improving the ability to fetch basic Revit file info (version, format etc)

Author: salaros

src/Extensions/RevitFileInfoExtensions.cs

@@ -1,4 +1,5 @@
-using System;
+using CodeCave.Revit.Toolkit.OLE;
+using System;
 using System.Collections.Generic;
 using System.Globalization;
 using System.Linq;
@@ -18,6 +19,8 @@ internal struct KnownRevitInfoProps
             public const string USERNAME = "Username";
             public const string CENTRAL_MODEL_PATH = "Central Model Path";
             public const string REVIT_BUILD = "Revit Build";
+            public const string BUILD = "Build";
+            public const string FORMAT = "Format";
             public const string LAST_SAVE_PATH = "Last Save Path";
             public const string OPEN_WORKSET_DEFAULT = "Open Workset Default";
             public const string PROJECT_SPARK_FILE = "Project Spark File";
@@ -36,6 +39,12 @@ internal struct KnownRevitInfoProps
                 RegexOptions.IgnoreCase |
                 RegexOptions.CultureInvariant);
 
+        private static readonly Regex versionExtractorSimple =
+            new Regex(@"(?<build>\d*)_(?<revision>\d*)(\((?<arch>\w*)\))?",
+                RegexOptions.Compiled |
+                RegexOptions.IgnoreCase |
+                RegexOptions.CultureInvariant);
+
         #endregion
 
         #region Methods
@@ -68,23 +77,28 @@ public static void ParseRevit(this RevitFileInfo revitFileInfo, Dictionary<strin
         {
             if (properties == null) throw new ArgumentNullException(nameof(properties));
 
+            if (properties.TryGetValue(KnownRevitInfoProps.FORMAT, out var formatRaw) && int.TryParse(formatRaw, out var format))
+                revitFileInfo.Format = format;
+
             // Parse Revit Build string
-            if (!properties.ContainsKey(KnownRevitInfoProps.REVIT_BUILD))
-                return;
+            var versionObj = default(Match);
+            if (properties.TryGetValue(KnownRevitInfoProps.REVIT_BUILD, out var buildInfo))
+                versionObj = versionExtractor.Match(buildInfo);
 
-            var versionObj = versionExtractor.Match(properties[KnownRevitInfoProps.REVIT_BUILD]);
-            if (!versionObj.Success)
+            if (properties.TryGetValue(KnownRevitInfoProps.BUILD, out buildInfo))
+                versionObj = versionExtractorSimple.Match(buildInfo);
+
+            if (!versionObj?.Success ?? false)
                 return;
 
-            revitFileInfo.Vendor = versionObj?.Groups["vendor"]?.Value;
-            revitFileInfo.Name = versionObj?.Groups["software"]?.Value;
             revitFileInfo.Is64Bit = versionObj.Groups["arch"]?.Value?.Contains("x64") ?? false;
+            revitFileInfo.ProductVersion = $"{versionObj.Groups["build"].Value}_{versionObj.Groups["revision"].Value}";
+
+            if (!string.IsNullOrWhiteSpace(versionObj?.Groups["vendor"]?.Value))
+                revitFileInfo.ProductVendor = versionObj?.Groups["vendor"]?.Value;
 
-            var versionString = string.Format("{0}.{1}.{2}",
-                versionObj.Groups["version"].Value,
-                versionObj.Groups["build"].Value,
-                versionObj.Groups["revision"].Value);
-            revitFileInfo.Version = new Version(versionString);
+            if (!string.IsNullOrWhiteSpace(versionObj?.Groups["software"]?.Value))
+                revitFileInfo.ProductName = versionObj?.Groups["software"]?.Value;
         }
 
         /// <summary>

src/OLE/RevitFileInfo.cs

@@ -5,11 +5,20 @@
 using System.IO;
 using System.Linq;
 using System.Text;
+using System.Text.RegularExpressions;
+using static CodeCave.Revit.Toolkit.RevitFileInfoExtensions;
 
-namespace CodeCave.Revit.Toolkit
+namespace CodeCave.Revit.Toolkit.OLE
 {
     public class RevitFileInfo
     {
+        static RevitFileInfo()
+        {
+#if !NET45
+            Encoding.RegisterProvider(CodePagesEncodingProvider.Instance);
+#endif
+        }
+
         #region Properties
 
         /// <summary>
@@ -58,22 +67,34 @@ public class RevitFileInfo
         /// </value>
         public FamilyType[] Types => PartAtom?.Family?.Parts;
 
-        public Version Version { get; internal set; }
+        public string ProductVendor { get; internal set; } = "Autodesk";
 
-        public string Name { get; internal set; }
+        public string ProductVersion { get; internal set; }
 
