Java Code Examples for org.apache.nifi.provenance.search.Query#setMaxResults()

/**
 * Here the event file is simply corrupted by virtue of not having any event
 * records while having correct headers
 */
@Test
public void testWithWithEventFileMissingRecord() throws Exception {
    assumeFalse(isWindowsEnvironment());
    File eventFile = this.prepCorruptedEventFileTests();

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "foo-*"));
    query.setMaxResults(100);

    DataOutputStream in = new DataOutputStream(new GZIPOutputStream(new FileOutputStream(eventFile)));
    in.writeUTF("BlahBlah");
    in.writeInt(4);
    in.close();
    assertTrue(eventFile.exists());
    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(10, result.getMatchingEvents().size());
}
 

/**
 * Here the event file is simply corrupted by virtue of not having any event
 * records while having correct headers
 */
@Test
public void testWithWithEventFileMissingRecord() throws Exception {
    File eventFile = this.prepCorruptedEventFileTests();

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "foo-*"));
    query.setMaxResults(100);

    DataOutputStream in = new DataOutputStream(new GZIPOutputStream(new FileOutputStream(eventFile)));
    in.writeUTF("BlahBlah");
    in.writeInt(4);
    in.close();
    assertTrue(eventFile.exists());
    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(10, result.getMatchingEvents().size());
}
 

private List<Document> runQuery(final File indexDirectory, final List<File> storageDirs, final String query) throws IOException, ParseException {
    assumeFalse(isWindowsEnvironment());
    try (final DirectoryReader directoryReader = DirectoryReader.open(FSDirectory.open(indexDirectory.toPath()))) {
        final IndexSearcher searcher = new IndexSearcher(directoryReader);

        final Analyzer analyzer = new SimpleAnalyzer();
        final org.apache.lucene.search.Query luceneQuery = new QueryParser("uuid", analyzer).parse(query);

        final Query q = new Query("");
        q.setMaxResults(1000);
        final TopDocs topDocs = searcher.search(luceneQuery, 1000);

        final List<Document> docs = new ArrayList<>();
        for (final ScoreDoc scoreDoc : topDocs.scoreDocs) {
            final int docId = scoreDoc.doc;
            final Document d = directoryReader.document(docId);
            docs.add(d);
        }

        return docs;
    }
}
 

/**
 * Here the event file is simply corrupted by virtue of being empty (0
 * bytes)
 */
@Test
public void testWithWithEventFileCorrupted() throws Exception {
    assumeFalse(isWindowsEnvironment());
    File eventFile = this.prepCorruptedEventFileTests();

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "foo-*"));
    query.setMaxResults(100);
    DataOutputStream in = new DataOutputStream(new GZIPOutputStream(new FileOutputStream(eventFile)));
    in.close();
    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(10, result.getMatchingEvents().size());
}
 

/**
 * Here the event file is simply corrupted by virtue of being empty (0
 * bytes)
 */
@Test
public void testWithWithEventFileCorrupted() throws Exception {
    File eventFile = this.prepCorruptedEventFileTests();

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "foo-*"));
    query.setMaxResults(100);
    DataOutputStream in = new DataOutputStream(new GZIPOutputStream(new FileOutputStream(eventFile)));
    in.close();
    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(10, result.getMatchingEvents().size());
}
 

@Test
public void testIndexAndCompressOnRolloverAndSubsequentSearchAsync() throws InterruptedException {
    repo = new VolatileProvenanceRepository(NiFiProperties.createBasicNiFiProperties(null, null));

    final String uuid = "00000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");

    for (int i = 0; i < 10; i++) {
        attributes.put("uuid", "00000000-0000-0000-0000-00000000000" + i);
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        repo.registerEvent(builder.build());
    }

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.FlowFileUUID, "00000*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.Filename, "file-*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "12?4"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.TransitURI, "nifi://*"));
    query.setMaxResults(100);

    final QuerySubmission submission = repo.submitQuery(query, createUser());
    while (!submission.getResult().isFinished()) {
        Thread.sleep(100L);
    }

    assertEquals(10, submission.getResult().getMatchingEvents().size());
    for (final ProvenanceEventRecord match : submission.getResult().getMatchingEvents()) {
        System.out.println(match);
    }
}
 

