/*
* Test a deletion policy that keeps last N commits.
*/
public void testKeepLastNDeletionPolicy() throws IOException {
final int N = 5;
for (int pass = 0; pass < 2; pass++) {
boolean useCompoundFile = (pass % 2) != 0;
Directory dir = newDirectory();
if (dir instanceof MockDirectoryWrapper) {
// test manually deletes files
((MockDirectoryWrapper) dir).setEnableVirusScanner(false);
}
KeepLastNDeletionPolicy policy = new KeepLastNDeletionPolicy(N);
for (int j = 0; j < N + 1; j++) {
IndexWriterConfig conf =
newIndexWriterConfig(new MockAnalyzer(random()))
.setOpenMode(OpenMode.CREATE)
.setIndexDeletionPolicy(policy)
.setMaxBufferedDocs(10);
MergePolicy mp = conf.getMergePolicy();
mp.setNoCFSRatio(useCompoundFile ? 1.0 : 0.0);
IndexWriter writer = new IndexWriter(dir, conf);
policy = (KeepLastNDeletionPolicy) writer.getConfig().getIndexDeletionPolicy();
for (int i = 0; i < 17; i++) {
addDoc(writer);
}
writer.forceMerge(1);
writer.close();
}
assertTrue(policy.numDelete > 0);
assertEquals(N + 1, policy.numOnInit);
assertEquals(N + 1, policy.numOnCommit);
// Simplistic check: just verify only the past N segments_N's still
// exist, and, I can open a reader on each:
long gen = SegmentInfos.getLastCommitGeneration(dir);
for (int i = 0; i < N + 1; i++) {
try {
IndexReader reader = DirectoryReader.open(dir);
reader.close();
if (i == N) {
fail("should have failed on commits prior to last " + N);
}
} catch (IOException e) {
if (i != N) {
throw e;
}
}
if (i < N) {
dir.deleteFile(IndexFileNames.fileNameFromGeneration(IndexFileNames.SEGMENTS, "", gen));
}
gen--;
}
dir.close();
}
}
public void testForceMergeNotNeeded() throws IOException {
try (Directory dir = newDirectory()) {
final AtomicBoolean mayMerge = new AtomicBoolean(true);
final MergeScheduler mergeScheduler =
new SerialMergeScheduler() {
@Override
public synchronized void merge(
IndexWriter writer, MergeTrigger trigger, boolean newMergesFound)
throws IOException {
if (mayMerge.get() == false) {
MergePolicy.OneMerge merge = writer.getNextMerge();
if (merge != null) {
System.out.println(
"TEST: we should not need any merging, yet merge policy returned merge "
+ merge);
throw new AssertionError();
}
}
super.merge(writer, trigger, newMergesFound);
}
};
MergePolicy mp = mergePolicy();
assumeFalse(
"this test cannot tolerate random forceMerges",
mp.toString().contains("MockRandomMergePolicy"));
mp.setNoCFSRatio(random().nextBoolean() ? 0 : 1);
IndexWriterConfig iwc = newIndexWriterConfig(new MockAnalyzer(random()));
iwc.setMergeScheduler(mergeScheduler);
iwc.setMergePolicy(mp);
IndexWriter writer = new IndexWriter(dir, iwc);
final int numSegments = TestUtil.nextInt(random(), 2, 20);
for (int i = 0; i < numSegments; ++i) {
final int numDocs = TestUtil.nextInt(random(), 1, 5);
for (int j = 0; j < numDocs; ++j) {
writer.addDocument(new Document());
}
writer.getReader().close();
}
for (int i = 5; i >= 0; --i) {
final int segmentCount = writer.getSegmentCount();
final int maxNumSegments = i == 0 ? 1 : TestUtil.nextInt(random(), 1, 10);
mayMerge.set(segmentCount > maxNumSegments);
if (VERBOSE) {
System.out.println(
"TEST: now forceMerge(maxNumSegments="
+ maxNumSegments
+ ") vs segmentCount="
+ segmentCount);
}
writer.forceMerge(maxNumSegments);
}
writer.close();
}
}
@Test
public void testNoMergePolicy() throws Exception {
MergePolicy mp = NoMergePolicy.NO_COMPOUND_FILES;
assertNull(mp.findMerges(null));
// LUCENE-3577
// assertNull(mp.findMergesToExpungeDeletes(null));
assertFalse(mp.useCompoundFile(null, null));
mp.close();
}
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();
}
@Override
public MergePolicy clone() {
MergePolicy clone;
try {
clone = (MergePolicy) super.clone();
} catch (CloneNotSupportedException e) {
// should not happen
throw new RuntimeException(e);
}
clone.writer = new SetOnce<IndexWriter>();
return clone;
}
/* Test keeping NO commit points. This is a viable and
* useful case eg where you want to build a big index and
* you know there are no readers.
