public virtual void TestDocsAndPositionsEnumStart()
{
Directory dir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir);
Document doc = new Document();
doc.Add(NewTextField("foo", "bar", Field.Store.NO));
writer.AddDocument(doc);
DirectoryReader reader = writer.Reader;
AtomicReader r = GetOnlySegmentReader(reader);
DocsAndPositionsEnum disi = r.TermPositionsEnum(new Term("foo", "bar"));
int docid = disi.DocID();
Assert.AreEqual(-1, docid);
Assert.IsTrue(disi.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
// now reuse and check again
TermsEnum te = r.Terms("foo").Iterator(null);
Assert.IsTrue(te.SeekExact(new BytesRef("bar")));
disi = te.DocsAndPositions(null, disi);
docid = disi.DocID();
Assert.AreEqual(-1, docid);
Assert.IsTrue(disi.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
writer.Dispose();
r.Dispose();
dir.Dispose();
}
public virtual void TestSeek()
{
Directory directory = NewDirectory();
IndexWriter writer = new IndexWriter(directory, NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random())));
for (int i = 0; i < 10; i++)
{
Document doc = new Document();
doc.Add(NewTextField(this.Field, "a b", Documents.Field.Store.YES));
writer.AddDocument(doc);
}
writer.Dispose();
IndexReader reader = DirectoryReader.Open(directory);
DocsAndPositionsEnum tp = MultiFields.GetTermPositionsEnum(reader, MultiFields.GetLiveDocs(reader), this.Field, new BytesRef("b"));
for (int i = 0; i < 10; i++)
{
tp.NextDoc();
Assert.AreEqual(tp.DocID(), i);
Assert.AreEqual(tp.NextPosition(), 1);
}
tp = MultiFields.GetTermPositionsEnum(reader, MultiFields.GetLiveDocs(reader), this.Field, new BytesRef("a"));
for (int i = 0; i < 10; i++)
{
tp.NextDoc();
Assert.AreEqual(tp.DocID(), i);
Assert.AreEqual(tp.NextPosition(), 0);
}
reader.Dispose();
directory.Dispose();
}
public virtual void CheckSkipTo(DocsAndPositionsEnum tp, int target, int maxCounter)
{
tp.Advance(target);
if (maxCounter < Counter)
{
Assert.Fail("Too many bytes read: " + Counter + " vs " + maxCounter);
}
Assert.AreEqual(target, tp.DocID(), "Wrong document " + tp.DocID() + " after skipTo target " + target);
Assert.AreEqual(1, tp.Freq(), "Frequency is not 1: " + tp.Freq());
tp.NextPosition();
BytesRef b = tp.Payload;
Assert.AreEqual(1, b.Length);
Assert.AreEqual((sbyte)target, (sbyte)b.Bytes[b.Offset], "Wrong payload for the target " + target + ": " + (sbyte)b.Bytes[b.Offset]);
}
public virtual void CheckSkipTo(DocsAndPositionsEnum tp, int target, int maxCounter)
{
tp.Advance(target);
if (maxCounter < Counter)
{
Assert.Fail("Too many bytes read: " + Counter + " vs " + maxCounter);
}
Assert.AreEqual(target, tp.DocID(), "Wrong document " + tp.DocID() + " after skipTo target " + target);
Assert.AreEqual(1, tp.Freq(), "Frequency is not 1: " + tp.Freq());
tp.NextPosition();
BytesRef b = tp.Payload;
Assert.AreEqual(1, b.Length);
Assert.AreEqual((sbyte)target, b.Bytes[b.Offset], "Wrong payload for the target " + target + ": " + b.Bytes[b.Offset]);
}
public virtual void TestTerms()
{
Fields fields = MultiFields.GetFields(Reader);
foreach (string field in fields)
{
Terms terms = fields.Terms(field);
Assert.IsNotNull(terms);
TermsEnum termsEnum = terms.Iterator(null);
while (termsEnum.Next() != null)
{
BytesRef term = termsEnum.Term();
Assert.IsTrue(term != null);
string fieldValue = (string)DocHelper.NameValues[field];
