use of org.apache.lucene.document.BinaryDocValuesField in project lucene-solr by apache.
the class TestMixedDocValuesUpdates method testManyReopensAndFields.
public void testManyReopensAndFields() throws Exception {
Directory dir = newDirectory();
final Random random = random();
IndexWriterConfig conf = newIndexWriterConfig(new MockAnalyzer(random));
LogMergePolicy lmp = newLogMergePolicy();
// merge often
lmp.setMergeFactor(3);
conf.setMergePolicy(lmp);
IndexWriter writer = new IndexWriter(dir, conf);
final boolean isNRT = random.nextBoolean();
DirectoryReader reader;
if (isNRT) {
reader = DirectoryReader.open(writer);
} else {
writer.commit();
reader = DirectoryReader.open(dir);
}
// 3-7
final int numFields = random.nextInt(4) + 3;
// 1-3
final int numNDVFields = random.nextInt(numFields / 2) + 1;
final long[] fieldValues = new long[numFields];
for (int i = 0; i < fieldValues.length; i++) {
fieldValues[i] = 1;
}
int numRounds = atLeast(15);
int docID = 0;
for (int i = 0; i < numRounds; i++) {
int numDocs = atLeast(5);
// System.out.println("TEST: round=" + i + ", numDocs=" + numDocs);
for (int j = 0; j < numDocs; j++) {
Document doc = new Document();
doc.add(new StringField("id", "doc-" + docID, Store.NO));
// update key
doc.add(new StringField("key", "all", Store.NO));
// add all fields with their current value
for (int f = 0; f < fieldValues.length; f++) {
if (f < numNDVFields) {
doc.add(new NumericDocValuesField("f" + f, fieldValues[f]));
} else {
doc.add(new BinaryDocValuesField("f" + f, TestBinaryDocValuesUpdates.toBytes(fieldValues[f])));
}
}
writer.addDocument(doc);
++docID;
}
int fieldIdx = random.nextInt(fieldValues.length);
String updateField = "f" + fieldIdx;
if (fieldIdx < numNDVFields) {
writer.updateNumericDocValue(new Term("key", "all"), updateField, ++fieldValues[fieldIdx]);
} else {
writer.updateBinaryDocValue(new Term("key", "all"), updateField, TestBinaryDocValuesUpdates.toBytes(++fieldValues[fieldIdx]));
}
if (random.nextDouble() < 0.2) {
// might also delete an already deleted document, ok!
int deleteDoc = random.nextInt(docID);
writer.deleteDocuments(new Term("id", "doc-" + deleteDoc));
// System.out.println("[" + Thread.currentThread().getName() + "]: deleted document: doc-" + deleteDoc);
}
// verify reader
if (!isNRT) {
writer.commit();
}
// System.out.println("[" + Thread.currentThread().getName() + "]: reopen reader: " + reader);
DirectoryReader newReader = DirectoryReader.openIfChanged(reader);
assertNotNull(newReader);
reader.close();
reader = newReader;
// System.out.println("[" + Thread.currentThread().getName() + "]: reopened reader: " + reader);
// we delete at most one document per round
assertTrue(reader.numDocs() > 0);
for (LeafReaderContext context : reader.leaves()) {
LeafReader r = context.reader();
// System.out.println(((SegmentReader) r).getSegmentName());
Bits liveDocs = r.getLiveDocs();
for (int field = 0; field < fieldValues.length; field++) {
String f = "f" + field;
BinaryDocValues bdv = r.getBinaryDocValues(f);
NumericDocValues ndv = r.getNumericDocValues(f);
if (field < numNDVFields) {
assertNotNull(ndv);
assertNull(bdv);
} else {
assertNull(ndv);
assertNotNull(bdv);
}
int maxDoc = r.maxDoc();
for (int doc = 0; doc < maxDoc; doc++) {
if (liveDocs == null || liveDocs.get(doc)) {
// System.out.println("doc=" + (doc + context.docBase) + " f='" + f + "' vslue=" + getValue(bdv, doc, scratch));
if (field < numNDVFields) {
assertEquals(doc, ndv.advance(doc));
assertEquals("invalid numeric value for doc=" + doc + ", field=" + f + ", reader=" + r, fieldValues[field], ndv.longValue());
} else {
assertEquals(doc, bdv.advance(doc));
assertEquals("invalid binary value for doc=" + doc + ", field=" + f + ", reader=" + r, fieldValues[field], TestBinaryDocValuesUpdates.getValue(bdv));
}
}
}
}
}
// System.out.println();
}
writer.close();
IOUtils.close(reader, dir);
}
use of org.apache.lucene.document.BinaryDocValuesField in project lucene-solr by apache.
