/// <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)
);
}
public virtual void Test()
{
IFieldCache cache = FieldCache.DEFAULT;
FieldCache.Doubles doubles = cache.GetDoubles(Reader, "theDouble", Random().NextBoolean());
Assert.AreSame(doubles, cache.GetDoubles(Reader, "theDouble", Random().NextBoolean()), "Second request to cache return same array");
Assert.AreSame(doubles, cache.GetDoubles(Reader, "theDouble", FieldCache.DEFAULT_DOUBLE_PARSER, Random().NextBoolean()), "Second request with explicit parser return same array");
for (int i = 0; i < NUM_DOCS; i++)
{
Assert.IsTrue(doubles.Get(i) == (double.MaxValue - i), doubles.Get(i) + " does not equal: " + (double.MaxValue - i));
}
FieldCache.Longs longs = cache.GetLongs(Reader, "theLong", Random().NextBoolean());
Assert.AreSame(longs, cache.GetLongs(Reader, "theLong", Random().NextBoolean()), "Second request to cache return same array");
Assert.AreSame(longs, cache.GetLongs(Reader, "theLong", FieldCache.DEFAULT_LONG_PARSER, Random().NextBoolean()), "Second request with explicit parser return same array");
for (int i = 0; i < NUM_DOCS; i++)
{
Assert.IsTrue(longs.Get(i) == (long.MaxValue - i), longs.Get(i) + " does not equal: " + (long.MaxValue - i) + " i=" + i);
}
FieldCache.Bytes bytes = cache.GetBytes(Reader, "theByte", Random().NextBoolean());
Assert.AreSame(bytes, cache.GetBytes(Reader, "theByte", Random().NextBoolean()), "Second request to cache return same array");
Assert.AreSame(bytes, cache.GetBytes(Reader, "theByte", FieldCache.DEFAULT_BYTE_PARSER, Random().NextBoolean()), "Second request with explicit parser return same array");
for (int i = 0; i < NUM_DOCS; i++)
{
Assert.IsTrue(bytes.Get(i) == (sbyte)(sbyte.MaxValue - i), bytes.Get(i) + " does not equal: " + (sbyte.MaxValue - i));
}
FieldCache.Shorts shorts = cache.GetShorts(Reader, "theShort", Random().NextBoolean());
Assert.AreSame(shorts, cache.GetShorts(Reader, "theShort", Random().NextBoolean()), "Second request to cache return same array");
Assert.AreSame(shorts, cache.GetShorts(Reader, "theShort", FieldCache.DEFAULT_SHORT_PARSER, Random().NextBoolean()), "Second request with explicit parser return same array");
for (int i = 0; i < NUM_DOCS; i++)
{
Assert.IsTrue(shorts.Get(i) == (short)(short.MaxValue - i), shorts.Get(i) + " does not equal: " + (short.MaxValue - i));
}
FieldCache.Ints ints = cache.GetInts(Reader, "theInt", Random().NextBoolean());
Assert.AreSame(ints, cache.GetInts(Reader, "theInt", Random().NextBoolean()), "Second request to cache return same array");
Assert.AreSame(ints, cache.GetInts(Reader, "theInt", FieldCache.DEFAULT_INT_PARSER, Random().NextBoolean()), "Second request with explicit parser return same array");
for (int i = 0; i < NUM_DOCS; i++)
{
Assert.IsTrue(ints.Get(i) == (int.MaxValue - i), ints.Get(i) + " does not equal: " + (int.MaxValue - i));
}
FieldCache.Floats floats = cache.GetFloats(Reader, "theFloat", Random().NextBoolean());
Assert.AreSame(floats, cache.GetFloats(Reader, "theFloat", Random().NextBoolean()), "Second request to cache return same array");
Assert.AreSame(floats, cache.GetFloats(Reader, "theFloat", FieldCache.DEFAULT_FLOAT_PARSER, Random().NextBoolean()), "Second request with explicit parser return same array");
for (int i = 0; i < NUM_DOCS; i++)
{
Assert.IsTrue(floats.Get(i) == (float.MaxValue - i), floats.Get(i) + " does not equal: " + (float.MaxValue - i));
}
Bits docsWithField = cache.GetDocsWithField(Reader, "theLong");
Assert.AreSame(docsWithField, cache.GetDocsWithField(Reader, "theLong"), "Second request to cache return same array");
Assert.IsTrue(docsWithField is Bits_MatchAllBits, "docsWithField(theLong) must be class Bits.MatchAllBits");
Assert.IsTrue(docsWithField.Length() == NUM_DOCS, "docsWithField(theLong) Size: " + docsWithField.Length() + " is not: " + NUM_DOCS);
for (int i = 0; i < docsWithField.Length(); i++)
{
Assert.IsTrue(docsWithField.Get(i));
