public override void Dump()
{
Debugging.CurrentPrice(Instrument);
Debugging.Dump(Instrument, range: new Range(-100, 1), indicators: new[] { MA0, MA1 });
Debugging.Dump(new SnR(Instrument));
}
//public static void main( string[] args ) throws Exception {
// Analyzer analyzer = new WhitespaceAnalyzer(Version.LUCENE_CURRENT);
// QueryParser parser = new QueryParser(Version.LUCENE_CURRENT, "f", analyzer );
// Query query = parser.parse( "a x:b" );
// FieldQuery fieldQuery = new FieldQuery( query, true, false );
// Directory dir = new RAMDirectory();
// IndexWriter writer = new IndexWriter(dir, new IndexWriterConfig(Version.LUCENE_CURRENT, analyzer));
// Document doc = new Document();
// IndexableFieldType ft = new IndexableFieldType(TextField.TYPE_STORED);
// ft.setStoreTermVectors(true);
// ft.setStoreTermVectorOffsets(true);
// ft.setStoreTermVectorPositions(true);
// doc.add( new Field( "f", ft, "a a a b b c a b b c d e f" ) );
// doc.add( new Field( "f", ft, "b a b a f" ) );
// writer.addDocument( doc );
// writer.close();
// IndexReader reader = IndexReader.open(dir1);
// new FieldTermStack( reader, 0, "f", fieldQuery );
// reader.close();
//}
/// <summary>
/// a constructor.
/// </summary>
/// <param name="reader"><see cref="IndexReader"/> of the index</param>
/// <param name="docId">document id to be highlighted</param>
/// <param name="fieldName">field of the document to be highlighted</param>
/// <param name="fieldQuery"><see cref="FieldQuery"/> object</param>
/// <exception cref="IOException">If there is a low-level I/O error</exception>
public FieldTermStack(IndexReader reader, int docId, string fieldName, FieldQuery fieldQuery)
{
this.fieldName = fieldName;
ISet <string> termSet = fieldQuery.GetTermSet(fieldName);
// just return to make null snippet if un-matched fieldName specified when fieldMatch == true
if (termSet == null)
{
return;
}
Fields vectors = reader.GetTermVectors(docId);
if (vectors == null)
{
// null snippet
return;
}
Terms vector = vectors.GetTerms(fieldName);
if (vector == null)
{
// null snippet
return;
}
CharsRef spare = new CharsRef();
TermsEnum termsEnum = vector.GetEnumerator();
DocsAndPositionsEnum dpEnum = null;
BytesRef text;
int numDocs = reader.MaxDoc;
while (termsEnum.MoveNext())
{
text = termsEnum.Term;
UnicodeUtil.UTF8toUTF16(text, spare);
string term = spare.ToString();
if (!termSet.Contains(term))
{
continue;
}
dpEnum = termsEnum.DocsAndPositions(null, dpEnum);
if (dpEnum == null)
{
// null snippet
return;
}
dpEnum.NextDoc();
// For weight look here: http://lucene.apache.org/core/3_6_0/api/core/org/apache/lucene/search/DefaultSimilarity.html
float weight = (float)(Math.Log(numDocs / (double)(reader.DocFreq(new Term(fieldName, text)) + 1)) + 1.0);
int freq = dpEnum.Freq;
for (int i = 0; i < freq; i++)
{
int pos = dpEnum.NextPosition();
if (dpEnum.StartOffset < 0)
{
return; // no offsets, null snippet
}
termList.Add(new TermInfo(term, dpEnum.StartOffset, dpEnum.EndOffset, pos, weight));
}
}
// sort by position
CollectionUtil.TimSort(termList);
// now look for dups at the same position, linking them together
int currentPos = -1;
TermInfo previous = null;
TermInfo first = null;
for (int i = 0; i < termList.Count;)
{
TermInfo current = termList[i];
if (current.Position == currentPos)
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(previous != null);
}
previous.SetNext(current);
previous = current;
//iterator.Remove();
// LUCENENET NOTE: Remove, but don't advance the i position (since removing will advance to the next item)
termList.RemoveAt(i);
}
else
{
if (previous != null)
{
previous.SetNext(first);
}
previous = first = current;
currentPos = current.Position;
// LUCENENET NOTE: Only increment the position if we don't do a delete.
i++;
}
}
if (previous != null)
{
previous.SetNext(first);
}
}
internal override void AddTerm(int termID)
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(docState.TestPoint("FreqProxTermsWriterPerField.addTerm start"));
}
FreqProxPostingsArray postings = (FreqProxPostingsArray)termsHashPerField.postingsArray;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(!hasFreq || postings.termFreqs[termID] > 0);
}
if (!hasFreq)
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(postings.termFreqs == null);
}
if (docState.docID != postings.lastDocIDs[termID])
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(docState.docID > postings.lastDocIDs[termID]);
}
termsHashPerField.WriteVInt32(0, postings.lastDocCodes[termID]);
postings.lastDocCodes[termID] = docState.docID - postings.lastDocIDs[termID];
postings.lastDocIDs[termID] = docState.docID;
fieldState.UniqueTermCount++;
}
}
else if (docState.docID != postings.lastDocIDs[termID])
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(docState.docID > postings.lastDocIDs[termID], "id: {0} postings ID: {1} termID: {2}", docState.docID, postings.lastDocIDs[termID], termID);
}
// Term not yet seen in the current doc but previously
// seen in other doc(s) since the last flush
// Now that we know doc freq for previous doc,
// write it & lastDocCode
if (1 == postings.termFreqs[termID])
{
termsHashPerField.WriteVInt32(0, postings.lastDocCodes[termID] | 1);
}
else
{
termsHashPerField.WriteVInt32(0, postings.lastDocCodes[termID]);
termsHashPerField.WriteVInt32(0, postings.termFreqs[termID]);
}
postings.termFreqs[termID] = 1;
fieldState.MaxTermFrequency = Math.Max(1, fieldState.MaxTermFrequency);
postings.lastDocCodes[termID] = (docState.docID - postings.lastDocIDs[termID]) << 1;
postings.lastDocIDs[termID] = docState.docID;
if (hasProx)
{
WriteProx(termID, fieldState.Position);
if (hasOffsets)
{
postings.lastOffsets[termID] = 0;
WriteOffsets(termID, fieldState.Offset);
}
}
else
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(!hasOffsets);
}
}
fieldState.UniqueTermCount++;
}
else
{
fieldState.MaxTermFrequency = Math.Max(fieldState.MaxTermFrequency, ++postings.termFreqs[termID]);
if (hasProx)
{
WriteProx(termID, fieldState.Position - postings.lastPositions[termID]);
}
if (hasOffsets)
{
WriteOffsets(termID, fieldState.Offset);
}
}
}
private void BtnCheckSeed_Click(object sender, EventArgs e)
{
if (!chkShowObtainable.Checked && !chkShowUnobtainable.Checked)
{
LBResult.Items.Clear();
return;
}
var logicCopy = Utility.CloneTrackerInstance(CheckerInstance);
foreach (var entry in logicCopy.Logic)
{
entry.Available = false;
entry.Checked = false;
entry.Aquired = false;
if (entry.SpoilerRandom > -1)
{
entry.RandomizedItem = entry.SpoilerRandom;
} //Make the items randomized item its spoiler item, just for consitancy sake
else if (entry.RandomizedItem > -1)
{
entry.SpoilerRandom = entry.RandomizedItem;
} //If the item doesn't have spoiler data, but does have a randomized item. set it's spoiler data to the randomized item
else if (entry.Unrandomized(2))
{
entry.SpoilerRandom = entry.ID; entry.RandomizedItem = entry.ID;
} //If the item doesn't have spoiler data or a randomized item and is unrandomized (manual), set it's spoiler item to it's self
}
LBResult.Items.Clear();
List <int> Ignored = new List <int>();
foreach (var item in LBIgnoredChecks.Items)
{
Ignored.Add((item as LogicObjects.ListItem).PathID);
}
var GameClearEntry = logicCopy.Logic.Find(x => x.DictionaryName == "MMRTGameClear");
if (GameClearEntry != null)
{
GameClearEntry.ItemName = (LogicObjects.MainTrackerInstance.IsMM()) ? "Defeat Majora" : "Beat the Game";
}
else if (LogicObjects.MainTrackerInstance.IsMM())
{
int GameClearID = PlaythroughGenerator.GetGameClearEntry(logicCopy.Logic, LogicObjects.MainTrackerInstance.IsEntranceRando());
logicCopy.Logic[GameClearID].ItemName = "Defeat Majora";
}
CheckSeed(logicCopy, true, Ignored);
List <string> obtainable = new List <string>();
List <string> unobtainable = new List <string>();
foreach (var item in LBNeededItems.Items)
{
bool Spoil = false;
var ListItem = item as LogicObjects.ListItem;
var iteminLogic = logicCopy.Logic[ListItem.PathID];
string ItemName = iteminLogic.ItemName ?? iteminLogic.DictionaryName;
var ItemsLocation = iteminLogic.GetItemsNewLocation(logicCopy.Logic);
string LocationFoundAt = (ItemsLocation != null) ? ItemsLocation.LocationName ?? ItemsLocation.DictionaryName : "";
string DisplayName = (Spoil) ? ItemName + ": " + LocationFoundAt : ItemName;
Debugging.Log(logicCopy.Logic[ListItem.PathID].DictionaryName + " " + logicCopy.Logic[ListItem.PathID].Aquired);
if (logicCopy.Logic[ListItem.PathID].Aquired)
{
obtainable.Add(DisplayName);
}
else
{
unobtainable.Add(DisplayName);
}
}
if (unobtainable.Count > 0 && chkShowUnobtainable.Checked)
{
LBResult.Items.Add("Unobtainable ==============================");
foreach (var i in unobtainable)
{
LBResult.Items.Add(i);
}
}
if (obtainable.Count > 0 && chkShowObtainable.Checked)
{
LBResult.Items.Add("Obtainable ==============================");
foreach (var i in obtainable)
{
LBResult.Items.Add(i);
}
}
}
/// <summary>
/// Safe (but, slowish) default method to write every
/// vector field in the document.
/// </summary>
protected void AddAllDocVectors(Fields vectors, MergeState mergeState)
{
if (vectors == null)
{
StartDocument(0);
FinishDocument();
return;
}
int numFields = vectors.Count;
if (numFields == -1)
{
// count manually! TODO: Maybe enforce that Fields.size() returns something valid?
numFields = 0;
//for (IEnumerator<string> it = vectors.Iterator(); it.hasNext();)
foreach (string it in vectors)
{
numFields++;
}
}
StartDocument(numFields);
string lastFieldName = null;
TermsEnum termsEnum = null;
DocsAndPositionsEnum docsAndPositionsEnum = null;
int fieldCount = 0;
foreach (string fieldName in vectors)
{
fieldCount++;
FieldInfo fieldInfo = mergeState.FieldInfos.FieldInfo(fieldName);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(lastFieldName == null || fieldName.CompareToOrdinal(lastFieldName) > 0, "lastFieldName={0} fieldName={1}", lastFieldName, fieldName);
}
lastFieldName = fieldName;
Terms terms = vectors.GetTerms(fieldName);
if (terms == null)
{
// FieldsEnum shouldn't lie...
