// DocValues updates
private void ApplyDocValuesUpdates <T1>(IEnumerable <T1> updates, ReadersAndUpdates rld, SegmentReader reader, DocValuesFieldUpdates.Container dvUpdatesContainer) where T1 : DocValuesUpdate
{
UninterruptableMonitor.Enter(this);
try
{
Fields fields = reader.Fields;
if (fields is null)
{
// this reader has no postings
return;
}
// TODO: we can process the updates per DV field, from last to first so that
// if multiple terms affect same document for the same field, we add an update
// only once (that of the last term). To do that, we can keep a bitset which
// marks which documents have already been updated. So e.g. if term T1
// updates doc 7, and then we process term T2 and it updates doc 7 as well,
// we don't apply the update since we know T1 came last and therefore wins
// the update.
// We can also use that bitset as 'liveDocs' to pass to TermEnum.docs(), so
// that these documents aren't even returned.
string currentField = null;
TermsEnum termsEnum = null;
DocsEnum docs = null;
//System.out.println(Thread.currentThread().getName() + " numericDVUpdate reader=" + reader);
foreach (DocValuesUpdate update in updates)
{
Term term = update.term;
int limit = update.docIDUpto;
// TODO: we traverse the terms in update order (not term order) so that we
// apply the updates in the correct order, i.e. if two terms udpate the
// same document, the last one that came in wins, irrespective of the
// terms lexical order.
// we can apply the updates in terms order if we keep an updatesGen (and
// increment it with every update) and attach it to each NumericUpdate. Note
// that we cannot rely only on docIDUpto because an app may send two updates
// which will get same docIDUpto, yet will still need to respect the order
// those updates arrived.
if (!string.Equals(term.Field, currentField, StringComparison.Ordinal))
{
// if we change the code to process updates in terms order, enable this assert
// assert currentField is null || currentField.CompareToOrdinal(term.Field) < 0;
currentField = term.Field;
Terms terms = fields.GetTerms(currentField);
if (terms != null)
{
termsEnum = terms.GetEnumerator(termsEnum);
}
else
{
termsEnum = null;
continue; // no terms in that field
}
}
if (termsEnum is null)
{
continue;
}
// System.out.println(" term=" + term);
if (termsEnum.SeekExact(term.Bytes))
{
// we don't need term frequencies for this
DocsEnum docsEnum = termsEnum.Docs(rld.LiveDocs, docs, DocsFlags.NONE);
//System.out.println("BDS: got docsEnum=" + docsEnum);
DocValuesFieldUpdates dvUpdates = dvUpdatesContainer.GetUpdates(update.field, update.type);
if (dvUpdates is null)
{
dvUpdates = dvUpdatesContainer.NewUpdates(update.field, update.type, reader.MaxDoc);
}
int doc;
while ((doc = docsEnum.NextDoc()) != DocIdSetIterator.NO_MORE_DOCS)
{
//System.out.println(Thread.currentThread().getName() + " numericDVUpdate term=" + term + " doc=" + docID);
if (doc >= limit)
{
break; // no more docs that can be updated for this term
}
// LUCENENET specific handling - dvUpdates handles getting the value so we don't
// have to deal with boxing/unboxing
dvUpdates.AddFromUpdate(doc, update);
}
}
}
}
finally
{
UninterruptableMonitor.Exit(this);
}
}
public virtual void WriteFieldUpdates(Directory dir, DocValuesFieldUpdates.Container dvUpdates)
{
UninterruptableMonitor.Enter(this);
try
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(UninterruptableMonitor.IsEntered(writer));
}
//System.out.println("rld.writeFieldUpdates: seg=" + info + " numericFieldUpdates=" + numericFieldUpdates);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(dvUpdates.Any());
}
// Do this so we can delete any created files on
// exception; this saves all codecs from having to do
// it:
TrackingDirectoryWrapper trackingDir = new TrackingDirectoryWrapper(dir);
FieldInfos fieldInfos = null;
bool success = false;
try
{
Codec codec = Info.Info.Codec;
// reader could be null e.g. for a just merged segment (from
// IndexWriter.commitMergedDeletes).
