public void testUpdateSameDoc() throws Exception {
final Directory dir = newDirectory();
final LineFileDocs docs = new LineFileDocs(random());
for (int r = 0; r < 3; r++) {
final IndexWriter w =
new IndexWriter(
dir, newIndexWriterConfig(new MockAnalyzer(random())).setMaxBufferedDocs(2));
final int numUpdates = atLeast(20);
int numThreads = TestUtil.nextInt(random(), 2, 6);
IndexingThread[] threads = new IndexingThread[numThreads];
for (int i = 0; i < numThreads; i++) {
threads[i] = new IndexingThread(docs, w, numUpdates);
threads[i].start();
}
for (int i = 0; i < numThreads; i++) {
threads[i].join();
}
w.close();
}
IndexReader open = DirectoryReader.open(dir);
assertEquals(1, open.numDocs());
open.close();
docs.close();
dir.close();
}
protected Fields getTermVectorsFromLucene(DirectoryReader directoryReader, TestDoc doc)
throws IOException {
IndexSearcher searcher = new IndexSearcher(directoryReader);
TopDocs search = searcher.search(new TermQuery(new Term("id", doc.id)), 1);
ScoreDoc[] scoreDocs = search.scoreDocs;
assert (scoreDocs.length == 1);
return directoryReader.getTermVectors(scoreDocs[0].doc);
}
private void doTest(Random random, PrintWriter out, boolean useCompoundFiles, int MAX_DOCS)
throws Exception {
Directory directory = newDirectory();
Analyzer analyzer = new MockAnalyzer(random);
IndexWriterConfig conf = newIndexWriterConfig(analyzer);
final MergePolicy mp = conf.getMergePolicy();
mp.setNoCFSRatio(useCompoundFiles ? 1.0 : 0.0);
IndexWriter writer = new IndexWriter(directory, conf);
if (VERBOSE) {
System.out.println("TEST: now build index MAX_DOCS=" + MAX_DOCS);
}
for (int j = 0; j < MAX_DOCS; j++) {
Document d = new Document();
d.add(newTextField(PRIORITY_FIELD, HIGH_PRIORITY, Field.Store.YES));
d.add(newTextField(ID_FIELD, Integer.toString(j), Field.Store.YES));
writer.addDocument(d);
}
writer.close();
// try a search without OR
IndexReader reader = DirectoryReader.open(directory);
IndexSearcher searcher = newSearcher(reader);
Query query = new TermQuery(new Term(PRIORITY_FIELD, HIGH_PRIORITY));
out.println("Query: " + query.toString(PRIORITY_FIELD));
if (VERBOSE) {
System.out.println("TEST: search query=" + query);
}
final Sort sort = new Sort(SortField.FIELD_SCORE, new SortField(ID_FIELD, SortField.Type.INT));
ScoreDoc[] hits = searcher.search(query, null, MAX_DOCS, sort).scoreDocs;
printHits(out, hits, searcher);
checkHits(hits, MAX_DOCS, searcher);
// try a new search with OR
searcher = newSearcher(reader);
hits = null;
BooleanQuery booleanQuery = new BooleanQuery();
booleanQuery.add(
new TermQuery(new Term(PRIORITY_FIELD, HIGH_PRIORITY)), BooleanClause.Occur.SHOULD);
booleanQuery.add(
new TermQuery(new Term(PRIORITY_FIELD, MED_PRIORITY)), BooleanClause.Occur.SHOULD);
out.println("Query: " + booleanQuery.toString(PRIORITY_FIELD));
hits = searcher.search(booleanQuery, null, MAX_DOCS, sort).scoreDocs;
printHits(out, hits, searcher);
checkHits(hits, MAX_DOCS, searcher);
reader.close();
directory.close();
}
protected DirectoryReader indexDocsWithLucene(TestDoc[] testDocs) throws IOException {
Map<String, Analyzer> mapping = new HashMap<String, Analyzer>();
for (TestFieldSetting field : testDocs[0].fieldSettings) {
if (field.storedPayloads) {
mapping.put(
field.name,
new Analyzer() {
@Override
protected TokenStreamComponents createComponents(String fieldName, Reader reader) {
Tokenizer tokenizer = new StandardTokenizer(Version.CURRENT.luceneVersion, reader);
TokenFilter filter = new LowerCaseFilter(Version.CURRENT.luceneVersion, tokenizer);
filter = new TypeAsPayloadTokenFilter(filter);