Assert.IsTrue(fieldValue.IndexOf(term.Utf8ToString()) != -1);
}
}
DocsEnum termDocs = TestUtil.Docs(Random(), Reader, DocHelper.TEXT_FIELD_1_KEY, new BytesRef("field"), MultiFields.GetLiveDocs(Reader), null, 0);
Assert.IsTrue(termDocs.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
termDocs = TestUtil.Docs(Random(), Reader, DocHelper.NO_NORMS_KEY, new BytesRef(DocHelper.NO_NORMS_TEXT), MultiFields.GetLiveDocs(Reader), null, 0);
Assert.IsTrue(termDocs.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
DocsAndPositionsEnum positions = MultiFields.GetTermPositionsEnum(Reader, MultiFields.GetLiveDocs(Reader), DocHelper.TEXT_FIELD_1_KEY, new BytesRef("field"));
// NOTE: prior rev of this test was failing to first
// call next here:
Assert.IsTrue(positions.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
Assert.IsTrue(positions.DocID() == 0);
Assert.IsTrue(positions.NextPosition() >= 0);
}
public AssertingDocsAndPositionsEnum(DocsAndPositionsEnum @in)
: base(@in)
{
int docid = @in.DocID();
Debug.Assert(docid == -1, "invalid initial doc id: " + docid);
Doc = -1;
}
public virtual void TestFilterIndexReader()
{
Directory directory = NewDirectory();
IndexWriter writer = new IndexWriter(directory, NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random())));
Document d1 = new Document();
d1.Add(NewTextField("default", "one two", Field.Store.YES));
writer.AddDocument(d1);
Document d2 = new Document();
d2.Add(NewTextField("default", "one three", Field.Store.YES));
writer.AddDocument(d2);
Document d3 = new Document();
d3.Add(NewTextField("default", "two four", Field.Store.YES));
writer.AddDocument(d3);
writer.Dispose();
Directory target = NewDirectory();
// We mess with the postings so this can fail:
((BaseDirectoryWrapper)target).CrossCheckTermVectorsOnClose = false;
writer = new IndexWriter(target, NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random())));
IndexReader reader = new TestReader(DirectoryReader.Open(directory));
writer.AddIndexes(reader);
writer.Dispose();
reader.Dispose();
reader = DirectoryReader.Open(target);
TermsEnum terms = MultiFields.GetTerms(reader, "default").Iterator(null);
while (terms.Next() != null)
{
Assert.IsTrue(terms.Term().Utf8ToString().IndexOf('e') != -1);
}
Assert.AreEqual(TermsEnum.SeekStatus.FOUND, terms.SeekCeil(new BytesRef("one")));
DocsAndPositionsEnum positions = terms.DocsAndPositions(MultiFields.GetLiveDocs(reader), null);
while (positions.NextDoc() != DocIdSetIterator.NO_MORE_DOCS)
{
Assert.IsTrue((positions.DocID() % 2) == 1);
}
reader.Dispose();
directory.Dispose();
target.Dispose();
}
public virtual void TestPositionsSimple()
{
Directory directory = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(Random(), directory, NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random())));
for (int i = 0; i < 39; i++)
{
Document doc = new Document();
FieldType customType = new FieldType(TextField.TYPE_NOT_STORED);
customType.OmitNorms = true;
doc.Add(NewField(FieldName, "1 2 3 4 5 6 7 8 9 10 " + "1 2 3 4 5 6 7 8 9 10 " + "1 2 3 4 5 6 7 8 9 10 " + "1 2 3 4 5 6 7 8 9 10", customType));
writer.AddDocument(doc);
}
IndexReader reader = writer.Reader;
writer.Dispose();
int num = AtLeast(13);
for (int i = 0; i < num; i++)
{
BytesRef bytes = new BytesRef("1");
IndexReaderContext topReaderContext = reader.Context;
