c# best way to store huge amount of data during runtime

Question

In my application I have a lot of small files that are saved in a certain path structure. I am creating a container file where I want to store all existing files starting with a header that contains information like offset of each file or file size. I am writing to this file with a BinaryWriter. But there are a lot of duplicates that are only added to my container file once. Therefore I create a hash value for each file and compare it to existing hash values in a data table. This works just like it should, but I want to know wether this is good practice or not because there could be a huge amount of data. Are there better or more performant ways to achieve my goal?

Here is my actual code:

// I parsed through my files and created my header
// all file paths were added to my tileList
DataTable dtImageInfos = new DataTable();
dtImageInfos.Columns.Add("tilename", typeof(String));
dtImageInfos.Columns.Add("hash", typeof(String));
dtImageInfos.Columns.Add("offset", typeof(long));

foreach (String tile in tileList)
{                            
    FileInfo f = new FileInfo(tile);
    int tileSize = Convert.ToInt32(f.Length);

    if (tileSize <= MAX_CHECK_SIZE)
    {
        Image tileImg = Image.FromFile(tile);
        String tileHash = createHashForImage(tileImg);
        DataTable dtCheck = dtImageInfos.Clone();                                 

        if (dtImageInfos.Rows.Count > 0)
            dtImageInfos.AsEnumerable().Where(t => t.Field<String>("hash").Equals(tileHash))
                            .CopyToDataTable(dtCheck, LoadOption.OverwriteChanges);c#                                

        if (dtCheck.Rows.Count == 0)
        {
            writer.Write(tileOffset);

            DataRow drNew = dtImageInfos.NewRow();
            drNew["tilename"] = tile;
            drNew["hash"] = tileHash;
            drNew["offset"] = tileOffset;
            dtImageInfos.Rows.Add(drNew);

            tileOffset += tileSize;
        }
        else
        {
            DataRow drCheck = dtCheck.Rows[0];
            writer.Write((long)drCheck["offset"]);
        }              
    }
    else
    {
        writer.Write(tileOffset);

        DataRow drNew = dtImageInfos.NewRow();
        drNew["tilename"] = tile;
        drNew["hash"] = "";
        drNew["offset"] = tileOffset;
        dtImageInfos.Rows.Add(drNew);

        tileOffset += tileSize;
    }
    writer.Write(tileSize);                            
}

foreach (DataRow drTile in dtImageInfos.Rows)
{
    byte[] tileData = File.ReadAllBytes(drTile["tilename"].ToString());
    writer.Write(tileData);
}

I hope I could make my issue understandable. Thanks in advance

Answer 1

Using the latest version of c#.NET you can avoid a lot of over head just using a Zip File.

System.IO.Compression.ZipFile.CreateFromDirectory(folderPathWithTiles, containerPath);

You can access the different file (tiles) in your zip archive (container file) with the .OpenRead method.

using (ZipArchive tiles = ZipFile.OpenRead(containerPath))
{
    ZipArchiveEntry tile = tiles.GetEntry(relativeTilePath);
    Image tileImage = Image.FromStream(tile.Open());
}

ZipFile: http://msdn.microsoft.com/en-us/library/system.io.compression.zipfile.aspx

ZipArchive: http://msdn.microsoft.com/en-us/library/system.io.compression.ziparchive.aspx

ZipArchiveEntry: http://msdn.microsoft.com/en-us/library/system.io.compression.ziparchiveentry.aspx