/// <summary>
/// Step 2: calculate crc
/// </summary>
/// <param name="fileList"></param>
/// <param name="option"></param>
/// <returns>List DuplicateArchiveInfo</returns>
private List <DuplicateArchiveInfo> CalculateCRC(List <FileInfo> fileList, DuplicateSearchOption option)
{
List <DuplicateArchiveInfo> list = new List <DuplicateArchiveInfo>();
int i = 0;
foreach (FileInfo f in fileList)
{
_pauseEvent.WaitOne(Timeout.Infinite);
if (_shutdownEvent.WaitOne(0))
{
break;
}
//NotifyCaller(f.FullName, OperationStatus.CALCULATING_CRC, curr: i, total: fileList.Count);
//string msg = String.Format("File {0} of {1}", i, fileList.Count);
NotifyCaller("", OperationStatus.CALCULATING_CRC, curr: i, total: fileList.Count);
try
{
DuplicateArchiveInfo item = Util.GetArchiveInfo(f.FullName, option);
item.FileSize = f.Length;
item.CreationTime = f.CreationTime;
list.Add(item);
}
catch (Exception ex)
{
string message = ex.Message + " (" + f.FullName + ")";
NotifyCaller(message, OperationStatus.ERROR);
}
++i;
}
NotifyCaller("Complete calculating CRC, total: " + list.Count, OperationStatus.CALCULATING_CRC, total: list.Count);
return(list);
}
/// <summary>
/// Check if file is duplicated
/// </summary>
/// <param name="original"></param>
/// <param name="duplicate"></param>
/// <param name="option"></param>
/// <returns></returns>
private bool Compare(ref DuplicateArchiveInfo original, ref DuplicateArchiveInfo duplicate, DuplicateSearchOption option)
{
lock (original)
{
//NotifyCaller("Comparing: " + Origin.Filename + " to " + Duplicate.Filename, OperationStatus.COMPARING);
// if the match type already changed from original, skip it
// most likely already validated by other task
if (original.MatchType != MatchType.ORIGINAL)
{
return(false);
}
// if item count is equal, try to check from crc strings.
original.MatchType = MatchType.ORIGINAL;
original.Percentage = 0.0;
if (original.NoMatches != null)
{
original.NoMatches.Clear();
}
if (original.Items.Count == duplicate.Items.Count)
{
if (original.ToCRCString() == duplicate.ToCRCString())
{
//NotifyCaller("CRC Strings are equal.", OperationStatus.COMPARING);
duplicate.Percentage = 100.0;
duplicate.MatchType = MatchType.EQUALCOUNT;
return(true);
}
else if (option.OnlyPerfectMatch)
{
return(false);
}
}
// Check each files in duplicate
int limitCount;
// if only have 'IgnoreLimit' files, then all must match
if (option.IgnoreLimit > duplicate.Items.Count)
{
limitCount = 0;
}
else
{
limitCount = duplicate.Items.Count - (duplicate.Items.Count * option.Limit / 100);
}
int skippedCount = 0;
int i = 0;
int j = 0;
while (i < original.Items.Count && j < duplicate.Items.Count && skippedCount <= limitCount)
{
// compare the from the biggest crc.
int result = string.Compare(original.Items[i].Crc, duplicate.Items[j].Crc, true, System.Globalization.CultureInfo.InvariantCulture);
if (result == 0)
{
++i; ++j;
}
else if (result > 0)
{
// Origin file skipped
++i;
}
else
{
// Duplicate file skipped, no match in Origin
++skippedCount;
if (duplicate.NoMatches == null)
{
duplicate.NoMatches = new List <ArchiveFileInfoSmall>();
}
duplicate.NoMatches.Add(duplicate.Items[j]);
++j;
}
}
if (j < duplicate.Items.Count)
{
if (duplicate.NoMatches == null)
{
duplicate.NoMatches = new List <ArchiveFileInfoSmall>();
}
duplicate.NoMatches.AddRange(duplicate.Items.GetRange(j, duplicate.Items.Count - j));
skippedCount = duplicate.NoMatches.Count;
}
double percent = (double)(duplicate.Items.Count - skippedCount) / duplicate.Items.Count * 100;
if (percent >= option.Limit && skippedCount < limitCount)
{
//NotifyCaller("Match: " + percent + "%", OperationStatus.COMPARING);
duplicate.Percentage = percent;
duplicate.MatchType = MatchType.SUBSET;
return(true);
}
//NotifyCaller("Not Match", OperationStatus.COMPARING);
if (duplicate.NoMatches != null)
{
duplicate.NoMatches.Clear();
}
return(false);
}
}
/// <summary>
/// Step 3: Build duplicate list
/// </summary>
/// <param name="list"></param>
/// <param name="limit"></param>