foreach (AtomicReaderContext atomicReaderContext in topReaderContext.Leaves)
{
DocsAndPositionsEnum docsAndPosEnum = GetDocsAndPositions((AtomicReader)atomicReaderContext.Reader, bytes, null);
Assert.IsNotNull(docsAndPosEnum);
if (atomicReaderContext.Reader.MaxDoc == 0)
{
continue;
}
int advance = docsAndPosEnum.Advance(Random().Next(atomicReaderContext.Reader.MaxDoc));
do
{
string msg = "Advanced to: " + advance + " current doc: " + docsAndPosEnum.DocID(); // TODO: + " usePayloads: " + usePayload;
Assert.AreEqual(4, docsAndPosEnum.Freq(), msg);
Assert.AreEqual(0, docsAndPosEnum.NextPosition(), msg);
Assert.AreEqual(4, docsAndPosEnum.Freq(), msg);
Assert.AreEqual(10, docsAndPosEnum.NextPosition(), msg);
Assert.AreEqual(4, docsAndPosEnum.Freq(), msg);
Assert.AreEqual(20, docsAndPosEnum.NextPosition(), msg);
Assert.AreEqual(4, docsAndPosEnum.Freq(), msg);
Assert.AreEqual(30, docsAndPosEnum.NextPosition(), msg);
} while (docsAndPosEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
}
}
reader.Dispose();
directory.Dispose();
}
private void PrintSegment(StreamWriter @out, SegmentCommitInfo si)
{
SegmentReader reader = new SegmentReader(si, DirectoryReader.DEFAULT_TERMS_INDEX_DIVISOR, NewIOContext(Random()));
for (int i = 0; i < reader.NumDocs; i++)
{
@out.WriteLine(reader.Document(i));
}
Fields fields = reader.Fields;
foreach (string field in fields)
{
Terms terms = fields.Terms(field);
Assert.IsNotNull(terms);
TermsEnum tis = terms.Iterator(null);
while (tis.Next() != null)
{
@out.Write(" term=" + field + ":" + tis.Term());
@out.WriteLine(" DF=" + tis.DocFreq());
DocsAndPositionsEnum positions = tis.DocsAndPositions(reader.LiveDocs, null);
while (positions.NextDoc() != DocIdSetIterator.NO_MORE_DOCS)
{
@out.Write(" doc=" + positions.DocID());
@out.Write(" TF=" + positions.Freq());
@out.Write(" pos=");
@out.Write(positions.NextPosition());
for (int j = 1; j < positions.Freq(); j++)
{
@out.Write("," + positions.NextPosition());
}
@out.WriteLine("");
}
}
}
reader.Dispose();
}
/// <summary>
/// checks docs + freqs + positions + payloads, sequentially
/// </summary>
public void AssertDocsAndPositionsEnumEquals(string info, DocsAndPositionsEnum leftDocs, DocsAndPositionsEnum rightDocs)
{
if (leftDocs == null || rightDocs == null)
{
Assert.IsNull(leftDocs);
Assert.IsNull(rightDocs);
return;
}
Assert.AreEqual(-1, leftDocs.DocID(), info);
Assert.AreEqual(-1, rightDocs.DocID(), info);
int docid;
while ((docid = leftDocs.NextDoc()) != DocIdSetIterator.NO_MORE_DOCS)
{
Assert.AreEqual(docid, rightDocs.NextDoc(), info);
int freq = leftDocs.Freq();
Assert.AreEqual(freq, rightDocs.Freq(), info);
for (int i = 0; i < freq; i++)
{
Assert.AreEqual(leftDocs.NextPosition(), rightDocs.NextPosition(), info);
Assert.AreEqual(leftDocs.Payload, rightDocs.Payload, info);
Assert.AreEqual(leftDocs.StartOffset(), rightDocs.StartOffset(), info);
Assert.AreEqual(leftDocs.EndOffset(), rightDocs.EndOffset(), info);
}
}
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, rightDocs.NextDoc(), info);
}
/// <summary>
/// checks docs + freqs + positions + payloads, sequentially
/// </summary>
public virtual void AssertDocsAndPositionsEnum(DocsAndPositionsEnum leftDocs, DocsAndPositionsEnum rightDocs)
{
if (leftDocs == null || rightDocs == null)
{
Assert.IsNull(leftDocs);
Assert.IsNull(rightDocs);
return;
}
Assert.AreEqual(-1, leftDocs.DocID());