SegmentReader reader = this.reader ?? new SegmentReader(Info, writer.Config.ReaderTermsIndexDivisor, IOContext.READ_ONCE);
try
{
// clone FieldInfos so that we can update their dvGen separately from
// the reader's infos and write them to a new fieldInfos_gen file
FieldInfos.Builder builder = new FieldInfos.Builder(writer.globalFieldNumberMap);
// cannot use builder.add(reader.getFieldInfos()) because it does not
// clone FI.attributes as well FI.dvGen
foreach (FieldInfo fi in reader.FieldInfos)
{
FieldInfo clone = builder.Add(fi);
// copy the stuff FieldInfos.Builder doesn't copy
if (fi.Attributes != null)
{
foreach (KeyValuePair <string, string> e in fi.Attributes)
{
clone.PutAttribute(e.Key, e.Value);
}
}
clone.DocValuesGen = fi.DocValuesGen;
}
// create new fields or update existing ones to have NumericDV type
foreach (string f in dvUpdates.numericDVUpdates.Keys)
{
builder.AddOrUpdate(f, NumericDocValuesField.TYPE);
}
// create new fields or update existing ones to have BinaryDV type
foreach (string f in dvUpdates.binaryDVUpdates.Keys)
{
builder.AddOrUpdate(f, BinaryDocValuesField.TYPE);
}
fieldInfos = builder.Finish();
long nextFieldInfosGen = Info.NextFieldInfosGen;
// LUCENENET specific: We created the segments names wrong in 4.8.0-beta00001 - 4.8.0-beta00015,
// so we added a switch to be able to read these indexes in later versions. This logic as well as an
// optimization on the first 100 segment values is implmeneted in SegmentInfos.SegmentNumberToString().
string segmentSuffix = SegmentInfos.SegmentNumberToString(nextFieldInfosGen);
SegmentWriteState state = new SegmentWriteState(null, trackingDir, Info.Info, fieldInfos, writer.Config.TermIndexInterval, null, IOContext.DEFAULT, segmentSuffix);
DocValuesFormat docValuesFormat = codec.DocValuesFormat;
DocValuesConsumer fieldsConsumer = docValuesFormat.FieldsConsumer(state);
bool fieldsConsumerSuccess = false;
try
{
// System.out.println("[" + Thread.currentThread().getName() + "] RLD.writeFieldUpdates: applying numeric updates; seg=" + info + " updates=" + numericFieldUpdates);
foreach (KeyValuePair <string, NumericDocValuesFieldUpdates> e in dvUpdates.numericDVUpdates)
{
string field = e.Key;
NumericDocValuesFieldUpdates fieldUpdates = e.Value;
FieldInfo fieldInfo = fieldInfos.FieldInfo(field);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(fieldInfo != null);
}
fieldInfo.DocValuesGen = nextFieldInfosGen;
// write the numeric updates to a new gen'd docvalues file
fieldsConsumer.AddNumericField(fieldInfo, GetInt64Enumerable(reader, field, fieldUpdates));
}
// System.out.println("[" + Thread.currentThread().getName() + "] RAU.writeFieldUpdates: applying binary updates; seg=" + info + " updates=" + dvUpdates.binaryDVUpdates);
foreach (KeyValuePair <string, BinaryDocValuesFieldUpdates> e in dvUpdates.binaryDVUpdates)
{
string field = e.Key;
BinaryDocValuesFieldUpdates dvFieldUpdates = e.Value;
FieldInfo fieldInfo = fieldInfos.FieldInfo(field);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(fieldInfo != null);
}
// System.out.println("[" + Thread.currentThread().getName() + "] RAU.writeFieldUpdates: applying binary updates; seg=" + info + " f=" + dvFieldUpdates + ", updates=" + dvFieldUpdates);
fieldInfo.DocValuesGen = nextFieldInfosGen;
// write the numeric updates to a new gen'd docvalues file
fieldsConsumer.AddBinaryField(fieldInfo, GetBytesRefEnumerable(reader, field, dvFieldUpdates));
}
