| public Document ( int n, Lucene.Net.Documents.FieldSelector fieldSelector ) : Lucene.Net.Documents.Document | ||
| n | int | |
| fieldSelector | Lucene.Net.Documents.FieldSelector | |
| return | Lucene.Net.Documents.Document | |
public virtual void Test([ValueSource(typeof(ConcurrentMergeSchedulerFactories), "Values")] Func <IConcurrentMergeScheduler> newScheduler)
{
MockDirectoryWrapper dir = new MockDirectoryWrapper(Random(), new MMapDirectory(CreateTempDir("4GBStoredFields")));
dir.Throttling = MockDirectoryWrapper.Throttling_e.NEVER;
var config = new IndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random()))
.SetMaxBufferedDocs(IndexWriterConfig.DISABLE_AUTO_FLUSH)
.SetRAMBufferSizeMB(256.0)
.SetMergeScheduler(newScheduler())
.SetMergePolicy(NewLogMergePolicy(false, 10))
.SetOpenMode(OpenMode.CREATE);
IndexWriter w = new IndexWriter(dir, config);
MergePolicy mp = w.Config.MergePolicy;
if (mp is LogByteSizeMergePolicy)
{
// 1 petabyte:
((LogByteSizeMergePolicy)mp).MaxMergeMB = 1024 * 1024 * 1024;
}
Document doc = new Document();
FieldType ft = new FieldType();
ft.IsIndexed = false;
ft.IsStored = true;
ft.Freeze();
int valueLength = RandomInts.NextIntBetween(Random(), 1 << 13, 1 << 20);
var value = new byte[valueLength];
for (int i = 0; i < valueLength; ++i)
{
// random so that even compressing codecs can't compress it
value[i] = (byte)Random().Next(256);
}
Field f = new Field("fld", value, ft);
doc.Add(f);
int numDocs = (int)((1L << 32) / valueLength + 100);
for (int i = 0; i < numDocs; ++i)
{
w.AddDocument(doc);
if (VERBOSE && i % (numDocs / 10) == 0)
{
Console.WriteLine(i + " of " + numDocs + "...");
}
}
w.ForceMerge(1);
w.Dispose();
if (VERBOSE)
{
bool found = false;
foreach (string file in dir.ListAll())
{
if (file.EndsWith(".fdt", StringComparison.Ordinal))
{
long fileLength = dir.FileLength(file);
if (fileLength >= 1L << 32)
{
found = true;
}
Console.WriteLine("File length of " + file + " : " + fileLength);
}
}
if (!found)
{
Console.WriteLine("No .fdt file larger than 4GB, test bug?");
}
}
DirectoryReader rd = DirectoryReader.Open(dir);
Document sd = rd.Document(numDocs - 1);
Assert.IsNotNull(sd);
Assert.AreEqual(1, sd.Fields.Count);
BytesRef valueRef = sd.GetBinaryValue("fld");
Assert.IsNotNull(valueRef);
Assert.AreEqual(new BytesRef(value), valueRef);
rd.Dispose();
dir.Dispose();
}
// [Test, LongRunningTest, Timeout(int.MaxValue)] // LUCENENET NOTE: For now, we are overriding this test in every subclass to pull it into the right context for the subclass
public virtual void TestBigDocuments()
{
// "big" as "much bigger than the chunk size"
// for this test we force a FS dir
// we can't just use newFSDirectory, because this test doesn't really index anything.