Assert.AreEqual(-1, rightDocs.DocID());
int docid;
while ((docid = leftDocs.NextDoc()) != DocIdSetIterator.NO_MORE_DOCS)
{
Assert.AreEqual(docid, rightDocs.NextDoc());
int freq = leftDocs.Freq();
Assert.AreEqual(freq, rightDocs.Freq());
for (int i = 0; i < freq; i++)
{
Assert.AreEqual(leftDocs.NextPosition(), rightDocs.NextPosition());
// we don't assert offsets/payloads, they are allowed to be different
}
}
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, rightDocs.NextDoc());
}
public virtual void TestRandomPositions()
{
Directory dir = NewDirectory();
RandomIndexWriter writer = new RandomIndexWriter(Random(), dir, NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random())).SetMergePolicy(NewLogMergePolicy()));
int numDocs = AtLeast(47);
int max = 1051;
int term = Random().Next(max);
int?[][] positionsInDoc = new int?[numDocs][];
FieldType customType = new FieldType(TextField.TYPE_NOT_STORED);
customType.OmitNorms = true;
for (int i = 0; i < numDocs; i++)
{
Document doc = new Document();
List <int?> positions = new List <int?>();
StringBuilder builder = new StringBuilder();
int num = AtLeast(131);
for (int j = 0; j < num; j++)
{
int nextInt = Random().Next(max);
builder.Append(nextInt).Append(" ");
if (nextInt == term)
{
positions.Add(Convert.ToInt32(j));
}
}
if (positions.Count == 0)
{
builder.Append(term);
positions.Add(num);
}
doc.Add(NewField(FieldName, builder.ToString(), customType));
positionsInDoc[i] = positions.ToArray();
writer.AddDocument(doc);
}
IndexReader reader = writer.Reader;
writer.Dispose();
int num_ = AtLeast(13);
for (int i = 0; i < num_; i++)
{
BytesRef bytes = new BytesRef("" + term);
IndexReaderContext topReaderContext = reader.Context;
foreach (AtomicReaderContext atomicReaderContext in topReaderContext.Leaves)
{
DocsAndPositionsEnum docsAndPosEnum = GetDocsAndPositions((AtomicReader)atomicReaderContext.Reader, bytes, null);
Assert.IsNotNull(docsAndPosEnum);
int initDoc = 0;
int maxDoc = atomicReaderContext.Reader.MaxDoc;
// initially advance or do next doc
if (Random().NextBoolean())
{
initDoc = docsAndPosEnum.NextDoc();
}
else
{
initDoc = docsAndPosEnum.Advance(Random().Next(maxDoc));
}
// now run through the scorer and check if all positions are there...
do
{
int docID = docsAndPosEnum.DocID();
if (docID == DocIdSetIterator.NO_MORE_DOCS)
{
break;
}
int?[] pos = positionsInDoc[atomicReaderContext.DocBase + docID];
Assert.AreEqual(pos.Length, docsAndPosEnum.Freq());
// number of positions read should be random - don't read all of them
// allways
int howMany = Random().Next(20) == 0 ? pos.Length - Random().Next(pos.Length) : pos.Length;
for (int j = 0; j < howMany; j++)
{
Assert.AreEqual(pos[j], docsAndPosEnum.NextPosition(), "iteration: " + i + " initDoc: " + initDoc + " doc: " + docID + " base: " + atomicReaderContext.DocBase + " positions: " + pos); /* TODO: + " usePayloads: "
+ usePayload*/
}
if (Random().Next(10) == 0) // once is a while advance
{
if (docsAndPosEnum.Advance(docID + 1 + Random().Next((maxDoc - docID))) == DocIdSetIterator.NO_MORE_DOCS)
{
break;
}
}
} while (docsAndPosEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
}
}
reader.Dispose();
dir.Dispose();
}
public virtual void TestPositionReader()
{
TermVectorsReader reader = Codec.Default.TermVectorsFormat().VectorsReader(Dir, Seg.Info, FieldInfos, NewIOContext(Random()));
BytesRef[] terms;
