public ScanResult FindDuplicateFiles(String dirPath)
{
worker.ReportProgress(0, "Preparing files ...");
FileInfo[] files = new DirectoryInfo(dirPath).GetFiles("*", SearchOption.AllDirectories);
Array.Sort(files, Comparer<FileInfo>.Create((a,b) => b.FullName.CompareTo(a.FullName)));
int total = files.Length;
double progress = 0;
IDictionary<long, IDictionary<FileInfo, IList<FileInfo>>> byteTable = new Dictionary<long, IDictionary<FileInfo, IList<FileInfo>>>();
foreach (FileInfo file in files) {
worker.ReportProgress((int)(++progress/total*100), String.Format("Scanning files... ({0}/{1})", progress, total));
// Compare size
long fileSize = file.Length;
if (!byteTable.ContainsKey(fileSize))
byteTable.Add(fileSize, new Dictionary<FileInfo, IList<FileInfo>>());
// Compare contents of the files with the same size
IDictionary<FileInfo, IList<FileInfo>> fileTable = byteTable[fileSize]; // All files in fileMap have the same size
bool foundDuplicate = false;
// Compare the current file to each file in fileTable
foreach (KeyValuePair<FileInfo, IList<FileInfo>> pair in fileTable) {
// If find a duplicate, add the file to the duplicate-files-list and break the iteration
if (FilesAreEqual(pair.Key, file)) {
foundDuplicate = true;
pair.Value.Add(file);
break;
}
}
// No duplicate found, create a new entry in fileTable
if (!foundDuplicate)
fileTable.Add(file, new List<FileInfo>());
}
// Build the result
worker.ReportProgress(100, "Build the result ...");
ScanResult result = new ScanResult();
int sum = 0;
foreach (IDictionary<FileInfo, IList<FileInfo>> fileTable in byteTable.Values) {
foreach (KeyValuePair<FileInfo, IList<FileInfo>> pair in fileTable) {
if (pair.Value.Count > 0) {
ISet<FileInfo> list = new SortedSet<FileInfo>(Comparer<FileInfo>.Create((a, b) => b.FullName.CompareTo(a.FullName)));
list.Add(pair.Key);
result.RemoveList.Add(pair.Key, false);
foreach (FileInfo file in pair.Value) {
list.Add(file);
result.RemoveList.Add(file, true);
}
result.FileList.Add(pair.Key, list);
sum += pair.Value.Count;
}
}
}
result.NumDuplicates = sum;
return result;
}
private void button_scan_Click(object sender, EventArgs e)
{
button_scan.Enabled = false;
listBox_details.DrawItem -= drawItemEventHandler;
duplicates = null;
listBox_overview.Items.Clear();
listBox_details.Items.Clear();
textBox_filePath.Clear();
toolStripProgressBar1.Value = 0;
toolStripStatusLabel1.Text = "";
backgroundWorker_scan.RunWorkerAsync();
}
private void backgroundWorker_scan_DoWork(object sender, DoWorkEventArgs e)
{
duplicates = new DuplicateFinder(sender as BackgroundWorker).FindDuplicateFiles(textBox_folder.Text);
}