/// <summary> /// Merges the sortedset docvalues from <paramref name="toMerge"/>. /// <para> /// The default implementation calls <see cref="AddSortedSetField(FieldInfo, IEnumerable{BytesRef}, IEnumerable{long?}, IEnumerable{long?})"/>, passing /// an <see cref="IEnumerable{T}"/> that merges ordinals and values and filters deleted documents.</para> /// </summary> public virtual void MergeSortedSetField(FieldInfo fieldInfo, MergeState mergeState, IList <SortedSetDocValues> toMerge) { var readers = mergeState.Readers.ToArray(); var dvs = toMerge.ToArray(); // step 1: iterate thru each sub and mark terms still in use var liveTerms = new TermsEnum[dvs.Length]; for (int sub = 0; sub < liveTerms.Length; sub++) { var reader = readers[sub]; var dv = dvs[sub]; var liveDocs = reader.LiveDocs; if (liveDocs == null) { liveTerms[sub] = dv.GetTermsEnum(); } else { var bitset = new Int64BitSet(dv.ValueCount); for (int i = 0; i < reader.MaxDoc; i++) { if (liveDocs.Get(i)) { dv.SetDocument(i); long ord; while ((ord = dv.NextOrd()) != SortedSetDocValues.NO_MORE_ORDS) { bitset.Set(ord); } } } liveTerms[sub] = new BitsFilteredTermsEnum(dv.GetTermsEnum(), bitset); } } // step 2: create ordinal map (this conceptually does the "merging") var map = new OrdinalMap(this, liveTerms); // step 3: add field AddSortedSetField(fieldInfo, GetMergeSortedSetValuesEnumerable(map, dvs), // doc -> ord count GetMergeSortedSetDocToOrdCountEnumerable(readers, dvs), // ords GetMergeSortedSetOrdsEnumerable(readers, dvs, map) ); }
/// <summary> /// Merges the sorted docvalues from <paramref name="toMerge"/>. /// <para> /// The default implementation calls <see cref="AddSortedField(FieldInfo, IEnumerable{BytesRef}, IEnumerable{long?})"/>, passing /// an <see cref="IEnumerable{T}"/> that merges ordinals and values and filters deleted documents.</para> /// </summary> public virtual void MergeSortedField(FieldInfo fieldInfo, MergeState mergeState, IList <SortedDocValues> toMerge) { AtomicReader[] readers = mergeState.Readers.ToArray(); SortedDocValues[] dvs = toMerge.ToArray(); // step 1: iterate thru each sub and mark terms still in use var liveTerms = new TermsEnum[dvs.Length]; for (int sub = 0; sub < liveTerms.Length; sub++) { AtomicReader reader = readers[sub]; SortedDocValues dv = dvs[sub]; IBits liveDocs = reader.LiveDocs; if (liveDocs == null) { liveTerms[sub] = dv.GetTermsEnum(); } else { var bitset = new Int64BitSet(dv.ValueCount); for (int i = 0; i < reader.MaxDoc; i++) { if (liveDocs.Get(i)) { int ord = dv.GetOrd(i); if (ord >= 0) { bitset.Set(ord); } } } liveTerms[sub] = new BitsFilteredTermsEnum(dv.GetTermsEnum(), bitset); } } // step 2: create ordinal map (this conceptually does the "merging") var map = new OrdinalMap(this, liveTerms); // step 3: add field AddSortedField(fieldInfo, GetMergeSortValuesEnumerable(map, dvs), // doc -> ord GetMergeSortedFieldDocToOrdEnumerable(readers, dvs, map) ); }
/* private class IterableAnonymousInnerClassHelper3 : IEnumerable<BytesRef> { private readonly DocValuesConsumer OuterInstance; private SortedDocValues[] Dvs; private OrdinalMap Map; public IterableAnonymousInnerClassHelper3(DocValuesConsumer outerInstance, SortedDocValues[] dvs, OrdinalMap map) { this.OuterInstance = outerInstance; this.Dvs = dvs; this.Map = map; } // ord -> value public virtual IEnumerator<BytesRef> GetEnumerator() { return new IteratorAnonymousInnerClassHelper3(this); } private class IteratorAnonymousInnerClassHelper3 : IEnumerator<BytesRef> { private readonly IterableAnonymousInnerClassHelper3 OuterInstance; public IteratorAnonymousInnerClassHelper3(IterableAnonymousInnerClassHelper3 outerInstance) { this.OuterInstance = outerInstance; scratch = new BytesRef(); } internal readonly BytesRef scratch; internal int currentOrd; public virtual bool HasNext() { return currentOrd < OuterInstance.Map.ValueCount; } public virtual BytesRef Next() { if (!HasNext()) { throw new Exception(); } int segmentNumber = OuterInstance.Map.GetFirstSegmentNumber(currentOrd); int segmentOrd = (int)OuterInstance.Map.GetFirstSegmentOrd(currentOrd); OuterInstance.Dvs[segmentNumber].LookupOrd(segmentOrd, scratch); currentOrd++; return scratch; } public virtual void Remove() { throw new System.NotSupportedException(); } } } private class IterableAnonymousInnerClassHelper4 : IEnumerable<Number> { private readonly DocValuesConsumer OuterInstance; private AtomicReader[] Readers; private SortedDocValues[] Dvs; private OrdinalMap Map; public