Add UTF-8 file name/path properties

#47
+semver: minor

Author: Krusen

BencodeNET.Tests/Torrents/TorrentParserTests.cs

@@ -24,13 +24,14 @@ public class TorrentParserTests
         public TorrentParserTests()
         {
             BencodeParser = Substitute.For<IBencodeParser>();
-            BencodeParser.Parse<BDictionary>((BencodeReader) null).ReturnsForAnyArgs(x => ParsedData);
+            BencodeParser.Parse<BDictionary>(null).ReturnsForAnyArgs(x => ParsedData);
 
             ValidSingleFileTorrentData = new BDictionary
             {
                 [TorrentFields.Info] = new BDictionary
                 {
                     [TorrentInfoFields.Name] = (BString) "",
+                    [TorrentInfoFields.NameUtf8] = (BString) "",
                     [TorrentInfoFields.Pieces] = (BString) "",
                     [TorrentInfoFields.PieceLength] = (BNumber) 0,
                     [TorrentInfoFields.Length] = (BNumber) 0
@@ -42,6 +43,7 @@ public TorrentParserTests()
                 [TorrentFields.Info] = new BDictionary
                 {
                     [TorrentInfoFields.Name] = (BString) "",
+                    [TorrentInfoFields.NameUtf8] = (BString) "",
                     [TorrentInfoFields.Pieces] = (BString) "",
                     [TorrentInfoFields.PieceLength] = (BNumber) 0,
                     [TorrentInfoFields.Files] = new BList()
@@ -394,13 +396,14 @@ public void AnnounceAndAnnounceList_DoesNotContainDuplicatesInPrimaryList(string
 
         [Theory]
         [AutoMockedData]
-        public void SingleFileInfo_IsParsed(long length, string fileName, string md5Sum)
+        public void SingleFileInfo_IsParsed(long length, string fileName,  string fileNameUtf8, string md5Sum)
         {
             // Arrange
             ParsedData = ValidSingleFileTorrentData;
             var info = ParsedData.Get<BDictionary>(TorrentFields.Info);
             info[TorrentInfoFields.Length] = (BNumber) length;
             info[TorrentInfoFields.Name] = (BString) fileName;
+            info[TorrentInfoFields.NameUtf8] = (BString) fileNameUtf8;
             info[TorrentInfoFields.Md5Sum] = (BString) md5Sum;
 
             // Act
@@ -414,12 +417,34 @@ public void SingleFileInfo_IsParsed(long length, string fileName, string md5Sum)
             torrent.File.Should().NotBeNull();
             torrent.File.FileSize.Should().Be(length);
             torrent.File.FileName.Should().Be(fileName);
+            torrent.File.FileNameUtf8.Should().Be(fileNameUtf8);
             torrent.File.Md5Sum.Should().Be(md5Sum);
         }
 
