7 +93,7 @@ public FilePart(string name, string fileName, Stream data, string contentType =
         public FilePart(string name, string fileName, Stream data, IDictionary<string, string> additionalProperties, string contentType = DefaultContentType, string contentDisposition = DontentDisposition)
         {
             Name = name;
-            FileName = fileName?.Split(Path.GetInvalidFileNameChars()).Last();
+            FileName = fileName?.Split(GetInvalidFileNameChars()).Last();
             Data = data;
             ContentType = contentType;
             ContentDisposition = contentDisposition;
@@ -136,5 +136,20 @@ public FilePart(string name, string fileName, Stream data, IDictionary<string, s
         public IReadOnlyDictionary<string, string> AdditionalProperties { get; private set; }
 
         #endregion
+
+        #region Private methods
+
+        // This is a cross-platform version of Path.GetInvalidFileNameChars() that returns the same array of characters regadless of the operating system.
+        // We use this method rather than the built-in Path.GetInvalidFileNameChars() because some unit tests were behaving differently on Ubuntu compared to Windows.
+        private static char[] GetInvalidFileNameChars() => new char[]
+        {
+            '\"', '<', '>', '|', '\0',
+            (char)1, (char)2, (char)3, (char)4, (char)5, (char)6, (char)7, (char)8, (char)9, (char)10,
+            (char)11, (char)12, (char)13, (char)14, (char)15, (char)16, (char)17, (char)18, (char)19, (char)20,
+            (char)21, (char)22, (char)23, (char)24, (char)25, (char)26, (char)27, (char)28, (char)29, (char)30,
+            (char)31, ':', '*', '?', '\\', '/'
+        };
+
+        #endregion
     }
 }