public void testMissingTerms()
{
String fieldName = "field1";
Directory rd = new RAMDirectory();
var w = new IndexWriter(rd, new KeywordAnalyzer(), IndexWriter.MaxFieldLength.UNLIMITED);
for (int i = 0; i < 100; i++)
{
var doc = new Document();
int term = i*10; //terms are units of 10;
doc.Add(new Field(fieldName, "" + term, Field.Store.YES, Field.Index.ANALYZED));
w.AddDocument(doc);
}
IndexReader reader = w.GetReader();
w.Close();
TermsFilter tf = new TermsFilter();
tf.AddTerm(new Term(fieldName, "19"));
FixedBitSet bits = (FixedBitSet) tf.GetDocIdSet(reader);
Assert.AreEqual(0, bits.Cardinality(), "Must match nothing");
tf.AddTerm(new Term(fieldName, "20"));
bits = (FixedBitSet) tf.GetDocIdSet(reader);
Assert.AreEqual(1, bits.Cardinality(), "Must match 1");
tf.AddTerm(new Term(fieldName, "10"));
bits = (FixedBitSet) tf.GetDocIdSet(reader);
Assert.AreEqual(2, bits.Cardinality(), "Must match 2");
tf.AddTerm(new Term(fieldName, "00"));
bits = (FixedBitSet) tf.GetDocIdSet(reader);
Assert.AreEqual(2, bits.Cardinality(), "Must match 2");
reader.Close();
rd.Close();
}
public void TestFieldNotPresent()
{
Directory dir = NewDirectory();
RandomIndexWriter w = new RandomIndexWriter(Random(), dir, Similarity, TimeZone);
int num = AtLeast(3);
int skip = Random().Next(num);
var terms = new List<Term>();
for (int i = 0; i < num; i++)
{
terms.Add(new Term("field" + i, "content1"));
Document doc = new Document();
if (skip == i)
{
continue;
}
doc.Add(NewStringField("field" + i, "content1", Field.Store.YES));
w.AddDocument(doc);
}
w.ForceMerge(1);
IndexReader reader = w.Reader;
w.Dispose();
assertEquals(1, reader.Leaves.size());
AtomicReaderContext context = reader.Leaves.First();
TermsFilter tf = new TermsFilter(terms);
FixedBitSet bits = (FixedBitSet)tf.GetDocIdSet(context, context.AtomicReader.LiveDocs);
assertEquals("Must be num fields - 1 since we skip only one field", num - 1, bits.Cardinality());
reader.Dispose();
dir.Dispose();
}
public void TestMissingField()
{
string fieldName = "field1";
Directory rd1 = NewDirectory();
RandomIndexWriter w1 = new RandomIndexWriter(
#if FEATURE_INSTANCE_TESTDATA_INITIALIZATION
this,
#endif
Random, rd1);
Document doc = new Document();
doc.Add(NewStringField(fieldName, "content1", Field.Store.YES));
w1.AddDocument(doc);
IndexReader reader1 = w1.GetReader();
w1.Dispose();
fieldName = "field2";
Directory rd2 = NewDirectory();
RandomIndexWriter w2 = new RandomIndexWriter(
#if FEATURE_INSTANCE_TESTDATA_INITIALIZATION
this,
#endif
Random, rd2);
doc = new Document();
doc.Add(NewStringField(fieldName, "content2", Field.Store.YES));
w2.AddDocument(doc);
IndexReader reader2 = w2.GetReader();
w2.Dispose();
TermsFilter tf = new TermsFilter(new Term(fieldName, "content1"));
MultiReader multi = new MultiReader(reader1, reader2);
foreach (AtomicReaderContext context in multi.Leaves)
{
DocIdSet docIdSet = tf.GetDocIdSet(context, context.AtomicReader.LiveDocs);
if (context.Reader.DocFreq(new Term(fieldName, "content1")) == 0)
{
assertNull(docIdSet);
}
else
{
FixedBitSet bits = (FixedBitSet)docIdSet;
assertTrue("Must be >= 0", bits.Cardinality >= 0);
}
}