return new TokenStreamComponents(tokenizer, filter);
}
});
}
}
PerFieldAnalyzerWrapper wrapper =
new PerFieldAnalyzerWrapper(
new StandardAnalyzer(Version.CURRENT.luceneVersion, CharArraySet.EMPTY_SET), mapping);
Directory dir = new RAMDirectory();
IndexWriterConfig conf = new IndexWriterConfig(Version.CURRENT.luceneVersion, wrapper);
conf.setOpenMode(IndexWriterConfig.OpenMode.CREATE);
IndexWriter writer = new IndexWriter(dir, conf);
for (TestDoc doc : testDocs) {
Document d = new Document();
d.add(new Field("id", doc.id, StringField.TYPE_STORED));
for (int i = 0; i < doc.fieldContent.length; i++) {
FieldType type = new FieldType(TextField.TYPE_STORED);
TestFieldSetting fieldSetting = doc.fieldSettings[i];
type.setStoreTermVectorOffsets(fieldSetting.storedOffset);
type.setStoreTermVectorPayloads(fieldSetting.storedPayloads);
type.setStoreTermVectorPositions(
fieldSetting.storedPositions
|| fieldSetting.storedPayloads
|| fieldSetting.storedOffset);
type.setStoreTermVectors(true);
type.freeze();
d.add(new Field(fieldSetting.name, doc.fieldContent[i], type));
}
writer.updateDocument(new Term("id", doc.id), d);
writer.commit();
}
writer.close();
return DirectoryReader.open(dir);
}
@Override
public void run() {
try {
DirectoryReader open = null;
for (int i = 0; i < num; i++) {
Document doc = new Document(); // docs.nextDoc();
BytesRef br = new BytesRef("test");
doc.add(newStringField("id", br, Field.Store.NO));
writer.updateDocument(new Term("id", br), doc);
if (random().nextInt(3) == 0) {
if (open == null) {
open = DirectoryReader.open(writer, true);
}
DirectoryReader reader = DirectoryReader.openIfChanged(open);
if (reader != null) {
open.close();
open = reader;
}
assertEquals(
"iter: "
+ i
+ " numDocs: "
+ open.numDocs()
+ " del: "
+ open.numDeletedDocs()
+ " max: "
+ open.maxDoc(),
1,
open.numDocs());
}
}
if (open != null) {
open.close();
}
} catch (Exception e) {
throw new RuntimeException(e);
}
}
@Before
public void setup() throws Exception {
super.setUp();
// setup field mappers
strMapper =
new StringFieldMapper.Builder("str_value")
.build(new Mapper.BuilderContext(null, new ContentPath(1)));
lngMapper =
new LongFieldMapper.Builder("lng_value")
.build(new Mapper.BuilderContext(null, new ContentPath(1)));
dblMapper =
new DoubleFieldMapper.Builder("dbl_value")
.build(new Mapper.BuilderContext(null, new ContentPath(1)));
// create index and fielddata service
ifdService = new IndexFieldDataService(new Index("test"), new DummyCircuitBreakerService());
MapperService mapperService =
MapperTestUtils.newMapperService(
ifdService.index(), ImmutableSettings.Builder.EMPTY_SETTINGS);
ifdService.setIndexService(new StubIndexService(mapperService));
writer =
new IndexWriter(
new RAMDirectory(),
new IndexWriterConfig(Lucene.VERSION, new StandardAnalyzer(Lucene.VERSION)));
int numDocs = 10;
for (int i = 0; i < numDocs; i++) {
Document d = new Document();
d.add(new StringField(strMapper.names().indexName(), "str" + i, Field.Store.NO));
d.add(new LongField(lngMapper.names().indexName(), i, Field.Store.NO));
d.add(new DoubleField(dblMapper.names().indexName(), Double.valueOf(i), Field.Store.NO));
writer.addDocument(d);
}
reader = SlowCompositeReaderWrapper.wrap(DirectoryReader.open(writer, true));
}
@Test
public void testMixedChecksums() throws IOException {
final ShardId shardId = new ShardId(new Index("index"), 1);
DirectoryService directoryService = new LuceneManagedDirectoryService(random());
Store store =
new Store(
shardId,
ImmutableSettings.EMPTY,
directoryService,
randomDistributor(directoryService),
new DummyShardLock(shardId));
// this time random codec....