continue;
}
bool hasPositions = terms.HasPositions;
bool hasOffsets = terms.HasOffsets;
bool hasPayloads = terms.HasPayloads;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(!hasPayloads || hasPositions);
}
int numTerms = (int)terms.Count;
if (numTerms == -1)
{
// count manually. It is stupid, but needed, as Terms.size() is not a mandatory statistics function
numTerms = 0;
termsEnum = terms.GetEnumerator(termsEnum);
while (termsEnum.MoveNext())
{
numTerms++;
}
}
StartField(fieldInfo, numTerms, hasPositions, hasOffsets, hasPayloads);
termsEnum = terms.GetEnumerator(termsEnum);
int termCount = 0;
while (termsEnum.MoveNext())
{
termCount++;
int freq = (int)termsEnum.TotalTermFreq;
StartTerm(termsEnum.Term, freq);
if (hasPositions || hasOffsets)
{
docsAndPositionsEnum = termsEnum.DocsAndPositions(null, docsAndPositionsEnum);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(docsAndPositionsEnum != null);
}
int docID = docsAndPositionsEnum.NextDoc();
if (Debugging.AssertsEnabled)
{
Debugging.Assert(docID != DocIdSetIterator.NO_MORE_DOCS);
Debugging.Assert(docsAndPositionsEnum.Freq == freq);
}
for (int posUpto = 0; posUpto < freq; posUpto++)
{
int pos = docsAndPositionsEnum.NextPosition();
int startOffset = docsAndPositionsEnum.StartOffset;
int endOffset = docsAndPositionsEnum.EndOffset;
BytesRef payload = docsAndPositionsEnum.GetPayload();
if (Debugging.AssertsEnabled)
{
Debugging.Assert(!hasPositions || pos >= 0);
}
AddPosition(pos, startOffset, endOffset, payload);
}
}
FinishTerm();
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(termCount == numTerms);
}
FinishField();
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(fieldCount == numFields);
}
FinishDocument();
}
public Lucene3xTermVectorsReader(Directory d, SegmentInfo si, FieldInfos fieldInfos, IOContext context)
{
string segment = Lucene3xSegmentInfoFormat.GetDocStoreSegment(si);
int docStoreOffset = Lucene3xSegmentInfoFormat.GetDocStoreOffset(si);
int size = si.DocCount;
bool success = false;
try
{
if (docStoreOffset != -1 && Lucene3xSegmentInfoFormat.GetDocStoreIsCompoundFile(si))
{
d = storeCFSReader = new CompoundFileDirectory(si.Dir, IndexFileNames.SegmentFileName(segment, "", Lucene3xCodec.COMPOUND_FILE_STORE_EXTENSION), context, false);
}
else
{
storeCFSReader = null;
}
string idxName = IndexFileNames.SegmentFileName(segment, "", VECTORS_INDEX_EXTENSION);
tvx = d.OpenInput(idxName, context);
format = CheckValidFormat(tvx);
string fn = IndexFileNames.SegmentFileName(segment, "", VECTORS_DOCUMENTS_EXTENSION);
tvd = d.OpenInput(fn, context);
int tvdFormat = CheckValidFormat(tvd);
fn = IndexFileNames.SegmentFileName(segment, "", VECTORS_FIELDS_EXTENSION);
tvf = d.OpenInput(fn, context);
int tvfFormat = CheckValidFormat(tvf);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(format == tvdFormat);
Debugging.Assert(format == tvfFormat);
}
numTotalDocs = (int)(tvx.Length >> 4);
if (-1 == docStoreOffset)
{
this.docStoreOffset = 0;
this.size = numTotalDocs;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(size == 0 || numTotalDocs == size);
}
}
else
{
this.docStoreOffset = docStoreOffset;
this.size = size;
// Verify the file is long enough to hold all of our
// docs
if (Debugging.AssertsEnabled)
{
Debugging.Assert(numTotalDocs >= size + docStoreOffset, "numTotalDocs={0} size={1} docStoreOffset={2}", numTotalDocs, size, docStoreOffset);
}
}
this.fieldInfos = fieldInfos;
success = true;
}
finally
{
// With lock-less commits, it's entirely possible (and
// fine) to hit a FileNotFound exception above. In
// this case, we want to explicitly close any subset
// of things that were opened so that we don't have to
// wait for a GC to do so.
if (!success)
{
try
{
Dispose();
} // keep our original exception
catch (Exception t) when(t.IsThrowable())
{
}
}
}
}
internal Cell?nextCell; //see getLeafDocs
/// <remarks>This is the primary algorithm; recursive. Returns null if finds none.</remarks>
/// <exception cref="IOException"></exception>
internal SmallDocSet?Visit(Cell cell, IBits acceptContains)
{
if (m_termsEnum is null)
{
//signals all done
return(null);
}
ContainsPrefixTreeFilter outerInstance = (ContainsPrefixTreeFilter)base.m_filter;
//Leaf docs match all query shape
SmallDocSet?leafDocs = GetLeafDocs(cell, acceptContains);
// Get the AND of all child results (into combinedSubResults)
SmallDocSet?combinedSubResults = null;
// Optimization: use null subCellsFilter when we know cell is within the query shape.
IShape?subCellsFilter = outerInstance.m_queryShape;
if (cell.Level != 0 && ((cell.ShapeRel == SpatialRelation.None || cell.ShapeRel == SpatialRelation.Within)))
{
subCellsFilter = null;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(cell.Shape.Relate(outerInstance.m_queryShape) == SpatialRelation.Within);
}
}
ICollection <Cell> subCells = cell.GetSubCells(subCellsFilter);
foreach (Cell subCell in subCells)
{
if (!SeekExact(subCell))
{
combinedSubResults = null;
}
else if (subCell.Level == outerInstance.m_detailLevel)
{
combinedSubResults = GetDocs(subCell, acceptContains);
}
else if (!outerInstance.m_multiOverlappingIndexedShapes &&
subCell.ShapeRel == SpatialRelation.Within)
{
combinedSubResults = GetLeafDocs(subCell, acceptContains); //recursion
}
else
{
combinedSubResults = Visit(subCell, acceptContains);
}
if (combinedSubResults is null)
{
break;
}
acceptContains = combinedSubResults;//has the 'AND' effect on next iteration
}
// Result: OR the leaf docs with AND of all child results
if (combinedSubResults != null)
{
if (leafDocs is null)
{
return(combinedSubResults);
}
return(leafDocs.Union(combinedSubResults));//union is 'or'
}
return(leafDocs);
}
public override SeekStatus SeekCeil(BytesRef term)
{
queue.Clear();
numTop = 0;
lastSeekExact = false;
bool seekOpt = false;
if (lastSeek != null && termComp.Compare(lastSeek, term) <= 0)
{
seekOpt = true;
}
lastSeekScratch.CopyBytes(term);
lastSeek = lastSeekScratch;
for (int i = 0; i < numSubs; i++)
{
SeekStatus status;
// LUCENE-2130: if we had just seek'd already, prior
// to this seek, and the new seek term is after the
// previous one, don't try to re-seek this sub if its
// current term is already beyond this new seek term.
// Doing so is a waste because this sub will simply
// seek to the same spot.
if (seekOpt)
{
BytesRef curTerm = currentSubs[i].Current;
if (curTerm != null)
{
int cmp = termComp.Compare(term, curTerm);
if (cmp == 0)
{
status = SeekStatus.FOUND;
}
else if (cmp < 0)
{
status = SeekStatus.NOT_FOUND;
}
else
{
status = currentSubs[i].Terms.SeekCeil(term);
}
}
else
{
status = SeekStatus.END;
}
}
else
{
status = currentSubs[i].Terms.SeekCeil(term);
}
if (status == SeekStatus.FOUND)
{
top[numTop++] = currentSubs[i];
current = currentSubs[i].Current = currentSubs[i].Terms.Term;
}
else
{
if (status == SeekStatus.NOT_FOUND)
{
currentSubs[i].Current = currentSubs[i].Terms.Term;
if (Debugging.AssertsEnabled) Debugging.Assert(currentSubs[i].Current != null);
queue.Add(currentSubs[i]);
}
else
{
// enum exhausted
currentSubs[i].Current = null;
}
}
}
if (numTop > 0)
{
// at least one sub had exact match to the requested term
return SeekStatus.FOUND;
}
else if (queue.Count > 0)
{
// no sub had exact match, but at least one sub found
// a term after the requested term -- advance to that
// next term:
PullTop();
return SeekStatus.NOT_FOUND;
}
else
{
return SeekStatus.END;
}
}
public override DocsEnum Docs(IBits liveDocs, DocsEnum reuse, DocsFlags flags)
{
// Can only reuse if incoming enum is also a MultiDocsEnum
// ... and was previously created w/ this MultiTermsEnum:
if (reuse is null || !(reuse is MultiDocsEnum docsEnum) || !docsEnum.CanReuse(this))
docsEnum = new MultiDocsEnum(this, subs.Length);
int upto = 0;
for (int i = 0; i < numTop; i++)
{
TermsEnumWithSlice entry = top[i];
IBits b;
if (liveDocs is MultiBits multiLiveDocs)
{
// optimize for common case: requested skip docs is a
// congruent sub-slice of MultiBits: in this case, we
// just pull the liveDocs from the sub reader, rather
// than making the inefficient
// Slice(Multi(sub-readers)):
MultiBits.SubResult sub = multiLiveDocs.GetMatchingSub(entry.SubSlice);
if (sub.Matches)
{
b = sub.Result;
}
else
{
// custom case: requested skip docs is foreign:
// must slice it on every access
b = new BitsSlice(liveDocs, entry.SubSlice);
}
}
else if (liveDocs != null)
{
b = new BitsSlice(liveDocs, entry.SubSlice);
}
else
{
// no deletions
b = null;
}
if (Debugging.AssertsEnabled) Debugging.Assert(entry.Index < docsEnum.subDocsEnum.Length, "{0} vs {1}; {2}", entry.Index, docsEnum.subDocsEnum.Length, subs.Length);
DocsEnum subDocsEnum = entry.Terms.Docs(b, docsEnum.subDocsEnum[entry.Index], flags);
if (subDocsEnum != null)
{
docsEnum.subDocsEnum[entry.Index] = subDocsEnum;
subDocs[upto].DocsEnum = subDocsEnum;
subDocs[upto].Slice = entry.SubSlice;
upto++;
}
else
{
// should this be an error?
if (Debugging.AssertsEnabled) Debugging.Assert(false, "One of our subs cannot provide a docsenum");
}
}
if (upto == 0)
{
return null;
}
else
{
return docsEnum.Reset(subDocs, upto);
}
}
internal long currentEntryIndex; // also indicates how many entries in the index are valid.
/// <summary>
/// Construct an Elias-Fano encoder.
/// After construction, call <see cref="EncodeNext(long)"/> <paramref name="numValues"/> times to encode
/// a non decreasing sequence of non negative numbers.
/// </summary>
/// <param name="numValues"> The number of values that is to be encoded. </param>
/// <param name="upperBound"> At least the highest value that will be encoded.
/// For space efficiency this should not exceed the power of two that equals
/// or is the first higher than the actual maximum.
/// <para/>When <c>numValues >= (upperBound/3)</c>
/// a <see cref="FixedBitSet"/> will take less space. </param>
/// <param name="indexInterval"> The number of high zero bits for which a single index entry is built.
