public void Find(IEnumerable <RootEntry> rootEntries)
{
if (VisitorFunc == null)
{
return;
}
int[] limitCount = { LimitResultCount };
if (ProgressFunc == null || ProgressModifier == 0)
{ // dummy func and huge progressModifier so wont call progressFunc anyway.
ProgressFunc = delegate { };
ProgressModifier = int.MaxValue;
}
// ReSharper disable PossibleMultipleEnumeration
ProgressEnd = rootEntries.TotalFileEntries();
// ReSharper restore PossibleMultipleEnumeration
ProgressFunc(_progressCount[0], ProgressEnd); // Start of process Progress report.
PatternMatcher = GetPatternMatcher();
var findFunc = GetFindFunc(_progressCount, limitCount);
// ReSharper disable PossibleMultipleEnumeration
CommonEntry.TraverseTreePair(rootEntries, findFunc);
ProgressFunc(_progressCount[0], ProgressEnd); // end of Progress
// ReSharper restore PossibleMultipleEnumeration
}
public IDictionary <long, List <PairDirEntry> > GetSizePairs(IEnumerable <RootEntry> rootEntries)
{
CommonEntry.TraverseTreePair(rootEntries, FindMatchesOnFileSize2);
_logger.LogDebug("Post TraverseMatchOnFileSize: {0}, dupeDictCount {1}", _applicationDiagnostics.GetMemoryAllocated().FormatAsBytes(), _duplicateFileSize.Count);
//Remove the single values from the dictionary. DOESNT SEEM TO CLEAR MEMORY ??? GC Force?
_duplicateFileSize.Where(kvp => kvp.Value.Count == 1).ToList().ForEach(x => _duplicateFileSize.Remove(x.Key));
_logger.LogDebug("Deleted entries from dictionary: {0}, dupeDictCount {1}", _applicationDiagnostics.GetMemoryAllocated().FormatAsBytes(), _duplicateFileSize.Count);
return(_duplicateFileSize);
}
public IList <KeyValuePair <DirEntry, List <PairDirEntry> > > GetDupePairs(IEnumerable <RootEntry> rootEntries)
{
CommonEntry.TraverseTreePair(rootEntries, BuildDuplicateList);
var moreThanOneFile = _duplicateFile.Where(d => d.Value.Count > 1).ToList();
_logger.LogInfo("Count of list of all hashes of files with same sizes {0}",
_duplicateFile.Count);
_logger.LogInfo("Count of list of all hashes of files with same sizes where more than 1 of that hash {0}",
moreThanOneFile.Count);
return(moreThanOneFile);
}
private void CheckDupesAndCompleteFullHash(IEnumerable <RootEntry> rootEntries)
{
_logger.LogDebug(string.Empty);
_logger.LogDebug("Checking duplicates and completing full hash.");
CommonEntry.TraverseTreePair(rootEntries, BuildDuplicateListIncludePartialHash);
var founddupes = _duplicateFile.Where(d => d.Value.Count > 1);
var totalEntriesInDupes = founddupes.Sum(x => x.Value.Count);
var longestListLength = founddupes.Any() ? founddupes.Max(x => x.Value.Count) : 0;
_logger.LogInfo("Found {0} duplication collections.", founddupes.Count());
_logger.LogInfo("Total files found with at least 1 other file duplicate {0}",
totalEntriesInDupes);
_logger.LogInfo("Longest list of duplicate files is {0}", longestListLength);
foreach (var keyValuePair in founddupes)
{
foreach (var pairDirEntry in keyValuePair.Value)
{
_dirEntriesRequiringFullHashing.Add(pairDirEntry.ChildDE);
}
}
CommonEntry.TraverseTreePair(rootEntries, CalculateFullMD5Hash);
}