public FileComparisonResult(FileInfoForComparison fileLeft,
FileInfoForComparison fileRight,
FileComparisonResultType type)
{
FileInfoLeft = fileLeft;
FileInfoRight = fileRight;
Type = type;
string showPath;
if (type == FileComparisonResultType.Same_Contents &&
FileInfoLeft.Filename != FileInfoRight.Filename)
{
showPath = StripInitialSlash(FileInfoLeft.Filename) + "; " +
StripInitialSlash(FileInfoRight.Filename);
}
else
{
showPath = StripInitialSlash(FileInfoLeft?.Filename ??
FileInfoRight.Filename);
}
ImageIndex = (int)type;
SubItems.Add(type.ToString().Replace("_", " "));
SubItems.Add(showPath);
if (type == FileComparisonResultType.Changed &&
fileRight.LastModifiedTime > fileLeft.LastModifiedTime)
{
// show a red icon if the file on the right is newer.
ImageIndex = 5;
}
}
public static List <FileComparisonResult> Go(SortFilesSettings settings)
{
var filesInLeft = new DirectoryInfo(settings.LeftDirectory).EnumerateFiles(
"*", SearchOption.AllDirectories);
var filesInRight = new DirectoryInfo(settings.RightDirectory).EnumerateFiles(
"*", SearchOption.AllDirectories);
// first, just make an index that simply maps filesizes to filenames.
// we don't need to compute any content-hashes yet, because if there
// is only one file with that filesize, we know it's not a duplicate.
var results = new List <FileComparisonResult>();
var indexLeft = MapFilesizesToFilenames(settings.LeftDirectory, filesInLeft);
// go through files on the right
foreach (var infoRight in filesInRight)
{
var filenameRight = infoRight.FullName.Substring(
settings.RightDirectory.Length);
var objLeft = FindInMap(indexLeft, settings.LeftDirectory, infoRight.FullName,
infoRight.Length, settings.SearchDuplicatesCanUseFiletimes,
infoRight.LastWriteTimeUtc, filenameRight);
if (objLeft != null)
{
// these are duplicates, they have the same hash and filesize.
var objRight = new FileInfoForComparison(filenameRight,
infoRight.Length, infoRight.LastWriteTimeUtc, objLeft.ContentHash);
results.Add(new FileComparisonResult(
objLeft, objRight, FileComparisonResultType.Same_Contents));
}
}
return(results);
}
public static List <FileComparisonResult> Go(SortFilesSettings settings)
{
var results = new List <FileComparisonResult>();
var filesInLeft = new Dictionary <string, FileInfoForComparison>(
StringComparer.OrdinalIgnoreCase);
// go through files in left
var diLeft = new DirectoryInfo(settings.LeftDirectory);
foreach (var info in diLeft.EnumerateFiles("*", SearchOption.AllDirectories))
{
var filename = info.FullName.Substring(settings.LeftDirectory.Length);
filesInLeft[filename] = new FileInfoForComparison(
filename, info.Length, info.LastWriteTimeUtc);
}
// go through files in right
var diRight = new DirectoryInfo(settings.RightDirectory);
foreach (var info in diRight.EnumerateFiles("*", SearchOption.AllDirectories))
{
var filenameRight = info.FullName.Substring(settings.RightDirectory.Length);
if (filesInLeft.TryGetValue(filenameRight, out FileInfoForComparison objLeft))
{
objLeft.MarkWhenVisited = true;
if (objLeft.FileSize != info.Length ||
!AreTimesEqual(objLeft.LastModifiedTime, info.LastWriteTimeUtc, settings))
{
// looks like a modified file. same path but different filesize/lmt.
var filename = info.FullName.Substring(settings.RightDirectory.Length);
var objRight = new FileInfoForComparison(
filename, info.Length, info.LastWriteTimeUtc);
results.Add(new FileComparisonResult(
objLeft, objRight, FileComparisonResultType.Changed));
}
}
else
{
// looks like a new file
var objRight = new FileInfoForComparison(
filenameRight, info.Length, info.LastWriteTimeUtc);
results.Add(new FileComparisonResult(
null, objRight, FileComparisonResultType.Right_Only));
}
}
// which files did we see in left but not in right?
foreach (var kvp in filesInLeft)
{
if (!kvp.Value.MarkWhenVisited)
{
// looks like a deleted file since it didn't show up on the right.
results.Add(new FileComparisonResult(
kvp.Value, null, FileComparisonResultType.Left_Only));
}
}
return(results);
}
public static Dictionary <long, List <FileInfoForComparison> > MapFilesizesToFilenames(
string dirName, IEnumerable <FileInfo> files)
{
// map filesize to List<FileInfoForComparison> or HashSet<FileInfoForComparison>?
// chose List<>; maintaining inserted order makes results that look nicer to the user.
var map = new Dictionary <long, List <FileInfoForComparison> >();
foreach (var info in files)
{
var filename = info.FullName.Substring(dirName.Length);
var obj = new FileInfoForComparison(
filename, info.Length, info.LastWriteTimeUtc);
if (!map.TryGetValue(obj.FileSize, out List <FileInfoForComparison> list))
{
list = map[obj.FileSize] = new List <FileInfoForComparison>();
}
list.Add(obj);
}
return(map);
}
static FileInfoForComparison FindInMap(Dictionary <long, List <FileInfoForComparison> > map,
string baseDir, string fullnameToFind, long lengthOfFileToFind,
bool useLmtShortcut, DateTime lmt, string partialFilename)
{
if (lengthOfFileToFind == 0)
{
// 0-length files compare unequal; often placeholders intentionally created by user
return(null);
}
if (map.TryGetValue(lengthOfFileToFind, out List <FileInfoForComparison> list))
{
// look for an entry with the same filename
FileInfoForComparison hasSameName = null;
foreach (var obj in list)
{
if (obj.Filename == partialFilename)
{
hasSameName = obj;
break;
}
}
// if enabled, treat files with the same filesize, lmt, and name as equal
if (useLmtShortcut && hasSameName != null &&
hasSameName.LastModifiedTime == lmt)
{
return(hasSameName);
}
// change order so that an entry with the same name is checked first. why?
// 1) results look nicer when paired this way
// 2) same order of results whether or not useLmtShortcut is on
// (as long as lmt times are accurate). Consider the case with two duplicates,
// one with the same name, and one that sorts earlier.
// We prefer to pick the file with the same name to show as the duplicate.
if (hasSameName != null)
{
list = new List <FileInfoForComparison>(list);
list.Insert(0, hasSameName);
}
// we found another file(s) with the same filesize, so
// let's compare hashes of the content to see if they're the same.
var hash = Utils.GetSha512(fullnameToFind);
foreach (var obj in list)
{
// compute the hash if it hasn't been computed already, then cache it
if (obj.ContentHash == null)
{
obj.ContentHash = Utils.GetSha512(
baseDir + obj.Filename);
}
if (obj.ContentHash == hash)
{
return(obj);
}
}
}
return(null);
}