/// The index will have at most <c>2 * numValues / indexInterval</c> entries
/// and each index entry will use at most <c>Ceil(Log2(3 * numValues))</c> bits,
/// see <see cref="EliasFanoEncoder"/>. </param>
/// <exception cref="ArgumentException"> when:
/// <list type="bullet">
/// <item><description><paramref name="numValues"/> is negative, or</description></item>
/// <item><description><paramref name="numValues"/> is non negative and <paramref name="upperBound"/> is negative, or</description></item>
/// <item><description>the low bits do not fit in a <c>long[]</c>:
/// <c>(L * numValues / 64) > System.Int32.MaxValue</c>, or</description></item>
/// <item><description>the high bits do not fit in a <c>long[]</c>:
/// <c>(2 * numValues / 64) > System.Int32.MaxValue</c>, or</description></item>
/// <item><description><c>indexInterval < 2</c>,</description></item>
/// <item><description>the index bits do not fit in a <c>long[]</c>:
/// <c>(numValues / indexInterval * ceil(2log(3 * numValues)) / 64) > System.Int32.MaxValue</c>.</description></item>
/// </list> </exception>
public EliasFanoEncoder(long numValues, long upperBound, long indexInterval)
{
if (numValues < 0L)
{
throw new ArgumentException("numValues should not be negative: " + numValues);
}
this.numValues = numValues;
if ((numValues > 0L) && (upperBound < 0L))
{
throw new ArgumentException("upperBound should not be negative: " + upperBound + " when numValues > 0");
}
this.upperBound = numValues > 0 ? upperBound : -1L; // if there is no value, -1 is the best upper bound
int nLowBits = 0;
if (this.numValues > 0) // nLowBits = max(0; floor(2log(upperBound/numValues)))
{
long lowBitsFac = this.upperBound / this.numValues;
if (lowBitsFac > 0)
{
nLowBits = 63 - lowBitsFac.LeadingZeroCount(); // see Long.numberOfLeadingZeros javadocs
}
}
this.numLowBits = nLowBits;
this.lowerBitsMask = (long)(unchecked ((ulong)long.MaxValue) >> (sizeof(long) * 8 - 1 - this.numLowBits));
long numLongsForLowBits = NumInt64sForBits(numValues * numLowBits);
if (numLongsForLowBits > int.MaxValue)
{
throw new ArgumentException("numLongsForLowBits too large to index a long array: " + numLongsForLowBits);
}
this.lowerLongs = new long[(int)numLongsForLowBits];
long numHighBitsClear = (long)((ulong)((this.upperBound > 0) ? this.upperBound : 0) >> this.numLowBits);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(numHighBitsClear <= (2 * this.numValues));
}
long numHighBitsSet = this.numValues;
long numLongsForHighBits = NumInt64sForBits(numHighBitsClear + numHighBitsSet);
if (numLongsForHighBits > int.MaxValue)
{
throw new ArgumentException("numLongsForHighBits too large to index a long array: " + numLongsForHighBits);
}
this.upperLongs = new long[(int)numLongsForHighBits];
if (indexInterval < 2)
{
throw new ArgumentException("indexInterval should at least 2: " + indexInterval);
}
// For the index:
long maxHighValue = (long)((ulong)upperBound >> this.numLowBits);
long nIndexEntries = maxHighValue / indexInterval; // no zero value index entry
this.numIndexEntries = (nIndexEntries >= 0) ? nIndexEntries : 0;
long maxIndexEntry = maxHighValue + numValues - 1; // clear upper bits, set upper bits, start at zero
this.nIndexEntryBits = (maxIndexEntry <= 0) ? 0 : (64 - maxIndexEntry.LeadingZeroCount());
long numLongsForIndexBits = NumInt64sForBits(numIndexEntries * nIndexEntryBits);
if (numLongsForIndexBits > int.MaxValue)
{
throw new ArgumentException("numLongsForIndexBits too large to index a long array: " + numLongsForIndexBits);
}
this.upperZeroBitPositionIndex = new long[(int)numLongsForIndexBits];
this.currentEntryIndex = 0;
this.indexInterval = indexInterval;
}
public override bool SeekExact(BytesRef term)
{
queue.Clear();
numTop = 0;
bool seekOpt = false;
if (lastSeek != null && termComp.Compare(lastSeek, term) <= 0)
{
seekOpt = true;
}
lastSeek = null;
lastSeekExact = true;
for (int i = 0; i < numSubs; i++)
{
bool status;
// LUCENE-2130: if we had just seek'd already, prior
// to this seek, and the new seek term is after the
// previous one, don't try to re-seek this sub if its
// current term is already beyond this new seek term.
// Doing so is a waste because this sub will simply
// seek to the same spot.
if (seekOpt)
{
BytesRef curTerm = currentSubs[i].Current;
if (curTerm != null)
{
int cmp = termComp.Compare(term, curTerm);
if (cmp == 0)
{
status = true;
}
else if (cmp < 0)
{
status = false;
}
else
{
status = currentSubs[i].Terms.SeekExact(term);
}
}
else
{
status = false;
}
}
else
{
status = currentSubs[i].Terms.SeekExact(term);
}
if (status)
{
top[numTop++] = currentSubs[i];
current = currentSubs[i].Current = currentSubs[i].Terms.Term;
if (Debugging.AssertsEnabled) Debugging.Assert(term.Equals(currentSubs[i].Current));
}
}
// if at least one sub had exact match to the requested
// term then we found match
return numTop > 0;
}
internal void FinishDocument()
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(docState.TestPoint("TermVectorsTermsWriterPerField.finish start"));
}
int numPostings = termsHashPerField.bytesHash.Count;
BytesRef flushTerm = termsWriter.flushTerm;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(numPostings >= 0);
}
if (numPostings > maxNumPostings)
{
maxNumPostings = numPostings;
}
// this is called once, after inverting all occurrences
// of a given field in the doc. At this point we flush
// our hash into the DocWriter.
if (Debugging.AssertsEnabled)
{
Debugging.Assert(termsWriter.VectorFieldsInOrder(fieldInfo));
}
TermVectorsPostingsArray postings = (TermVectorsPostingsArray)termsHashPerField.postingsArray;
TermVectorsWriter tv = termsWriter.writer;
int[] termIDs = termsHashPerField.SortPostings(tv.Comparer);
tv.StartField(fieldInfo, numPostings, doVectorPositions, doVectorOffsets, hasPayloads);
ByteSliceReader posReader = doVectorPositions ? termsWriter.vectorSliceReaderPos : null;
ByteSliceReader offReader = doVectorOffsets ? termsWriter.vectorSliceReaderOff : null;
ByteBlockPool termBytePool = termsHashPerField.termBytePool;
for (int j = 0; j < numPostings; j++)
{
int termID = termIDs[j];
int freq = postings.freqs[termID];
// Get BytesRef
termBytePool.SetBytesRef(flushTerm, postings.textStarts[termID]);
tv.StartTerm(flushTerm, freq);
if (doVectorPositions || doVectorOffsets)
{
if (posReader != null)
{
termsHashPerField.InitReader(posReader, termID, 0);
}
if (offReader != null)
{
termsHashPerField.InitReader(offReader, termID, 1);
}
tv.AddProx(freq, posReader, offReader);
}
tv.FinishTerm();
}
tv.FinishField();
termsHashPerField.Reset();
fieldInfo.SetStoreTermVectors();
}
/// <summary>
/// Sole constructor. </summary>
public CompressingTermVectorsReader(Directory d, SegmentInfo si, string segmentSuffix, FieldInfos fn, IOContext context, string formatName, CompressionMode compressionMode)
{
this.compressionMode = compressionMode;
string segment = si.Name;
bool success = false;
fieldInfos = fn;
numDocs = si.DocCount;
ChecksumIndexInput indexStream = null;
try
{
// Load the index into memory
string indexStreamFN = IndexFileNames.SegmentFileName(segment, segmentSuffix, CompressingTermVectorsWriter.VECTORS_INDEX_EXTENSION);
indexStream = d.OpenChecksumInput(indexStreamFN, context);
string codecNameIdx = formatName + CompressingTermVectorsWriter.CODEC_SFX_IDX;
version = CodecUtil.CheckHeader(indexStream, codecNameIdx, CompressingTermVectorsWriter.VERSION_START, CompressingTermVectorsWriter.VERSION_CURRENT);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(CodecUtil.HeaderLength(codecNameIdx) == indexStream.GetFilePointer());
}
indexReader = new CompressingStoredFieldsIndexReader(indexStream, si);
if (version >= CompressingTermVectorsWriter.VERSION_CHECKSUM)
{
indexStream.ReadVInt64(); // the end of the data file
CodecUtil.CheckFooter(indexStream);
}
else
{
#pragma warning disable 612, 618
CodecUtil.CheckEOF(indexStream);
#pragma warning restore 612, 618
}
indexStream.Dispose();
indexStream = null;
// Open the data file and read metadata
string vectorsStreamFN = IndexFileNames.SegmentFileName(segment, segmentSuffix, CompressingTermVectorsWriter.VECTORS_EXTENSION);
vectorsStream = d.OpenInput(vectorsStreamFN, context);
string codecNameDat = formatName + CompressingTermVectorsWriter.CODEC_SFX_DAT;
int version2 = CodecUtil.CheckHeader(vectorsStream, codecNameDat, CompressingTermVectorsWriter.VERSION_START, CompressingTermVectorsWriter.VERSION_CURRENT);
if (version != version2)
{
throw new Exception("Version mismatch between stored fields index and data: " + version + " != " + version2);
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(CodecUtil.HeaderLength(codecNameDat) == vectorsStream.GetFilePointer());
}
packedIntsVersion = vectorsStream.ReadVInt32();
chunkSize = vectorsStream.ReadVInt32();
decompressor = compressionMode.NewDecompressor();
this.reader = new BlockPackedReaderIterator(vectorsStream, packedIntsVersion, CompressingTermVectorsWriter.BLOCK_SIZE, 0);
success = true;
}
finally
{
if (!success)
{
IOUtils.DisposeWhileHandlingException(this, indexStream);
}
}
}
public override Fields Get(int doc)
{
EnsureOpen();
// seek to the right place
{
long startPointer = indexReader.GetStartPointer(doc);
vectorsStream.Seek(startPointer);
}
// decode
// - docBase: first doc ID of the chunk
// - chunkDocs: number of docs of the chunk
int docBase = vectorsStream.ReadVInt32();
int chunkDocs = vectorsStream.ReadVInt32();
if (doc < docBase || doc >= docBase + chunkDocs || docBase + chunkDocs > numDocs)
{
throw new CorruptIndexException("docBase=" + docBase + ",chunkDocs=" + chunkDocs + ",doc=" + doc + " (resource=" + vectorsStream + ")");
}
int skip; // number of fields to skip
int numFields; // number of fields of the document we're looking for
int totalFields; // total number of fields of the chunk (sum for all docs)
if (chunkDocs == 1)
{
skip = 0;
numFields = totalFields = vectorsStream.ReadVInt32();
}
else
{
reader.Reset(vectorsStream, chunkDocs);
int sum = 0;
for (int i = docBase; i < doc; ++i)
{
sum += (int)reader.Next();
}
skip = sum;
numFields = (int)reader.Next();
sum += numFields;
for (int i = doc + 1; i < docBase + chunkDocs; ++i)
{
sum += (int)reader.Next();
}
totalFields = sum;
}
if (numFields == 0)
{
// no vectors
return(null);
}
// read field numbers that have term vectors
int[] fieldNums;
{
int token = vectorsStream.ReadByte() & 0xFF;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(token != 0); // means no term vectors, cannot happen since we checked for numFields == 0
}
int bitsPerFieldNum = token & 0x1F;
int totalDistinctFields = (int)((uint)token >> 5);
if (totalDistinctFields == 0x07)
{
totalDistinctFields += vectorsStream.ReadVInt32();
}
++totalDistinctFields;
PackedInt32s.IReaderIterator it = PackedInt32s.GetReaderIteratorNoHeader(vectorsStream, PackedInt32s.Format.PACKED, packedIntsVersion, totalDistinctFields, bitsPerFieldNum, 1);
fieldNums = new int[totalDistinctFields];
for (int i = 0; i < totalDistinctFields; ++i)
{
fieldNums[i] = (int)it.Next();
}
}
// read field numbers and flags
int[] fieldNumOffs = new int[numFields];
PackedInt32s.Reader flags;
{
int bitsPerOff = PackedInt32s.BitsRequired(fieldNums.Length - 1);
PackedInt32s.Reader allFieldNumOffs = PackedInt32s.GetReaderNoHeader(vectorsStream, PackedInt32s.Format.PACKED, packedIntsVersion, totalFields, bitsPerOff);