// so if we get NRTCachingDir+SimpleText, we make massive stored fields and OOM (LUCENE-4484)
Directory dir = new MockDirectoryWrapper(Random(), new MMapDirectory(CreateTempDir("testBigDocuments")));
IndexWriterConfig iwConf = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random()));
iwConf.SetMaxBufferedDocs(RandomInts.NextIntBetween(Random(), 2, 30));
RandomIndexWriter iw = new RandomIndexWriter(Random(), dir, iwConf);
if (dir is MockDirectoryWrapper)
{
((MockDirectoryWrapper)dir).Throttling = MockDirectoryWrapper.Throttling_e.NEVER;
}
Document emptyDoc = new Document(); // emptyDoc
Document bigDoc1 = new Document(); // lot of small fields
Document bigDoc2 = new Document(); // 1 very big field
Field idField = new StringField("id", "", Field.Store.NO);
emptyDoc.Add(idField);
bigDoc1.Add(idField);
bigDoc2.Add(idField);
FieldType onlyStored = new FieldType(StringField.TYPE_STORED);
onlyStored.IsIndexed = false;
Field smallField = new Field("fld", RandomByteArray(Random().Next(10), 256), onlyStored);
int numFields = RandomInts.NextIntBetween(Random(), 500000, 1000000);
for (int i = 0; i < numFields; ++i)
{
bigDoc1.Add(smallField);
}
Field bigField = new Field("fld", RandomByteArray(RandomInts.NextIntBetween(Random(), 1000000, 5000000), 2), onlyStored);
bigDoc2.Add(bigField);
int numDocs = AtLeast(5);
Document[] docs = new Document[numDocs];
for (int i = 0; i < numDocs; ++i)
{
docs[i] = RandomInts.RandomFrom(Random(), Arrays.AsList(emptyDoc, bigDoc1, bigDoc2));
}
for (int i = 0; i < numDocs; ++i)
{
idField.SetStringValue("" + i);
iw.AddDocument(docs[i]);
if (Random().Next(numDocs) == 0)
{
iw.Commit();
}
}
iw.Commit();
iw.ForceMerge(1); // look at what happens when big docs are merged
DirectoryReader rd = DirectoryReader.Open(dir);
IndexSearcher searcher = new IndexSearcher(rd);
for (int i = 0; i < numDocs; ++i)
{
Query query = new TermQuery(new Term("id", "" + i));
TopDocs topDocs = searcher.Search(query, 1);
Assert.AreEqual(1, topDocs.TotalHits, "" + i);
Document doc = rd.Document(topDocs.ScoreDocs[0].Doc);
Assert.IsNotNull(doc);
IIndexableField[] fieldValues = doc.GetFields("fld");
Assert.AreEqual(docs[i].GetFields("fld").Length, fieldValues.Length);
if (fieldValues.Length > 0)
{
Assert.AreEqual(docs[i].GetFields("fld")[0].GetBinaryValue(), fieldValues[0].GetBinaryValue());
}
}
rd.Dispose();
iw.Dispose();
dir.Dispose();
}
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.GetTerms(idField);
if (terms1 == null)
{
Assert.IsTrue(MultiFields.GetFields(r2) == null || MultiFields.GetFields(r2).GetTerms(idField) == null);
return;
}
TermsEnum termsEnum = terms1.GetIterator(null);
IBits liveDocs1 = MultiFields.GetLiveDocs(r1);
IBits 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):
IBits liveDocs = MultiFields.GetLiveDocs(r1);
DocsEnum docs = null;
while (termsEnum.Next() != null)
{
docs = TestUtil.Docs(Random(), termsEnum, liveDocs, docs, DocsFlags.NONE);
while (docs.NextDoc() != DocIdSetIterator.NO_MORE_DOCS)
{
Assert.Fail("r1 is not empty but r2 is");
}
}
return;
}
Terms terms2 = fields.GetTerms(idField);
TermsEnum termsEnum2 = terms2.GetIterator(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, DocsFlags.NONE);
if (termsEnum2.SeekExact(term))
{
termDocs2 = TestUtil.Docs(Random(), termsEnum2, liveDocs2, termDocs2, DocsFlags.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.GetTerms(field);
Assert.IsNotNull(terms3);
TermsEnum termsEnum3 = terms3.GetIterator(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, DocsFlags.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.GetTerms(field);