-        public string Vendor { get; internal set; }
+        public string ProductName { get; internal set; } = "Revit";
 
-        public object Architecture { get; internal set; }
+        public int Format { get; internal set; }
 
-        public bool Is64Bit { get; internal set; }
+        public bool Is64Bit { get; internal set; } = true;
 
         public string FilePath { get; private set; }
 
         #endregion
 
         #region Methods
 
+        public static int GetFormatFromFile(string filePath)
+        {
+            if (string.IsNullOrWhiteSpace(filePath) || !File.Exists(filePath))
+                throw new ArgumentException("Please supply a valid path to a Revit document", nameof(filePath));
+
+            var properties = GetProperties(filePath);
+            if (properties.TryGetValue(KnownRevitInfoProps.FORMAT, out var formatRaw) && int.TryParse(formatRaw, out var format))
+                return format;
+
+            throw new InvalidDataException($"Couldn't read format information from the followin file: '{filePath}'");
+        }
+
         /// <summary>
         /// Gets a <see cref="RevitFileInfo" /> instance from the given file path.
         /// </summary>
@@ -84,8 +105,8 @@ public class RevitFileInfo
         /// <exception cref="T:System.ArgumentException">filePath is invalid</exception>
         public static RevitFileInfo GetFromFile(string filePath)
         {
-            if (string.IsNullOrWhiteSpace(filePath))
-                throw new ArgumentException("filePath is invalid");
+            if (string.IsNullOrWhiteSpace(filePath) || !File.Exists(filePath))
+                throw new ArgumentException("Please supply a valid path to a Revit document", nameof(filePath));
 
             var rfi = new RevitFileInfo
             {
@@ -130,28 +151,25 @@ public static RevitFileInfo GetFromFile(string filePath)
         /// <returns></returns>
         internal static Dictionary<string, string> GetProperties(string filePath)
         {
-            var basicInfo = OleDataReader.GetRawBytes(filePath, "BasicFileInfo");
+            var basicInfo = OleDataReader.GetRawBytes(filePath, RevitFileMap.OleStreams.BASIC_FILE_INFO);
             var fileProps = new Dictionary<string, string>();
-            var supportedEncodeings = new[]
-            {
-                Encoding.Unicode,
-                Encoding.BigEndianUnicode,
-                Encoding.UTF8,
-                Encoding.ASCII,
-                Encoding.Default,
-            };
 
-            foreach (var encoding in supportedEncodeings)
+            var asd = new Regex(@"\p{IsCJKUnifiedIdeographs}");
+
+            foreach (var encoding in RevitFileMap.SupportedEncodings)
             {
                 var basicInfoString = encoding.GetString(basicInfo)?.TrimEnd('\u0a0d');
-                using var basicInfoReader = new StringReader(basicInfoString);
+                if (asd.IsMatch(basicInfoString))
+                    continue;
+
+                using var basicInfoReader = new StringReader(basicInfoString?.Replace("\0", string.Empty));
 
                 // ReSharper disable once RedundantAssignment
                 var stringLine = basicInfoReader.ReadLine(); // skip the first line
                 while (!string.IsNullOrWhiteSpace(stringLine = basicInfoReader.ReadLine()))
                 {
-                    var parts = stringLine.Split(new[] { ":" }, 2, StringSplitOptions.None);
-                    if (parts.Length != 2)
+                    var parts = stringLine.Split(new[] { ":" }, StringSplitOptions.None);
+                    if (parts.Length != 2 || IsInvalidProperty(parts.FirstOrDefault()))
                         continue;
                     fileProps.Add(parts[0].Trim(), parts[1].Trim());
                 }
@@ -163,6 +181,11 @@ internal static Dictionary<string, string> GetProperties(string filePath)
             return fileProps;
         }
 
+        private static bool IsInvalidProperty(string input)
+        {
+            return input?.Any(c => char.IsControl(c)) ?? true;
+        }
+
         #endregion
     }
 }

src/OLE/RevitFileMap.cs

@@ -0,0 +1,37 @@
+using System.Text;
+
+namespace CodeCave.Revit.Toolkit.OLE
+{
+    internal struct RevitFileMap
+    {
+        internal struct OleStreams
+        {
+            public const string IMAGE_STREAM = "RevitPreview4.0";
+            public const string BASIC_FILE_INFO = "BasicFileInfo";
+        }
+
+        internal struct PngImageMarker
+        {
+            public const int MARKER_10 = 10;  // 0x0A
+            public const int MARKER_13 = 13;  // 0x0D
+            public const int MARKER_26 = 26;  // 0x1A
+            public const int MARKER_71 = 71;  // 0x47
+            public const int MARKER_78 = 78;  // 0x4E
+            public const int MARKER_80 = 80;  // 0x50
+            public const int MARKER_137 = 137; // 0x89
+        }
+
+        public static readonly Encoding[] SupportedEncodings = new[]
+        {
+            Encoding.BigEndianUnicode,
+            Encoding.UTF8,
+            Encoding.Unicode,
+            Encoding.ASCII,
+            Encoding.GetEncoding(1251),
+            Encoding.UTF32,
+            Encoding.GetEncoding(1252),
+            Encoding.UTF7,
+            Encoding.Default
+        };
+    }
+}