switch (vectorsStream.ReadVInt32())
{
case 0:
PackedInt32s.Reader fieldFlags = PackedInt32s.GetReaderNoHeader(vectorsStream, PackedInt32s.Format.PACKED, packedIntsVersion, fieldNums.Length, CompressingTermVectorsWriter.FLAGS_BITS);
PackedInt32s.Mutable f = PackedInt32s.GetMutable(totalFields, CompressingTermVectorsWriter.FLAGS_BITS, PackedInt32s.COMPACT);
for (int i = 0; i < totalFields; ++i)
{
int fieldNumOff = (int)allFieldNumOffs.Get(i);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(fieldNumOff >= 0 && fieldNumOff < fieldNums.Length);
}
int fgs = (int)fieldFlags.Get(fieldNumOff);
f.Set(i, fgs);
}
flags = f;
break;
case 1:
flags = PackedInt32s.GetReaderNoHeader(vectorsStream, PackedInt32s.Format.PACKED, packedIntsVersion, totalFields, CompressingTermVectorsWriter.FLAGS_BITS);
break;
default:
throw new Exception();
}
for (int i = 0; i < numFields; ++i)
{
fieldNumOffs[i] = (int)allFieldNumOffs.Get(skip + i);
}
}
// number of terms per field for all fields
PackedInt32s.Reader numTerms;
int totalTerms;
{
int bitsRequired = vectorsStream.ReadVInt32();
numTerms = PackedInt32s.GetReaderNoHeader(vectorsStream, PackedInt32s.Format.PACKED, packedIntsVersion, totalFields, bitsRequired);
int sum = 0;
for (int i = 0; i < totalFields; ++i)
{
sum += (int)numTerms.Get(i);
}
totalTerms = sum;
}
// term lengths
int docOff = 0, docLen = 0, totalLen;
int[] fieldLengths = new int[numFields];
int[][] prefixLengths = new int[numFields][];
int[][] suffixLengths = new int[numFields][];
{
reader.Reset(vectorsStream, totalTerms);
// skip
int toSkip = 0;
for (int i = 0; i < skip; ++i)
{
toSkip += (int)numTerms.Get(i);
}
reader.Skip(toSkip);
// read prefix lengths
for (int i = 0; i < numFields; ++i)
{
int termCount = (int)numTerms.Get(skip + i);
int[] fieldPrefixLengths = new int[termCount];
prefixLengths[i] = fieldPrefixLengths;
for (int j = 0; j < termCount;)
{
Int64sRef next = reader.Next(termCount - j);
for (int k = 0; k < next.Length; ++k)
{
fieldPrefixLengths[j++] = (int)next.Int64s[next.Offset + k];
}
}
}
reader.Skip(totalTerms - reader.Ord);
reader.Reset(vectorsStream, totalTerms);
// skip
toSkip = 0;
for (int i = 0; i < skip; ++i)
{
for (int j = 0; j < numTerms.Get(i); ++j)
{
docOff += (int)reader.Next();
}
}
for (int i = 0; i < numFields; ++i)
{
int termCount = (int)numTerms.Get(skip + i);
int[] fieldSuffixLengths = new int[termCount];
suffixLengths[i] = fieldSuffixLengths;
for (int j = 0; j < termCount;)
{
Int64sRef next = reader.Next(termCount - j);
for (int k = 0; k < next.Length; ++k)
{
fieldSuffixLengths[j++] = (int)next.Int64s[next.Offset + k];
}
}
fieldLengths[i] = Sum(suffixLengths[i]);
docLen += fieldLengths[i];
}
totalLen = docOff + docLen;
for (int i = skip + numFields; i < totalFields; ++i)
{
for (int j = 0; j < numTerms.Get(i); ++j)
{
totalLen += (int)reader.Next();
}
}
}
// term freqs
int[] termFreqs = new int[totalTerms];
{
reader.Reset(vectorsStream, totalTerms);
for (int i = 0; i < totalTerms;)
{
Int64sRef next = reader.Next(totalTerms - i);
for (int k = 0; k < next.Length; ++k)
{
termFreqs[i++] = 1 + (int)next.Int64s[next.Offset + k];
}
}
}
// total number of positions, offsets and payloads
int totalPositions = 0, totalOffsets = 0, totalPayloads = 0;
for (int i = 0, termIndex = 0; i < totalFields; ++i)
{
int f = (int)flags.Get(i);
int termCount = (int)numTerms.Get(i);
for (int j = 0; j < termCount; ++j)
{
int freq = termFreqs[termIndex++];
if ((f & CompressingTermVectorsWriter.POSITIONS) != 0)
{
totalPositions += freq;
}
if ((f & CompressingTermVectorsWriter.OFFSETS) != 0)
{
totalOffsets += freq;
}
if ((f & CompressingTermVectorsWriter.PAYLOADS) != 0)
{
totalPayloads += freq;
}
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(i != totalFields - 1 || termIndex == totalTerms, () => termIndex + " " + totalTerms);
}
}
int[][] positionIndex = PositionIndex(skip, numFields, numTerms, termFreqs);
int[][] positions, startOffsets, lengths;
if (totalPositions > 0)
{
positions = ReadPositions(skip, numFields, flags, numTerms, termFreqs, CompressingTermVectorsWriter.POSITIONS, totalPositions, positionIndex);
}
else
{
positions = new int[numFields][];
}
if (totalOffsets > 0)
{
// average number of chars per term
float[] charsPerTerm = new float[fieldNums.Length];
for (int i = 0; i < charsPerTerm.Length; ++i)
{
charsPerTerm[i] = J2N.BitConversion.Int32BitsToSingle(vectorsStream.ReadInt32());
}
startOffsets = ReadPositions(skip, numFields, flags, numTerms, termFreqs, CompressingTermVectorsWriter.OFFSETS, totalOffsets, positionIndex);
lengths = ReadPositions(skip, numFields, flags, numTerms, termFreqs, CompressingTermVectorsWriter.OFFSETS, totalOffsets, positionIndex);
for (int i = 0; i < numFields; ++i)
{
int[] fStartOffsets = startOffsets[i];
int[] fPositions = positions[i];
// patch offsets from positions
if (fStartOffsets != null && fPositions != null)
{
float fieldCharsPerTerm = charsPerTerm[fieldNumOffs[i]];
for (int j = 0; j < startOffsets[i].Length; ++j)
{
fStartOffsets[j] += (int)(fieldCharsPerTerm * fPositions[j]);
}
}
if (fStartOffsets != null)
{
int[] fPrefixLengths = prefixLengths[i];
int[] fSuffixLengths = suffixLengths[i];
int[] fLengths = lengths[i];
for (int j = 0, end = (int)numTerms.Get(skip + i); j < end; ++j)
{
// delta-decode start offsets and patch lengths using term lengths
int termLength = fPrefixLengths[j] + fSuffixLengths[j];
lengths[i][positionIndex[i][j]] += termLength;
for (int k = positionIndex[i][j] + 1; k < positionIndex[i][j + 1]; ++k)
{
fStartOffsets[k] += fStartOffsets[k - 1];
fLengths[k] += termLength;
}
}
}
}
}
else
{
startOffsets = lengths = new int[numFields][];
}
if (totalPositions > 0)
{
// delta-decode positions
for (int i = 0; i < numFields; ++i)
{
int[] fPositions = positions[i];
int[] fpositionIndex = positionIndex[i];
if (fPositions != null)
{
for (int j = 0, end = (int)numTerms.Get(skip + i); j < end; ++j)
{
// delta-decode start offsets
for (int k = fpositionIndex[j] + 1; k < fpositionIndex[j + 1]; ++k)
{
fPositions[k] += fPositions[k - 1];
}
}
}
}
}
// payload lengths
int[][] payloadIndex = new int[numFields][];
int totalPayloadLength = 0;
int payloadOff = 0;
int payloadLen = 0;
if (totalPayloads > 0)
{
reader.Reset(vectorsStream, totalPayloads);
// skip
int termIndex = 0;
for (int i = 0; i < skip; ++i)
{
int f = (int)flags.Get(i);
int termCount = (int)numTerms.Get(i);
if ((f & CompressingTermVectorsWriter.PAYLOADS) != 0)
{
for (int j = 0; j < termCount; ++j)
{
int freq = termFreqs[termIndex + j];
for (int k = 0; k < freq; ++k)
{
int l = (int)reader.Next();
payloadOff += l;
}
}
}
termIndex += termCount;
}
totalPayloadLength = payloadOff;
// read doc payload lengths
for (int i = 0; i < numFields; ++i)
{
int f = (int)flags.Get(skip + i);
int termCount = (int)numTerms.Get(skip + i);
if ((f & CompressingTermVectorsWriter.PAYLOADS) != 0)
{
int totalFreq = positionIndex[i][termCount];
payloadIndex[i] = new int[totalFreq + 1];
int posIdx = 0;
payloadIndex[i][posIdx] = payloadLen;
for (int j = 0; j < termCount; ++j)
{
int freq = termFreqs[termIndex + j];
for (int k = 0; k < freq; ++k)
{
int payloadLength = (int)reader.Next();
payloadLen += payloadLength;
payloadIndex[i][posIdx + 1] = payloadLen;
++posIdx;
}
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(posIdx == totalFreq);
}
}
termIndex += termCount;
}
totalPayloadLength += payloadLen;
for (int i = skip + numFields; i < totalFields; ++i)
{
int f = (int)flags.Get(i);
int termCount = (int)numTerms.Get(i);
if ((f & CompressingTermVectorsWriter.PAYLOADS) != 0)
{
for (int j = 0; j < termCount; ++j)
{
int freq = termFreqs[termIndex + j];
for (int k = 0; k < freq; ++k)
{
totalPayloadLength += (int)reader.Next();
}
}
}
termIndex += termCount;
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(termIndex == totalTerms, () => termIndex + " " + totalTerms);
}
}
// decompress data
BytesRef suffixBytes = new BytesRef();
decompressor.Decompress(vectorsStream, totalLen + totalPayloadLength, docOff + payloadOff, docLen + payloadLen, suffixBytes);
suffixBytes.Length = docLen;
BytesRef payloadBytes = new BytesRef(suffixBytes.Bytes, suffixBytes.Offset + docLen, payloadLen);
int[] FieldFlags = new int[numFields];
for (int i = 0; i < numFields; ++i)
{
FieldFlags[i] = (int)flags.Get(skip + i);
}
int[] fieldNumTerms = new int[numFields];
for (int i = 0; i < numFields; ++i)
{
fieldNumTerms[i] = (int)numTerms.Get(skip + i);
}
int[][] fieldTermFreqs = new int[numFields][];
{
int termIdx = 0;
for (int i = 0; i < skip; ++i)
{
termIdx += (int)numTerms.Get(i);
}
for (int i = 0; i < numFields; ++i)
{
int termCount = (int)numTerms.Get(skip + i);
fieldTermFreqs[i] = new int[termCount];
for (int j = 0; j < termCount; ++j)
{
fieldTermFreqs[i][j] = termFreqs[termIdx++];
}
}
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(Sum(fieldLengths) == docLen, () => Sum(fieldLengths) + " != " + docLen);
}
return(new TVFields(this, fieldNums, FieldFlags, fieldNumOffs, fieldNumTerms, fieldLengths, prefixLengths, suffixLengths, fieldTermFreqs, positionIndex, positions, startOffsets, lengths, payloadBytes, payloadIndex, suffixBytes));
}
internal readonly bool isSegmentPrivate; // set to true iff this frozen packet represents
// a segment private deletes. in that case is should
// only have Queries
public FrozenBufferedUpdates(BufferedUpdates deletes, bool isSegmentPrivate)
{
this.isSegmentPrivate = isSegmentPrivate;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(!isSegmentPrivate || deletes.terms.Count == 0, () => "segment private package should only have del queries");
}
Term[] termsArray = deletes.terms.Keys.ToArray(/*new Term[deletes.terms.Count]*/);
termCount = termsArray.Length;
ArrayUtil.TimSort(termsArray);
PrefixCodedTerms.Builder builder = new PrefixCodedTerms.Builder();
foreach (Term term in termsArray)
{
builder.Add(term);
}
terms = builder.Finish();
queries = new Query[deletes.queries.Count];
queryLimits = new int[deletes.queries.Count];
int upto = 0;
foreach (KeyValuePair <Query, int?> ent in deletes.queries)
{
queries[upto] = ent.Key;
if (ent.Value.HasValue)
{
queryLimits[upto] = ent.Value.Value;
}
else
{
// LUCENENET NOTE: According to this: http://stackoverflow.com/a/13914344
// we are supposed to throw an exception in this case, rather than
// silently fail.
throw new NullReferenceException();
}
upto++;
}
// TODO if a Term affects multiple fields, we could keep the updates key'd by Term
// so that it maps to all fields it affects, sorted by their docUpto, and traverse
// that Term only once, applying the update to all fields that still need to be
// updated.
IList <NumericDocValuesUpdate> allNumericUpdates = new List <NumericDocValuesUpdate>();
int numericUpdatesSize = 0;
foreach (var numericUpdates in deletes.numericUpdates.Values)
{
foreach (NumericDocValuesUpdate update in numericUpdates.Values)
{
allNumericUpdates.Add(update);
numericUpdatesSize += update.GetSizeInBytes();
}
}
numericDVUpdates = allNumericUpdates.ToArray();
// TODO if a Term affects multiple fields, we could keep the updates key'd by Term
// so that it maps to all fields it affects, sorted by their docUpto, and traverse
// that Term only once, applying the update to all fields that still need to be
// updated.