*/
public void testKeepNoneOnInitDeletionPolicy() throws IOException {
for (int pass = 0; pass < 2; pass++) {
boolean useCompoundFile = (pass % 2) != 0;
Directory dir = newDirectory();
IndexWriterConfig conf =
newIndexWriterConfig(new MockAnalyzer(random()))
.setOpenMode(OpenMode.CREATE)
.setIndexDeletionPolicy(new KeepNoneOnInitDeletionPolicy())
.setMaxBufferedDocs(10);
MergePolicy mp = conf.getMergePolicy();
mp.setNoCFSRatio(useCompoundFile ? 1.0 : 0.0);
IndexWriter writer = new IndexWriter(dir, conf);
KeepNoneOnInitDeletionPolicy policy =
(KeepNoneOnInitDeletionPolicy) writer.getConfig().getIndexDeletionPolicy();
for (int i = 0; i < 107; i++) {
addDoc(writer);
}
writer.close();
conf =
newIndexWriterConfig(new MockAnalyzer(random()))
.setOpenMode(OpenMode.APPEND)
.setIndexDeletionPolicy(policy);
mp = conf.getMergePolicy();
mp.setNoCFSRatio(1.0);
writer = new IndexWriter(dir, conf);
policy = (KeepNoneOnInitDeletionPolicy) writer.getConfig().getIndexDeletionPolicy();
writer.forceMerge(1);
writer.close();
assertEquals(2, policy.numOnInit);
// If we are not auto committing then there should
// be exactly 2 commits (one per close above):
assertEquals(2, policy.numOnCommit);
// Simplistic check: just verify the index is in fact
// readable:
IndexReader reader = DirectoryReader.open(dir);
reader.close();
dir.close();
}
}
public MergeSpecification findForcedMerges(
SegmentInfos segmentInfos,
int maxSegmentCount,
Map<SegmentCommitInfo, Boolean> segmentsToMerge)
throws IOException {
return inner.findForcedMerges(segmentInfos, maxSegmentCount, segmentsToMerge);
}
// LUCENE-2790: tests that the non CFS files were deleted by addIndexes
public void testNonCFSLeftovers() throws Exception {
Directory[] dirs = new Directory[2];
for (int i = 0; i < dirs.length; i++) {
dirs[i] = new RAMDirectory();
IndexWriter w =
new IndexWriter(
dirs[i], new IndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(random())));
Document d = new Document();
FieldType customType = new FieldType(TextField.TYPE_STORED);
customType.setStoreTermVectors(true);
d.add(new Field("c", "v", customType));
w.addDocument(d);
w.close();
}
IndexReader[] readers =
new IndexReader[] {DirectoryReader.open(dirs[0]), DirectoryReader.open(dirs[1])};
Directory dir = new MockDirectoryWrapper(random(), new RAMDirectory());
IndexWriterConfig conf =
new IndexWriterConfig(TEST_VERSION_CURRENT, new MockAnalyzer(random()))
.setMergePolicy(newLogMergePolicy(true));
MergePolicy lmp = conf.getMergePolicy();
// Force creation of CFS:
lmp.setNoCFSRatio(1.0);
lmp.setMaxCFSSegmentSizeMB(Double.POSITIVE_INFINITY);
IndexWriter w3 = new IndexWriter(dir, conf);
w3.addIndexes(readers);
w3.close();
// we should now see segments_X,
// segments.gen,_Y.cfs,_Y.cfe, _Z.si
assertEquals(
"Only one compound segment should exist, but got: " + Arrays.toString(dir.listAll()),
5,
dir.listAll().length);
dir.close();
}
/*
* Test a silly deletion policy that keeps all commits around.
*/
public void testKeepAllDeletionPolicy() throws IOException {
for (int pass = 0; pass < 2; pass++) {
if (VERBOSE) {
System.out.println("TEST: cycle pass="******"TEST: open writer for forceMerge");
}
writer = new IndexWriter(dir, conf);
policy = (KeepAllDeletionPolicy) writer.getConfig().getIndexDeletionPolicy();
writer.forceMerge(1);
writer.close();
}
assertEquals(needsMerging ? 2 : 1, policy.numOnInit);
// If we are not auto committing then there should
// be exactly 2 commits (one per close above):
assertEquals(1 + (needsMerging ? 1 : 0), policy.numOnCommit);
// Test listCommits
Collection<IndexCommit> commits = DirectoryReader.listCommits(dir);
// 2 from closing writer
assertEquals(1 + (needsMerging ? 1 : 0), commits.size());
// Make sure we can open a reader on each commit:
for (final IndexCommit commit : commits) {
IndexReader r = DirectoryReader.open(commit);
r.close();
}
// Simplistic check: just verify all segments_N's still
// exist, and, I can open a reader on each:
long gen = SegmentInfos.getLastCommitGeneration(dir);
while (gen > 0) {
IndexReader reader = DirectoryReader.open(dir);
reader.close();
dir.deleteFile(IndexFileNames.fileNameFromGeneration(IndexFileNames.SEGMENTS, "", gen));
gen--;
if (gen > 0) {
// Now that we've removed a commit point, which
// should have orphan'd at least one index file.