Terms vector = reader.Get(0).Terms(TestFields[0]);
Assert.IsNotNull(vector);
Assert.AreEqual(TestTerms.Length, vector.Size());
TermsEnum termsEnum = vector.Iterator(null);
DocsAndPositionsEnum dpEnum = null;
for (int i = 0; i < TestTerms.Length; i++)
{
BytesRef text = termsEnum.Next();
Assert.IsNotNull(text);
string term = text.Utf8ToString();
//System.out.println("Term: " + term);
Assert.AreEqual(TestTerms[i], term);
dpEnum = termsEnum.DocsAndPositions(null, dpEnum);
Assert.IsNotNull(dpEnum);
int doc = dpEnum.DocID();
Assert.AreEqual(-1, doc);
Assert.IsTrue(dpEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
Assert.AreEqual(dpEnum.Freq(), Positions[i].Length);
for (int j = 0; j < Positions[i].Length; j++)
{
Assert.AreEqual(Positions[i][j], dpEnum.NextPosition());
}
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, dpEnum.NextDoc());
dpEnum = termsEnum.DocsAndPositions(null, dpEnum);
doc = dpEnum.DocID();
Assert.AreEqual(-1, doc);
Assert.IsTrue(dpEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
Assert.IsNotNull(dpEnum);
Assert.AreEqual(dpEnum.Freq(), Positions[i].Length);
for (int j = 0; j < Positions[i].Length; j++)
{
Assert.AreEqual(Positions[i][j], dpEnum.NextPosition());
Assert.AreEqual(j * 10, dpEnum.StartOffset());
Assert.AreEqual(j * 10 + TestTerms[i].Length, dpEnum.EndOffset());
}
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, dpEnum.NextDoc());
}
Terms freqVector = reader.Get(0).Terms(TestFields[1]); //no pos, no offset
Assert.IsNotNull(freqVector);
Assert.AreEqual(TestTerms.Length, freqVector.Size());
termsEnum = freqVector.Iterator(null);
Assert.IsNotNull(termsEnum);
for (int i = 0; i < TestTerms.Length; i++)
{
BytesRef text = termsEnum.Next();
Assert.IsNotNull(text);
string term = text.Utf8ToString();
//System.out.println("Term: " + term);
Assert.AreEqual(TestTerms[i], term);
Assert.IsNotNull(termsEnum.Docs(null, null));
Assert.IsNull(termsEnum.DocsAndPositions(null, null)); // no pos
}
reader.Dispose();
}
public AssertingDocsAndPositionsEnum(DocsAndPositionsEnum @in)
: base(@in)
{
int docid = @in.DocID();
Debug.Assert(docid == -1, "invalid initial doc id: " + docid);
Doc = -1;
}
public virtual void VerifyEquals(DirectoryReader r1, DirectoryReader r2, string idField)
{
if (VERBOSE)
{
Console.WriteLine("\nr1 docs:");
PrintDocs(r1);
Console.WriteLine("\nr2 docs:");
PrintDocs(r2);
}
if (r1.NumDocs != r2.NumDocs)
{
Debug.Assert(false, "r1.NumDocs=" + r1.NumDocs + " vs r2.NumDocs=" + r2.NumDocs);
}
bool hasDeletes = !(r1.MaxDoc == r2.MaxDoc && r1.NumDocs == r1.MaxDoc);
int[] r2r1 = new int[r2.MaxDoc]; // r2 id to r1 id mapping
// create mapping from id2 space to id2 based on idField
Fields f1 = MultiFields.GetFields(r1);
if (f1 == null)
{
// make sure r2 is empty
Assert.IsNull(MultiFields.GetFields(r2));
return;
}
Terms terms1 = f1.Terms(idField);
if (terms1 == null)
{
Assert.IsTrue(MultiFields.GetFields(r2) == null || MultiFields.GetFields(r2).Terms(idField) == null);
return;
}
TermsEnum termsEnum = terms1.Iterator(null);
Bits liveDocs1 = MultiFields.GetLiveDocs(r1);
Bits liveDocs2 = MultiFields.GetLiveDocs(r2);
Fields fields = MultiFields.GetFields(r2);
if (fields == null)
{
// make sure r1 is in fact empty (eg has only all
// deleted docs):
Bits liveDocs = MultiFields.GetLiveDocs(r1);
DocsEnum docs = null;
while (termsEnum.Next() != null)
{
