private void _Finish()
{
if (_doThreading) {
_SortAndWriteThreaded();
// wait on all previous threads, sorts and writes
foreach (Thread t in _threads) {
t.Join();
}
_threads.Clear();
}
else {
// don't write the last one to disk
_recordsToSort.Sort();
_sortedRecords = _recordsToSort;
}
int numToMerge = _numTempFiles;
if (_sortedRecords != null && _sortedRecords.NumRecords != 0) {
numToMerge++;
}
if (numToMerge > 1) {
_merger = new SortedRecordMerger();
if (_sortedRecords != null && _sortedRecords.NumRecords != 0) {
_merger.AddInput(_sortedRecords);
}
for (int i = 0; i < _numTempFiles; i++) {
InternalRecordSource source = new InternalRecordFileReader(GetTempFilename(i));
_merger.AddInput(source);
}
_output = _merger;
}
else {
_output = _sortedRecords;
}
// set up reduction filter for reduction across merged sources
if (_internalReductionEnabled) {
ReduceFilter reducer = new ReduceFilter();
RecordFilterDriver filterDriver = new RecordFilterDriver(reducer);
filterDriver.AddInput(_output);
_output = filterDriver;
}
// this is kind of a hack till i figure out how these should be set
TotalRecordBytesEstimate = _output.TotalRecordBytesEstimate;
TotalRecordsEstimate = _output.TotalRecordsEstimate;
}
// write this chunk to disk. This is where reduction happens if it happens.
private void _WriteSortedRecords()
{
if (!Directory.Exists(_tempDir)) {
Directory.CreateDirectory(_tempDir);
}
InternalRecordSource output = _recordsToWrite;
// set up reduction filter. The whole reason we have a sorterReducer
// and not just a sorter is so that we can reduce before we write our
// temp files to disk.
if (_internalReductionEnabled) {
ReduceFilter rf = new ReduceFilter();
RecordFilterDriver filterDriver = new RecordFilterDriver(rf);
filterDriver.AddInput(output);
output = filterDriver;
}
RecordFileWriter recordFileWriter = new RecordFileWriter(GetTempFilename(_numTempFiles));
_numTempFiles++;
try {
recordFileWriter.Write(output);
}
finally {
_recordsToWrite.Close();
}
}
private void _SortReduce(bool sortAscending, bool reductionEnabled)
{
// lets change sort needs to a number for easy comparison.
// no sort = 0, sort ascending = 1, sort descending = 2
int askSortNum = 1; // we are asking for sorting on our output
if (!sortAscending) askSortNum = 2;
bool inputIsSorted = InternalSource.Sorting.IsSorted;
bool inputIsSortedAscending = InternalSource.Sorting.IsSortedAscending;
int inputSortNum = 0;
if (inputIsSorted)
inputSortNum = 1;
if (inputIsSorted && !inputIsSortedAscending)
inputSortNum = 2;
// (aside: one might ask, why not just have a separate sorter
// and reducer. Reduction is combined in the sorter so that
// reduction can happen in memory before temp files are written
// to disk.
if (askSortNum == inputSortNum && !reductionEnabled) return; // don't insert a sorter or reducer
if (askSortNum != inputSortNum) {
RecordSorterReducer sr = new RecordSorterReducer(Processor.TempDir, sortAscending, reductionEnabled);
sr.MaxMemorySize = Processor.MaxMemorySize;
sr.DeleteTempFiles = Processor.DeleteTempFiles;
InternalRecordSource temp = InternalSource; // grab our input
sr.AddInput(temp); // pipe it into the sorterReducer
this.InternalSource = sr; // make the sorterReducer the output of this source
}
else {
IRecordFilter reducer = new ReduceFilter();
RecordFilterDriver driver = new RecordFilterDriver(reducer);
driver.AddInput(InternalSource);
InternalSource = driver;
}
}