+        [Fact]
+        public void SingleFileInfo_NameUtf8_IsParsedAsUtf8EncodingIndependentlyOfActualEncoding()
+        {
+            var encoding = "ISO-8859-1";
+            var fileName = "øæå"; // Use characters with different byte values for UTF8 and ISO-8859-1
+
+            ParsedData = ValidSingleFileTorrentData;
+            ParsedData[TorrentFields.Encoding] = (BString) encoding;
+            var info = ParsedData.Get<BDictionary>(TorrentFields.Info);
+            info[TorrentInfoFields.Name] = new BString(fileName, Encoding.GetEncoding(encoding));
+            info[TorrentInfoFields.NameUtf8] = new BString(fileName, Encoding.UTF8);
+
+            // Act
+            var parser = new TorrentParser(BencodeParser);
+            var torrent = parser.Parse((BencodeReader)null);
+
+            // Assert
+            torrent.File.FileName.Should().Be(fileName);
+            torrent.File.FileNameUtf8.Should().Be(fileName);
+        }
+
         [Theory]
         [AutoMockedData]
-        public void MultiFileInfo_IsParsed(string directoryName, long length1, IList<string> paths1, string md5Sum1, long length2, IList<string> paths2, string md5Sum2)
+        public void MultiFileInfo_IsParsed(string directoryName, long length1, IList<string> paths1, IList<string> paths1Utf8, string md5Sum1, long length2, IList<string> paths2, IList<string> paths2Utf8, string md5Sum2)
         {
             // Arrange
             ParsedData = ValidMultiFileTorrentData;
@@ -431,12 +456,14 @@ public void MultiFileInfo_IsParsed(string directoryName, long length1, IList<str
                 {
                     [TorrentFilesFields.Length] = (BNumber) length1,
                     [TorrentFilesFields.Path] = new BList(paths1),
+                    [TorrentFilesFields.PathUtf8] = new BList(paths1Utf8),
                     [TorrentFilesFields.Md5Sum] = (BString) md5Sum1
                 },
                 new BDictionary
                 {
                     [TorrentFilesFields.Length] = (BNumber) length2,
                     [TorrentFilesFields.Path] = new BList(paths2),
+                    [TorrentFilesFields.PathUtf8] = new BList(paths2Utf8),
                     [TorrentFilesFields.Md5Sum] = (BString) md5Sum2
                 }
             };
@@ -453,9 +480,11 @@ public void MultiFileInfo_IsParsed(string directoryName, long length1, IList<str
             torrent.Files.Should().HaveCount(2);
             torrent.Files[0].FileSize.Should().Be(length1);
             torrent.Files[0].Path.Should().BeEquivalentTo(paths1);
+            torrent.Files[0].PathUtf8.Should().BeEquivalentTo(paths1Utf8);
             torrent.Files[0].Md5Sum.Should().Be(md5Sum1);
             torrent.Files[1].FileSize.Should().Be(length2);
             torrent.Files[1].Path.Should().BeEquivalentTo(paths2);
+            torrent.Files[1].PathUtf8.Should().BeEquivalentTo(paths2Utf8);
             torrent.Files[1].Md5Sum.Should().Be(md5Sum2);
         }
 

BencodeNET/Torrents/MultiFileInfo.cs

@@ -16,7 +16,12 @@ public class MultiFileInfo
         /// <summary>
         /// The file name. It just returns the last part of <see cref="Path"/>.
         /// </summary>
-        public string FileName => Path.LastOrDefault();
+        public string FileName => Path?.LastOrDefault();
+
+        /// <summary>
+        /// The UTF-8 encoded file name. It just returns the last part of <see cref="PathUtf8"/>.
+        /// </summary>
+        public string FileNameUtf8 => PathUtf8?.LastOrDefault();
 
         /// <summary>
         /// The file size in bytes.
@@ -36,6 +41,11 @@ public class MultiFileInfo
         /// </summary>
         public IList<string> Path { get; set; } = new List<string>();
 
+        /// <summary>
+        /// A list of UTF-8 encoded file path elements.
+        /// </summary>
+        public IList<string> PathUtf8 { get; set; } = new List<string>();
+
         /// <summary>
         /// The full path of the file including file name.
         /// </summary>
@@ -44,5 +54,14 @@ public string FullPath
             get => string.Join(System.IO.Path.DirectorySeparatorChar.ToString(), Path);
             set => Path = value.Split(new[] { System.IO.Path.DirectorySeparatorChar }, StringSplitOptions.RemoveEmptyEntries);
         }
+
+        /// <summary>
+        /// The full UTF-8 encoded path of the file including file name.
+        /// </summary>
+        public string FullPathUtf8
+        {
+            get => string.Join(System.IO.Path.DirectorySeparatorChar.ToString(), PathUtf8);
+            set => PathUtf8 = value.Split(new[] { System.IO.Path.DirectorySeparatorChar }, StringSplitOptions.RemoveEmptyEntries);
+        }
     }
 }