multi.Dispose();
reader1.Dispose();
reader2.Dispose();
rd1.Dispose();
rd2.Dispose();
}
public void testMissingTerms()
{
String fieldName = "field1";
Directory rd = new RAMDirectory();
var w = new IndexWriter(rd, new KeywordAnalyzer(), IndexWriter.MaxFieldLength.UNLIMITED);
for (int i = 0; i < 100; i++)
{
var doc = new Document();
int term = i * 10; //terms are units of 10;
doc.Add(new Field(fieldName, "" + term, Field.Store.YES, Field.Index.ANALYZED));
w.AddDocument(doc);
}
IndexReader reader = w.GetReader();
w.Close();
TermsFilter tf = new TermsFilter();
tf.AddTerm(new Term(fieldName, "19"));
FixedBitSet bits = (FixedBitSet)tf.GetDocIdSet(reader);
Assert.AreEqual(0, bits.Cardinality(), "Must match nothing");
tf.AddTerm(new Term(fieldName, "20"));
bits = (FixedBitSet)tf.GetDocIdSet(reader);
Assert.AreEqual(1, bits.Cardinality(), "Must match 1");
tf.AddTerm(new Term(fieldName, "10"));
bits = (FixedBitSet)tf.GetDocIdSet(reader);
Assert.AreEqual(2, bits.Cardinality(), "Must match 2");
tf.AddTerm(new Term(fieldName, "00"));
bits = (FixedBitSet)tf.GetDocIdSet(reader);
Assert.AreEqual(2, bits.Cardinality(), "Must match 2");
reader.Close();
rd.Close();
}
public void TestSkipField()
{
Directory dir = NewDirectory();
RandomIndexWriter w = new RandomIndexWriter(
#if FEATURE_INSTANCE_TESTDATA_INITIALIZATION
this,
#endif
Random, dir);
int num = AtLeast(10);
var terms = new JCG.HashSet <Term>();
for (int i = 0; i < num; i++)
{
string field = "field" + Random.Next(100);
terms.Add(new Term(field, "content1"));
Document doc = new Document();
doc.Add(NewStringField(field, "content1", Field.Store.YES));
w.AddDocument(doc);
}
int randomFields = Random.Next(10);
for (int i = 0; i < randomFields; i++)
{
while (true)
{
string field = "field" + Random.Next(100);
Term t = new Term(field, "content1");
if (!terms.Contains(t))
{
terms.Add(t);
break;
}
}
}
w.ForceMerge(1);
IndexReader reader = w.GetReader();
w.Dispose();
assertEquals(1, reader.Leaves.size());
AtomicReaderContext context = reader.Leaves[0];
TermsFilter tf = new TermsFilter(terms.ToList());
FixedBitSet bits = (FixedBitSet)tf.GetDocIdSet(context, context.AtomicReader.LiveDocs);
assertEquals(context.Reader.NumDocs, bits.Cardinality);
reader.Dispose();
dir.Dispose();
}
public void TestFieldNotPresent()
{
Directory dir = NewDirectory();
RandomIndexWriter w = new RandomIndexWriter(
#if FEATURE_INSTANCE_TESTDATA_INITIALIZATION
this,
#endif
Random, dir);
int num = AtLeast(3);
int skip = Random.Next(num);
var terms = new JCG.List <Term>();
for (int i = 0; i < num; i++)
{
terms.Add(new Term("field" + i, "content1"));
Document doc = new Document();
if (skip == i)
{
continue;
}
doc.Add(NewStringField("field" + i, "content1", Field.Store.YES));
w.AddDocument(doc);
}
w.ForceMerge(1);
IndexReader reader = w.GetReader();
w.Dispose();
assertEquals(1, reader.Leaves.size());
AtomicReaderContext context = reader.Leaves[0];
TermsFilter tf = new TermsFilter(terms);
FixedBitSet bits = (FixedBitSet)tf.GetDocIdSet(context, context.AtomicReader.LiveDocs);
assertEquals("Must be num fields - 1 since we skip only one field", num - 1, bits.Cardinality);
