public static IFileInfo[][] FindDuplicates(DedupContext dup, Action <int, int, string> reportProgress = null)
{
List <IFileInfo> files = new List <IFileInfo>();
reportProgress?.Invoke(0, 100, "building list..");
foreach (var d in dup.Dirs)
{
Stuff.GetAllFiles(d, files);
}
files.AddRange(dup.Files);
reportProgress?.Invoke(25, 100, "filtering");
files = files.Where(z => z.Exist && z.Length > 0).ToList();
reportProgress?.Invoke(50, 100, "grouping 1");
var grp1 = files.GroupBy(z => z.Length).Where(z => z.Count() > 1).ToArray();
List <IFileInfo[]> groups = new List <IFileInfo[]>();
foreach (var item in grp1)
{
reportProgress?.Invoke(75, 100, "grouping 2");
var arr0 = item.GroupBy(z => Stuff.CalcPartMD5(z, 1024 * 1024)).ToArray();
var cnt0 = arr0.Count(z => z.Count() > 1);
if (cnt0 == 0)
{
continue;
}
groups.AddRange(arr0.Select(z => z.ToArray()).ToArray());
}
//todo: binary compare candidates
return(groups.ToArray());
}
public void SetGroups(DedupContext ctx, IFileInfo[][] groups)
{
Context = ctx;
listView1.Items.Clear();
foreach (var fileInfo in groups.OrderByDescending(z => z.First().Length *z.Length))
{
listView1.Items.Add(new ListViewItem(new string[] { fileInfo.First().Name,
Stuff.CalcPartMD5(fileInfo.First(), 1024 * 1024),
fileInfo.Length + "",
Stuff.GetUserFriendlyFileSize((fileInfo.First().Length *(fileInfo.Length - 1))) })
{
Tag = fileInfo
});
}
label1.Text = "Total repeats groups: " + groups.Length;
label2.Text = "Total memory overhead: " + Stuff.GetUserFriendlyFileSize(groups.Sum(z => z.First().Length *(z.Length - 1)));
listView1.AutoResizeColumns(ColumnHeaderAutoResizeStyle.ColumnContent);
listView1.AutoResizeColumns(ColumnHeaderAutoResizeStyle.HeaderSize);
}