Assert.IsNotNull(terms3);
TermsEnum termsEnum3 = terms3.GetIterator(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, DocsFlags.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.GetTerms(field1);
if (terms == null)
{
continue;
}
termsEnum1 = terms.GetIterator(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, DocsFlags.FREQS);
while (docs1.NextDoc() != DocIdSetIterator.NO_MORE_DOCS)
{
int d = docs1.DocID;
int f = docs1.Freq;
info1[len1] = (((long)d) << 32) | (uint)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.GetTerms(field2);
if (terms == null)
{
continue;
}
termsEnum2 = terms.GetIterator(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, DocsFlags.FREQS);
while (docs2.NextDoc() != DocIdSetIterator.NO_MORE_DOCS)
{
int d = r2r1[docs2.DocID];
int f = docs2.Freq;
info2[len2] = (((long)d) << 32) | (uint)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());
}
}
}
// [Test] // LUCENENET NOTE: For now, we are overriding this test in every subclass to pull it into the right context for the subclass
public virtual void TestWriteReadMerge()
{
// get another codec, other than the default: so we are merging segments across different codecs
Codec otherCodec;
if ("SimpleText".Equals(Codec.Default.Name, StringComparison.Ordinal))
{
otherCodec = new Lucene46Codec();
}
else
{
otherCodec = new SimpleTextCodec();
}
Directory dir = NewDirectory();
IndexWriterConfig iwConf = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random()));
iwConf.SetMaxBufferedDocs(RandomInts.NextIntBetween(Random(), 2, 30));
RandomIndexWriter iw = new RandomIndexWriter(Random(), dir, (IndexWriterConfig)iwConf.Clone());
int docCount = AtLeast(200);
var data = new byte[docCount][][];
for (int i = 0; i < docCount; ++i)
{
int fieldCount = Rarely() ? RandomInts.NextIntBetween(Random(), 1, 500) : RandomInts.NextIntBetween(Random(), 1, 5);
data[i] = new byte[fieldCount][];
for (int j = 0; j < fieldCount; ++j)
{
int length = Rarely() ? Random().Next(1000) : Random().Next(10);
int max = Rarely() ? 256 : 2;
data[i][j] = RandomByteArray(length, max);
}
}
FieldType type = new FieldType(StringField.TYPE_STORED);
type.IsIndexed = false;
type.Freeze();
Int32Field id = new Int32Field("id", 0, Field.Store.YES);
for (int i = 0; i < data.Length; ++i)
{
Document doc = new Document();
doc.Add(id);
id.SetInt32Value(i);
for (int j = 0; j < data[i].Length; ++j)
{
Field f = new Field("bytes" + j, data[i][j], type);
doc.Add(f);
}
iw.w.AddDocument(doc);
if (Random().NextBoolean() && (i % (data.Length / 10) == 0))
{
iw.w.Dispose();
// test merging against a non-compressing codec
if (iwConf.Codec == otherCodec)
{
iwConf.SetCodec(Codec.Default);
}
else
{
iwConf.SetCodec(otherCodec);
}
iw = new RandomIndexWriter(Random(), dir, (IndexWriterConfig)iwConf.Clone());
}
}
for (int i = 0; i < 10; ++i)
{
int min = Random().Next(data.Length);
int max = min + Random().Next(20);
iw.DeleteDocuments(NumericRangeQuery.NewInt32Range("id", min, max, true, false));
}
iw.ForceMerge(2); // force merges with deletions
iw.Commit();
DirectoryReader ir = DirectoryReader.Open(dir);
Assert.IsTrue(ir.NumDocs > 0);
int numDocs = 0;
for (int i = 0; i < ir.MaxDoc; ++i)
{
Document doc = ir.Document(i);
if (doc == null)
{
continue;
}
++numDocs;
int docId = (int)doc.GetField("id").GetNumericValue();
Assert.AreEqual(data[docId].Length + 1, doc.Fields.Count);
for (int j = 0; j < data[docId].Length; ++j)
{
var arr = data[docId][j];
BytesRef arr2Ref = doc.GetBinaryValue("bytes" + j);