codec.FieldInfosFormat.FieldInfosWriter.Write(trackingDir, Info.Info.Name, segmentSuffix, fieldInfos, IOContext.DEFAULT);
fieldsConsumerSuccess = true;
}
finally
{
if (fieldsConsumerSuccess)
{
fieldsConsumer.Dispose();
}
else
{
IOUtils.DisposeWhileHandlingException(fieldsConsumer);
}
}
}
finally
{
if (reader != this.reader)
{
// System.out.println("[" + Thread.currentThread().getName() + "] RLD.writeLiveDocs: closeReader " + reader);
reader.Dispose();
}
}
success = true;
}
finally
{
if (!success)
{
// Advance only the nextWriteDocValuesGen so that a 2nd
// attempt to write will write to a new file
Info.AdvanceNextWriteFieldInfosGen();
// Delete any partially created file(s):
foreach (string fileName in trackingDir.CreatedFiles)
{
try
{
dir.DeleteFile(fileName);
}
catch (Exception t) when(t.IsThrowable())
{
// Ignore so we throw only the first exc
}
}
}
}
Info.AdvanceFieldInfosGen();
// copy all the updates to mergingUpdates, so they can later be applied to the merged segment
if (isMerging)
{
foreach (KeyValuePair <string, NumericDocValuesFieldUpdates> e in dvUpdates.numericDVUpdates)
{
if (!mergingDVUpdates.TryGetValue(e.Key, out DocValuesFieldUpdates updates))
{
mergingDVUpdates[e.Key] = e.Value;
}
else
{
updates.Merge(e.Value);
}
}
foreach (KeyValuePair <string, BinaryDocValuesFieldUpdates> e in dvUpdates.binaryDVUpdates)
{
if (!mergingDVUpdates.TryGetValue(e.Key, out DocValuesFieldUpdates updates))
{
mergingDVUpdates[e.Key] = e.Value;
}
else
{
updates.Merge(e.Value);
}
}
}
// create a new map, keeping only the gens that are in use
IDictionary <long, ISet <string> > genUpdatesFiles = Info.UpdatesFiles;
IDictionary <long, ISet <string> > newGenUpdatesFiles = new Dictionary <long, ISet <string> >();
long fieldInfosGen = Info.FieldInfosGen;
foreach (FieldInfo fi in fieldInfos)
{
long dvGen = fi.DocValuesGen;
if (dvGen != -1 && !newGenUpdatesFiles.ContainsKey(dvGen))
{
if (dvGen == fieldInfosGen)
{
newGenUpdatesFiles[fieldInfosGen] = trackingDir.CreatedFiles;
}
else
{
newGenUpdatesFiles[dvGen] = genUpdatesFiles[dvGen];
}
}
}
Info.SetGenUpdatesFiles(newGenUpdatesFiles);
// wrote new files, should checkpoint()
writer.Checkpoint();
// if there is a reader open, reopen it to reflect the updates
if (reader != null)
{
SegmentReader newReader = new SegmentReader(Info, reader, liveDocs, Info.Info.DocCount - Info.DelCount - pendingDeleteCount);
bool reopened = false;
try
{
reader.DecRef();
reader = newReader;
reopened = true;
}
finally
{
if (!reopened)
{
newReader.DecRef();
}
}
}
}
finally
{
UninterruptableMonitor.Exit(this);
}
}
public virtual ApplyDeletesResult ApplyDeletesAndUpdates(IndexWriter.ReaderPool readerPool, IList <SegmentCommitInfo> infos)
{
UninterruptableMonitor.Enter(this);
try
{
long t0 = J2N.Time.NanoTime() / J2N.Time.MillisecondsPerNanosecond; // LUCENENET: Use NanoTime() rather than CurrentTimeMilliseconds() for more accurate/reliable results
if (infos.Count == 0)
{
return(new ApplyDeletesResult(false, nextGen++, null));
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(CheckDeleteStats());
}
if (!Any())
{
if (infoStream.IsEnabled("BD"))
{
infoStream.Message("BD", "applyDeletes: no deletes; skipping");
}
return(new ApplyDeletesResult(false, nextGen++, null));
}
if (infoStream.IsEnabled("BD"))
{
infoStream.Message("BD", "applyDeletes: infos=" + string.Format(J2N.Text.StringFormatter.InvariantCulture, "{0}", infos) + " packetCount=" + updates.Count);
}
long gen = nextGen++;
JCG.List <SegmentCommitInfo> infos2 = new JCG.List <SegmentCommitInfo>();
infos2.AddRange(infos);