IterableAnonymousInnerClassHelper4(DocValuesConsumer outerInstance, AtomicReader[] readers, SortedDocValues[] dvs, OrdinalMap map) { this.OuterInstance = outerInstance; this.Readers = readers; this.Dvs = dvs; this.Map = map; } public virtual IEnumerator<Number> GetEnumerator() { return new IteratorAnonymousInnerClassHelper4(this); } private class IteratorAnonymousInnerClassHelper4 : IEnumerator<Number> { private readonly IterableAnonymousInnerClassHelper4 OuterInstance; public IteratorAnonymousInnerClassHelper4(IterableAnonymousInnerClassHelper4 outerInstance) { this.OuterInstance = outerInstance; readerUpto = -1; } internal int readerUpto; internal int docIDUpto; internal int nextValue; internal AtomicReader currentReader; internal Bits currentLiveDocs; internal bool nextIsSet; public virtual bool HasNext() { return nextIsSet || SetNext(); } public virtual void Remove() { throw new System.NotSupportedException(); } public virtual Number Next() { if (!HasNext()) { throw new NoSuchElementException(); } Debug.Assert(nextIsSet); nextIsSet = false; // TODO make a mutable number return nextValue; } private bool SetNext() { while (true) { if (readerUpto == OuterInstance.Readers.Length) { return false; } if (currentReader == null || docIDUpto == currentReader.MaxDoc) { readerUpto++; if (readerUpto < OuterInstance.Readers.Length) { currentReader = OuterInstance.Readers[readerUpto]; currentLiveDocs = currentReader.LiveDocs; } docIDUpto = 0; continue; } if (currentLiveDocs == null || currentLiveDocs.get(docIDUpto)) { nextIsSet = true; int segOrd = OuterInstance.Dvs[readerUpto].GetOrd(docIDUpto); nextValue = segOrd == -1 ? - 1 : (int) OuterInstance.Map.GetGlobalOrd(readerUpto, segOrd); docIDUpto++; return true; } docIDUpto++; } } } }*/ /// <summary> /// Merges the sortedset docvalues from <code>toMerge</code>. /// <p> /// The default implementation calls <seealso cref="#addSortedSetField"/>, passing /// an Iterable that merges ordinals and values and filters deleted documents . /// </summary> public virtual void MergeSortedSetField(FieldInfo fieldInfo, MergeState mergeState, IList<SortedSetDocValues> toMerge) { var readers = mergeState.Readers.ToArray(); var dvs = toMerge.ToArray(); // step 1: iterate thru each sub and mark terms still in use var liveTerms = new TermsEnum[dvs.Length]; for (int sub = 0; sub < liveTerms.Length; sub++) { var reader = readers[sub]; var dv = dvs[sub]; var liveDocs = reader.LiveDocs; if (liveDocs == null) { liveTerms[sub] = dv.TermsEnum(); } else { var bitset = new LongBitSet(dv.ValueCount); for (int i = 0; i < reader.MaxDoc; i++) { if (liveDocs.Get(i)) { dv.Document = i; long ord; while ((ord = dv.NextOrd()) != SortedSetDocValues.NO_MORE_ORDS) { bitset.Set(ord); } } } liveTerms[sub] = new BitsFilteredTermsEnum(dv.TermsEnum(), bitset); } } // step 2: create ordinal map (this conceptually does the "merging") var map = new OrdinalMap(this, liveTerms); // step 3: add field AddSortedSetField(fieldInfo, GetMergeSortedSetValuesEnumerable(map, dvs), // doc -> ord count GetMergeSortedSetDocToOrdCountEnumerable(readers, dvs), // ords GetMergeSortedSetOrdsEnumerable(readers, dvs, map) ); }
/// <summary> /// Merges the sorted docvalues from <code>toMerge</code>. /// <p> /// The default implementation calls <seealso cref="#addSortedField"/>, passing /// an Iterable that merges ordinals and values and filters deleted documents.</p> /// </summary> public virtual void MergeSortedField(FieldInfo fieldInfo, MergeState mergeState, IList<SortedDocValues> toMerge) { AtomicReader[] readers = mergeState.Readers.ToArray(); SortedDocValues[] dvs = toMerge.ToArray(); // step 1: iterate thru each sub and mark terms still in use var liveTerms = new TermsEnum[dvs.Length]; for (int sub = 0; sub < liveTerms.Length; sub++) { AtomicReader reader = readers[sub]; SortedDocValues dv = dvs[sub]; Bits liveDocs = reader.LiveDocs; if (liveDocs == null) { liveTerms[sub] = dv.TermsEnum(); } else { var bitset = new LongBitSet(dv.ValueCount); for (int i = 0; i < reader.MaxDoc; i++) { if (liveDocs.Get(i)) { int ord = dv.GetOrd(i); if (ord >= 0) { bitset.Set(ord); } } } liveTerms[sub] = new BitsFilteredTermsEnum(dv.TermsEnum(), bitset); } } // step 2: create ordinal map (this conceptually does the "merging") var map = new OrdinalMap(this, liveTerms); // step 3: add field AddSortedField(fieldInfo, GetMergeSortValuesEnumerable(map, dvs), // doc -> ord GetMergeSortedFieldDocToOrdEnumerable(readers, dvs, map) ); }