Unzip from a SQL Server text column to an image column

I have images of various formats (.png, .jpg, .bmp, etc.) stored as compressed text in a text column in a SQL Server 2005 table. I need to read the row, unzip the image and store it in an image column in another table.

I am using the SharpZip library, and all of the examples deal with file sources and destinations. I can't find anything that covers unzipping from a variable to another variable. A code snippet illustrating this or a link to a relevant resource would be much appreciated.

EDIT: A bit more information - the data is stored in a TEXT column. It appears as follows (text column abbreviated for display):

ImageID  ImageData
1       FORMAT-ZIPV3 UEsDBBQAAAAIAOV6wzxdTnDvshs...
2       FORMAT-ZIPV3 UEsDBBQAAAAIAAF2yjxGncjOLgA...
3       FORMAT-ZIPV3 UEsDBBQAAAAIAKd6yjyjnQNr6gg...
4       FORMAT-ZIPV3 UEsDBBQAAAAIALdNyzyrPC8EMJw...
5       FORMAT-ZIPV3 UEsDBBQAAAAIAA1rOD1nZY1t0f0...
6       FORMAT-ZIPV3 UEsDBBQAAAAIANZplj2seyJ+VmM...
7       FORMAT-ZIPV3 UEsDBBQAAAAIAC5vhD27LPbPcv8...
8       FORMAT-ZIPV3 UEsDBBQAAAAIAK1qKz5DJNH3xMg...
9       FORMAT-ZIPV3 UEsDBBQAAAAIAHVkEztC3th/9hs...
10     FORMAT-ZIPV3 UEsDBBQAAAAIAEtXKz7DXHUdvow...

What I know for certain is that the images were compressed at some point in the process using SharpZip before being inserted into the table. It appears that the format information was added to the beginning of the data prior to inserting.

Looking at this data, would anyone have any insight on how this image data has been manipulated? Again, I need to get the uncompressed image data into a column of a data type conducive to reading for display on a web page.

EDIT: Ok, I'm stumped. Executing the following code produces the error, "Failed to convert parameter value from a Int32 to a Byte[]". It appears to be placing the length of the byte array into the byte array's value...

        commandUncompressed.Connection = connectionUncompressed;
        commandUncompressed.Parameters.Add("@Image_k", SqlDbType.VarChar, 10);
        commandUncompressed.Parameters.Add("@ImageContents", SqlDbType.Image);
        commandUncompressed.CommandText = sqlSaveImage;

        connectionUncompressed.Open();
        reader = command.ExecuteReader();

        if (reader.HasRows)
        {
            while (reader.Read())
            {
                Console.WriteLine(reader["Image_k"].ToString());  // Merely for testing
                String format = reader["ImageContents_Compressed"].ToString().Substring(0, 12);
                var offset = 13; //"FORMAT-ZIPV3 ".Length;
                var s = reader["ImageContents_Compressed"].ToString().Substring(offset);
                var bytes = Convert.FromBase64String(s);
                if (format == "FORMAT-ZIPV2 ")
                {
                    bytes = ConvertStringToBytes(s);    // Not a Base-64 encoded string? External conversion function utilized.
                }

                using (var zis = new ZipInputStream(new MemoryStream(bytes)))
                {
                    ZipEntry zipEntry = zis.GetNextEntry();   // Doesn't seem to work unless an entry has been referenced
                    byte[] buffer = new byte[zis.Length];
                    commandUncompressed.Parameters["@Image_k"].Value = reader["Image_k"].ToString();
                    commandUncompressed.Parameters["@ImageContents"].Value = zis.Read(buffer, 0, buffer.Length);
                    commandUncompressed.ExecuteNonQuery();

                }
            }
        }

It appears to be reading the data from the source text column just fine. I just cannot figure out how to get that into the image type parameter. The value for buffer variable shows the length of the byte array, rather than the actual bytes. Maybe that's what the value property typically shows for byte arrays? I'm so close and yet so far away. :/

EDIT: Ok, I'm a knucklehead. I made the following correction, and it works!

zis.Read(buffer, 0, buffer.Length)                            
commandUncompressed.Parameters["@ImageContents"].Value = buffer;

At this point I am only able to process FORMAT-ZIPV3 data, as I haven't figured out how to decode the FORMAT-ZIP2 strings yet. Following is a sampling of the V2 data. If anyone is able to determine the encoding, let me know. Would it be different if zipped using BZIP instead of ZIP format?

ImageID ImageData
1          FORMAT-ZIPV2 504B03041400020008005157422A2E25FDBAF26701008D6901000E...
2          FORMAT-ZIPV2 504开发者_如何学JAVAB03041400020008009159422A7FC94BA2B2540500D35705000E...
3          FORMAT-ZIPV2 504B0304140002000800685A422A0CAA51F4473A0600B97206000E...
4          FORMAT-ZIPV2 504B03041400020008001D5D422A770BD3ED201902002C4A02000E...
5          FORMAT-ZIPV2 504B0304140002000800325E422A4B6C2FB4045001001C6E01000E...
6          FORMAT-ZIPV2 504B03041400020008006F72422A5F793AC1A1F00200ECF302000E...
7          FORMAT-ZIPV2 504B0304140002000800D572422A1B348A731DE5000085EB00000E...
8          FORMAT-ZIPV2 504B03041400020008003D73422A8AEBB7F855640300DD1B04000E...
9          FORMAT-ZIPV2 504B03041400020008006368D528C5D0A6BA794900004A2502000E...
10         FORMAT-ZIPV2 504B03041400020008008E5B6C2A2D9E9C33D7AF05005CEC05000E...