var arr2 = Arrays.CopyOfRange(arr2Ref.Bytes, arr2Ref.Offset, arr2Ref.Offset + arr2Ref.Length);
Assert.AreEqual(arr, arr2);
}
}
Assert.IsTrue(ir.NumDocs <= numDocs);
ir.Dispose();
iw.DeleteAll();
iw.Commit();
iw.ForceMerge(1);
iw.Dispose();
dir.Dispose();
}
private void DoTestReopenWithCommit(Random random, Directory dir, bool withReopen)
{
IndexWriter iwriter = new IndexWriter(dir, NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(random)).SetOpenMode(OpenMode.CREATE).SetMergeScheduler(new SerialMergeScheduler()).SetMergePolicy(NewLogMergePolicy()));
iwriter.Commit();
DirectoryReader reader = DirectoryReader.Open(dir);
try
{
int M = 3;
FieldType customType = new FieldType(TextField.TYPE_STORED);
customType.IsTokenized = false;
FieldType customType2 = new FieldType(TextField.TYPE_STORED);
customType2.IsTokenized = false;
customType2.OmitNorms = true;
FieldType customType3 = new FieldType();
customType3.IsStored = true;
for (int i = 0; i < 4; i++)
{
for (int j = 0; j < M; j++)
{
Document doc = new Document();
doc.Add(NewField("id", i + "_" + j, customType));
doc.Add(NewField("id2", i + "_" + j, customType2));
doc.Add(NewField("id3", i + "_" + j, customType3));
iwriter.AddDocument(doc);
if (i > 0)
{
int k = i - 1;
int n = j + k * M;
Document prevItereationDoc = reader.Document(n);
Assert.IsNotNull(prevItereationDoc);
string id = prevItereationDoc.Get("id");
Assert.AreEqual(k + "_" + j, id);
}
}
iwriter.Commit();
if (withReopen)
{
// reopen
DirectoryReader r2 = DirectoryReader.OpenIfChanged(reader);
if (r2 != null)
{
reader.Dispose();
reader = r2;
}
}
else
{
// recreate
reader.Dispose();
reader = DirectoryReader.Open(dir);
}
}
}
finally
{
iwriter.Dispose();
reader.Dispose();
}
}
public virtual void TestStressAdvance_Mem()
{
for (int iter = 0; iter < 3; iter++)
{
if (VERBOSE)
{
Console.WriteLine("\nTEST: iter=" + iter);
}
Directory dir = NewDirectory();
RandomIndexWriter w = new RandomIndexWriter(Random(), dir);
HashSet <int> aDocs = new HashSet <int>();
Documents.Document doc = new Documents.Document();
Field f = NewStringField("field", "", Field.Store.NO);
doc.Add(f);
Field idField = NewStringField("id", "", Field.Store.YES);
doc.Add(idField);
int num = AtLeast(4097);
if (VERBOSE)
{
Console.WriteLine("\nTEST: numDocs=" + num);
}
for (int id = 0; id < num; id++)
{
if (Random().Next(4) == 3)
{
f.StringValue = "a";
aDocs.Add(id);
}
else
{
f.StringValue = "b";
}
idField.StringValue = "" + id;
w.AddDocument(doc);
if (VERBOSE)
{
Console.WriteLine("\nTEST: doc upto " + id);
}
}
w.ForceMerge(1);
IList <int> aDocIDs = new List <int>();
IList <int> bDocIDs = new List <int>();
DirectoryReader r = w.Reader;
int[] idToDocID = new int[r.MaxDoc];
for (int docID = 0; docID < idToDocID.Length; docID++)
{
int id = Convert.ToInt32(r.Document(docID).Get("id"));
if (aDocs.Contains(id))
{
aDocIDs.Add(docID);
}
else
{
bDocIDs.Add(docID);
}
}
TermsEnum te = GetOnlySegmentReader(r).Fields.Terms("field").Iterator(null);
DocsEnum de = null;
for (int iter2 = 0; iter2 < 10; iter2++)
{
if (VERBOSE)
{
Console.WriteLine("\nTEST: iter=" + iter + " iter2=" + iter2);
}
Assert.AreEqual(TermsEnum.SeekStatus.FOUND, te.SeekCeil(new BytesRef("a")));
de = TestUtil.Docs(Random(), te, null, de, DocsEnum.FLAG_NONE);
TestOne(de, aDocIDs);
Assert.AreEqual(TermsEnum.SeekStatus.FOUND, te.SeekCeil(new BytesRef("b")));
de = TestUtil.Docs(Random(), te, null, de, DocsEnum.FLAG_NONE);
TestOne(de, bDocIDs);
}
w.Dispose();
r.Dispose();
dir.Dispose();
}
}
// TODO: maybe this can reuse the logic of test dueling codecs?