IList <BinaryDocValuesUpdate> allBinaryUpdates = new List <BinaryDocValuesUpdate>();
int binaryUpdatesSize = 0;
foreach (var binaryUpdates in deletes.binaryUpdates.Values)
{
foreach (BinaryDocValuesUpdate update in binaryUpdates.Values)
{
allBinaryUpdates.Add(update);
binaryUpdatesSize += update.GetSizeInBytes();
}
}
binaryDVUpdates = allBinaryUpdates.ToArray();
bytesUsed = (int)terms.GetSizeInBytes() + queries.Length * BYTES_PER_DEL_QUERY + numericUpdatesSize + numericDVUpdates.Length * RamUsageEstimator.NUM_BYTES_OBJECT_REF + binaryUpdatesSize + binaryDVUpdates.Length * RamUsageEstimator.NUM_BYTES_OBJECT_REF;
numTermDeletes = deletes.numTermDeletes;
}
public override DocsAndPositionsEnum DocsAndPositions(IBits liveDocs, DocsAndPositionsEnum reuse, DocsAndPositionsFlags flags)
{
// Can only reuse if incoming enum is also a MultiDocsAndPositionsEnum
// ... and was previously created w/ this MultiTermsEnum:
if (reuse is null || !(reuse is MultiDocsAndPositionsEnum docsAndPositionsEnum) || !docsAndPositionsEnum.CanReuse(this))
docsAndPositionsEnum = new MultiDocsAndPositionsEnum(this, subs.Length);
int upto = 0;
for (int i = 0; i < numTop; i++)
{
TermsEnumWithSlice entry = top[i];
IBits b;
if (liveDocs is MultiBits multiLiveDocs)
{
// Optimize for common case: requested skip docs is a
// congruent sub-slice of MultiBits: in this case, we
// just pull the liveDocs from the sub reader, rather
// than making the inefficient
// Slice(Multi(sub-readers)):
MultiBits.SubResult sub = multiLiveDocs.GetMatchingSub(top[i].SubSlice);
if (sub.Matches)
{
b = sub.Result;
}
else
{
// custom case: requested skip docs is foreign:
// must slice it on every access (very
// inefficient)
b = new BitsSlice(liveDocs, top[i].SubSlice);
}
}
else if (liveDocs != null)
{
b = new BitsSlice(liveDocs, top[i].SubSlice);
}
else
{
// no deletions
b = null;
}
if (Debugging.AssertsEnabled) Debugging.Assert(entry.Index < docsAndPositionsEnum.subDocsAndPositionsEnum.Length, "{0} vs {1}; {2}", entry.Index, docsAndPositionsEnum.subDocsAndPositionsEnum.Length, subs.Length);
DocsAndPositionsEnum subPostings = entry.Terms.DocsAndPositions(b, docsAndPositionsEnum.subDocsAndPositionsEnum[entry.Index], flags);
if (subPostings != null)
{
docsAndPositionsEnum.subDocsAndPositionsEnum[entry.Index] = subPostings;
subDocsAndPositions[upto].DocsAndPositionsEnum = subPostings;
subDocsAndPositions[upto].Slice = entry.SubSlice;
upto++;
}
else
{
if (entry.Terms.Docs(b, null, DocsFlags.NONE) != null)
{
// At least one of our subs does not store
// offsets or positions -- we can't correctly
// produce a MultiDocsAndPositions enum
return null;
}
}
}
if (upto == 0)
{
return null;
}
else
{
return docsAndPositionsEnum.Reset(subDocsAndPositions, upto);
}
}
/// <exception cref="IOException"/>
protected override void WriteSkipData(int level, IndexOutput skipBuffer)
{
// To efficiently store payloads in the posting lists we do not store the length of
// every payload. Instead we omit the length for a payload if the previous payload had
// the same length.
// However, in order to support skipping the payload length at every skip point must be known.
// So we use the same length encoding that we use for the posting lists for the skip data as well:
// Case 1: current field does not store payloads
// SkipDatum --> DocSkip, FreqSkip, ProxSkip
// DocSkip,FreqSkip,ProxSkip --> VInt
// DocSkip records the document number before every SkipInterval th document in TermFreqs.
// Document numbers are represented as differences from the previous value in the sequence.
// Case 2: current field stores payloads
// SkipDatum --> DocSkip, PayloadLength?, FreqSkip,ProxSkip
// DocSkip,FreqSkip,ProxSkip --> VInt
// PayloadLength --> VInt
// In this case DocSkip/2 is the difference between
// the current and the previous value. If DocSkip
// is odd, then a PayloadLength encoded as VInt follows,
// if DocSkip is even, then it is assumed that the
// current payload length equals the length at the previous
// skip point
if (Debugging.AssertsEnabled)
{
Debugging.Assert(indexOptions == IndexOptions.DOCS_AND_FREQS_AND_POSITIONS || !curStorePayloads);
}
if (curStorePayloads)
{
int delta = curDoc - lastSkipDoc[level];
if (curPayloadLength == lastSkipPayloadLength[level])
{
// the current payload length equals the length at the previous skip point,
// so we don't store the length again
skipBuffer.WriteVInt32(delta << 1);
}
else
{
// the payload length is different from the previous one. We shift the DocSkip,
// set the lowest bit and store the current payload length as VInt.
skipBuffer.WriteVInt32(delta << 1 | 1);
skipBuffer.WriteVInt32(curPayloadLength);
lastSkipPayloadLength[level] = curPayloadLength;
}
}
else
{
// current field does not store payloads
skipBuffer.WriteVInt32(curDoc - lastSkipDoc[level]);
}
if (indexOptions != IndexOptions.DOCS_ONLY)
{
freqIndex[level].Mark();
freqIndex[level].Write(skipBuffer, false);
}
docIndex[level].Mark();
docIndex[level].Write(skipBuffer, false);
if (indexOptions == IndexOptions.DOCS_AND_FREQS_AND_POSITIONS)
{
posIndex[level].Mark();
posIndex[level].Write(skipBuffer, false);
if (curStorePayloads)
{
skipBuffer.WriteVInt32((int)(curPayloadPointer - lastSkipPayloadPointer[level]));
}
}
lastSkipDoc[level] = curDoc;
lastSkipPayloadPointer[level] = curPayloadPointer;
}
public TermsEnumWithSlice(int index, ReaderSlice subSlice)
{
this.SubSlice = subSlice;
this.Index = index;
if (Debugging.AssertsEnabled) Debugging.Assert(subSlice.Length >= 0,"length={0}", subSlice.Length);
}
private IEnumerable <long?> GetMergeSortedSetOrdsEnumerable(AtomicReader[] readers, SortedSetDocValues[] dvs, OrdinalMap map)
{
int readerUpto = -1;
int docIDUpto = 0;
AtomicReader currentReader = null;
IBits currentLiveDocs = null;
var ords = new long[8];
int ordUpto = 0;
int ordLength = 0;
while (true)
{
if (readerUpto == readers.Length)
{
yield break;
}
if (ordUpto < ordLength)
{
var value = ords[ordUpto];
ordUpto++;
yield return(value);
continue;
}
if (currentReader == null || docIDUpto == currentReader.MaxDoc)
{
readerUpto++;
if (readerUpto < readers.Length)
{
currentReader = readers[readerUpto];
currentLiveDocs = currentReader.LiveDocs;
}
docIDUpto = 0;
continue;
}
if (currentLiveDocs == null || currentLiveDocs.Get(docIDUpto))
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(docIDUpto < currentReader.MaxDoc);
}
SortedSetDocValues dv = dvs[readerUpto];
dv.SetDocument(docIDUpto);
ordUpto = ordLength = 0;
long ord;
while ((ord = dv.NextOrd()) != SortedSetDocValues.NO_MORE_ORDS)
{
if (ordLength == ords.Length)
{
ords = ArrayUtil.Grow(ords, ordLength + 1);
}
ords[ordLength] = map.GetGlobalOrd(readerUpto, ord);
ordLength++;
}
docIDUpto++;
continue;
}
docIDUpto++;
}
}
internal SegmentCoreReaders(SegmentReader owner, Directory dir, SegmentCommitInfo si, IOContext context, int termsIndexDivisor)
{
fieldsReaderLocal = new DisposableThreadLocal <StoredFieldsReader>(()
=> (StoredFieldsReader)fieldsReaderOrig.Clone());
termVectorsLocal = new DisposableThreadLocal <TermVectorsReader>(()
=> (termVectorsReaderOrig == null) ? null : (TermVectorsReader)termVectorsReaderOrig.Clone());
if (termsIndexDivisor == 0)
{
throw new ArgumentException("indexDivisor must be < 0 (don't load terms index) or greater than 0 (got 0)");
}
Codec codec = si.Info.Codec;
Directory cfsDir; // confusing name: if (cfs) its the cfsdir, otherwise its the segment's directory.
bool success = false;
try
{
if (si.Info.UseCompoundFile)
{
cfsDir = cfsReader = new CompoundFileDirectory(dir, IndexFileNames.SegmentFileName(si.Info.Name, "", IndexFileNames.COMPOUND_FILE_EXTENSION), context, false);
}
else
{
cfsReader = null;
cfsDir = dir;
}
FieldInfos fieldInfos = owner.FieldInfos;
this.termsIndexDivisor = termsIndexDivisor;
PostingsFormat format = codec.PostingsFormat;
SegmentReadState segmentReadState = new SegmentReadState(cfsDir, si.Info, fieldInfos, context, termsIndexDivisor);
// Ask codec for its Fields
fields = format.FieldsProducer(segmentReadState);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(fields != null);
}
// ask codec for its Norms:
// TODO: since we don't write any norms file if there are no norms,
// kinda jaky to assume the codec handles the case of no norms file at all gracefully?!