the class TestBinaryDocValuesUpdates method doc.
private Document doc(int id) {
Document doc = new Document();
doc.add(new StringField("id", "doc-" + id, Store.NO));
doc.add(new BinaryDocValuesField("val", toBytes(id + 1)));
return doc;
}
use of org.apache.lucene.document.BinaryDocValuesField in project lucene-solr by apache.
the class TestBinaryDocValuesUpdates method testUpdateBinaryDVFieldWithSameNameAsPostingField.
public void testUpdateBinaryDVFieldWithSameNameAsPostingField() throws Exception {
// this used to fail because FieldInfos.Builder neglected to update
// globalFieldMaps.docValuesTypes map
Directory dir = newDirectory();
IndexWriterConfig conf = newIndexWriterConfig(new MockAnalyzer(random()));
IndexWriter writer = new IndexWriter(dir, conf);
Document doc = new Document();
doc.add(new StringField("f", "mock-value", Store.NO));
doc.add(new BinaryDocValuesField("f", toBytes(5L)));
writer.addDocument(doc);
writer.commit();
writer.updateBinaryDocValue(new Term("f", "mock-value"), "f", toBytes(17L));
writer.close();
DirectoryReader r = DirectoryReader.open(dir);
BinaryDocValues bdv = r.leaves().get(0).reader().getBinaryDocValues("f");
assertEquals(0, bdv.nextDoc());
assertEquals(17, getValue(bdv));
r.close();
dir.close();
}
use of org.apache.lucene.document.BinaryDocValuesField in project lucene-solr by apache.
the class TestBinaryDocValuesUpdates method testUpdateTwoNonexistingTerms.
public void testUpdateTwoNonexistingTerms() throws Exception {
Directory dir = newDirectory();
IndexWriterConfig conf = newIndexWriterConfig(new MockAnalyzer(random()));
IndexWriter writer = new IndexWriter(dir, conf);
Document doc = new Document();
doc.add(new StringField("id", "doc", Store.NO));
doc.add(new BinaryDocValuesField("f1", toBytes(1L)));
writer.addDocument(doc);
// update w/ multiple nonexisting terms in same field
writer.updateBinaryDocValue(new Term("c", "foo"), "f1", toBytes(2L));
writer.updateBinaryDocValue(new Term("c", "bar"), "f1", toBytes(2L));
writer.close();
DirectoryReader reader = DirectoryReader.open(dir);
assertEquals(1, reader.leaves().size());
BinaryDocValues bdv = reader.leaves().get(0).reader().getBinaryDocValues("f1");
assertEquals(0, bdv.nextDoc());
assertEquals(1L, getValue(bdv));
reader.close();
dir.close();
}
use of org.apache.lucene.document.BinaryDocValuesField in project lucene-solr by apache.
the class TestIndexWriterOnVMError method doTest.
// just one thread, serial merge policy, hopefully debuggable
private void doTest(MockDirectoryWrapper.Failure failOn) throws Exception {
// log all exceptions we hit, in case we fail (for debugging)
ByteArrayOutputStream exceptionLog = new ByteArrayOutputStream();
PrintStream exceptionStream = new PrintStream(exceptionLog, true, "UTF-8");
//PrintStream exceptionStream = System.out;
final long analyzerSeed = random().nextLong();
final Analyzer analyzer = new Analyzer() {
@Override
protected TokenStreamComponents createComponents(String fieldName) {
MockTokenizer tokenizer = new MockTokenizer(MockTokenizer.WHITESPACE, false);
// we are gonna make it angry
tokenizer.setEnableChecks(false);
TokenStream stream = tokenizer;
// emit some payloads
if (fieldName.contains("payloads")) {
stream = new MockVariableLengthPayloadFilter(new Random(analyzerSeed), stream);
}
return new TokenStreamComponents(tokenizer, stream);
}
};
MockDirectoryWrapper dir = null;
final int numIterations = TEST_NIGHTLY ? atLeast(100) : atLeast(5);
STARTOVER: for (int iter = 0; iter < numIterations; iter++) {
try {
// close from last run
if (dir != null) {
dir.close();
}
// disable slow things: we don't rely upon sleeps here.
dir = newMockDirectory();
dir.setThrottling(MockDirectoryWrapper.Throttling.NEVER);
dir.setUseSlowOpenClosers(false);
IndexWriterConfig conf = newIndexWriterConfig(analyzer);
// just for now, try to keep this test reproducible
conf.setMergeScheduler(new SerialMergeScheduler());
// test never makes it this far...