public static void AssertIndexEquals(DirectoryReader index1, DirectoryReader index2)
{
Assert.AreEqual(index1.NumDocs, index2.NumDocs, "IndexReaders have different values for numDocs.");
Assert.AreEqual(index1.MaxDoc, index2.MaxDoc, "IndexReaders have different values for maxDoc.");
Assert.AreEqual(index1.HasDeletions, index2.HasDeletions, "Only one IndexReader has deletions.");
Assert.AreEqual(index1.Leaves.Count == 1, index2.Leaves.Count == 1, "Single segment test differs.");
// check field names
FieldInfos fieldInfos1 = MultiFields.GetMergedFieldInfos(index1);
FieldInfos fieldInfos2 = MultiFields.GetMergedFieldInfos(index2);
Assert.AreEqual(fieldInfos1.Size(), fieldInfos2.Size(), "IndexReaders have different numbers of fields.");
int numFields = fieldInfos1.Size();
for (int fieldID = 0; fieldID < numFields; fieldID++)
{
FieldInfo fieldInfo1 = fieldInfos1.FieldInfo(fieldID);
FieldInfo fieldInfo2 = fieldInfos2.FieldInfo(fieldID);
Assert.AreEqual(fieldInfo1.Name, fieldInfo2.Name, "Different field names.");
}
// check norms
foreach (FieldInfo fieldInfo in fieldInfos1)
{
string curField = fieldInfo.Name;
NumericDocValues norms1 = MultiDocValues.GetNormValues(index1, curField);
NumericDocValues norms2 = MultiDocValues.GetNormValues(index2, curField);
if (norms1 != null && norms2 != null)
{
// todo: generalize this (like TestDuelingCodecs assert)
for (int i = 0; i < index1.MaxDoc; i++)
{
Assert.AreEqual(norms1.Get(i), norms2.Get(i), "Norm different for doc " + i + " and field '" + curField + "'.");
}
}
else
{
Assert.IsNull(norms1);
Assert.IsNull(norms2);
}
}
// check deletions
Bits liveDocs1 = MultiFields.GetLiveDocs(index1);
Bits liveDocs2 = MultiFields.GetLiveDocs(index2);
for (int i = 0; i < index1.MaxDoc; i++)
{
Assert.AreEqual(liveDocs1 == null || !liveDocs1.Get(i), liveDocs2 == null || !liveDocs2.Get(i), "Doc " + i + " only deleted in one index.");
}
// check stored fields
for (int i = 0; i < index1.MaxDoc; i++)
{
if (liveDocs1 == null || liveDocs1.Get(i))
{
Document doc1 = index1.Document(i);
Document doc2 = index2.Document(i);
IList<IndexableField> field1 = doc1.Fields;
IList<IndexableField> field2 = doc2.Fields;
Assert.AreEqual(field1.Count, field2.Count, "Different numbers of fields for doc " + i + ".");
IEnumerator<IndexableField> itField1 = field1.GetEnumerator();
IEnumerator<IndexableField> itField2 = field2.GetEnumerator();
while (itField1.MoveNext())
{
Field curField1 = (Field)itField1.Current;
itField2.MoveNext();
Field curField2 = (Field)itField2.Current;
Assert.AreEqual(curField1.Name(), curField2.Name(), "Different fields names for doc " + i + ".");
Assert.AreEqual(curField1.StringValue, curField2.StringValue, "Different field values for doc " + i + ".");
}
}
}
// check dictionary and posting lists
Fields fields1 = MultiFields.GetFields(index1);
Fields fields2 = MultiFields.GetFields(index2);
IEnumerator<string> fenum2 = fields2.GetEnumerator();