docs = TestUtil.Docs(Random(), termsEnum, liveDocs, docs, DocsEnum.FLAG_NONE);
while (docs.NextDoc() != DocIdSetIterator.NO_MORE_DOCS)
{
Assert.Fail("r1 is not empty but r2 is");
}
}
return;
}
Terms terms2 = fields.Terms(idField);
TermsEnum termsEnum2 = terms2.Iterator(null);
DocsEnum termDocs1 = null;
DocsEnum termDocs2 = null;
while (true)
{
BytesRef term = termsEnum.Next();
//System.out.println("TEST: match id term=" + term);
if (term == null)
{
break;
}
termDocs1 = TestUtil.Docs(Random(), termsEnum, liveDocs1, termDocs1, DocsEnum.FLAG_NONE);
if (termsEnum2.SeekExact(term))
{
termDocs2 = TestUtil.Docs(Random(), termsEnum2, liveDocs2, termDocs2, DocsEnum.FLAG_NONE);
}
else
{
termDocs2 = null;
}
if (termDocs1.NextDoc() == DocIdSetIterator.NO_MORE_DOCS)
{
// this doc is deleted and wasn't replaced
Assert.IsTrue(termDocs2 == null || termDocs2.NextDoc() == DocIdSetIterator.NO_MORE_DOCS);
continue;
}
int id1 = termDocs1.DocID();
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, termDocs1.NextDoc());
Assert.IsTrue(termDocs2.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
int id2 = termDocs2.DocID();
Assert.AreEqual(DocIdSetIterator.NO_MORE_DOCS, termDocs2.NextDoc());
r2r1[id2] = id1;
// verify stored fields are equivalent
try
{
VerifyEquals(r1.Document(id1), r2.Document(id2));
}
catch (Exception t)
{
Console.WriteLine("FAILED id=" + term + " id1=" + id1 + " id2=" + id2 + " term=" + term);
Console.WriteLine(" d1=" + r1.Document(id1));
Console.WriteLine(" d2=" + r2.Document(id2));
throw t;
}
try
{
// verify term vectors are equivalent
VerifyEquals(r1.GetTermVectors(id1), r2.GetTermVectors(id2));
}
catch (Exception e)
{
Console.WriteLine("FAILED id=" + term + " id1=" + id1 + " id2=" + id2);
Fields tv1 = r1.GetTermVectors(id1);
Console.WriteLine(" d1=" + tv1);
if (tv1 != null)
{
DocsAndPositionsEnum dpEnum = null;
DocsEnum dEnum = null;
foreach (string field in tv1)
{
Console.WriteLine(" " + field + ":");
Terms terms3 = tv1.Terms(field);
Assert.IsNotNull(terms3);
TermsEnum termsEnum3 = terms3.Iterator(null);
BytesRef term2;
while ((term2 = termsEnum3.Next()) != null)
{
Console.WriteLine(" " + term2.Utf8ToString() + ": freq=" + termsEnum3.TotalTermFreq());
dpEnum = termsEnum3.DocsAndPositions(null, dpEnum);
if (dpEnum != null)
{
Assert.IsTrue(dpEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
int freq = dpEnum.Freq();
Console.WriteLine(" doc=" + dpEnum.DocID() + " freq=" + freq);
for (int posUpto = 0; posUpto < freq; posUpto++)
{
Console.WriteLine(" pos=" + dpEnum.NextPosition());
}
}
else
{
dEnum = TestUtil.Docs(Random(), termsEnum3, null, dEnum, DocsEnum.FLAG_FREQS);
Assert.IsNotNull(dEnum);
Assert.IsTrue(dEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
int freq = dEnum.Freq();
Console.WriteLine(" doc=" + dEnum.DocID() + " freq=" + freq);
}
}
}
}
Fields tv2 = r2.GetTermVectors(id2);
Console.WriteLine(" d2=" + tv2);
if (tv2 != null)
{
DocsAndPositionsEnum dpEnum = null;
DocsEnum dEnum = null;
foreach (string field in tv2)
{
Console.WriteLine(" " + field + ":");
Terms terms3 = tv2.Terms(field);
Assert.IsNotNull(terms3);
TermsEnum termsEnum3 = terms3.Iterator(null);
BytesRef term2;
while ((term2 = termsEnum3.Next()) != null)
{
Console.WriteLine(" " + term2.Utf8ToString() + ": freq=" + termsEnum3.TotalTermFreq());
dpEnum = termsEnum3.DocsAndPositions(null, dpEnum);
if (dpEnum != null)