/// <param name="ignoreLimit"></param>
private List <DuplicateArchiveInfoList> BuildDuplicateList(List <DuplicateArchiveInfo> list, DuplicateSearchOption option)
{
NotifyCaller("Start building duplicate list.", OperationStatus.BUILDING_DUPLICATE_LIST);
List <DuplicateArchiveInfoList> dupList = new List <DuplicateArchiveInfoList>();
list.Sort(new DuplicateArchiveInfoItemCountComparer());
int totalCount = list.Count;
int i = 0;
while (list.Count > 0)
{
_pauseEvent.WaitOne(Timeout.Infinite);
if (_shutdownEvent.WaitOne(0))
{
NotifyCaller("Stopping...", OperationStatus.BUILDING_DUPLICATE_LIST);
break;
}
++i;
DuplicateArchiveInfoList dup = new DuplicateArchiveInfoList();
DuplicateArchiveInfo original = list[0];
list.RemoveAt(0);
dup.Original = original;
string message = "Checking: " + original.Filename + " ( Duplicate group found: " + i + " Remaining: " + list.Count + ")";
//NotifyCaller(message, OperationStatus.BUILDING_DUPLICATE_LIST, curr: i, total: totalCount);
NotifyCaller("", OperationStatus.BUILDING_DUPLICATE_LIST, curr: i, total: totalCount);
// parallel method
if (option.TaskLimit > 1)
{
var taskScheduler = new Nandaka.Common.LimitedConcurrencyLevelTaskScheduler(option.TaskLimit, 16);
var pOption = new ParallelOptions()
{
TaskScheduler = taskScheduler
};
Parallel.For(0, list.Count, pOption, (innerIdx) =>
{
DuplicateArchiveInfo curr = list[innerIdx];
if (curr.IsRemoved)
{
return;
}
if (Compare(ref original, ref curr, option))
{
if (dup.Duplicates == null)
{
dup.Duplicates = new List <DuplicateArchiveInfo>();
}
// remove from the source list.
lock (list)
{
curr.IsRemoved = true;
}
dup.Duplicates.Add(curr);
}
});
}
else
{
// check for other possible dups.
int index = 0;
while (list.Count > index)
{
DuplicateArchiveInfo curr = list[index];
if (Compare(ref original, ref curr, option))
{
if (dup.Duplicates == null)
{
dup.Duplicates = new List <DuplicateArchiveInfo>();
}
dup.Duplicates.Add(curr);
// remove from the source list.
list.Remove(curr);
--totalCount;
}
else
{
++index;
}
}
}
if (dup.Duplicates != null && dup.Duplicates.Count > 0)
{
dupList.Add(dup);
}
}
foreach (DuplicateArchiveInfoList dup in dupList)
{
if (dup.Duplicates != null)
{
dup.Duplicates.Sort(new DuplicateArchiveInfoPercentageComparer());
}
}
NotifyCaller("Building Duplicate List Complete.", OperationStatus.BUILDING_DUPLICATE_LIST);
return(dupList);
}
/// <summary>
/// Build DuplicateArchiveInfo containing the files' crc, sorted.
/// </summary>
/// <param name="filename"></param>
/// <param name="blackListPattern"></param>
/// <returns></returns>
public static DuplicateArchiveInfo GetArchiveInfo(string filename, DuplicateSearchOption option)
{
Regex re = new Regex(option.BlacklistPattern, option.BlacklistCaseInsensitive ? RegexOptions.IgnoreCase : RegexOptions.None);
DuplicateArchiveInfo info = new DuplicateArchiveInfo();
SevenZipExtractor.SetLibraryPath(option.SevenZipPath);
using (SevenZipExtractor extractor = new SevenZipExtractor(filename))
{
info.Filename = filename;
info.Items = new List <ArchiveFileInfoSmall>();
info.RealSize = extractor.UnpackedSize;
info.ArchivedSize = extractor.PackedSize;
ulong countedSize = 0;
foreach (ArchiveFileInfo af in extractor.ArchiveFileData)
{
if (af.IsDirectory)
{
info.DirectoryCount++;
continue;
}
ArchiveFileInfoSmall item = new ArchiveFileInfoSmall()
{
Crc = ConvertToHexString(af.Crc),
Filename = af.FileName,
Size = af.Size
};
if (!String.IsNullOrWhiteSpace(option.BlacklistPattern) && re.IsMatch(af.FileName))
{
if (info.Skipped == null)
{
info.Skipped = new List <ArchiveFileInfoSmall>();
}
item.Remark = "Blacklisted";
info.Skipped.Add(item);
}
else if (option.IgnoreSmallFile && item.Size < option.SmallFileSizeLimit)
{
if (info.Skipped == null)
{
info.Skipped = new List <ArchiveFileInfoSmall>();
}
item.Remark = "SmallFileSizeLimit";
info.Skipped.Add(item);
}
else
{
item.Remark = "";
info.Items.Add(item);
}
countedSize += af.Size;
}
if (info.RealSize == -1)
{
info.RealSize = Convert.ToInt64(countedSize);
}
info.SortItems();
}
return(info);
}