reader.Dispose();
dir.Dispose();
}
public void TestFieldNotPresent()
{
Directory dir = NewDirectory();
RandomIndexWriter w = new RandomIndexWriter(Random(), dir, Similarity, TimeZone);
int num = AtLeast(3);
int skip = Random().Next(num);
var terms = new List <Term>();
for (int i = 0; i < num; i++)
{
terms.Add(new Term("field" + i, "content1"));
Document doc = new Document();
if (skip == i)
{
continue;
}
doc.Add(NewStringField("field" + i, "content1", Field.Store.YES));
w.AddDocument(doc);
}
w.ForceMerge(1);
IndexReader reader = w.Reader;
w.Dispose();
assertEquals(1, reader.Leaves.size());
AtomicReaderContext context = reader.Leaves.First();
TermsFilter tf = new TermsFilter(terms);
FixedBitSet bits = (FixedBitSet)tf.GetDocIdSet(context, context.AtomicReader.LiveDocs);
assertEquals("Must be num fields - 1 since we skip only one field", num - 1, bits.Cardinality());
reader.Dispose();
dir.Dispose();
}
public void TestMissingField()
{
string fieldName = "field1";
Directory rd1 = NewDirectory();
RandomIndexWriter w1 = new RandomIndexWriter(Random(), rd1, Similarity, TimeZone);
Document doc = new Document();
doc.Add(NewStringField(fieldName, "content1", Field.Store.YES));
w1.AddDocument(doc);
IndexReader reader1 = w1.Reader;
w1.Dispose();
fieldName = "field2";
Directory rd2 = NewDirectory();
RandomIndexWriter w2 = new RandomIndexWriter(Random(), rd2, Similarity, TimeZone);
doc = new Document();
doc.Add(NewStringField(fieldName, "content2", Field.Store.YES));
w2.AddDocument(doc);
IndexReader reader2 = w2.Reader;
w2.Dispose();
TermsFilter tf = new TermsFilter(new Term(fieldName, "content1"));
MultiReader multi = new MultiReader(reader1, reader2);
foreach (AtomicReaderContext context in multi.Leaves)
{
DocIdSet docIdSet = tf.GetDocIdSet(context, context.AtomicReader.LiveDocs);
if (context.Reader.DocFreq(new Term(fieldName, "content1")) == 0)
{
assertNull(docIdSet);
}
else
{
FixedBitSet bits = (FixedBitSet)docIdSet;
assertTrue("Must be >= 0", bits.Cardinality() >= 0);
}
}
multi.Dispose();
reader1.Dispose();
reader2.Dispose();
rd1.Dispose();
rd2.Dispose();
}
public void TestSkipField()
{
Directory dir = NewDirectory();
RandomIndexWriter w = new RandomIndexWriter(Random(), dir, Similarity, TimeZone);
int num = AtLeast(10);
var terms = new HashSet<Term>();
for (int i = 0; i < num; i++)
{
string field = "field" + Random().Next(100);
terms.Add(new Term(field, "content1"));
Document doc = new Document();
doc.Add(NewStringField(field, "content1", Field.Store.YES));
w.AddDocument(doc);
}
int randomFields = Random().Next(10);
for (int i = 0; i < randomFields; i++)
{
while (true)
{
string field = "field" + Random().Next(100);
Term t = new Term(field, "content1");
if (!terms.Contains(t))
{
terms.Add(t);
break;
}
}
}
w.ForceMerge(1);
IndexReader reader = w.Reader;
w.Dispose();
assertEquals(1, reader.Leaves.size());
AtomicReaderContext context = reader.Leaves.First();
TermsFilter tf = new TermsFilter(terms.ToList());
FixedBitSet bits = (FixedBitSet)tf.GetDocIdSet(context, context.AtomicReader.LiveDocs);
assertEquals(context.Reader.NumDocs, bits.Cardinality());
reader.Dispose();
dir.Dispose();
}