/*
* 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();
}
}
// Verifies no *.prx exists when all fields omit term positions:
public void testNoPrxFile() throws Throwable {
Directory ram = newDirectory();
if (ram instanceof MockDirectoryWrapper) {
// we verify some files get deleted
((MockDirectoryWrapper) ram).setEnableVirusScanner(false);
}
Analyzer analyzer = new MockAnalyzer(random());
IndexWriter writer =
new IndexWriter(
ram,
newIndexWriterConfig(analyzer)
.setMaxBufferedDocs(3)
.setMergePolicy(newLogMergePolicy()));
LogMergePolicy lmp = (LogMergePolicy) writer.getConfig().getMergePolicy();
lmp.setMergeFactor(2);
lmp.setNoCFSRatio(0.0);
Document d = new Document();
FieldType ft = new FieldType(TextField.TYPE_NOT_STORED);
ft.setIndexOptions(IndexOptions.DOCS_AND_FREQS);
Field f1 = newField("f1", "This field has term freqs", ft);
d.add(f1);
for (int i = 0; i < 30; i++) writer.addDocument(d);
writer.commit();
assertNoPrx(ram);
// now add some documents with positions, and check there is no prox after optimization
d = new Document();
f1 = newTextField("f1", "This field has positions", Field.Store.NO);
d.add(f1);
for (int i = 0; i < 30; i++) writer.addDocument(d);
// force merge
writer.forceMerge(1);
// flush
writer.close();
assertNoPrx(ram);
ram.close();
}
/*
* 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();
}
}
/*
* 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();
}
@Test
public void testRollingUpdates() throws Exception {
Random random = new Random(random().nextLong());
final BaseDirectoryWrapper dir = newDirectory();
// test checks for no unref'ed files with the IW helper method, which isn't aware of "tried to
// delete files"
if (dir instanceof MockDirectoryWrapper) {
((MockDirectoryWrapper) dir).setEnableVirusScanner(false);
}
final LineFileDocs docs = new LineFileDocs(random, true);
// provider.register(new MemoryCodec());
if (random().nextBoolean()) {
Codec.setDefault(
TestUtil.alwaysPostingsFormat(
new MemoryPostingsFormat(random().nextBoolean(), random.nextFloat())));
}
MockAnalyzer analyzer = new MockAnalyzer(random());
analyzer.setMaxTokenLength(TestUtil.nextInt(random(), 1, IndexWriter.MAX_TERM_LENGTH));
final IndexWriter w = new IndexWriter(dir, newIndexWriterConfig(analyzer));
final int SIZE = atLeast(20);
int id = 0;
IndexReader r = null;
IndexSearcher s = null;
final int numUpdates =
(int)
(SIZE * (2 + (TEST_NIGHTLY ? 200 * random().nextDouble() : 5 * random().nextDouble())));
if (VERBOSE) {
System.out.println("TEST: numUpdates=" + numUpdates);
}
int updateCount = 0;
// TODO: sometimes update ids not in order...
for (int docIter = 0; docIter < numUpdates; docIter++) {
final Document doc = docs.nextDoc();
final String myID = Integer.toString(id);
if (id == SIZE - 1) {
id = 0;
} else {
id++;
}
if (VERBOSE) {
System.out.println(" docIter=" + docIter + " id=" + id);
}
((Field) doc.getField("docid")).setStringValue(myID);
Term idTerm = new Term("docid", myID);
final boolean doUpdate;
if (s != null && updateCount < SIZE) {
TopDocs hits = s.search(new TermQuery(idTerm), 1);
assertEquals(1, hits.totalHits);
doUpdate = !w.tryDeleteDocument(r, hits.scoreDocs[0].doc);
if (VERBOSE) {
if (doUpdate) {
System.out.println(" tryDeleteDocument failed");
} else {
System.out.println(" tryDeleteDocument succeeded");
}
}
} else {
doUpdate = true;
if (VERBOSE) {
System.out.println(" no searcher: doUpdate=true");
}
}
updateCount++;
if (doUpdate) {
if (random().nextBoolean()) {
w.updateDocument(idTerm, doc);
} else {
// It's OK to not be atomic for this test (no separate thread reopening readers):
w.deleteDocuments(new TermQuery(idTerm));
w.addDocument(doc);
}
} else {
w.addDocument(doc);
}
if (docIter >= SIZE && random().nextInt(50) == 17) {
if (r != null) {
r.close();
}
final boolean applyDeletions = random().nextBoolean();
if (VERBOSE) {
System.out.println("TEST: reopen applyDeletions=" + applyDeletions);
}
r = w.getReader(applyDeletions);
if (applyDeletions) {
s = newSearcher(r);
} else {
s = null;
}
assertTrue(
"applyDeletions=" + applyDeletions + " r.numDocs()=" + r.numDocs() + " vs SIZE=" + SIZE,
!applyDeletions || r.numDocs() == SIZE);
updateCount = 0;
}
}
if (r != null) {
r.close();
}
w.commit();
assertEquals(SIZE, w.numDocs());
w.close();
TestIndexWriter.assertNoUnreferencedFiles(dir, "leftover files after rolling updates");
docs.close();
// LUCENE-4455:
SegmentInfos infos = SegmentInfos.readLatestCommit(dir);
long totalBytes = 0;
for (SegmentCommitInfo sipc : infos) {
totalBytes += sipc.sizeInBytes();
}
long totalBytes2 = 0;
for (String fileName : dir.listAll()) {
if (IndexFileNames.CODEC_FILE_PATTERN.matcher(fileName).matches()) {
totalBytes2 += dir.fileLength(fileName);
}
}
assertEquals(totalBytes2, totalBytes);
dir.close();
}