public void testSorted() throws Exception {
Directory dir = newDirectory();
Document doc = new Document();
Field field = new SortedDocValuesField("bytes", new BytesRef());
doc.add(field);
IndexWriterConfig iwc = newIndexWriterConfig(random(), null);
iwc.setMergePolicy(newLogMergePolicy());
RandomIndexWriter iw = new RandomIndexWriter(random(), dir, iwc);
int numDocs = TEST_NIGHTLY ? atLeast(500) : atLeast(50);
for (int i = 0; i < numDocs; i++) {
BytesRef ref = new BytesRef(TestUtil.randomUnicodeString(random()));
field.setBytesValue(ref);
if (random().nextInt(7) == 0) {
iw.addDocument(new Document());
}
iw.addDocument(doc);
if (random().nextInt(17) == 0) {
iw.commit();
}
}
DirectoryReader ir = iw.getReader();
iw.forceMerge(1);
DirectoryReader ir2 = iw.getReader();
LeafReader merged = getOnlyLeafReader(ir2);
iw.close();
SortedDocValues multi = MultiDocValues.getSortedValues(ir, "bytes");
SortedDocValues single = merged.getSortedDocValues("bytes");
assertEquals(single.getValueCount(), multi.getValueCount());
for (int i = 0; i < numDocs; i++) {
// check ord
assertEquals(single.getOrd(i), multi.getOrd(i));
// check value
final BytesRef expected = BytesRef.deepCopyOf(single.get(i));
final BytesRef actual = multi.get(i);
assertEquals(expected, actual);
}
ir.close();
ir2.close();
dir.close();
}
public void testDocValuesMemoryIndexVsNormalIndex() throws Exception {
Document doc = new Document();
long randomLong = random().nextLong();
doc.add(new NumericDocValuesField("numeric", randomLong));
if (random().nextBoolean()) {
doc.add(new LegacyLongField("numeric", randomLong, Field.Store.NO));
}
int numValues = atLeast(5);
for (int i = 0; i < numValues; i++) {
randomLong = random().nextLong();
doc.add(new SortedNumericDocValuesField("sorted_numeric", randomLong));
if (random().nextBoolean()) {
// randomly duplicate field/value
doc.add(new SortedNumericDocValuesField("sorted_numeric", randomLong));
}
if (random().nextBoolean()) {
doc.add(new LegacyLongField("numeric", randomLong, Field.Store.NO));
}
}
BytesRef randomTerm = new BytesRef(randomTerm());
doc.add(new BinaryDocValuesField("binary", randomTerm));
if (random().nextBoolean()) {
doc.add(new StringField("binary", randomTerm, Field.Store.NO));
}
randomTerm = new BytesRef(randomTerm());
doc.add(new SortedDocValuesField("sorted", randomTerm));
if (random().nextBoolean()) {
doc.add(new StringField("sorted", randomTerm, Field.Store.NO));
}
numValues = atLeast(5);
for (int i = 0; i < numValues; i++) {
randomTerm = new BytesRef(randomTerm());
doc.add(new SortedSetDocValuesField("sorted_set", randomTerm));
if (random().nextBoolean()) {
// randomly duplicate field/value
doc.add(new SortedSetDocValuesField("sorted_set", randomTerm));
}
if (random().nextBoolean()) {
// randomily just add a normal string field
doc.add(new StringField("sorted_set", randomTerm, Field.Store.NO));
}
}
MockAnalyzer mockAnalyzer = new MockAnalyzer(random());
MemoryIndex memoryIndex = MemoryIndex.fromDocument(doc, mockAnalyzer);
IndexReader indexReader = memoryIndex.createSearcher().getIndexReader();
LeafReader leafReader = indexReader.leaves().get(0).reader();
Directory dir = newDirectory();
IndexWriter writer = new IndexWriter(dir, newIndexWriterConfig(random(), mockAnalyzer));
writer.addDocument(doc);
writer.close();
IndexReader controlIndexReader = DirectoryReader.open(dir);
LeafReader controlLeafReader = controlIndexReader.leaves().get(0).reader();
NumericDocValues numericDocValues = leafReader.getNumericDocValues("numeric");
NumericDocValues controlNumericDocValues = controlLeafReader.getNumericDocValues("numeric");
assertEquals(controlNumericDocValues.get(0), numericDocValues.get(0));
SortedNumericDocValues sortedNumericDocValues =
leafReader.getSortedNumericDocValues("sorted_numeric");
sortedNumericDocValues.setDocument(0);
SortedNumericDocValues controlSortedNumericDocValues =
controlLeafReader.getSortedNumericDocValues("sorted_numeric");
controlSortedNumericDocValues.setDocument(0);
assertEquals(controlSortedNumericDocValues.count(), sortedNumericDocValues.count());
for (int i = 0; i < controlSortedNumericDocValues.count(); i++) {
assertEquals(controlSortedNumericDocValues.valueAt(i), sortedNumericDocValues.valueAt(i));
}
BinaryDocValues binaryDocValues = leafReader.getBinaryDocValues("binary");
BinaryDocValues controlBinaryDocValues = controlLeafReader.getBinaryDocValues("binary");
assertEquals(controlBinaryDocValues.get(0), binaryDocValues.get(0));
SortedDocValues sortedDocValues = leafReader.getSortedDocValues("sorted");
SortedDocValues controlSortedDocValues = controlLeafReader.getSortedDocValues("sorted");
assertEquals(controlSortedDocValues.getValueCount(), sortedDocValues.getValueCount());
assertEquals(controlSortedDocValues.get(0), sortedDocValues.get(0));
assertEquals(controlSortedDocValues.getOrd(0), sortedDocValues.getOrd(0));
assertEquals(controlSortedDocValues.lookupOrd(0), sortedDocValues.lookupOrd(0));
SortedSetDocValues sortedSetDocValues = leafReader.getSortedSetDocValues("sorted_set");
sortedSetDocValues.setDocument(0);
SortedSetDocValues controlSortedSetDocValues =
controlLeafReader.getSortedSetDocValues("sorted_set");
controlSortedSetDocValues.setDocument(0);
assertEquals(controlSortedSetDocValues.getValueCount(), sortedSetDocValues.getValueCount());
for (long controlOrd = controlSortedSetDocValues.nextOrd();
controlOrd != SortedSetDocValues.NO_MORE_ORDS;
controlOrd = controlSortedSetDocValues.nextOrd()) {
assertEquals(controlOrd, sortedSetDocValues.nextOrd());
assertEquals(
controlSortedSetDocValues.lookupOrd(controlOrd),
sortedSetDocValues.lookupOrd(controlOrd));
}
assertEquals(SortedSetDocValues.NO_MORE_ORDS, sortedSetDocValues.nextOrd());
indexReader.close();
controlIndexReader.close();
dir.close();
}