What is the "best" way to compress FileStream?

[JumpyNET]

I tried using "System.IO.Compression" to compress a stream I am saving to harddisk. It reduced the file size a bit, but when I recompressed the files again with an arhiving software the results amazed me. The uncompressed files squeezed in to much smaller files.

 

Here are the file sizes:

Uncompressed 256 KB

Uncompressed --> rar --> 64 KB

Uncompressed --> zip --> 75 KB

GZipStreamed 197 KB

GZipStreamed --> rar --> 107 KB

GZipStreamed --> zip --> 110 KB

 

What do you suggest? Should I save the files uncompressed and compress the files with some algorithm? Or do you know a better way to compress a FileStream?

 

 

Here is the code I am using:

   Public Sub SaveAllFunctionsToFile(ByVal SavePath As String, Optional ByVal Compress As Boolean = False)

       Select Case Compress
           Case True
               Dim fs As New System.IO.FileStream(SavePath, IO.FileMode.Create)
               Dim s As New System.Runtime.Serialization.Formatters.Binary.BinaryFormatter
               Dim Compresser As New System.IO.Compression.GZipStream(fs, IO.Compression.CompressionMode.Compress, False)
               s.Serialize(Compresser, AllFunctions)
               Compresser.Close()
               fs.Close()
           Case False
               Dim fs As New System.IO.FileStream(SavePath, IO.FileMode.Create)
               Dim s As New System.Runtime.Serialization.Formatters.Binary.BinaryFormatter
               s.Serialize(fs, AllFunctions)
               fs.Close()
       End Select

   End Sub

 

 

PS: I am using VB 2005 Express.

[PlausiblyDamp]

Looking at the classes under the System.IO.Compression namespace and as far as I can see neither the Defalte or the GZip streams give any means to set the compression level.

 

http://www.icsharpcode.net/OpenSource/SharpZipLib/Default.aspx provide a free compression library that does expose the compression level however and is just as easy to use...

 

e.g. the following is a simple cut and paste of the same routine

   Private Sub Button1_Click(ByVal sender As System.Object, ByVal e As System.EventArgs) Handles Button1.Click
       Dim st As String
       Dim sr As New System.IO.StreamReader("c:\setupapi.log.0.old")

       st = sr.ReadToEnd()

       Dim fs As New System.IO.FileStream("c:\test1.bin", IO.FileMode.Create)
       Dim s As New System.Runtime.Serialization.Formatters.Binary.BinaryFormatter
       Dim Compresser As New System.IO.Compression.GZipStream(fs, IO.Compression.CompressionMode.Compress, False)
       s.Serialize(Compresser, st)
       Compresser.Close()
       fs.Close()

   End Sub

   Private Sub Button2_Click(ByVal sender As System.Object, ByVal e As System.EventArgs) Handles Button2.Click
       Dim st As String
       Dim sr As New System.IO.StreamReader("c:\setupapi.log.0.old")

       st = sr.ReadToEnd()

       Dim fs As New System.IO.FileStream("c:\test2.bin", IO.FileMode.Create)
       Dim s As New System.Runtime.Serialization.Formatters.Binary.BinaryFormatter

       Dim Compresser As New ICSharpCode.SharpZipLib.GZip.GZipOutputStream(fs)
       Compresser.SetLevel(9)
       s.Serialize(Compresser, st)
       Compresser.Close()
       fs.Close()
   End Sub

 

the c:\setupapi.log.0.old file is a 4.5M text file found in my windows folder which using the built in GZip compressed to 590K, using the ICSharp library set to level 9 compression came out as 155K - a fairly impressive saving.

 

Given the ease of use of the ICSharp library I would be tempted to go that route rather than saving an uncompressed stream and then using some other tool to compress it later.

 

As an aside the file sizes you were getting are reflective of the fact that an uncompressed file has more redundancy than a 'partially compressed' file and can be compressed better.

 

EDIT: just tried it with the ICSharp's BZip2 routine

Dim Compresser As New ICSharpCode.SharpZipLib.BZip2.BZip2OutputStream(fs)

 

and the resultant file is now only 124K

[JumpyNET]

Thanks. That sure seems easy. The CSharp in the librarys name scared me of before. It is good to know that SharpZipLib works also in VB.