IndexWriter writer =
new IndexWriter(
store.directory(),
newIndexWriterConfig(random(), new MockAnalyzer(random()))
.setCodec(actualDefaultCodec()));
int docs = 1 + random().nextInt(100);
for (int i = 0; i < docs; i++) {
Document doc = new Document();
doc.add(
new TextField("id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
doc.add(
new TextField(
"body",
TestUtil.randomRealisticUnicodeString(random()),
random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
doc.add(
new SortedDocValuesField(
"dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random()))));
writer.addDocument(doc);
}
if (random().nextBoolean()) {
for (int i = 0; i < docs; i++) {
if (random().nextBoolean()) {
Document doc = new Document();
doc.add(
new TextField(
"id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
doc.add(
new TextField(
"body",
TestUtil.randomRealisticUnicodeString(random()),
random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
writer.updateDocument(new Term("id", "" + i), doc);
}
}
}
if (random().nextBoolean()) {
DirectoryReader.open(writer, random().nextBoolean()).close(); // flush
}
Store.MetadataSnapshot metadata;
// check before we committed
try {
store.getMetadata();
fail("no index present - expected exception");
} catch (IndexNotFoundException ex) {
// expected
}
assertThat(store.getMetadataOrEmpty(), is(Store.MetadataSnapshot.EMPTY)); // nothing committed
writer.commit();
writer.close();
Store.LegacyChecksums checksums = new Store.LegacyChecksums();
metadata = store.getMetadata();
assertThat(metadata.asMap().isEmpty(), is(false));
for (StoreFileMetaData meta : metadata) {
try (IndexInput input = store.directory().openInput(meta.name(), IOContext.DEFAULT)) {
if (meta.checksum() == null) {
String checksum = null;
try {
CodecUtil.retrieveChecksum(input);
fail("expected a corrupt index - posting format has not checksums");
} catch (CorruptIndexException
| IndexFormatTooOldException
| IndexFormatTooNewException ex) {
try (ChecksumIndexInput checksumIndexInput =
store.directory().openChecksumInput(meta.name(), IOContext.DEFAULT)) {
checksumIndexInput.seek(meta.length());
checksum = Store.digestToString(checksumIndexInput.getChecksum());
}
// fine - it's a postings format without checksums
checksums.add(new StoreFileMetaData(meta.name(), meta.length(), checksum, null));
}
} else {
String checksum = Store.digestToString(CodecUtil.retrieveChecksum(input));
assertThat(
"File: " + meta.name() + " has a different checksum",
meta.checksum(),
equalTo(checksum));
assertThat(meta.hasLegacyChecksum(), equalTo(false));
assertThat(meta.writtenBy(), equalTo(Version.LATEST));
}
}
}
assertConsistent(store, metadata);
checksums.write(store);
metadata = store.getMetadata();
assertThat(metadata.asMap().isEmpty(), is(false));
for (StoreFileMetaData meta : metadata) {
assertThat(
"file: " + meta.name() + " has a null checksum", meta.checksum(), not(nullValue()));
if (meta.hasLegacyChecksum()) {
try (ChecksumIndexInput checksumIndexInput =
store.directory().openChecksumInput(meta.name(), IOContext.DEFAULT)) {
checksumIndexInput.seek(meta.length());
assertThat(
meta.checksum(), equalTo(Store.digestToString(checksumIndexInput.getChecksum())));
}
} else {
try (IndexInput input = store.directory().openInput(meta.name(), IOContext.DEFAULT)) {
String checksum = Store.digestToString(CodecUtil.retrieveChecksum(input));
assertThat(
"File: " + meta.name() + " has a different checksum",
