PR 351: Add support for Filename field in GZip

Author: piksel

src/ICSharpCode.SharpZipLib/GZip/GZipConstants.cs

@@ -1,3 +1,6 @@
+using System;
+using System.Text;
+
 namespace ICSharpCode.SharpZipLib.GZip
 {
 	/// <summary>
@@ -7,53 +10,69 @@ namespace ICSharpCode.SharpZipLib.GZip
 	sealed public class GZipConstants
 	{
 		/// <summary>
-		/// Magic number found at start of GZIP header
+		/// First GZip identification byte
 		/// </summary>
-		public const int GZIP_MAGIC = 0x1F8B;
+		public const byte ID1 = 0x1F;
 
-		/*  The flag byte is divided into individual bits as follows:
+		/// <summary>
+		/// Second GZip identification byte
+		/// </summary>
+		public const byte ID2 = 0x8B;
 
-			bit 0   FTEXT
-			bit 1   FHCRC
-			bit 2   FEXTRA
-			bit 3   FNAME
-			bit 4   FCOMMENT
-			bit 5   reserved
-			bit 6   reserved
-			bit 7   reserved
-		 */
+		/// <summary>
+		/// Deflate compression method
+		/// </summary>
+		public const byte CompressionMethodDeflate = 0x8;
 
 		/// <summary>
-		/// Flag bit mask for text
+		/// Get the GZip specified encoding (CP-1252 if supported, otherwise ASCII)
 		/// </summary>
-		public const int FTEXT = 0x1;
+		public static Encoding Encoding
+		{
+			get
+			{
+				try
+				{
+					return Encoding.GetEncoding(1252);
+				}
+				catch
+				{
+					return Encoding.ASCII;
+				}
+			}
+		}
 
+	}
+
+	/// <summary>
+	/// GZip header flags
+	/// </summary>
+	[Flags]
+	public enum GZipFlags: byte
+	{
 		/// <summary>
-		/// Flag bitmask for Crc
+		/// Text flag hinting that the file is in ASCII
 		/// </summary>
-		public const int FHCRC = 0x2;
+		FTEXT = 0x1 << 0,
 
 		/// <summary>
-		/// Flag bit mask for extra
+		/// CRC flag indicating that a CRC16 preceeds the data
 		/// </summary>
-		public const int FEXTRA = 0x4;
+		FHCRC = 0x1 << 1,
 
 		/// <summary>
-		/// flag bitmask for name
+		/// Extra flag indicating that extra fields are present
 		/// </summary>
-		public const int FNAME = 0x8;
+		FEXTRA = 0x1 << 2,
 
 		/// <summary>
-		/// flag bit mask indicating comment is present
+		/// Filename flag indicating that the original filename is present
 		/// </summary>
-		public const int FCOMMENT = 0x10;
+		FNAME = 0x1 << 3,
 
 		/// <summary>
-		/// Initialise default instance.
+		/// Flag bit mask indicating that a comment is present
 		/// </summary>
-		/// <remarks>Constructor is private to prevent instances being created.</remarks>
-		private GZipConstants()
-		{
-		}
+		FCOMMENT = 0x1 << 4,
 	}
 }

src/ICSharpCode.SharpZipLib/GZip/GzipInputStream.cs

@@ -3,6 +3,7 @@
 using ICSharpCode.SharpZipLib.Zip.Compression.Streams;
 using System;
 using System.IO;
+using System.Text;
 
 namespace ICSharpCode.SharpZipLib.GZip
 {
@@ -54,6 +55,8 @@ public class GZipInputStream : InflaterInputStream
 		/// </summary>
 		private bool completedLastBlock;
 
+		private string fileName;
+
 		#endregion Instance Fields
 
 		#region Constructors
@@ -149,6 +152,15 @@ public override int Read(byte[] buffer, int offset, int count)
 			}
 		}
 
+		/// <summary>
+		/// Retrieves the filename header field for the block last read
+		/// </summary>
+		/// <returns></returns>
+		public string GetFilename()
+		{
+			return fileName;
+		}
+
 		#endregion Stream overrides
 
 		#region Support routines
@@ -170,149 +182,108 @@ private bool ReadHeader()
 				}
 			}
 
-			// 1. Check the two magic bytes
 			var headCRC = new Crc32();
-			int magic = inputBuffer.ReadLeByte();
 
-			if (magic < 0)
-			{
-				throw new EndOfStreamException("EOS reading GZIP header");
-			}
+			// 1. Check the two magic bytes
 
+			var magic = inputBuffer.ReadLeByte();
 			headCRC.Update(magic);
-			if (magic != (GZipConstants.GZIP_MAGIC >> 8))
+			if (magic != GZipConstants.ID1)
 			{
 				throw new GZipException("Error GZIP header, first magic byte doesn't match");
 			}
 