{
Assert.IsTrue(dpEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
int freq = dpEnum.Freq();
Console.WriteLine(" doc=" + dpEnum.DocID() + " freq=" + freq);
for (int posUpto = 0; posUpto < freq; posUpto++)
{
Console.WriteLine(" pos=" + dpEnum.NextPosition());
}
}
else
{
dEnum = TestUtil.Docs(Random(), termsEnum3, null, dEnum, DocsEnum.FLAG_FREQS);
Assert.IsNotNull(dEnum);
Assert.IsTrue(dEnum.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
int freq = dEnum.Freq();
Console.WriteLine(" doc=" + dEnum.DocID() + " freq=" + freq);
}
}
}
}
throw e;
}
}
//System.out.println("TEST: done match id");
// Verify postings
//System.out.println("TEST: create te1");
Fields fields1 = MultiFields.GetFields(r1);
IEnumerator <string> fields1Enum = fields1.GetEnumerator();
Fields fields2 = MultiFields.GetFields(r2);
IEnumerator <string> fields2Enum = fields2.GetEnumerator();
string field1 = null, field2 = null;
TermsEnum termsEnum1 = null;
termsEnum2 = null;
DocsEnum docs1 = null, docs2 = null;
// pack both doc and freq into single element for easy sorting
long[] info1 = new long[r1.NumDocs];
long[] info2 = new long[r2.NumDocs];
for (; ;)
{
BytesRef term1 = null, term2 = null;
// iterate until we get some docs
int len1;
for (; ;)
{
len1 = 0;
if (termsEnum1 == null)
{
if (!fields1Enum.MoveNext())
{
break;
}
field1 = fields1Enum.Current;
Terms terms = fields1.Terms(field1);
if (terms == null)
{
continue;
}
termsEnum1 = terms.Iterator(null);
}
term1 = termsEnum1.Next();
if (term1 == null)
{
// no more terms in this field
termsEnum1 = null;
continue;
}
//System.out.println("TEST: term1=" + term1);
docs1 = TestUtil.Docs(Random(), termsEnum1, liveDocs1, docs1, DocsEnum.FLAG_FREQS);
while (docs1.NextDoc() != DocIdSetIterator.NO_MORE_DOCS)
{
int d = docs1.DocID();
int f = docs1.Freq();
info1[len1] = (((long)d) << 32) | f;
len1++;
}
if (len1 > 0)
{
break;
}
}
// iterate until we get some docs
int len2;
for (; ;)
{
len2 = 0;
if (termsEnum2 == null)
{
if (!fields2Enum.MoveNext())
{
break;
}
field2 = fields2Enum.Current;
Terms terms = fields2.Terms(field2);
if (terms == null)
{
continue;
}
termsEnum2 = terms.Iterator(null);
}
term2 = termsEnum2.Next();
if (term2 == null)
{
// no more terms in this field
termsEnum2 = null;
continue;
}
//System.out.println("TEST: term1=" + term1);
docs2 = TestUtil.Docs(Random(), termsEnum2, liveDocs2, docs2, DocsEnum.FLAG_FREQS);
while (docs2.NextDoc() != DocIdSetIterator.NO_MORE_DOCS)
{
int d = r2r1[docs2.DocID()];
int f = docs2.Freq();
info2[len2] = (((long)d) << 32) | f;
len2++;
}
if (len2 > 0)
{
break;
}
}
Assert.AreEqual(len1, len2);
if (len1 == 0) // no more terms
{
break;
}
Assert.AreEqual(field1, field2);
Assert.IsTrue(term1.BytesEquals(term2));
if (!hasDeletes)
{
Assert.AreEqual(termsEnum1.DocFreq(), termsEnum2.DocFreq());
}
Assert.AreEqual(term1, term2, "len1=" + len1 + " len2=" + len2 + " deletes?=" + hasDeletes);
// sort info2 to get it into ascending docid
Array.Sort(info2, 0, len2);
// now compare
for (int i = 0; i < len1; i++)
{
Assert.AreEqual(info1[i], info2[i], "i=" + i + " len=" + len1 + " d1=" + ((long)((ulong)info1[i] >> 32)) + " f1=" + (info1[i] & int.MaxValue) + " d2=" + ((long)((ulong)info2[i] >> 32)) + " f2=" + (info2[i] & int.MaxValue) + " field=" + field1 + " term=" + term1.Utf8ToString());
}
}
}
public override int DocID()
{
return(@in.DocID());
}