if (fieldInfos.HasNorms)
{
normsProducer = codec.NormsFormat.NormsProducer(segmentReadState);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(normsProducer != null);
}
}
else
{
normsProducer = null;
}
fieldsReaderOrig = si.Info.Codec.StoredFieldsFormat.FieldsReader(cfsDir, si.Info, fieldInfos, context);
if (fieldInfos.HasVectors) // open term vector files only as needed
{
termVectorsReaderOrig = si.Info.Codec.TermVectorsFormat.VectorsReader(cfsDir, si.Info, fieldInfos, context);
}
else
{
termVectorsReaderOrig = null;
}
success = true;
}
finally
{
if (!success)
{
DecRef();
}
}
}
public override void Write(Directory directory, string segmentName, string segmentSuffix, FieldInfos infos,
IOContext context)
{
var fileName = IndexFileNames.SegmentFileName(segmentName, segmentSuffix, FIELD_INFOS_EXTENSION);
var output = directory.CreateOutput(fileName, context);
var scratch = new BytesRef();
var success = false;
try
{
SimpleTextUtil.Write(output, NUMFIELDS);
SimpleTextUtil.Write(output, infos.Count.ToString(CultureInfo.InvariantCulture), scratch);
SimpleTextUtil.WriteNewline(output);
foreach (FieldInfo fi in infos)
{
SimpleTextUtil.Write(output, NAME);
SimpleTextUtil.Write(output, fi.Name, scratch);
SimpleTextUtil.WriteNewline(output);
SimpleTextUtil.Write(output, NUMBER);
SimpleTextUtil.Write(output, fi.Number.ToString(CultureInfo.InvariantCulture), scratch);
SimpleTextUtil.WriteNewline(output);
SimpleTextUtil.Write(output, ISINDEXED);
SimpleTextUtil.Write(output, CultureInfo.InvariantCulture.TextInfo.ToLower(fi.IsIndexed.ToString()), scratch);
SimpleTextUtil.WriteNewline(output);
if (fi.IsIndexed)
{
// LUCENENET specific - to avoid boxing, changed from CompareTo() to IndexOptionsComparer.Compare()
if (Debugging.AssertsEnabled)
{
Debugging.Assert(IndexOptionsComparer.Default.Compare(fi.IndexOptions, IndexOptions.DOCS_AND_FREQS_AND_POSITIONS) >= 0 || !fi.HasPayloads);
}
SimpleTextUtil.Write(output, INDEXOPTIONS);
SimpleTextUtil.Write(output,
fi.IndexOptions != IndexOptions.NONE ? fi.IndexOptions.ToString() : string.Empty,
scratch);
SimpleTextUtil.WriteNewline(output);
}
SimpleTextUtil.Write(output, STORETV);
SimpleTextUtil.Write(output, CultureInfo.InvariantCulture.TextInfo.ToLower(fi.HasVectors.ToString()), scratch);
SimpleTextUtil.WriteNewline(output);
SimpleTextUtil.Write(output, PAYLOADS);
SimpleTextUtil.Write(output, CultureInfo.InvariantCulture.TextInfo.ToLower(fi.HasPayloads.ToString()), scratch);
SimpleTextUtil.WriteNewline(output);
SimpleTextUtil.Write(output, NORMS);
SimpleTextUtil.Write(output, CultureInfo.InvariantCulture.TextInfo.ToLower((!fi.OmitsNorms).ToString()), scratch);
SimpleTextUtil.WriteNewline(output);
SimpleTextUtil.Write(output, NORMS_TYPE);
SimpleTextUtil.Write(output, GetDocValuesType(fi.NormType), scratch);
SimpleTextUtil.WriteNewline(output);
SimpleTextUtil.Write(output, DOCVALUES);
SimpleTextUtil.Write(output, GetDocValuesType(fi.DocValuesType), scratch);
SimpleTextUtil.WriteNewline(output);
SimpleTextUtil.Write(output, DOCVALUES_GEN);
SimpleTextUtil.Write(output, fi.DocValuesGen.ToString(CultureInfo.InvariantCulture), scratch);
SimpleTextUtil.WriteNewline(output);
IDictionary <string, string> atts = fi.Attributes;
int numAtts = atts is null ? 0 : atts.Count;
SimpleTextUtil.Write(output, NUM_ATTS);
SimpleTextUtil.Write(output, numAtts.ToString(CultureInfo.InvariantCulture), scratch);
SimpleTextUtil.WriteNewline(output);
if (numAtts <= 0 || atts is null)
{
continue;
}
foreach (var entry in atts)
{
SimpleTextUtil.Write(output, ATT_KEY);
SimpleTextUtil.Write(output, entry.Key, scratch);
SimpleTextUtil.WriteNewline(output);
SimpleTextUtil.Write(output, ATT_VALUE);
SimpleTextUtil.Write(output, entry.Value, scratch);
SimpleTextUtil.WriteNewline(output);
}
}
SimpleTextUtil.WriteChecksum(output, scratch);
success = true;
}
finally
{
if (success)
{
output.Dispose();
}
else
{
IOUtils.DisposeWhileHandlingException(output);
}
}
}
public override ThreadState GetAndLock(Thread requestingThread, DocumentsWriter documentsWriter)
{
ThreadState threadState = null;
if (NumThreadStatesActive == 0)
{
lock (this)
{
if (NumThreadStatesActive == 0)
{
threadState = states[0] = NewThreadState();
return(threadState);
}
}
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(NumThreadStatesActive > 0);
}
for (int i = 0; i < maxRetry; i++)
{
int ord = random.Next(NumThreadStatesActive);
lock (this)
{
threadState = states[ord];
if (Debugging.AssertsEnabled)
{
Debugging.Assert(threadState != null);
}
}
if (threadState.TryLock())
{
return(threadState);
}
if (random.Next(20) == 0)
{
break;
}
}
/*
* only try to create a new threadstate if we can not lock the randomly
* selected state. this is important since some tests rely on a single
* threadstate in the single threaded case. Eventually it would be nice if
* we would not have this limitation but for now we just make sure we only
* allocate one threadstate if indexing is single threaded
*/
lock (this)
{
ThreadState newThreadState = NewThreadState();
if (newThreadState != null) // did we get a new state?
{
threadState = states[NumThreadStatesActive - 1] = newThreadState;
//if (Debugging.AssertsEnabled) Debugging.Assert(threadState.HeldByCurrentThread);
return(threadState);
}
// if no new state is available lock the random one
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(threadState != null);
}
threadState.@Lock();
return(threadState);
}
/// <summary>
/// Determinizes the given automaton.
/// <para/>
/// Worst case complexity: exponential in number of states.
/// </summary>
public static void Determinize(Automaton a)
{
if (a.IsDeterministic || a.IsSingleton)
{
return;
}
State[] allStates = a.GetNumberedStates();
// subset construction
bool initAccept = a.initial.accept;
int initNumber = a.initial.number;
a.initial = new State();
SortedInt32Set.FrozenInt32Set initialset = new SortedInt32Set.FrozenInt32Set(initNumber, a.initial);
Queue <SortedInt32Set.FrozenInt32Set> worklist = new Queue <SortedInt32Set.FrozenInt32Set>(); // LUCENENET specific - Queue is much more performant than LinkedList
IDictionary <SortedInt32Set.FrozenInt32Set, State> newstate = new Dictionary <SortedInt32Set.FrozenInt32Set, State>();
worklist.Enqueue(initialset);
a.initial.accept = initAccept;
newstate[initialset] = a.initial;
int newStateUpto = 0;
State[] newStatesArray = new State[5];
newStatesArray[newStateUpto] = a.initial;
a.initial.number = newStateUpto;
newStateUpto++;
// like Set<Integer,PointTransitions>
PointTransitionSet points = new PointTransitionSet();
// like SortedMap<Integer,Integer>
SortedInt32Set statesSet = new SortedInt32Set(5);
while (worklist.Count > 0)
{
SortedInt32Set.FrozenInt32Set s = worklist.Dequeue();
//worklist.Remove(s);
// Collate all outgoing transitions by min/1+max:
for (int i = 0; i < s.values.Length; i++)
{
State s0 = allStates[s.values[i]];
for (int j = 0; j < s0.numTransitions; j++)
{
points.Add(s0.TransitionsArray[j]);
}
}
if (points.count == 0)
{
// No outgoing transitions -- skip it
continue;
}
points.Sort();
int lastPoint = -1;
int accCount = 0;
State r = s.state;
for (int i = 0; i < points.count; i++)
{
int point = points.points[i].point;
if (statesSet.upto > 0)
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(lastPoint != -1);
}
statesSet.ComputeHash();
if (!newstate.TryGetValue(statesSet.ToFrozenInt32Set(), out State q) || q == null)
{
q = new State();
SortedInt32Set.FrozenInt32Set p = statesSet.Freeze(q);
worklist.Enqueue(p);
if (newStateUpto == newStatesArray.Length)
{
// LUCENENET: Resize rather than copy
Array.Resize(ref newStatesArray, ArrayUtil.Oversize(1 + newStateUpto, RamUsageEstimator.NUM_BYTES_OBJECT_REF));
}
newStatesArray[newStateUpto] = q;
q.number = newStateUpto;
newStateUpto++;
q.accept = accCount > 0;
newstate[p] = q;
}
else
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert((accCount > 0) == q.accept, "accCount={0} vs existing accept={1} states={2}", accCount, q.accept, statesSet);
}
}
r.AddTransition(new Transition(lastPoint, point - 1, q));
}
// process transitions that end on this point
// (closes an overlapping interval)
Transition[] transitions = points.points[i].ends.transitions;
int limit = points.points[i].ends.count;
for (int j = 0; j < limit; j++)
{
Transition t = transitions[j];
int num = t.to.number;
statesSet.Decr(num);
accCount -= t.to.accept ? 1 : 0;
}
points.points[i].ends.count = 0;
// process transitions that start on this point
// (opens a new interval)
transitions = points.points[i].starts.transitions;
limit = points.points[i].starts.count;
for (int j = 0; j < limit; j++)
{
Transition t = transitions[j];
int num = t.to.number;
statesSet.Incr(num);
accCount += t.to.accept ? 1 : 0;
}
lastPoint = point;
points.points[i].starts.count = 0;
}
points.Reset();
if (Debugging.AssertsEnabled)
{
Debugging.Assert(statesSet.upto == 0, "upto={0}", statesSet.upto);
}
}
a.deterministic = true;
a.SetNumberedStates(newStatesArray, newStateUpto);
}
/// <summary>
/// Builds an <see cref="SynonymMap"/> and returns it.
/// </summary>
public virtual SynonymMap Build()
{
ByteSequenceOutputs outputs = ByteSequenceOutputs.Singleton;
// TODO: are we using the best sharing options?
var builder = new Builder <BytesRef>(FST.INPUT_TYPE.BYTE4, outputs);
BytesRef scratch = new BytesRef(64);
ByteArrayDataOutput scratchOutput = new ByteArrayDataOutput();
ISet <int> dedupSet;
if (dedup)
{
dedupSet = new JCG.HashSet <int>();
}
else
{
dedupSet = null;
}
var spare = new byte[5];
ICollection <CharsRef> keys = workingSet.Keys;
CharsRef[] sortedKeys = new CharsRef[keys.Count];
keys.CopyTo(sortedKeys, 0);
#pragma warning disable 612, 618
System.Array.Sort(sortedKeys, CharsRef.UTF16SortedAsUTF8Comparer);
#pragma warning restore 612, 618
Int32sRef scratchIntsRef = new Int32sRef();
//System.out.println("fmap.build");
for (int keyIdx = 0; keyIdx < sortedKeys.Length; keyIdx++)
{
CharsRef input = sortedKeys[keyIdx];
MapEntry output = workingSet[input];
int numEntries = output.ords.Count;
// output size, assume the worst case
int estimatedSize = 5 + numEntries * 5; // numEntries + one ord for each entry
scratch.Grow(estimatedSize);
scratchOutput.Reset(scratch.Bytes, scratch.Offset, scratch.Bytes.Length);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(scratch.Offset == 0);
}
// now write our output data:
int count = 0;
for (int i = 0; i < numEntries; i++)
{
if (dedupSet != null)
{
// LUCENENET specific - no boxing happening here
int ent = output.ords[i];
if (dedupSet.Contains(ent))
{
continue;
}
dedupSet.Add(ent);
}
scratchOutput.WriteVInt32(output.ords[i]);
count++;
}
int pos = scratchOutput.Position;
scratchOutput.WriteVInt32(count << 1 | (output.includeOrig ? 0 : 1));
int pos2 = scratchOutput.Position;
int vIntLen = pos2 - pos;
// Move the count + includeOrig to the front of the byte[]:
Array.Copy(scratch.Bytes, pos, spare, 0, vIntLen);
Array.Copy(scratch.Bytes, 0, scratch.Bytes, vIntLen, pos);
Array.Copy(spare, 0, scratch.Bytes, 0, vIntLen);
if (dedupSet != null)
{
dedupSet.Clear();
}
scratch.Length = scratchOutput.Position - scratch.Offset;
//System.out.println(" add input=" + input + " output=" + scratch + " offset=" + scratch.offset + " length=" + scratch.length + " count=" + count);
builder.Add(Lucene.Net.Util.Fst.Util.ToUTF32(input.ToString(), scratchIntsRef), BytesRef.DeepCopyOf(scratch));
}
FST <BytesRef> fst = builder.Finish();
return(new SynonymMap(fst, words, maxHorizontalContext));
}
public override void Merge(IndexWriter writer, MergeTrigger trigger, bool newMergesFound)
{
UninterruptableMonitor.Enter(this);
try
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(!UninterruptableMonitor.IsEntered(writer));
}
this.m_writer = writer;
InitMergeThreadPriority();
m_dir = writer.Directory;
// First, quickly run through the newly proposed merges
// and add any orthogonal merges (ie a merge not
// involving segments already pending to be merged) to
// the queue. If we are way behind on merging, many of
// these newly proposed merges will likely already be
// registered.
if (IsVerbose)
{
Message("now merge");
Message(" index: " + writer.SegString());
}
// Iterate, pulling from the IndexWriter's queue of
// pending merges, until it's empty:
while (true)
{
long startStallTime = 0;
while (writer.HasPendingMerges() && MergeThreadCount >= maxMergeCount)
{
// this means merging has fallen too far behind: we
// have already created maxMergeCount threads, and
// now there's at least one more merge pending.