-			//magic = baseInputStream.ReadByte();
 			magic = inputBuffer.ReadLeByte();
-
-			if (magic < 0)
-			{
-				throw new EndOfStreamException("EOS reading GZIP header");
-			}
-
-			if (magic != (GZipConstants.GZIP_MAGIC & 0xFF))
+			if (magic != GZipConstants.ID2)
 			{
 				throw new GZipException("Error GZIP header,  second magic byte doesn't match");
 			}
-
 			headCRC.Update(magic);
 
 			// 2. Check the compression type (must be 8)
-			int compressionType = inputBuffer.ReadLeByte();
-
-			if (compressionType < 0)
-			{
-				throw new EndOfStreamException("EOS reading GZIP header");
-			}
+			var compressionType = inputBuffer.ReadLeByte();
 
-			if (compressionType != 8)
+			if (compressionType != GZipConstants.CompressionMethodDeflate)
 			{
 				throw new GZipException("Error GZIP header, data not in deflate format");
 			}
 			headCRC.Update(compressionType);
 
 			// 3. Check the flags
-			int flags = inputBuffer.ReadLeByte();
-			if (flags < 0)
-			{
-				throw new EndOfStreamException("EOS reading GZIP header");
-			}
-			headCRC.Update(flags);
-
-			/*    This flag byte is divided into individual bits as follows:
+			var flagsByte = inputBuffer.ReadLeByte();
 
-			bit 0   FTEXT
-			bit 1   FHCRC
-			bit 2   FEXTRA
-			bit 3   FNAME
-			bit 4   FCOMMENT
-			bit 5   reserved
-			bit 6   reserved
-			bit 7   reserved
-			*/
+			headCRC.Update(flagsByte);
 
 			// 3.1 Check the reserved bits are zero
 
-			if ((flags & 0xE0) != 0)
+			if ((flagsByte & 0xE0) != 0)
 			{
 				throw new GZipException("Reserved flag bits in GZIP header != 0");
 			}
 
+			var flags = (GZipFlags)flagsByte;
+
 			// 4.-6. Skip the modification time, extra flags, and OS type
 			for (int i = 0; i < 6; i++)
 			{
-				int readByte = inputBuffer.ReadLeByte();
-				if (readByte < 0)
-				{
-					throw new EndOfStreamException("EOS reading GZIP header");
-				}
-				headCRC.Update(readByte);
+				headCRC.Update(inputBuffer.ReadLeByte());
 			}
 
 			// 7. Read extra field
-			if ((flags & GZipConstants.FEXTRA) != 0)
+			if (flags.HasFlag(GZipFlags.FEXTRA))
 			{
 				// XLEN is total length of extra subfields, we will skip them all
-				int len1, len2;
-				len1 = inputBuffer.ReadLeByte();
-				len2 = inputBuffer.ReadLeByte();
-				if ((len1 < 0) || (len2 < 0))
-				{
-					throw new EndOfStreamException("EOS reading GZIP header");
-				}
+				var len1 = inputBuffer.ReadLeByte();
+				var len2 = inputBuffer.ReadLeByte();
+
 				headCRC.Update(len1);
 				headCRC.Update(len2);
 
 				int extraLen = (len2 << 8) | len1;      // gzip is LSB first
 				for (int i = 0; i < extraLen; i++)
 				{
-					int readByte = inputBuffer.ReadLeByte();
-					if (readByte < 0)
-					{
-						throw new EndOfStreamException("EOS reading GZIP header");
-					}
-					headCRC.Update(readByte);
+					headCRC.Update(inputBuffer.ReadLeByte());
 				}
 			}
 
 			// 8. Read file name
-			if ((flags & GZipConstants.FNAME) != 0)
+			if (flags.HasFlag(GZipFlags.FNAME))
 			{
+				var fname = new byte[1024];
+				var fnamePos = 0;
 				int readByte;
 				while ((readByte = inputBuffer.ReadLeByte()) > 0)
 				{
+					if (fnamePos < 1024)
+					{
+						fname[fnamePos++] = (byte)readByte;
+					}
 					headCRC.Update(readByte);
 				}
 
-				if (readByte < 0)
-				{
-					throw new EndOfStreamException("EOS reading GZIP header");
-				}
 				headCRC.Update(readByte);
+
+				fileName = GZipConstants.Encoding.GetString(fname, 0, fnamePos);
+			}
+			else
+			{
+				fileName = null;
 			}
 
 			// 9. Read comment
-			if ((flags & GZipConstants.FCOMMENT) != 0)
+			if (flags.HasFlag(GZipFlags.FCOMMENT))
 			{
 				int readByte;
 				while ((readByte = inputBuffer.ReadLeByte()) > 0)
 				{
 					headCRC.Update(readByte);
 				}
 
-				if (readByte < 0)
-				{
-					throw new EndOfStreamException("EOS reading GZIP header");
-				}
-
 				headCRC.Update(readByte);
 			}
 
 			// 10. Read header CRC
-			if ((flags & GZipConstants.FHCRC) != 0)
+			if (flags.HasFlag(GZipFlags.FHCRC))
 			{
 				int tempByte;
 				int crcval = inputBuffer.ReadLeByte();