infos2.Sort(sortSegInfoByDelGen);
CoalescedUpdates coalescedUpdates = null;
bool anyNewDeletes = false;
int infosIDX = infos2.Count - 1;
int delIDX = updates.Count - 1;
IList <SegmentCommitInfo> allDeleted = null;
while (infosIDX >= 0)
{
//System.out.println("BD: cycle delIDX=" + delIDX + " infoIDX=" + infosIDX);
FrozenBufferedUpdates packet = delIDX >= 0 ? updates[delIDX] : null;
SegmentCommitInfo info = infos2[infosIDX];
long segGen = info.BufferedDeletesGen;
if (packet != null && segGen < packet.DelGen)
{
// System.out.println(" coalesce");
if (coalescedUpdates is null)
{
coalescedUpdates = new CoalescedUpdates();
}
if (!packet.isSegmentPrivate)
{
/*
* Only coalesce if we are NOT on a segment private del packet: the segment private del packet
* must only applied to segments with the same delGen. Yet, if a segment is already deleted
* from the SI since it had no more documents remaining after some del packets younger than
* its segPrivate packet (higher delGen) have been applied, the segPrivate packet has not been
* removed.
*/
coalescedUpdates.Update(packet);
}
delIDX--;
}
else if (packet != null && segGen == packet.DelGen)
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(packet.isSegmentPrivate, "Packet and Segments deletegen can only match on a segment private del packet gen={0}", segGen);
}
//System.out.println(" eq");
// Lock order: IW -> BD -> RP
if (Debugging.AssertsEnabled)
{
Debugging.Assert(readerPool.InfoIsLive(info));
}
ReadersAndUpdates rld = readerPool.Get(info, true);
SegmentReader reader = rld.GetReader(IOContext.READ);
int delCount = 0;
bool segAllDeletes;
try
{
DocValuesFieldUpdates.Container dvUpdates = new DocValuesFieldUpdates.Container();
if (coalescedUpdates != null)
{
//System.out.println(" del coalesced");
delCount += (int)ApplyTermDeletes(coalescedUpdates.TermsIterable(), rld, reader);
delCount += (int)ApplyQueryDeletes(coalescedUpdates.QueriesIterable(), rld, reader);
ApplyDocValuesUpdates(coalescedUpdates.numericDVUpdates, rld, reader, dvUpdates);
ApplyDocValuesUpdates(coalescedUpdates.binaryDVUpdates, rld, reader, dvUpdates);
}
//System.out.println(" del exact");
// Don't delete by Term here; DocumentsWriterPerThread
// already did that on flush:
delCount += (int)ApplyQueryDeletes(packet.GetQueriesEnumerable(), rld, reader);
ApplyDocValuesUpdates(packet.numericDVUpdates, rld, reader, dvUpdates);
ApplyDocValuesUpdates(packet.binaryDVUpdates, rld, reader, dvUpdates);
if (dvUpdates.Any())
{
rld.WriteFieldUpdates(info.Info.Dir, dvUpdates);
}
int fullDelCount = rld.Info.DelCount + rld.PendingDeleteCount;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(fullDelCount <= rld.Info.Info.DocCount);
}
segAllDeletes = fullDelCount == rld.Info.Info.DocCount;
}
finally
{
rld.Release(reader);
readerPool.Release(rld);
}
anyNewDeletes |= delCount > 0;
if (segAllDeletes)
{
if (allDeleted is null)
{
allDeleted = new JCG.List <SegmentCommitInfo>();
}
allDeleted.Add(info);
}
if (infoStream.IsEnabled("BD"))
{
infoStream.Message("BD", "seg=" + info + " segGen=" + segGen + " segDeletes=[" + packet + "]; coalesced deletes=[" + (coalescedUpdates is null ? "null" : coalescedUpdates.ToString()) + "] newDelCount=" + delCount + (segAllDeletes ? " 100% deleted" : ""));
}
if (coalescedUpdates is null)
{
coalescedUpdates = new CoalescedUpdates();
}
/*
* Since we are on a segment private del packet we must not
* update the coalescedDeletes here! We can simply advance to the
* next packet and seginfo.