// Note that only maxThreadCount of
// those created merge threads will actually be
// running; the rest will be paused (see
// updateMergeThreads). We stall this producer
// thread to prevent creation of new segments,
// until merging has caught up:
startStallTime = J2N.Time.NanoTime() / J2N.Time.MillisecondsPerNanosecond; // LUCENENET: Use NanoTime() rather than CurrentTimeMilliseconds() for more accurate/reliable results
if (IsVerbose)
{
Message(" too many merges; stalling...");
}
try
{
UninterruptableMonitor.Wait(this);
}
catch (Exception ie) when(ie.IsInterruptedException())
{
throw new Util.ThreadInterruptedException(ie);
}
}
if (IsVerbose)
{
if (startStallTime != 0)
{
Message(" stalled for " + ((J2N.Time.NanoTime() / J2N.Time.MillisecondsPerNanosecond) - startStallTime) + " msec"); // LUCENENET: Use NanoTime() rather than CurrentTimeMilliseconds() for more accurate/reliable results
}
}
MergePolicy.OneMerge merge = writer.NextMerge();
if (merge == null)
{
if (IsVerbose)
{
Message(" no more merges pending; now return");
}
return;
}
bool success = false;
try
{
if (IsVerbose)
{
Message(" consider merge " + writer.SegString(merge.Segments));
}
// OK to spawn a new merge thread to handle this
// merge:
MergeThread merger = GetMergeThread(writer, merge);
m_mergeThreads.Add(merger);
if (IsVerbose)
{
Message(" launch new thread [" + merger.Name + "]");
}
merger.Start();
// Must call this after starting the thread else
// the new thread is removed from mergeThreads
// (since it's not alive yet):
UpdateMergeThreads();
success = true;
}
finally
{
if (!success)
{
writer.MergeFinish(merge);
}
}
}
}
finally
{
UninterruptableMonitor.Exit(this);
}
}
/// <summary>
/// Compute the intersection of the provided sets. This method is much faster than
/// computing the intersection manually since it operates directly at the byte level.
/// </summary>
public static WAH8DocIdSet Intersect(ICollection <WAH8DocIdSet> docIdSets, int indexInterval)
{
// LUCENENET: Added guard clause for null
if (docIdSets is null)
{
throw new ArgumentNullException(nameof(docIdSets));
}
switch (docIdSets.Count)
{
case 0:
throw new ArgumentException("There must be at least one set to intersect");
case 1:
return(docIdSets.First());
}
// The logic below is similar to ConjunctionScorer
int numSets = docIdSets.Count;
var iterators = new Iterator[numSets];
int i = 0;
foreach (WAH8DocIdSet set in docIdSets)
{
var it = (Iterator)set.GetIterator();
iterators[i++] = it;
}
Array.Sort(iterators, SERIALIZED_LENGTH_COMPARER);
WordBuilder builder = (WordBuilder)(new WordBuilder()).SetIndexInterval(indexInterval);
int wordNum = 0;
while (true)
{
// Advance the least costly iterator first
iterators[0].AdvanceWord(wordNum);
wordNum = iterators[0].wordNum;
if (wordNum == DocIdSetIterator.NO_MORE_DOCS)
{
break;
}
byte word = iterators[0].word;
for (i = 1; i < numSets; ++i)
{
if (iterators[i].wordNum < wordNum)
{
iterators[i].AdvanceWord(wordNum);
}
if (iterators[i].wordNum > wordNum)
{
wordNum = iterators[i].wordNum;
goto mainContinue;
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(iterators[i].wordNum == wordNum);
}
word &= iterators[i].word;
if (word == 0)
{
// There are common words, but they don't share any bit
++wordNum;
goto mainContinue;
}
}
// Found a common word
if (Debugging.AssertsEnabled)
{
Debugging.Assert(word != 0);
}
builder.AddWord(wordNum, word);
++wordNum;
mainContinue :;
}
//mainBreak:
return(builder.Build());
}
/// <summary>
/// Given a <paramref name="target"/> value, advance the decoding index to the first bigger or equal value
/// and return it if it is available. Otherwise return <see cref="NO_MORE_VALUES"/>.
/// <para/>
/// The current implementation uses the index on the upper zero bit positions.
/// </summary>
public virtual long AdvanceToValue(long target)
{
efIndex += 1;
if (efIndex >= numEncoded)
{
return(NO_MORE_VALUES);
}
setBitForIndex += 1; // the high bit at setBitForIndex belongs to the unary code for efIndex
int highIndex = (int)((long)((ulong)setBitForIndex >> LOG2_INT64_SIZE));
long upperLong = efEncoder.upperLongs[highIndex];
curHighLong = (long)((ulong)upperLong >> ((int)(setBitForIndex & ((sizeof(long) * 8) - 1)))); // may contain the unary 1 bit for efIndex
// determine index entry to advance to
long highTarget = (long)((ulong)target >> efEncoder.numLowBits);
long indexEntryIndex = (highTarget / efEncoder.indexInterval) - 1;
if (indexEntryIndex >= 0) // not before first index entry
{
if (indexEntryIndex >= numIndexEntries)
{
indexEntryIndex = numIndexEntries - 1; // no further than last index entry
}
long indexHighValue = (indexEntryIndex + 1) * efEncoder.indexInterval;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(indexHighValue <= highTarget);
}
if (indexHighValue > (setBitForIndex - efIndex)) // advance to just after zero bit position of index entry.
{
setBitForIndex = UnPackValue(efEncoder.upperZeroBitPositionIndex, efEncoder.nIndexEntryBits, indexEntryIndex, indexMask);
efIndex = setBitForIndex - indexHighValue; // the high bit at setBitForIndex belongs to the unary code for efIndex
highIndex = (int)(((ulong)setBitForIndex >> LOG2_INT64_SIZE));
upperLong = efEncoder.upperLongs[highIndex];
curHighLong = (long)((ulong)upperLong >> ((int)(setBitForIndex & ((sizeof(long) * 8) - 1)))); // may contain the unary 1 bit for efIndex
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(efIndex < numEncoded); // there is a high value to be found.
}
}
int curSetBits = curHighLong.PopCount(); // shifted right.
int curClearBits = (sizeof(long) * 8) - curSetBits - ((int)(setBitForIndex & ((sizeof(long) * 8) - 1))); // subtract right shift, may be more than encoded
while (((setBitForIndex - efIndex) + curClearBits) < highTarget)
{
// curHighLong has not enough clear bits to reach highTarget
efIndex += curSetBits;
if (efIndex >= numEncoded)
{
return(NO_MORE_VALUES);
}
setBitForIndex += (sizeof(long) * 8) - (setBitForIndex & ((sizeof(long) * 8) - 1));
// highIndex = (int)(setBitForIndex >>> LOG2_LONG_SIZE);
if (Debugging.AssertsEnabled)
{
Debugging.Assert((highIndex + 1) == (int)((long)((ulong)setBitForIndex >> LOG2_INT64_SIZE)));
}
highIndex += 1;
upperLong = efEncoder.upperLongs[highIndex];
curHighLong = upperLong;
curSetBits = curHighLong.PopCount();
curClearBits = (sizeof(long) * 8) - curSetBits;
}
// curHighLong has enough clear bits to reach highTarget, and may not have enough set bits.
while (curHighLong == 0L)
{
setBitForIndex += (sizeof(long) * 8) - (setBitForIndex & ((sizeof(long) * 8) - 1));
if (Debugging.AssertsEnabled)
{
Debugging.Assert((highIndex + 1) == (int)((ulong)setBitForIndex >> LOG2_INT64_SIZE));
}
highIndex += 1;
upperLong = efEncoder.upperLongs[highIndex];
curHighLong = upperLong;
}
// curHighLong has enough clear bits to reach highTarget, has at least 1 set bit, and may not have enough set bits.
int rank = (int)(highTarget - (setBitForIndex - efIndex)); // the rank of the zero bit for highValue.
if (Debugging.AssertsEnabled)
{
Debugging.Assert((rank <= (sizeof(long) * 8)), "rank {0}", rank);
}
if (rank >= 1)
{
long invCurHighLong = ~curHighLong;
int clearBitForValue = (rank <= 8) ? BroadWord.SelectNaive(invCurHighLong, rank) : BroadWord.Select(invCurHighLong, rank);
if (Debugging.AssertsEnabled)
{
Debugging.Assert(clearBitForValue <= ((sizeof(long) * 8) - 1));
}
setBitForIndex += clearBitForValue + 1; // the high bit just before setBitForIndex is zero
int oneBitsBeforeClearBit = clearBitForValue - rank + 1;
efIndex += oneBitsBeforeClearBit; // the high bit at setBitForIndex and belongs to the unary code for efIndex
if (efIndex >= numEncoded)
{
return(NO_MORE_VALUES);
}
if ((setBitForIndex & ((sizeof(long) * 8) - 1)) == 0L) // exhausted curHighLong
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert((highIndex + 1) == (int)((ulong)setBitForIndex >> LOG2_INT64_SIZE));
}
highIndex += 1;
upperLong = efEncoder.upperLongs[highIndex];
curHighLong = upperLong;
}
else
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(highIndex == (int)((ulong)setBitForIndex >> LOG2_INT64_SIZE));
}
curHighLong = (long)((ulong)upperLong >> ((int)(setBitForIndex & ((sizeof(long) * 8) - 1))));
}
// curHighLong has enough clear bits to reach highTarget, and may not have enough set bits.