meta.checksum(),
equalTo(checksum));
assertThat(meta.hasLegacyChecksum(), equalTo(false));
assertThat(meta.writtenBy(), equalTo(Version.LATEST));
}
}
}
assertConsistent(store, metadata);
TestUtil.checkIndex(store.directory());
assertDeleteContent(store, directoryService);
IOUtils.close(store);
}
@Test
public void testNewChecksums() throws IOException {
final ShardId shardId = new ShardId(new Index("index"), 1);
DirectoryService directoryService = new LuceneManagedDirectoryService(random());
Store store =
new Store(
shardId,
ImmutableSettings.EMPTY,
directoryService,
randomDistributor(directoryService),
new DummyShardLock(shardId));
// set default codec - all segments need checksums
IndexWriter writer =
new IndexWriter(
store.directory(),
newIndexWriterConfig(random(), new MockAnalyzer(random()))
.setCodec(actualDefaultCodec()));
int docs = 1 + random().nextInt(100);
for (int i = 0; i < docs; i++) {
Document doc = new Document();
doc.add(
new TextField("id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
doc.add(
new TextField(
"body",
TestUtil.randomRealisticUnicodeString(random()),
random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
doc.add(
new SortedDocValuesField(
"dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random()))));
writer.addDocument(doc);
}
if (random().nextBoolean()) {
for (int i = 0; i < docs; i++) {
if (random().nextBoolean()) {
Document doc = new Document();
doc.add(
new TextField(
"id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
doc.add(
new TextField(
"body",
TestUtil.randomRealisticUnicodeString(random()),
random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
writer.updateDocument(new Term("id", "" + i), doc);
}
}
}
if (random().nextBoolean()) {
DirectoryReader.open(writer, random().nextBoolean()).close(); // flush
}
Store.MetadataSnapshot metadata;
// check before we committed
try {
store.getMetadata();
fail("no index present - expected exception");
} catch (IndexNotFoundException ex) {
// expected
}
assertThat(store.getMetadataOrEmpty(), is(Store.MetadataSnapshot.EMPTY)); // nothing committed
writer.commit();
writer.close();
metadata = store.getMetadata();
assertThat(metadata.asMap().isEmpty(), is(false));
for (StoreFileMetaData meta : metadata) {
try (IndexInput input = store.directory().openInput(meta.name(), IOContext.DEFAULT)) {
String checksum = Store.digestToString(CodecUtil.retrieveChecksum(input));
assertThat(
"File: " + meta.name() + " has a different checksum",
meta.checksum(),
equalTo(checksum));
assertThat(meta.hasLegacyChecksum(), equalTo(false));
assertThat(meta.writtenBy(), equalTo(Version.LATEST));
if (meta.name().endsWith(".si") || meta.name().startsWith("segments_")) {
assertThat(meta.hash().length, greaterThan(0));
}
}
}
assertConsistent(store, metadata);
TestUtil.checkIndex(store.directory());
assertDeleteContent(store, directoryService);
IOUtils.close(store);
}
// IF THIS TEST FAILS ON UPGRADE GO LOOK AT THE
// OldSIMockingCodec!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
@Test
public void testWriteLegacyChecksums() throws IOException {
final ShardId shardId = new ShardId(new Index("index"), 1);
DirectoryService directoryService = new LuceneManagedDirectoryService(random());
Store store =
new Store(
shardId,
ImmutableSettings.EMPTY,
directoryService,
randomDistributor(directoryService),
new DummyShardLock(shardId));
// set default codec - all segments need checksums