int numDocs = atLeast(2000);
IndexWriter iw = new IndexWriter(dir, conf);
// ensure there is always a commit
iw.commit();
dir.failOn(failOn);
for (int i = 0; i < numDocs; i++) {
Document doc = new Document();
doc.add(newStringField("id", Integer.toString(i), Field.Store.NO));
doc.add(new NumericDocValuesField("dv", i));
doc.add(new BinaryDocValuesField("dv2", new BytesRef(Integer.toString(i))));
doc.add(new SortedDocValuesField("dv3", new BytesRef(Integer.toString(i))));
doc.add(new SortedSetDocValuesField("dv4", new BytesRef(Integer.toString(i))));
doc.add(new SortedSetDocValuesField("dv4", new BytesRef(Integer.toString(i - 1))));
doc.add(new SortedNumericDocValuesField("dv5", i));
doc.add(new SortedNumericDocValuesField("dv5", i - 1));
doc.add(newTextField("text1", TestUtil.randomAnalysisString(random(), 20, true), Field.Store.NO));
// ensure we store something
doc.add(new StoredField("stored1", "foo"));
doc.add(new StoredField("stored1", "bar"));
// ensure we get some payloads
doc.add(newTextField("text_payloads", TestUtil.randomAnalysisString(random(), 6, true), Field.Store.NO));
// ensure we get some vectors
FieldType ft = new FieldType(TextField.TYPE_NOT_STORED);
ft.setStoreTermVectors(true);
doc.add(newField("text_vectors", TestUtil.randomAnalysisString(random(), 6, true), ft));
doc.add(new IntPoint("point", random().nextInt()));
doc.add(new IntPoint("point2d", random().nextInt(), random().nextInt()));
if (random().nextInt(10) > 0) {
// single doc
try {
iw.addDocument(doc);
// we made it, sometimes delete our doc, or update a dv
int thingToDo = random().nextInt(4);
if (thingToDo == 0) {
iw.deleteDocuments(new Term("id", Integer.toString(i)));
} else if (thingToDo == 1) {
iw.updateNumericDocValue(new Term("id", Integer.toString(i)), "dv", i + 1L);
} else if (thingToDo == 2) {
iw.updateBinaryDocValue(new Term("id", Integer.toString(i)), "dv2", new BytesRef(Integer.toString(i + 1)));
}
} catch (VirtualMachineError | AlreadyClosedException disaster) {
getTragedy(disaster, iw, exceptionStream);
continue STARTOVER;
}
} else {
// block docs
Document doc2 = new Document();
doc2.add(newStringField("id", Integer.toString(-i), Field.Store.NO));
doc2.add(newTextField("text1", TestUtil.randomAnalysisString(random(), 20, true), Field.Store.NO));
doc2.add(new StoredField("stored1", "foo"));
doc2.add(new StoredField("stored1", "bar"));
doc2.add(newField("text_vectors", TestUtil.randomAnalysisString(random(), 6, true), ft));
try {
iw.addDocuments(Arrays.asList(doc, doc2));
// we made it, sometimes delete our docs
if (random().nextBoolean()) {
iw.deleteDocuments(new Term("id", Integer.toString(i)), new Term("id", Integer.toString(-i)));
}
} catch (VirtualMachineError | AlreadyClosedException disaster) {
getTragedy(disaster, iw, exceptionStream);
continue STARTOVER;
}
}
if (random().nextInt(10) == 0) {
// trigger flush:
try {
if (random().nextBoolean()) {
DirectoryReader ir = null;
try {
ir = DirectoryReader.open(iw, random().nextBoolean(), false);
TestUtil.checkReader(ir);
} finally {
IOUtils.closeWhileHandlingException(ir);
}
} else {
iw.commit();
}
if (DirectoryReader.indexExists(dir)) {
TestUtil.checkIndex(dir);
}
} catch (VirtualMachineError | AlreadyClosedException disaster) {
getTragedy(disaster, iw, exceptionStream);
continue STARTOVER;
}
}
}
try {
iw.close();
} catch (VirtualMachineError | AlreadyClosedException disaster) {
getTragedy(disaster, iw, exceptionStream);
continue STARTOVER;
}
} catch (Throwable t) {
System.out.println("Unexpected exception: dumping fake-exception-log:...");
exceptionStream.flush();
System.out.println(exceptionLog.toString("UTF-8"));
System.out.flush();
Rethrow.rethrow(t);
}
}
dir.close();
if (VERBOSE) {
System.out.println("TEST PASSED: dumping fake-exception-log:...");
System.out.println(exceptionLog.toString("UTF-8"));
}
}
Aggregations