public SessionFileFindHelper(BackupInfo currentBkp, List <Tuple <BackupFileInfo, BackupInfo> > prevBackupFiles)
{
_currentBkp = currentBkp;
_prevBackupFilesLookupByLengthByHash = prevBackupFiles
.GroupBy(x => x.Item1.Length)
.ToDictionary(x => x.Key, x => x.GroupBy(y => y.Item1.Hash).ToDictionary(y => y.Key, y => y.First()));
_prevBackupFilesLookupBypathByLength = prevBackupFiles
.GroupBy(x => x.Item1.Path)
.ToDictionary(x => x.Key, x => x.GroupBy(y => y.Item1.Length).ToDictionary(y => y.Key, y => y.First()));
}
public static IEnumerable <BackupInfo> DiscoverBackups(string path)
{
var dirs = Directory.EnumerateDirectories(path);
foreach (var dir in dirs)
{
var info = new BackupInfo(dir);
try
{
var bkpInfoDir = Directory.EnumerateDirectories(dir).FirstOrDefault(d => new DirectoryInfo(d).Name == BackupInfoDir);
if (string.IsNullOrEmpty(bkpInfoDir))
{
continue;
}
var bkpInfoFile = Path.Combine(bkpInfoDir, BackupInfoFile);
if (!File.Exists(bkpInfoFile))
{
continue;
}
using (var file = File.OpenText(bkpInfoFile))
{
var headerRaw = file.ReadLine();
if (string.IsNullOrEmpty(headerRaw))
{
continue;
}
var headerParts = headerRaw.Contains("|") ? headerRaw.Split('|') : new[] { headerRaw };
if (headerParts.Length == 0)
{
continue;
}
if (DateTime.TryParseExact(headerParts[0], DateSerializationFormat, CultureInfo.InvariantCulture, DateTimeStyles.None, out var dt))
{
info.DateTime = dt;
}
else
{
continue;
}
if (headerParts.Length > 2)
{
continue;
}
info.AttributesAvailable = headerParts.Length == 2 && headerParts[1] == "CM";
while (!file.EndOfStream)
{
var line = file.ReadLine();
var parts = line?.Split('|');
if (parts == null)
{
throw new Exception("Failed to parse backup files list. Empty file line");
}
if (parts.Length != 3 && parts.Length != 5)
{
throw new Exception($"Failed to parse backup files list. Unknown file line '{line}'");
}
var fi = new BackupFileInfo
{
Path = parts[0],
Hash = parts[1],
Length = long.Parse(parts[2]),
};
if (parts.Length == 5)
{
fi.Created = DateTime.ParseExact(parts[3], DateSerializationFormat, CultureInfo.InvariantCulture, DateTimeStyles.None);
fi.Modified = DateTime.ParseExact(parts[4], DateSerializationFormat, CultureInfo.InvariantCulture, DateTimeStyles.None);
}
info._files.Add(fi);
}
}
}
catch (Exception)
{
continue;
}
yield return(info);
}
}
private void ProcessSmallFiles(List <BackupFileModel> smallFiles, ref int category, string smallFilesTarPath, string currentBkpDir, SessionFileFindHelper findHelper, int filesCount, BackupInfo currentBkp, ref int linkedCount, ref int processed)
{
if (smallFiles.Count == 0)
{
return;
}
WriteLog("Small files hash calculation...", ++category);
var cnt = smallFiles
.AsParallel()
.Select(x =>
{
try
{
return(x.FileInfo.FastHashStr);
}
catch (Exception exception)
{
Console.WriteLine(exception);
return("invalid hash: " + exception.Message);
}
})
.Count(x => x.StartsWith("invalid hash"));
if (cnt > 0)
{
WriteLog($"Found {cnt} invalid records", ++category);
}
WriteLog($"{smallFiles.Count} files will be transferred in a batch as tar.gz", ++category);
var sw = System.Diagnostics.Stopwatch.StartNew();
bool created;
var tmpTarPath = smallFilesTarPath + ".tmp";
var archivedCount = 0;
using (var tar = new TarGzHelper(tmpTarPath))
{
foreach (var file in smallFiles)
{
try
{
var processedLocal = Interlocked.Increment(ref processed);
var newFileWin = Path.Combine(currentBkpDir, file.RelativePathWin);
var newFileRelativeName = newFileWin.Replace(currentBkpDir, string.Empty);
var existingFileWin = findHelper.FindByLengthAndHash(file.FileInfo);
if (existingFileWin != null)
{
_hardLinkHelper.AddHardLinkToQueue(existingFileWin, newFileWin);
linkedCount++;
WriteLog($"[{processedLocal} of {filesCount}] {{link}} {newFileRelativeName} to {existingFileWin}", Interlocked.Increment(ref category));
}
else
{
var relFileName = file.RelativePathUnix;
using (var fl = file.FileInfo.FileInfo.OpenRead())
{
tar.AddFile(relFileName, fl);
++archivedCount;
}
WriteLog($"[{processedLocal} of {filesCount}] {{tar}} {newFileRelativeName} ", Interlocked.Increment(ref category));
}
var o = new BackupFileInfo
{
Path = newFileRelativeName,
Hash = file.FileInfo.FastHashStr,
Length = file.FileInfo.FileInfo.Length,
IsLink = existingFileWin != null,
};
currentBkp.AddFile(o);
}
catch (Exception e)
{
Console.WriteLine(e);
}
}
created = tar.IsArchiveCreated;
}
sw.Stop();
if (created)
{
var tarAndSendDuration = sw.Elapsed;
WriteLog("Unpacking small files", Interlocked.Increment(ref category));
sw = System.Diagnostics.Stopwatch.StartNew();
File.Move(tmpTarPath, smallFilesTarPath);
_hardLinkHelper.UnpackTar(smallFilesTarPath);