// Open & close a writer and assert that it
// actually removed something:
int preCount = dir.listAll().length;
writer =
new IndexWriter(
dir,
newIndexWriterConfig(new MockAnalyzer(random()))
.setOpenMode(OpenMode.APPEND)
.setIndexDeletionPolicy(policy));
writer.close();
int postCount = dir.listAll().length;
assertTrue(postCount < preCount);
}
}
dir.close();
}
}
/*
* Test "by time expiration" deletion policy:
*/
public void testExpirationTimeDeletionPolicy() throws IOException, InterruptedException {
final double SECONDS = 2.0;
Directory dir = newDirectory();
if (dir instanceof MockDirectoryWrapper) {
// test manually deletes files
((MockDirectoryWrapper) dir).setEnableVirusScanner(false);
}
IndexWriterConfig conf =
newIndexWriterConfig(new MockAnalyzer(random()))
.setIndexDeletionPolicy(new ExpirationTimeDeletionPolicy(dir, SECONDS));
MergePolicy mp = conf.getMergePolicy();
mp.setNoCFSRatio(1.0);
IndexWriter writer = new IndexWriter(dir, conf);
ExpirationTimeDeletionPolicy policy =
(ExpirationTimeDeletionPolicy) writer.getConfig().getIndexDeletionPolicy();
Map<String, String> commitData = new HashMap<>();
commitData.put("commitTime", String.valueOf(System.currentTimeMillis()));
writer.setCommitData(commitData);
writer.commit();
writer.close();
long lastDeleteTime = 0;
final int targetNumDelete = TestUtil.nextInt(random(), 1, 5);
while (policy.numDelete < targetNumDelete) {
// Record last time when writer performed deletes of
// past commits
lastDeleteTime = System.currentTimeMillis();
conf =
newIndexWriterConfig(new MockAnalyzer(random()))
.setOpenMode(OpenMode.APPEND)
.setIndexDeletionPolicy(policy);
mp = conf.getMergePolicy();
mp.setNoCFSRatio(1.0);
writer = new IndexWriter(dir, conf);
policy = (ExpirationTimeDeletionPolicy) writer.getConfig().getIndexDeletionPolicy();
for (int j = 0; j < 17; j++) {
addDoc(writer);
}
commitData = new HashMap<>();
commitData.put("commitTime", String.valueOf(System.currentTimeMillis()));
writer.setCommitData(commitData);
writer.commit();
writer.close();
Thread.sleep((int) (1000.0 * (SECONDS / 5.0)));
}
// Then simplistic check: just verify that the
// segments_N's that still exist are in fact within SECONDS
// seconds of the last one's mod time, and, that I can
// open a reader on each:
long gen = SegmentInfos.getLastCommitGeneration(dir);
String fileName = IndexFileNames.fileNameFromGeneration(IndexFileNames.SEGMENTS, "", gen);
boolean oneSecondResolution = true;
while (gen > 0) {
try {
IndexReader reader = DirectoryReader.open(dir);
reader.close();
fileName = IndexFileNames.fileNameFromGeneration(IndexFileNames.SEGMENTS, "", gen);
// if we are on a filesystem that seems to have only
// 1 second resolution, allow +1 second in commit
// age tolerance:
SegmentInfos sis = SegmentInfos.readCommit(dir, fileName);
long modTime = Long.parseLong(sis.getUserData().get("commitTime"));
oneSecondResolution &= (modTime % 1000) == 0;
final long leeway = (long) ((SECONDS + (oneSecondResolution ? 1.0 : 0.0)) * 1000);
assertTrue(
"commit point was older than "
+ SECONDS
+ " seconds ("
+ (lastDeleteTime - modTime)
+ " msec) but did not get deleted ",
lastDeleteTime - modTime <= leeway);
} catch (IOException e) {
// OK
break;
}
dir.deleteFile(IndexFileNames.fileNameFromGeneration(IndexFileNames.SEGMENTS, "", gen));
gen--;
}
dir.close();
}
public boolean useCompoundFile(SegmentInfos infos, SegmentCommitInfo mergedInfo)
throws IOException {
return inner.useCompoundFile(infos, mergedInfo);
}
public MergeSpecification findMerges(MergeTrigger mergeTrigger, SegmentInfos segmentInfos)
throws IOException {
return inner.findMerges(mergeTrigger, segmentInfos);
}
public MergeSpecification findForcedDeletesMerges(SegmentInfos segmentInfos) throws IOException {
return inner.findForcedDeletesMerges(segmentInfos);
}
public void close() {
inner.close();
}
private RandomMergePolicy(MergePolicy inner) {
super(inner.getNoCFSRatio(), (long) (inner.getMaxCFSSegmentSizeMB() * 1024 * 1024));
this.inner = inner;
log.info("RandomMergePolicy wrapping {}: {}", inner.getClass(), inner);
}
/*
* Test a deletion policy that keeps last N commits
* around, through creates.
