public static void BeforeClassSortingAtomicReaderTest()
{
// sort the index by id (as integer, in NUMERIC_DV_FIELD)
Sort sort = new Sort(new SortField(NUMERIC_DV_FIELD, SortField.Type_e.INT));
Sorter.DocMap docMap = new Sorter(sort).Sort(reader);
// Sorter.compute also sorts the values
NumericDocValues dv = reader.GetNumericDocValues(NUMERIC_DV_FIELD);
sortedValues = new int[reader.MaxDoc];
for (int i = 0; i < reader.MaxDoc; ++i)
{
sortedValues[docMap.OldToNew(i)] = (int)dv.Get(i);
}
if (VERBOSE)
{
Console.WriteLine("docMap: " + docMap);
Console.WriteLine("sortedValues: " + Arrays.ToString(sortedValues));
}
// sort the index by id (as integer, in NUMERIC_DV_FIELD)
reader = SortingAtomicReader.Wrap(reader, sort);
if (VERBOSE)
{
Console.WriteLine("mapped-deleted-docs: ");
Bits mappedLiveDocs = reader.LiveDocs;
for (int i = 0; i < mappedLiveDocs.Length(); i++)
{
if (!mappedLiveDocs.Get(i))
{
Console.WriteLine(i + " ");
}
}
Console.WriteLine();
}
TestUtil.CheckReader(reader);
}
public void Test()
{
RandomIndexWriter writer;
DirectoryReader indexReader;
int numParents = AtLeast(200);
IndexWriterConfig cfg = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random()));
cfg.SetMergePolicy(NewLogMergePolicy());
using (writer = new RandomIndexWriter(Random(), NewDirectory(), cfg))
{
Document parentDoc = new Document();
NumericDocValuesField parentVal = new NumericDocValuesField("parent_val", 0L);
parentDoc.Add(parentVal);
StringField parent = new StringField("parent", "true", Field.Store.YES);
parentDoc.Add(parent);
for (int i = 0; i < numParents; ++i)
{
List<Document> documents = new List<Document>();
int numChildren = Random().nextInt(10);
for (int j = 0; j < numChildren; ++j)
{
Document childDoc = new Document();
childDoc.Add(new NumericDocValuesField("child_val", Random().nextInt(5)));
documents.Add(childDoc);
}
parentVal.LongValue = (Random().nextInt(50));
documents.Add(parentDoc);
writer.AddDocuments(documents);
}
writer.ForceMerge(1);
indexReader = writer.Reader;
}
AtomicReader reader = GetOnlySegmentReader(indexReader);
Filter parentsFilter = new FixedBitSetCachingWrapperFilter(new QueryWrapperFilter(new TermQuery(new Term("parent", "true"))));
FixedBitSet parentBits = (FixedBitSet)parentsFilter.GetDocIdSet(reader.AtomicContext, null);
NumericDocValues parentValues = reader.GetNumericDocValues("parent_val");
NumericDocValues childValues = reader.GetNumericDocValues("child_val");
Sort parentSort = new Sort(new SortField("parent_val", SortField.Type_e.LONG));
Sort childSort = new Sort(new SortField("child_val", SortField.Type_e.LONG));
Sort sort = new Sort(new SortField("custom", new BlockJoinComparatorSource(parentsFilter, parentSort, childSort)));
Sorter sorter = new Sorter(sort);
Sorter.DocMap docMap = sorter.Sort(reader);
assertEquals(reader.MaxDoc, docMap.Count);
int[] children = new int[1];
int numChildren2 = 0;
int previousParent = -1;
for (int i = 0; i < docMap.Count; ++i)
{
int oldID = docMap.NewToOld(i);
if (parentBits.Get(oldID))
{
// check that we have the right children
for (int j = 0; j < numChildren2; ++j)
{
assertEquals(oldID, parentBits.NextSetBit(children[j]));
}
// check that children are sorted
for (int j = 1; j < numChildren2; ++j)
{
int doc1 = children[j - 1];
int doc2 = children[j];
if (childValues.Get(doc1) == childValues.Get(doc2))
{
assertTrue(doc1 < doc2); // sort is stable
}
else
{
assertTrue(childValues.Get(doc1) < childValues.Get(doc2));
}
}
// check that parents are sorted
if (previousParent != -1)
{
if (parentValues.Get(previousParent) == parentValues.Get(oldID))
{
assertTrue(previousParent < oldID);
}
else
{
assertTrue(parentValues.Get(previousParent) < parentValues.Get(oldID));
}
}
// reset
previousParent = oldID;
numChildren2 = 0;
}
else
{
children = ArrayUtil.Grow(children, numChildren2 + 1);
children[numChildren2++] = oldID;
}
}
indexReader.Dispose();
writer.w.Directory.Dispose();
}
internal readonly Sorter.DocMap docMap; // pkg-protected to avoid synthetic accessor methods
private SortingAtomicReader(AtomicReader @in, Sorter.DocMap docMap) : base(@in)
{
this.docMap = docMap;
}
/// <summary>
/// Create a new <see cref="MergePolicy"/> that sorts documents with the given <paramref name="sort"/>.