In response to a similar question, someone on sqlmonster.com provided a nifty VarBinaryStream class. It works with a column type of varbinary(max).

If your data is stored in a varbinary(max), and is in zip format, you could use that class to instantiate a VarBinaryStream, then instantiate a ZipInputStream around that, and ba-da-boom, you're there. Just read from the ZipInputStream.

In C# it might look like this

using (var imageSrc = new VarBinarySource(connection, 
                                          "Table.Name", 
                                          "Column",
                                          "KeyColName",
                                          1))
{
    using (var s = new VarBinaryStream(imageSrc))
    {
        using(var zis = new ZipInputStream(s))
        {
           ....
        }
    }
}

If the images are small, then you probably wouldn't want all this streaming stuff. If the column is a binary(n) or a varbinary(n) where n is less than 8000, just use the SqlBinary type and read in all the data into memory, then instantiate a MemoryStream around that. Simpler. In VB.NET it looks something like this:

Dim bytes as Bytes()
bytes = dr.GetSqlBinary(columnNumber)
Using ms As New MemoryStream(bytes)
    Using zis As New ZipInputStream(ms)
        ...
    End Using
End Using 

Finally, I'm going to question the wisdom of applying zip compression to .jpg images, and similar. The jpg format is already compressed; compressing it again before putting the data into SQL Server won't cause the data to become appreciably smaller. It only increases processing time. If possible, I'd suggest you reconsider your design for storage of compressed images.

ok, with the update you provided, containing the data format, you can draw some conclusions.

The data is an actual string. Suspecting that it is a Base64-encoded string, I did a small test and used Convert.ToBase64String() on a byte stream that contains a zip file. It looks like this: UEsDBBQAAAAIAJJyYyk3M56F+QIAA...

Aha! you have a base64-encoded (string) version of the byte data for a bonafide zip file. To decode it, strip the prefix and then use FromBase64String() to get the byte array, insert into a MemoryStream, then read it with ZipInputStream.

something like this:

var offset = "FORMAT-ZIPV3 ".Length(); 
var s = sqlReader["CompressedImage"].ToString().Substring(offset);
var bytes = Convert.FromBase64String(s);    
using (var zis = new ZipInputStream(new MemoryStream(bytes))) 
{
    ...
    zis.Read(...);
    ...
}

If the data is "really long", you're going to want to stream it out of that table, rather than just read it into a big string and convert it. I don't know how large text columns can be, but supposing that it could be 500mb, you don't want a 500mb string, and you don't want to do a conversion of a 500mb string with Convert.FromBase64String(). In that case You need to use a Base64Stream, or the FromBase64Transform class in the System.Security.Cryptography namespace.

Editorial comment. It is sort of backwards to zip-compress image data. The images are probably compressed already. But to compound that backwardsness by then doing a base64 encode, thereby expanding the data... ??? That is triple backwards. That makes noooooo sense at all. I understand that's how your vendor supplied it.

Ok, with your furhter update, using this as the format:

ImageID ImageData
1          FORMAT-ZIPV2 504B03041400020008005157422A2E25FDBAF26701008D6901000E...
2          FORMAT-ZIPV2 504B03041400020008009159422A7FC94BA2B2540500D35705000E...

That data is still zipfile data, but it is encoded as simple hex digits. You need to convert that to a byte array. Here's some code to do it.

public static class ConvertEx
{
    static readonly String prefix= "FORMAT-ZIPV2 ";

    public static string ToHexString(byte[] b)
    {
        System.Text.StringBuilder sb1 = new System.Text.StringBuilder();
        int i = 0;
        for (i = 0; i < b.Length; i++)
        {
            sb1.Append(System.String.Format("{0:X2}", b[i]));
        }
        return sb1.ToString().ToLower();
    }

    public static byte[] ToByteArray(string s)
    {
        if (s.StartsWith(prefix))
        {
            System.Console.WriteLine("removing prefix");
            s = s.Substring(prefix.Length);
        }
        s= s.Trim(); // whitespace
        System.Console.WriteLine("length: {0}", s.Length);

        var r= new byte[s.Length/2];
        for (int i = 0; i < s.Length; i+=2)
        {
            r[i/2] = (byte) Convert.ToUInt32(s.Substring(i,2), 16);
        }
        return r;
    }
}

You can use that this way:

        string s = GetStringContentFromDatabase()
        var decoded = ConvertEx.ToByteArray(s);

        using (var ms = new MemoryStream(decoded))
        {
            // use DotNetZip to read the zip file
            // SharpZipLib is something similar...
            using (var zip = ZipFile.Read(ms))
            {
                // print out the list of entries in the zipfile
                foreach (var e in zip)
                {
                    System.Console.WriteLine("{0}", e.FileName);
                }
            }
        }

The examples on the SharpZip Wiki use Stream objects - while the sample does use a File, you could easily use a MemoryStream object here and the sample would work the same.