*/
delIDX--;
infosIDX--;
info.SetBufferedDeletesGen(gen);
}
else
{
//System.out.println(" gt");
if (coalescedUpdates != null)
{
// Lock order: IW -> BD -> RP
if (Debugging.AssertsEnabled)
{
Debugging.Assert(readerPool.InfoIsLive(info));
}
ReadersAndUpdates rld = readerPool.Get(info, true);
SegmentReader reader = rld.GetReader(IOContext.READ);
int delCount = 0;
bool segAllDeletes;
try
{
delCount += (int)ApplyTermDeletes(coalescedUpdates.TermsIterable(), rld, reader);
delCount += (int)ApplyQueryDeletes(coalescedUpdates.QueriesIterable(), rld, reader);
DocValuesFieldUpdates.Container dvUpdates = new DocValuesFieldUpdates.Container();
ApplyDocValuesUpdates(coalescedUpdates.numericDVUpdates, rld, reader, dvUpdates);
ApplyDocValuesUpdates(coalescedUpdates.binaryDVUpdates, rld, reader, dvUpdates);
if (dvUpdates.Any())
{
rld.WriteFieldUpdates(info.Info.Dir, dvUpdates);
}
int fullDelCount = rld.Info.DelCount + rld.PendingDeleteCount;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(fullDelCount <= rld.Info.Info.DocCount);
}
segAllDeletes = fullDelCount == rld.Info.Info.DocCount;
}
finally
{
rld.Release(reader);
readerPool.Release(rld);
}
anyNewDeletes |= delCount > 0;
if (segAllDeletes)
{
if (allDeleted is null)
{
allDeleted = new JCG.List <SegmentCommitInfo>();
}
allDeleted.Add(info);
}
if (infoStream.IsEnabled("BD"))
{
infoStream.Message("BD", "seg=" + info + " segGen=" + segGen + " coalesced deletes=[" + coalescedUpdates + "] newDelCount=" + delCount + (segAllDeletes ? " 100% deleted" : ""));
}
}
info.SetBufferedDeletesGen(gen);
infosIDX--;
}
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(CheckDeleteStats());
}
if (infoStream.IsEnabled("BD"))
{
infoStream.Message("BD", "applyDeletes took " + ((J2N.Time.NanoTime() / J2N.Time.MillisecondsPerNanosecond) - t0) + " msec"); // LUCENENET: Use NanoTime() rather than CurrentTimeMilliseconds() for more accurate/reliable results
}
// assert infos != segmentInfos || !any() : "infos=" + infos + " segmentInfos=" + segmentInfos + " any=" + any;
return(new ApplyDeletesResult(anyNewDeletes, gen, allDeleted));
}
finally
{
UninterruptableMonitor.Exit(this);
}
}