while (curHighLong == 0L)
{
setBitForIndex += (sizeof(long) * 8) - (setBitForIndex & ((sizeof(long) * 8) - 1));
if (Debugging.AssertsEnabled)
{
Debugging.Assert((highIndex + 1) == (int)((ulong)setBitForIndex >> LOG2_INT64_SIZE));
}
highIndex += 1;
upperLong = efEncoder.upperLongs[highIndex];
curHighLong = upperLong;
}
}
setBitForIndex += curHighLong.TrailingZeroCount();
if (Debugging.AssertsEnabled)
{
Debugging.Assert((setBitForIndex - efIndex) >= highTarget); // highTarget reached
}
// Linear search also with low values
long currentValue = CombineHighLowValues((setBitForIndex - efIndex), CurrentLowValue());
while (currentValue < target)
{
currentValue = NextValue();
if (currentValue == NO_MORE_VALUES)
{
return(NO_MORE_VALUES);
}
}
return(currentValue);
}
internal virtual void AddWord(int wordNum, byte word)
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(wordNum > lastWordNum);
Debugging.Assert(word != 0);
}
if (!reverse)
{
if (lastWordNum == -1)
{
clean = 2 + wordNum; // special case for the 1st sequence
dirtyWords.WriteByte(word);
}
else
{
switch (wordNum - lastWordNum)
{
case 1:
if (word == 0xFF && (byte)dirtyWords.Bytes[dirtyWords.Length - 1] == 0xFF)
{
--dirtyWords.Length;
WriteSequence();
reverse = true;
clean = 2;
}
else
{
dirtyWords.WriteByte(word);
}
break;
case 2:
dirtyWords.WriteByte(0);
dirtyWords.WriteByte(word);
break;
default:
WriteSequence();
clean = wordNum - lastWordNum - 1;
dirtyWords.WriteByte(word);
break;
}
}
}
else
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(lastWordNum >= 0);
}
switch (wordNum - lastWordNum)
{
case 1:
if (word == 0xFF)
{
if (dirtyWords.Length == 0)
{
++clean;
}
else if ((byte)dirtyWords.Bytes[dirtyWords.Length - 1] == 0xFF)
{
--dirtyWords.Length;
WriteSequence();
clean = 2;
}
else
{
dirtyWords.WriteByte(word);
}
}
else
{
dirtyWords.WriteByte(word);
}
break;
case 2:
dirtyWords.WriteByte(0);
dirtyWords.WriteByte(word);
break;
default:
WriteSequence();
reverse = false;
clean = wordNum - lastWordNum - 1;
dirtyWords.WriteByte(word);
break;
}
}
lastWordNum = wordNum;
cardinality += BitUtil.BitCount(word);
}
internal void Flush(string fieldName, FieldsConsumer consumer, SegmentWriteState state)
{
if (!fieldInfo.IsIndexed)
{
return; // nothing to flush, don't bother the codec with the unindexed field
}
TermsConsumer termsConsumer = consumer.AddField(fieldInfo);
IComparer <BytesRef> termComp = termsConsumer.Comparer;
// CONFUSING: this.indexOptions holds the index options
// that were current when we first saw this field. But
// it's possible this has changed, eg when other
// documents are indexed that cause a "downgrade" of the
// IndexOptions. So we must decode the in-RAM buffer
// according to this.indexOptions, but then write the
// new segment to the directory according to
// currentFieldIndexOptions:
IndexOptions currentFieldIndexOptions = fieldInfo.IndexOptions;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(currentFieldIndexOptions != IndexOptions.NONE);
}
// LUCENENET specific - to avoid boxing, changed from CompareTo() to IndexOptionsComparer.Compare()
bool writeTermFreq = IndexOptionsComparer.Default.Compare(currentFieldIndexOptions, IndexOptions.DOCS_AND_FREQS) >= 0;
bool writePositions = IndexOptionsComparer.Default.Compare(currentFieldIndexOptions, IndexOptions.DOCS_AND_FREQS_AND_POSITIONS) >= 0;
bool writeOffsets = IndexOptionsComparer.Default.Compare(currentFieldIndexOptions, IndexOptions.DOCS_AND_FREQS_AND_POSITIONS_AND_OFFSETS) >= 0;
bool readTermFreq = this.hasFreq;
bool readPositions = this.hasProx;
bool readOffsets = this.hasOffsets;
//System.out.println("flush readTF=" + readTermFreq + " readPos=" + readPositions + " readOffs=" + readOffsets);
// Make sure FieldInfo.update is working correctly!:
if (Debugging.AssertsEnabled)
{
Debugging.Assert(!writeTermFreq || readTermFreq);
Debugging.Assert(!writePositions || readPositions);
Debugging.Assert(!writeOffsets || readOffsets);
Debugging.Assert(!writeOffsets || writePositions);
}
IDictionary <Term, int?> segDeletes;
if (state.SegUpdates != null && state.SegUpdates.terms.Count > 0)
{
segDeletes = state.SegUpdates.terms;
}
else
{
segDeletes = null;
}
int[] termIDs = termsHashPerField.SortPostings(termComp);
int numTerms = termsHashPerField.bytesHash.Count;
BytesRef text = new BytesRef();
FreqProxPostingsArray postings = (FreqProxPostingsArray)termsHashPerField.postingsArray;
ByteSliceReader freq = new ByteSliceReader();
ByteSliceReader prox = new ByteSliceReader();
FixedBitSet visitedDocs = new FixedBitSet(state.SegmentInfo.DocCount);
long sumTotalTermFreq = 0;
long sumDocFreq = 0;
Term protoTerm = new Term(fieldName);
for (int i = 0; i < numTerms; i++)
{
int termID = termIDs[i];
// Get BytesRef
int textStart = postings.textStarts[termID];
termsHashPerField.bytePool.SetBytesRef(text, textStart);
termsHashPerField.InitReader(freq, termID, 0);
if (readPositions || readOffsets)
{
termsHashPerField.InitReader(prox, termID, 1);
}
// TODO: really TermsHashPerField should take over most
// of this loop, including merge sort of terms from
// multiple threads and interacting with the
// TermsConsumer, only calling out to us (passing us the
// DocsConsumer) to handle delivery of docs/positions
PostingsConsumer postingsConsumer = termsConsumer.StartTerm(text);
int?delDocLimit;
if (segDeletes != null)
{
protoTerm.Bytes = text;
if (segDeletes.TryGetValue(protoTerm, out int?docIDUpto) && docIDUpto != null)
{
delDocLimit = docIDUpto;
}
else
{
delDocLimit = 0;
}
}
else
{
delDocLimit = 0;
}
// Now termStates has numToMerge FieldMergeStates
// which all share the same term. Now we must
// interleave the docID streams.
int docFreq = 0;
long totalTermFreq = 0;
int docID = 0;
while (true)
{
//System.out.println(" cycle");
int termFreq;
if (freq.Eof())
{
if (postings.lastDocCodes[termID] != -1)
{
// Return last doc
docID = postings.lastDocIDs[termID];
if (readTermFreq)
{
termFreq = postings.termFreqs[termID];
}
else
{
termFreq = -1;
}
postings.lastDocCodes[termID] = -1;
}
else
{
// EOF
break;
}
}
else
{
int code = freq.ReadVInt32();
if (!readTermFreq)
{
docID += code;
termFreq = -1;
}
else
{
docID += code.TripleShift(1);
if ((code & 1) != 0)
{
termFreq = 1;
}
else
{
termFreq = freq.ReadVInt32();
}
}
if (Debugging.AssertsEnabled)
{
Debugging.Assert(docID != postings.lastDocIDs[termID]);
}
}
docFreq++;
if (Debugging.AssertsEnabled)
{
Debugging.Assert(docID < state.SegmentInfo.DocCount, "doc={0} maxDoc={1}", docID, state.SegmentInfo.DocCount);
}
// NOTE: we could check here if the docID was
// deleted, and skip it. However, this is somewhat
// dangerous because it can yield non-deterministic
// behavior since we may see the docID before we see
// the term that caused it to be deleted. this
// would mean some (but not all) of its postings may
// make it into the index, which'd alter the docFreq
// for those terms. We could fix this by doing two
// passes, ie first sweep marks all del docs, and
// 2nd sweep does the real flush, but I suspect
// that'd add too much time to flush.
visitedDocs.Set(docID);
postingsConsumer.StartDoc(docID, writeTermFreq ? termFreq : -1);
if (docID < delDocLimit)
{
// Mark it deleted. TODO: we could also skip
// writing its postings; this would be
// deterministic (just for this Term's docs).
// TODO: can we do this reach-around in a cleaner way????
if (state.LiveDocs == null)
{
state.LiveDocs = docState.docWriter.codec.LiveDocsFormat.NewLiveDocs(state.SegmentInfo.DocCount);
}
if (state.LiveDocs.Get(docID))
{
state.DelCountOnFlush++;
state.LiveDocs.Clear(docID);
}
}
totalTermFreq += termFreq;
// Carefully copy over the prox + payload info,
// changing the format to match Lucene's segment
// format.
if (readPositions || readOffsets)
{
// we did record positions (& maybe payload) and/or offsets
int position = 0;
int offset = 0;
for (int j = 0; j < termFreq; j++)
{
BytesRef thisPayload;
if (readPositions)
{
int code = prox.ReadVInt32();
position += code.TripleShift(1);
if ((code & 1) != 0)
{
// this position has a payload
int payloadLength = prox.ReadVInt32();
if (payload == null)
{
payload = new BytesRef();
payload.Bytes = new byte[payloadLength];
}
else if (payload.Bytes.Length < payloadLength)
{
payload.Grow(payloadLength);
}
prox.ReadBytes(payload.Bytes, 0, payloadLength);
payload.Length = payloadLength;
thisPayload = payload;
}
else
{
thisPayload = null;
}
if (readOffsets)
{
int startOffset = offset + prox.ReadVInt32();
int endOffset = startOffset + prox.ReadVInt32();
if (writePositions)
{
if (writeOffsets)
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(startOffset >= 0 && endOffset >= startOffset, "startOffset={0},endOffset={1},offset={2}", startOffset, endOffset, offset);
}
postingsConsumer.AddPosition(position, thisPayload, startOffset, endOffset);
}
else
{
postingsConsumer.AddPosition(position, thisPayload, -1, -1);
}
}
offset = startOffset;
}
else if (writePositions)
{
postingsConsumer.AddPosition(position, thisPayload, -1, -1);
}
}
}
}
postingsConsumer.FinishDoc();
}
termsConsumer.FinishTerm(text, new TermStats(docFreq, writeTermFreq ? totalTermFreq : -1));
sumTotalTermFreq += totalTermFreq;
sumDocFreq += docFreq;
}
termsConsumer.Finish(writeTermFreq ? sumTotalTermFreq : -1, sumDocFreq, visitedDocs.Cardinality());
}
/// <summary>
/// Build a new <see cref="WAH8DocIdSet"/>. </summary>
public virtual WAH8DocIdSet Build()
{
if (cardinality == 0)
{
if (Debugging.AssertsEnabled)
{
Debugging.Assert(lastWordNum == -1);
}
return(EMPTY);
}
WriteSequence();
byte[] data = Arrays.CopyOf(@out.Bytes, @out.Length);
// Now build the index
int valueCount = (numSequences - 1) / indexInterval + 1;
MonotonicAppendingInt64Buffer indexPositions, indexWordNums;
if (valueCount <= 1)
{
indexPositions = indexWordNums = SINGLE_ZERO_BUFFER;
}
else
{
const int pageSize = 128;
int initialPageCount = (valueCount + pageSize - 1) / pageSize;
MonotonicAppendingInt64Buffer positions = new MonotonicAppendingInt64Buffer(initialPageCount, pageSize, PackedInt32s.COMPACT);
MonotonicAppendingInt64Buffer wordNums = new MonotonicAppendingInt64Buffer(initialPageCount, pageSize, PackedInt32s.COMPACT);
positions.Add(0L);
wordNums.Add(0L);
Iterator it = new Iterator(data, cardinality, int.MaxValue, SINGLE_ZERO_BUFFER, SINGLE_ZERO_BUFFER);
if (Debugging.AssertsEnabled)
{
Debugging.Assert([email protected] == 0);
Debugging.Assert(it.wordNum == -1);
}
for (int i = 1; i < valueCount; ++i)
{
// skip indexInterval sequences
for (int j = 0; j < indexInterval; ++j)
{
bool readSequence = it.ReadSequence();
if (Debugging.AssertsEnabled)
{
Debugging.Assert(readSequence);
}
it.SkipDirtyBytes();
}
int position = [email protected];
int wordNum = it.wordNum;
positions.Add(position);
wordNums.Add(wordNum + 1);
}
positions.Freeze();
wordNums.Freeze();
indexPositions = positions;
indexWordNums = wordNums;
}
return(new WAH8DocIdSet(data, cardinality, indexInterval, indexPositions, indexWordNums));
}
public skyetek_hid(String path)
{
MyDebugging = new Debugging(); // For viewing results of API calls via Debug.Write.
MyDeviceManagement = new DeviceManagement();
MyHid = new Hid();
readBuffer = new Queue<byte>(65);
ReadFinished = false;
ReadStart = 0;
myDevicePathName = path;
}
private void CorProcess_OnBreakpoint(object sender, Debugging.CorDebug.CorBreakpointEventArgs e)
{
//executeMDbgCommand("w");
// var sync = new AutoResetEvent(false);
/* O2Thread.mtaThread(() =>
{
var activeThread = CommandBase.Debugger.Processes.Active.Threads.Active;
var filename = activeThread.BottomFrame.SourcePosition.Path;
var line = activeThread.BottomFrame.SourcePosition.Line;
O2Messages.raiseO2MDbgBreakEvent(filename, line);
//sync.Set();
});*/
// first find the current thread (we need to use this.ActiveProcess.Threads since e.Thread doesn't give us the info we need
//MDbgThread e.Thread.id
// if (DateTime.Now.Millisecond < 100)
/* if (DI.o2MDbg.LogBreakpointEvent)
DI.log.info("*** BREAKPOINT >> " + O2MDbgBreakPoint.getActiveFrameFunctionName(e));
CorFrame activeFrame = e.Thread.ActiveFrame;
activeFrame.
//var activeThread = this.ActiveProcess.Threads.Active;
// MDbgSourcePosition pos = CommandBase.Debugger.Processes.Active.Threads.Active.CurrentSourcePosition;
activeFrame.Code.*/
// var filename = "filenameOfBreak";
// int line = 11;
//sync.WaitOne();
e.Continue = DI.o2MDbg.AutoContinueOnBreakPointEvent; // can't do this here since CommandBase.Debugger.Processes.Active.Threads.Active is only set after this is executed
}