*/
public void testKeepLastNDeletionPolicyWithCreates() throws IOException {
final int N = 10;
for (int pass = 0; pass < 2; pass++) {
boolean useCompoundFile = (pass % 2) != 0;
Directory dir = newDirectory();
if (dir instanceof MockDirectoryWrapper) {
// test manually deletes files
((MockDirectoryWrapper) dir).setEnableVirusScanner(false);
}
IndexWriterConfig conf =
newIndexWriterConfig(new MockAnalyzer(random()))
.setOpenMode(OpenMode.CREATE)
.setIndexDeletionPolicy(new KeepLastNDeletionPolicy(N))
.setMaxBufferedDocs(10);
MergePolicy mp = conf.getMergePolicy();
mp.setNoCFSRatio(useCompoundFile ? 1.0 : 0.0);
IndexWriter writer = new IndexWriter(dir, conf);
KeepLastNDeletionPolicy policy =
(KeepLastNDeletionPolicy) writer.getConfig().getIndexDeletionPolicy();
writer.close();
Term searchTerm = new Term("content", "aaa");
Query query = new TermQuery(searchTerm);
for (int i = 0; i < N + 1; i++) {
conf =
newIndexWriterConfig(new MockAnalyzer(random()))
.setOpenMode(OpenMode.APPEND)
.setIndexDeletionPolicy(policy)
.setMaxBufferedDocs(10);
mp = conf.getMergePolicy();
mp.setNoCFSRatio(useCompoundFile ? 1.0 : 0.0);
writer = new IndexWriter(dir, conf);
policy = (KeepLastNDeletionPolicy) writer.getConfig().getIndexDeletionPolicy();
for (int j = 0; j < 17; j++) {
addDocWithID(writer, i * (N + 1) + j);
}
// this is a commit
writer.close();
conf =
new IndexWriterConfig(new MockAnalyzer(random()))
.setIndexDeletionPolicy(policy)
.setMergePolicy(NoMergePolicy.INSTANCE);
writer = new IndexWriter(dir, conf);
policy = (KeepLastNDeletionPolicy) writer.getConfig().getIndexDeletionPolicy();
writer.deleteDocuments(new Term("id", "" + (i * (N + 1) + 3)));
// this is a commit
writer.close();
IndexReader reader = DirectoryReader.open(dir);
IndexSearcher searcher = newSearcher(reader);
ScoreDoc[] hits = searcher.search(query, 1000).scoreDocs;
assertEquals(16, hits.length);
reader.close();
writer =
new IndexWriter(
dir,
newIndexWriterConfig(new MockAnalyzer(random()))
.setOpenMode(OpenMode.CREATE)
.setIndexDeletionPolicy(policy));
policy = (KeepLastNDeletionPolicy) writer.getConfig().getIndexDeletionPolicy();
// This will not commit: there are no changes
// pending because we opened for "create":
writer.close();
}
assertEquals(3 * (N + 1) + 1, policy.numOnInit);
assertEquals(3 * (N + 1) + 1, policy.numOnCommit);
IndexReader rwReader = DirectoryReader.open(dir);
IndexSearcher searcher = newSearcher(rwReader);
ScoreDoc[] hits = searcher.search(query, 1000).scoreDocs;
assertEquals(0, hits.length);
// Simplistic check: just verify only the past N segments_N's still
// exist, and, I can open a reader on each:
long gen = SegmentInfos.getLastCommitGeneration(dir);
int expectedCount = 0;
rwReader.close();
for (int i = 0; i < N + 1; i++) {
try {
IndexReader reader = DirectoryReader.open(dir);
// Work backwards in commits on what the expected
// count should be.
searcher = newSearcher(reader);
hits = searcher.search(query, 1000).scoreDocs;
assertEquals(expectedCount, hits.length);
if (expectedCount == 0) {
expectedCount = 16;
} else if (expectedCount == 16) {
expectedCount = 17;
} else if (expectedCount == 17) {
expectedCount = 0;
}
reader.close();
if (i == N) {
fail("should have failed on commits before last " + N);
}
} catch (IOException e) {
if (i != N) {
throw e;
}
}
if (i < N) {
dir.deleteFile(IndexFileNames.fileNameFromGeneration(IndexFileNames.SEGMENTS, "", gen));
}
gen--;
}
dir.close();
}
}
public void setIndexWriter(IndexWriter writer) {
inner.setIndexWriter(writer);
}