sw.Stop();
WriteLog($"{archivedCount} files archived and transferred in {tarAndSendDuration:g} and unpacked in {sw.Elapsed:g}", ++category);
}
}
public async Task DoBackup()
{
await Task.Yield();
Validate();
var category = 0;
var localFiles = GetFilesToBackup(ref category);
WriteLog("Discovering backups...", ++category);
var newBkpDate = DateTime.Now;
var newBkpName = newBkpDate.ToString(BackupFolderNamingDateFormat, CultureInfo.InvariantCulture);
var prevBkps = BackupInfo.DiscoverBackups(_destination).ToList();
if (_backupRoots != null)
{
foreach (var root in _backupRoots)
{
prevBkps.AddRange(BackupInfo.DiscoverBackups(root));
}
}
WriteLog($"Found {prevBkps.Count} backups", ++category);
var currentBkpDir = Path.Combine(_destination, newBkpName);
var filesCount = localFiles.Count;
var currentBkp = new BackupInfo(currentBkpDir)
{
DateTime = newBkpDate,
AttributesAvailable = true,
};
var prevBackupFiles = GetFilesFromPrevBackups(prevBkps, ref category);
var copiedCount = 0;
var linkedCount = 0;
var svcDir = currentBkp.CreateFolders();
currentBkp.CreateIncompleteAttribute();
var smallFilesTarPath = Path.Combine(svcDir, "small-files.tar.gz");
var directoriesTarPath = Path.Combine(svcDir, "dir-tree.tar.gz");
var findHelper = new SessionFileFindHelper(currentBkp, prevBackupFiles);
WriteLog("Backing up...", ++category);
var processed = 0;
{ // collect changes
var byLength = prevBackupFiles
.GroupBy(x => x.Item1.Hash)
.SelectMany(x => x)
.GroupBy(x => x.Item1.Length)
.ToDictionary(x => x.Key, x => x.ToList());
var lengthMatch = 0;
var lengthFailMatch = 0;
var lengthMismatch = 0;
var hashMatch = 0;
var fails = 0;
foreach (var localFileInfo in localFiles)
{
if (byLength.TryGetValue(localFileInfo.FileInfo.FileInfo.Length, out var files))
{
lengthMatch++;
try
{
var backupFile = files.FirstOrDefault(x => x.Item1.Hash == localFileInfo.FileInfo.FastHashStr);
if (backupFile == null)
{
lengthFailMatch++;
}
else
{
hashMatch++;
}
}
catch
{
fails++;
}
}
else
{
lengthMismatch++;
}
}
WriteLog($"match: {lengthMatch}; mismatch: {lengthMismatch}; fail match: {lengthFailMatch}; hash: {hashMatch}; fails: {fails}", Interlocked.Increment(ref category));
}
throw null;
CreateDirectories(localFiles, directoriesTarPath, ref category);
var smallFiles = GetFilesForCompression(localFiles);
ProcessSmallFiles(smallFiles, ref category, smallFilesTarPath, currentBkpDir, findHelper, filesCount, currentBkp, ref linkedCount, ref processed);
foreach (var localFileInfo in localFiles)
{
try
{
var processedLocal = Interlocked.Increment(ref processed);
var newFile = Path.Combine(currentBkpDir, localFileInfo.RelativePathWin);
var newFileRelativeName = newFile.Replace(currentBkpDir, string.Empty);
var newDir = Path.GetDirectoryName(newFile);
if (newDir == null)
{
throw new InvalidOperationException("Cannot get file's directory");
}
if (!Directory.Exists(newDir))
{
Directory.CreateDirectory(newDir);
}
var existingFile = findHelper.FindByLengthAndHash(localFileInfo.FileInfo);
if (existingFile != null)
{
WriteLog($"[{processedLocal} of {filesCount}] {{link}} {localFileInfo.RelativePathWin} ", Interlocked.Increment(ref category));
_hardLinkHelper.AddHardLinkToQueue(existingFile, newFile);
linkedCount++;
}
else
{
void ProgressCallback(double progress)
{
WriteLogExt($"{progress:F2} %");
}
WriteLog($"[{processedLocal} of {filesCount}] {localFileInfo.RelativePathWin} ", Interlocked.Increment(ref category));
var copiedHash = HashSumHelper.CopyUnbufferedAndComputeHashAsyncXX(localFileInfo.FileInfo.FileName, newFile, ProgressCallback, _allowSimultaneousReadWrite).Result;
if (localFileInfo.FileInfo.FastHashStr == string.Concat(copiedHash.Select(b => $"{b:X}")))
{
copiedCount++;
}
else
{
WriteLog($"{localFileInfo.RelativePathWin} copy failed", Interlocked.Increment(ref category));
System.Diagnostics.Debugger.Break();
}
new FileInfo(newFile).Attributes |= FileAttributes.ReadOnly;
}
var o = new BackupFileInfo
{
Path = newFileRelativeName,
Hash = localFileInfo.FileInfo.FastHashStr,
Length = localFileInfo.FileInfo.FileInfo.Length,
IsLink = existingFile != null,
};
currentBkp.AddFile(o);
}
catch (Exception e)
{
Console.WriteLine();
Console.WriteLine(e);
Console.WriteLine();
}
}
WriteLog("Writing hardlinks to target", Interlocked.Increment(ref category));
try
{
_hardLinkHelper.CreateHardLinks();
}
finally
{
currentBkp.WriteToDisk();
currentBkp.DeleteIncompleteAttribute();
}
var log = "Backup done.";
if (copiedCount > 0)
{
log += $" {copiedCount} files copied";
}
if (linkedCount > 0)
{
log += $" {linkedCount} files linked";
}
WriteLog(log, ++category);
}