}
docsWithField = cache.GetDocsWithField(Reader, "sparse");
Assert.AreSame(docsWithField, cache.GetDocsWithField(Reader, "sparse"), "Second request to cache return same array");
Assert.IsFalse(docsWithField is Bits_MatchAllBits, "docsWithField(sparse) must not be class Bits.MatchAllBits");
Assert.IsTrue(docsWithField.Length() == NUM_DOCS, "docsWithField(sparse) Size: " + docsWithField.Length() + " is not: " + NUM_DOCS);
for (int i = 0; i < docsWithField.Length(); i++)
{
Assert.AreEqual(i % 2 == 0, docsWithField.Get(i));
}
// getTermsIndex
SortedDocValues termsIndex = cache.GetTermsIndex(Reader, "theRandomUnicodeString");
Assert.AreSame(termsIndex, cache.GetTermsIndex(Reader, "theRandomUnicodeString"), "Second request to cache return same array");
BytesRef br = new BytesRef();
for (int i = 0; i < NUM_DOCS; i++)
{
BytesRef term;
int ord = termsIndex.GetOrd(i);
if (ord == -1)
{
term = null;
}
else
{
termsIndex.LookupOrd(ord, br);
term = br;
}
string s = term == null ? null : term.Utf8ToString();
Assert.IsTrue(UnicodeStrings[i] == null || UnicodeStrings[i].Equals(s), "for doc " + i + ": " + s + " does not equal: " + UnicodeStrings[i]);
}
int nTerms = termsIndex.ValueCount;
TermsEnum tenum = termsIndex.TermsEnum();
BytesRef val = new BytesRef();
for (int i = 0; i < nTerms; i++)
{
BytesRef val1 = tenum.Next();
termsIndex.LookupOrd(i, val);
// System.out.println("i="+i);
Assert.AreEqual(val, val1);
}
// seek the enum around (note this isn't a great test here)
int num = AtLeast(100);
for (int i = 0; i < num; i++)
{
int k = Random().Next(nTerms);
termsIndex.LookupOrd(k, val);
Assert.AreEqual(TermsEnum.SeekStatus.FOUND, tenum.SeekCeil(val));
Assert.AreEqual(val, tenum.Term());
}
for (int i = 0; i < nTerms; i++)
{
termsIndex.LookupOrd(i, val);
Assert.AreEqual(TermsEnum.SeekStatus.FOUND, tenum.SeekCeil(val));
Assert.AreEqual(val, tenum.Term());
}
// test bad field
termsIndex = cache.GetTermsIndex(Reader, "bogusfield");
// getTerms
BinaryDocValues terms = cache.GetTerms(Reader, "theRandomUnicodeString", true);
Assert.AreSame(terms, cache.GetTerms(Reader, "theRandomUnicodeString", true), "Second request to cache return same array");
Bits bits = cache.GetDocsWithField(Reader, "theRandomUnicodeString");
for (int i = 0; i < NUM_DOCS; i++)
{
terms.Get(i, br);
BytesRef term;
if (!bits.Get(i))
{
term = null;
}
else
{
term = br;
}
string s = term == null ? null : term.Utf8ToString();
Assert.IsTrue(UnicodeStrings[i] == null || UnicodeStrings[i].Equals(s), "for doc " + i + ": " + s + " does not equal: " + UnicodeStrings[i]);
}
// test bad field
terms = cache.GetTerms(Reader, "bogusfield", false);
// getDocTermOrds
SortedSetDocValues termOrds = cache.GetDocTermOrds(Reader, "theRandomUnicodeMultiValuedField");
int numEntries = cache.CacheEntries.Length;
// ask for it again, and check that we didnt create any additional entries:
termOrds = cache.GetDocTermOrds(Reader, "theRandomUnicodeMultiValuedField");
Assert.AreEqual(numEntries, cache.CacheEntries.Length);
for (int i = 0; i < NUM_DOCS; i++)
{
termOrds.Document = i;
// this will remove identical terms. A DocTermOrds doesn't return duplicate ords for a docId
IList <BytesRef> values = new List <BytesRef>(new /*Linked*/ HashSet <BytesRef>(Arrays.AsList(MultiValued[i])));
foreach (BytesRef v in values)
{
if (v == null)
{
// why does this test use null values... instead of an empty list: confusing
break;
}
long ord = termOrds.NextOrd();
Debug.Assert(ord != SortedSetDocValues.NO_MORE_ORDS);
BytesRef scratch = new BytesRef();
termOrds.LookupOrd(ord, scratch);
Assert.AreEqual(v, scratch);
}
Assert.AreEqual(SortedSetDocValues.NO_MORE_ORDS, termOrds.NextOrd());
}
// test bad field
termOrds = cache.GetDocTermOrds(Reader, "bogusfield");
Assert.IsTrue(termOrds.ValueCount == 0);
FieldCache.DEFAULT.PurgeByCacheKey(Reader.CoreCacheKey);
}
public override TermsEnum Iterator(TermsEnum reuse)
{
return(Fcsi.TermsEnum());
}