src/OLE/RevitFileVersionInfo.cs

@@ -10,7 +10,7 @@ namespace CodeCave.Revit.Toolkit.Thumbnails
     /// http://thebuildingcoder.typepad.com/blog/2008/10/rvt-file-version.html
     /// </summary>
     /// <seealso cref="ThumbnailExtractor" />
-    public class RfaTumbnailExtractor : ThumbnailExtractor
+    public partial class RevitTumbnailExtractor : ThumbnailExtractor
     {
         #region Methods
 
@@ -60,7 +60,7 @@ public override MemoryStream ExtractStream(string pathToFile)
         {
             try
             {
-                var thumbnailBytes = OleDataReader.GetRawBytes(pathToFile, RevitFileMap.OleStorage.IMAGE_STREAM);
+                var thumbnailBytes = OleDataReader.GetRawBytes(pathToFile, RevitFileMap.OleStreams.IMAGE_STREAM);
                 return ExtractStream(thumbnailBytes);
             }
             catch (Exception ex)
@@ -73,7 +73,7 @@ public override MemoryStream ExtractStream(MemoryStream memoryStream)
         {
             try
             {
-                var thumbnailBytes = OleDataReader.GetRawBytes(memoryStream, RevitFileMap.OleStorage.IMAGE_STREAM);
+                var thumbnailBytes = OleDataReader.GetRawBytes(memoryStream, RevitFileMap.OleStreams.IMAGE_STREAM);
                 return ExtractStream(thumbnailBytes);
             }
             catch (Exception ex)
@@ -177,28 +177,5 @@ private static int GetPngOffset(byte[] thumbnailBytes)
         }
 
         #endregion Helpers
-
-        #region Revit file map
-
-        internal struct RevitFileMap
-        {
-            internal struct OleStorage
-            {
-                public const string IMAGE_STREAM = "RevitPreview4.0";
-            }
-
-            internal struct PngImageMarker
-            {
-                public const int MARKER_10 = 10;  // 0x0A
-                public const int MARKER_13 = 13;  // 0x0D
-                public const int MARKER_26 = 26;  // 0x1A
-                public const int MARKER_71 = 71;  // 0x47
-                public const int MARKER_78 = 78;  // 0x4E
-                public const int MARKER_80 = 80;  // 0x50
-                public const int MARKER_137 = 137; // 0x89
-            }
-        }
-
-        #endregion
     }
 }

src/Thumbnails/RfaTumbnailExtractor.cs src/OLE/RevitTumbnailExtractor.cssrc/Thumbnails/RfaTumbnailExtractor.cs renamed to src/OLE/RevitTumbnailExtractor.cs

@@ -61,7 +61,7 @@
     <PackageReference Include="OpenMcdf-2" Version="2.1.2" />
     <PackageReference Include="System.ComponentModel.Annotations" Version="4.5.0" />
     <PackageReference Include="System.Drawing.Common" Version="4.5.1" Condition="'$(TargetFramework)' == 'netstandard2.0' Or '$(TargetFramework)' == 'netcoreapp2.1'" />
-    <PackageReference Include="System.Text.Encoding.CodePages" Version="4.5.1" Condition="'$(TargetFramework)' != 'net45'" />
+    <PackageReference Include="System.Text.Encoding.CodePages" Version="4.7.0" Condition="'$(TargetFramework)' != 'net45'" />
   </ItemGroup>
 
   <ItemGroup>

src/Revit.Toolkit.csproj

@@ -30,7 +30,7 @@ public void RfaThumbnailIsGenerated(string rfaRelativePath)
             var rfaThumbnailPath = Path.ChangeExtension(Path.Combine(Environment.CurrentDirectory, Path.GetFileName(rfaFilePath)), "png");
 
             // act
-            var rfaThumbnailer = new RfaTumbnailExtractor();
+            var rfaThumbnailer = new RevitTumbnailExtractor();
             var fileExtracted = rfaThumbnailer.TryExtractFile(rfaFilePath, rfaThumbnailPath);
 
             // assert