public static Stats Run(IEnumerable <TSV_DATA> findA, IEnumerable <TSV_DATA> findB, DeltaWriter writer,
out List <string> newDirs, out List <string> delDirs,
string sortedAfilename, string sortedBfilename)
{
newDirs = null;
delDirs = null;
Stats stats = null;
List <TSV_DATA> newFiles;
List <TSV_DATA> delFiles;
Console.Error.WriteLine("trying diff on given data");
bool wasSorted = true;
try
{
stats = DoDiff(findA, findB, writer, out newDirs, out delDirs, out newFiles, out delFiles);
}
catch (ApplicationException appEx)
{
wasSorted = false;
Console.Error.WriteLine(appEx.Message);
Console.Error.WriteLine("data was not sorted!");
}
Task writeSortedA = Task.CompletedTask;
Task writeSortedB = Task.CompletedTask;
if (!wasSorted)
{
Console.Error.WriteLine("reading files again and sorting");
Task <List <TSV_DATA> > sortedA = SortData(findA, "A");
Task <List <TSV_DATA> > sortedB = SortData(findB, "B");
while (!Task.WaitAll(new Task[] { sortedA, sortedB }, millisecondsTimeout: 5000))
{
try
{
var proc = System.Diagnostics.Process.GetCurrentProcess();
Console.Error.WriteLine($"virtMem: {Misc.GetPrettyFilesize(proc.VirtualMemorySize64)}");
}
catch { }
}
Console.Error.WriteLine($"number sorted items A/B: {sortedA.Result.Count}/{sortedB.Result.Count}");
writeSortedA = (sortedAfilename == null) ? Task.CompletedTask : WriteTsvData(sortedAfilename, sortedA.Result);
writeSortedB = (sortedBfilename == null) ? Task.CompletedTask : WriteTsvData(sortedBfilename, sortedB.Result);
Console.Error.WriteLine("running diff on sorted data");
stats = DoDiff(sortedA.Result, sortedB.Result, writer, out newDirs, out delDirs, out newFiles, out delFiles);
}
File.WriteAllLines(@".\DelDirsBeforeCompress.txt", delDirs);
delDirs.Sort();
IEnumerable <string> compressDelDirs = CompressToBaseDirs(delDirs);
delDirs = compressDelDirs.ToList();
Task.WaitAll(new Task[] { writeSortedA, writeSortedB });
return(stats);
}
private static Stats DoDiff(IEnumerable <TSV_DATA> sortedA, IEnumerable <TSV_DATA> sortedB, DeltaWriter writer, out List <string> newDirs, out List <string> delDirs,
out List <TSV_DATA> newFiles, out List <TSV_DATA> delFiles)
{
List <string> tmpNewDirs = new List <string>();
List <string> tmpDelDirs = new List <string>();
List <TSV_DATA> tmpNewFiles = new List <TSV_DATA>();
List <TSV_DATA> tmpDelFiles = new List <TSV_DATA>();
Stats stats = new Stats();
uint diff =
Spi.Data.Diff.DiffSortedEnumerables <TSV_DATA>(
sortedA,
sortedB,
KeyComparer: (TSV_DATA a, TSV_DATA b) =>
{
int cmp = String.Compare(a.relativeFilename, b.relativeFilename, StringComparison.OrdinalIgnoreCase);
if (cmp != 0)
{
return(cmp);
}
bool KindOfA = Spi.Misc.IsDirectoryFlagSet(a.dwFileAttributes);
bool KindOfB = Spi.Misc.IsDirectoryFlagSet(b.dwFileAttributes);
return(KindOfA == KindOfB
? 0 // two directories OR two files --> same name --> return 0
: -1); // one dir AND one file --> same name --> return -1 to represent the difference
},
AttributeComparer: (TSV_DATA a, TSV_DATA b) =>
{
if (Misc.IsDirectoryFlagSet(a.dwFileAttributes) && Misc.IsDirectoryFlagSet(b.dwFileAttributes))
{
return(0);
}
long cmp;
if ((cmp = (a.timeModified - b.timeModified)) != 0)
{
return((int)cmp);
}
if ((cmp = (long)(a.size - b.size)) != 0)
{
return((int)cmp);
}
return(0);
},
OnCompared: (DIFF_STATE state, TSV_DATA a, TSV_DATA b) =>
{
switch (state)
{
case DIFF_STATE.NEW:
if (Misc.IsDirectoryFlagSet(b.dwFileAttributes))
{
tmpNewDirs.Add(b.relativeFilename);
}
else
{
writer.newFilesWriter.WriteLine(b.relativeFilename);
tmpNewFiles.Add(b);
stats.newFiles += 1;
stats.newFilesSize += b.size;
}
break;
case DIFF_STATE.MODIFY:
writer.modFilesWriter.WriteLine(b.relativeFilename);
stats.modFiles += 1;
stats.modFilesSize += b.size;
break;
case DIFF_STATE.DELETE:
if (Misc.IsDirectoryFlagSet(a.dwFileAttributes))
{
tmpDelDirs.Add(a.relativeFilename);
}
else
{
writer.delFilesWriter.WriteLine(a.relativeFilename);
tmpDelFiles.Add(a);
stats.delFiles += 1;
stats.delFilesSize += a.size;
}
break;
}
},
checkSortOrder: true);
newDirs = tmpNewDirs;
delDirs = tmpDelDirs;
newFiles = tmpNewFiles;
delFiles = tmpDelFiles;
return(stats);
}