final boolean usesOldCodec = randomBoolean();
IndexWriter writer =
new IndexWriter(
store.directory(),
newIndexWriterConfig(random(), new MockAnalyzer(random()))
.setCodec(usesOldCodec ? new OldSIMockingCodec() : actualDefaultCodec()));
int docs = 1 + random().nextInt(100);
for (int i = 0; i < docs; i++) {
Document doc = new Document();
doc.add(
new TextField("id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
doc.add(
new TextField(
"body",
TestUtil.randomRealisticUnicodeString(random()),
random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
doc.add(
new SortedDocValuesField(
"dv", new BytesRef(TestUtil.randomRealisticUnicodeString(random()))));
writer.addDocument(doc);
}
if (random().nextBoolean()) {
for (int i = 0; i < docs; i++) {
if (random().nextBoolean()) {
Document doc = new Document();
doc.add(
new TextField(
"id", "" + i, random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
doc.add(
new TextField(
"body",
TestUtil.randomRealisticUnicodeString(random()),
random().nextBoolean() ? Field.Store.YES : Field.Store.NO));
writer.updateDocument(new Term("id", "" + i), doc);
}
}
}
if (random().nextBoolean()) {
DirectoryReader.open(writer, random().nextBoolean()).close(); // flush
}
Store.MetadataSnapshot metadata;
// check before we committed
try {
store.getMetadata();
fail("no index present - expected exception");
} catch (IndexNotFoundException ex) {
// expected
}
assertThat(store.getMetadataOrEmpty(), is(Store.MetadataSnapshot.EMPTY)); // nothing committed
writer.close();
Store.LegacyChecksums checksums = new Store.LegacyChecksums();
Map<String, StoreFileMetaData> legacyMeta = new HashMap<>();
for (String file : store.directory().listAll()) {
if (file.equals("write.lock") || file.equals(IndexFileNames.OLD_SEGMENTS_GEN)) {
continue;
}
BytesRef hash = new BytesRef();
if (file.startsWith("segments")) {
hash = Store.MetadataSnapshot.hashFile(store.directory(), file);
}
StoreFileMetaData storeFileMetaData =
new StoreFileMetaData(
file, store.directory().fileLength(file), file + "checksum", null, hash);
legacyMeta.put(file, storeFileMetaData);
checksums.add(storeFileMetaData);
}
checksums.write(store);
metadata = store.getMetadata();
Map<String, StoreFileMetaData> stringStoreFileMetaDataMap = metadata.asMap();
assertThat(legacyMeta.size(), equalTo(stringStoreFileMetaDataMap.size()));
if (usesOldCodec) {
for (StoreFileMetaData meta : legacyMeta.values()) {
assertTrue(meta.toString(), stringStoreFileMetaDataMap.containsKey(meta.name()));
assertEquals(meta.name() + "checksum", meta.checksum());
assertTrue(
meta + " vs. " + stringStoreFileMetaDataMap.get(meta.name()),
stringStoreFileMetaDataMap.get(meta.name()).isSame(meta));
}
} else {
// even if we have a legacy checksum - if we use a new codec we should reuse
for (StoreFileMetaData meta : legacyMeta.values()) {
assertTrue(meta.toString(), stringStoreFileMetaDataMap.containsKey(meta.name()));
assertFalse(
meta + " vs. " + stringStoreFileMetaDataMap.get(meta.name()),
stringStoreFileMetaDataMap.get(meta.name()).isSame(meta));
StoreFileMetaData storeFileMetaData = metadata.get(meta.name());
try (IndexInput input =
store.openVerifyingInput(meta.name(), IOContext.DEFAULT, storeFileMetaData)) {
assertTrue(storeFileMetaData.toString(), input instanceof Store.VerifyingIndexInput);
input.seek(meta.length());
Store.verify(input);
}
}
}
assertDeleteContent(store, directoryService);
IOUtils.close(store);
}