BencodeNET/Torrents/SingleFileInfo.cs

@@ -8,7 +8,6 @@
     /// </remarks>
     public class SingleFileInfo
     {
-        // Name
         /// <summary>
         /// The file name. This is purely advisory.
         /// </summary>
@@ -17,6 +16,14 @@ public class SingleFileInfo
         /// </remarks>
         public string FileName { get; set; }
 
+        /// <summary>
+        /// The UTF-8 encoded file name. This is purely advisory.
+        /// </summary>
+        /// <remarks>
+        /// Corresponds to the 'name.utf-8' field.
+        /// </remarks>
+        public string FileNameUtf8 { get; set; }
+
         /// <summary>
         /// The file size in bytes.
         /// </summary>

BencodeNET/Torrents/TorrentFields.cs

@@ -34,6 +34,7 @@ public static class TorrentFields
     public static class TorrentInfoFields
     {
         public const string Name = "name";
+        public const string NameUtf8 = "name.utf-8";
         public const string Private = "private";
         public const string PieceLength = "piece length";
         public const string Pieces = "pieces";
@@ -44,6 +45,7 @@ public static class TorrentInfoFields
         public static readonly BString[] Keys =
         {
             Name,
+            NameUtf8,
             Private,
             PieceLength,
             Pieces,
@@ -60,12 +62,14 @@ public static class TorrentFilesFields
     {
         public const string Length = "length";
         public const string Path = "path";
+        public const string PathUtf8 = "path.utf-8";
         public const string Md5Sum = "md5sum";
 
         public static readonly BString[] Keys =
         {
             Length,
             Path,
+            PathUtf8,
             Md5Sum
         };
     }

BencodeNET/Torrents/TorrentParser.cs

@@ -180,7 +180,7 @@ protected void EnsureValidTorrentData(BDictionary data)
             }
         }
 
-        private static void EnsureFields(IEnumerable<string> requiredFields, BDictionary data, string message = null)
+        private static void EnsureFields(IList<string> requiredFields, BDictionary data, string message = null)
         {
             message = message ?? "Torrent is missing required field.";
 
@@ -190,7 +190,7 @@ private static void EnsureFields(IEnumerable<string> requiredFields, BDictionary
             }
         }
 
-        private static void EnsureFields(IEnumerable<string> requiredFields, IEnumerable<BDictionary> list, string message = null)
+        private static void EnsureFields(IList<string> requiredFields, IEnumerable<BDictionary> list, string message = null)
         {
             message = message ?? "Torrent is missing required field.";
 
@@ -217,6 +217,7 @@ protected virtual SingleFileInfo ParseSingleFileInfo(BDictionary info, Encoding
             return new SingleFileInfo
             {
                 FileName = info.Get<BString>(TorrentInfoFields.Name)?.ToString(encoding),
+                FileNameUtf8 = info.Get<BString>(TorrentInfoFields.NameUtf8)?.ToString(Encoding.UTF8),
                 FileSize = info.Get<BNumber>(TorrentInfoFields.Length),
                 Md5Sum = info.Get<BString>(TorrentInfoFields.Md5Sum)?.ToString(encoding)
             };
@@ -243,6 +244,7 @@ protected virtual MultiFileInfoList ParseMultiFileInfo(BDictionary info, Encodin
                 {
                     FileSize = x.Get<BNumber>(TorrentFilesFields.Length),
                     Path = x.Get<BList>(TorrentFilesFields.Path)?.AsStrings(encoding).ToList(),
+                    PathUtf8 = x.Get<BList>(TorrentFilesFields.PathUtf8)?.AsStrings(Encoding.UTF8).ToList(),
                     Md5Sum = x.Get<BString>(TorrentFilesFields.Md5Sum)?.ToString(encoding)
                 });
 

CHANGELOG.md

@@ -5,6 +5,9 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
 ## [Unreleased]
+### Added
+- Added `FileNameUtf8` and `PathUtf8` and `FullPathUtf8` properties to `SingleFileInfo`/`MultiFileInfo` 
+  - These properties reads from `name.utf-8` and `path.utf-8` fields.
 
 ## [3.0.1] - 2019-10-17
 ### Fixed