Bits liveDocs = MultiFields.GetLiveDocs(index1);
foreach (string field1 in fields1)
{
fenum2.MoveNext();
Assert.AreEqual(field1, fenum2.Current, "Different fields");
Terms terms1 = fields1.Terms(field1);
if (terms1 == null)
{
Assert.IsNull(fields2.Terms(field1));
continue;
}
TermsEnum enum1 = terms1.Iterator(null);
Terms terms2 = fields2.Terms(field1);
Assert.IsNotNull(terms2);
TermsEnum enum2 = terms2.Iterator(null);
while (enum1.Next() != null)
{
Assert.AreEqual(enum1.Term(), enum2.Next(), "Different terms");
DocsAndPositionsEnum tp1 = enum1.DocsAndPositions(liveDocs, null);
DocsAndPositionsEnum tp2 = enum2.DocsAndPositions(liveDocs, null);
while (tp1.NextDoc() != DocIdSetIterator.NO_MORE_DOCS)
{
Assert.IsTrue(tp2.NextDoc() != DocIdSetIterator.NO_MORE_DOCS);
Assert.AreEqual(tp1.DocID(), tp2.DocID(), "Different doc id in postinglist of term " + enum1.Term() + ".");
Assert.AreEqual(tp1.Freq(), tp2.Freq(), "Different term frequence in postinglist of term " + enum1.Term() + ".");
for (int i = 0; i < tp1.Freq(); i++)
{
Assert.AreEqual(tp1.NextPosition(), tp2.NextPosition(), "Different positions in postinglist of term " + enum1.Term() + ".");
}
}
}
}
Assert.IsFalse(fenum2.MoveNext());
}
public virtual void TestReadSkip()
{
using Directory dir = NewDirectory();
IndexWriterConfig iwConf = NewIndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(Random));
iwConf.SetMaxBufferedDocs(RandomInts.RandomInt32Between(Random, 2, 30));
using RandomIndexWriter iw = new RandomIndexWriter(Random, dir, iwConf);
FieldType ft = new FieldType();
ft.IsStored = true;
ft.Freeze();
string @string = TestUtil.RandomSimpleString(Random, 50);
var bytes = @string.GetBytes(Encoding.UTF8);
long l = Random.NextBoolean() ? Random.Next(42) : Random.NextInt64();
int i = Random.NextBoolean() ? Random.Next(42) : Random.Next();
float f = Random.NextSingle();
double d = Random.NextDouble();
Field[] fields = new Field[]
{
new Field("bytes", bytes, ft),
new Field("string", @string, ft),
new Int64Field("long", l, Field.Store.YES),
new Int32Field("int", i, Field.Store.YES),
new SingleField("float", f, Field.Store.YES),
new DoubleField("double", d, Field.Store.YES)
};
for (int k = 0; k < 100; ++k)
{
Document doc = new Document();
foreach (Field fld in fields)
{
doc.Add(fld);
}
iw.IndexWriter.AddDocument(doc);
}
iw.Commit();
using DirectoryReader reader = DirectoryReader.Open(dir);
int docID = Random.Next(100);
foreach (Field fld in fields)
{
string fldName = fld.Name;
Document sDoc = reader.Document(docID, new JCG.HashSet <string> {
fldName
});
IIndexableField sField = sDoc.GetField(fldName);
if (typeof(Field) == fld.GetType())
{
Assert.AreEqual(fld.GetBinaryValue(), sField.GetBinaryValue());
Assert.AreEqual(fld.GetStringValue(), sField.GetStringValue());
}
else
{
#pragma warning disable 612, 618
Assert.AreEqual(fld.GetNumericValue(), sField.GetNumericValue());
#pragma warning restore 612, 618
}
}
}