/// </summary>
public SortingMergePolicy(MergePolicy @in, Sort sort)
{
this.@in = @in;
this.sorter = new Sorter(sort);
this.sort = sort;
}
// LUCENENET TODO: Make the FieldInfo.IndexOptions non nullable
public SortingTerms(Terms @in, FieldInfo.IndexOptions? indexOptions, Sorter.DocMap docMap) : base(@in)
{
this.docMap = docMap;
this.indexOptions = indexOptions;
}
/// <summary>
/// Expert: same as <see cref="Wrap(AtomicReader, Sort)"/> but operates directly on a <see cref="Sorter.DocMap"/>.
/// </summary>
internal static AtomicReader Wrap(AtomicReader reader, Sorter.DocMap docMap)
{
if (docMap == null)
{
// the reader is already sorter
return reader;
}
if (reader.MaxDoc != docMap.Count)
{
throw new System.ArgumentException("reader.maxDoc() should be equal to docMap.size(), got" + reader.MaxDoc + " != " + docMap.Count);
}
Debug.Assert(Sorter.IsConsistent(docMap));
return new SortingAtomicReader(reader, docMap);
}
public SortingFields(Fields @in, FieldInfos infos, Sorter.DocMap docMap)
: base(@in)
{
this.docMap = docMap;
this.infos = infos;
}
internal SortingDocsAndPositionsEnum(int maxDoc, SortingDocsAndPositionsEnum reuse, DocsAndPositionsEnum @in, Sorter.DocMap docMap, bool storeOffsets)
: base(@in)
{
this.maxDoc = maxDoc;
this.storeOffsets = storeOffsets;
if (reuse != null)
{
docs = reuse.docs;
offsets = reuse.offsets;
payload = reuse.payload;
file = reuse.file;
if (reuse.maxDoc == maxDoc)
{
sorter = reuse.sorter;
}
else
{
sorter = new DocOffsetSorter(maxDoc);
}
}
else
{
docs = new int[32];
offsets = new long[32];
payload = new BytesRef(32);
file = new RAMFile();
sorter = new DocOffsetSorter(maxDoc);
}
using (IndexOutput @out = new RAMOutputStream(file))
{
int doc;
int i = 0;
while ((doc = @in.NextDoc()) != DocIdSetIterator.NO_MORE_DOCS)
{
if (i == docs.Length)
{
int newLength = ArrayUtil.Oversize(i + 1, 4);
docs = Arrays.CopyOf(docs, newLength);
offsets = Arrays.CopyOf(offsets, newLength);
}
docs[i] = docMap.OldToNew(doc);
offsets[i] = @out.FilePointer;
AddPositions(@in, @out);
i++;
}
upto = i;
sorter.Reset(docs, offsets);
sorter.Sort(0, upto);
}
this.postingInput = new RAMInputStream("", file);
}
internal SortingSortedSetDocValues(SortedSetDocValues @in, Sorter.DocMap docMap)
{
this.@in = @in;
this.docMap = docMap;
}
internal SortingDocsEnum(int maxDoc, SortingDocsEnum reuse, DocsEnum @in, bool withFreqs, Sorter.DocMap docMap)
: base(@in)
{
this.maxDoc = maxDoc;
this.withFreqs = withFreqs;
if (reuse != null)
{
if (reuse.maxDoc == maxDoc)
{
sorter = reuse.sorter;
}
else
{
sorter = new DocFreqSorter(maxDoc);
}
docs = reuse.docs;
freqs = reuse.freqs; // maybe null
}
else
{
docs = new int[64];
sorter = new DocFreqSorter(maxDoc);
}
docIt = -1;
int i = 0;
int doc;
if (withFreqs)
{
if (freqs == null || freqs.Length < docs.Length)
{
freqs = new int[docs.Length];
}
while ((doc = @in.NextDoc()) != DocIdSetIterator.NO_MORE_DOCS)
{
if (i >= docs.Length)
{
docs = ArrayUtil.Grow(docs, docs.Length + 1);
freqs = ArrayUtil.Grow(freqs, freqs.Length + 1);
}
docs[i] = docMap.OldToNew(doc);
freqs[i] = @in.Freq();
++i;
}
}
else
{
freqs = null;
while ((doc = @in.NextDoc()) != DocIdSetIterator.NO_MORE_DOCS)
{
if (i >= docs.Length)
{
docs = ArrayUtil.Grow(docs, docs.Length + 1);
}
docs[i++] = docMap.OldToNew(doc);
}
}
// TimSort can save much time compared to other sorts in case of
// reverse sorting, or when sorting a concatenation of sorted readers
sorter.Reset(docs, freqs);
sorter.Sort(0, i);
upto = i;
}
public SortingBits(Bits @in, Sorter.DocMap docMap)
{
this.@in = @in;
this.docMap = docMap;
}
public SortingNumericDocValues(NumericDocValues @in, Sorter.DocMap docMap)
{
this.@in = @in;
this.docMap = docMap;
}
internal SortingBinaryDocValues(BinaryDocValues @in, Sorter.DocMap docMap)
{
this.@in = @in;
this.docMap = docMap;
}
public SortingTermsEnum(TermsEnum @in, Sorter.DocMap docMap, FieldInfo.IndexOptions? indexOptions)
: base(@in)
{
this.docMap = docMap;
this.indexOptions = indexOptions;
}
public void Sort <TSorter>(ref Sorter <T, TSorter> sorter) where TSorter : struct, IComparer <T>
{
sorter.Sort(_items, 0, (int)_size);
}