@Test
public void testIndexAndCompressOnRolloverAndSubsequentEmptySearch() throws IOException, InterruptedException, ParseException {
    assumeFalse(isWindowsEnvironment());
    final RepositoryConfiguration config = createConfiguration();
    config.setMaxRecordLife(30, TimeUnit.SECONDS);
    config.setMaxStorageCapacity(1024L * 1024L);
    config.setMaxEventFileLife(500, TimeUnit.MILLISECONDS);
    config.setMaxEventFileCapacity(1024L * 1024L);
    config.setSearchableFields(new ArrayList<>(SearchableFields.getStandardFields()));

    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    final String uuid = "00000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    attributes.put("uuid", uuid);
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");

    for (int i = 0; i < 10; i++) {
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        repo.registerEvent(builder.build());
    }

    // Give time for rollover to happen
    repo.waitForRollover();

    final Query query = new Query(UUID.randomUUID().toString());
    query.setMaxResults(100);

    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(10, result.getMatchingEvents().size());
    for (final ProvenanceEventRecord match : result.getMatchingEvents()) {
        System.out.println(match);
    }

    Thread.sleep(2000L);

    config.setMaxStorageCapacity(100L);
    config.setMaxRecordLife(500, TimeUnit.MILLISECONDS);
    repo.purgeOldEvents();

    Thread.sleep(1000L);

    final QueryResult newRecordSet = repo.queryEvents(query, createUser());
    assertTrue(newRecordSet.getMatchingEvents().isEmpty());
}
 

@Test
public void testIndexAndCompressOnRolloverAndSubsequentSearchMultipleStorageDirs() throws IOException, InterruptedException, ParseException {
    assumeFalse(isWindowsEnvironment());
    final RepositoryConfiguration config = createConfiguration();
    config.addStorageDirectory("2", new File("target/storage/" + UUID.randomUUID().toString()));
    config.setMaxRecordLife(30, TimeUnit.SECONDS);
    config.setMaxStorageCapacity(1024L * 1024L);
    config.setMaxEventFileLife(1, TimeUnit.SECONDS);
    config.setMaxEventFileCapacity(1024L * 1024L);
    config.setSearchableFields(new ArrayList<>(SearchableFields.getStandardFields()));

    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    final String uuid = "00000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    for (int j = 0; j < 3; j++) {
        attributes.put("iteration", String.valueOf(j));

        builder.setEventTime(System.currentTimeMillis());
        builder.setEventType(ProvenanceEventType.RECEIVE);
        builder.setTransitUri("nifi://unit-test");
        builder.setComponentId("1234");
        builder.setComponentType("dummy processor");
        builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));

        for (int i = 0; i < 10; i++) {
            String uuidSuffix = String.valueOf(i + j * 10);
            if (uuidSuffix.length() < 2) {
                uuidSuffix = "0" + uuidSuffix;
            }

            attributes.put("uuid", "00000000-0000-0000-0000-0000000000" + uuidSuffix);
            builder.fromFlowFile(createFlowFile(i + j * 10, 3000L, attributes));
            repo.registerEvent(builder.build());
        }

        repo.waitForRollover();
    }

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.Filename, "file-*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "12?4"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.TransitURI, "nifi://*"));
    query.setMaxResults(100);

    final QuerySubmission submission = repo.submitQuery(query, createUser());
    while (!submission.getResult().isFinished()) {
        Thread.sleep(100L);
    }

    assertEquals(30, submission.getResult().getMatchingEvents().size());
    final Map<String, Integer> counts = new HashMap<>();
    for (final ProvenanceEventRecord match : submission.getResult().getMatchingEvents()) {
        System.out.println(match);

        final String index = match.getAttributes().get("iteration");
        Integer count = counts.get(index);
        if (count == null) {
            count = 0;
        }
        counts.put(index, count + 1);
    }

    assertEquals(3, counts.size());
    assertEquals(10, counts.get("0").intValue());
    assertEquals(10, counts.get("1").intValue());
    assertEquals(10, counts.get("2").intValue());

    config.setMaxRecordLife(1, TimeUnit.MILLISECONDS);

    repo.purgeOldEvents();

    Thread.sleep(2000L); // purge is async. Give it time to do its job.

    query.setMaxResults(100);
    final QuerySubmission noResultSubmission = repo.submitQuery(query, createUser());
    while (!noResultSubmission.getResult().isFinished()) {
        Thread.sleep(10L);
    }

    assertEquals(0, noResultSubmission.getResult().getTotalHitCount());
}
 

@Test
public void testIndexAndCompressOnRolloverAndSubsequentSearchAsync() throws InterruptedException {
    repo = new VolatileProvenanceRepository(NiFiProperties.createBasicNiFiProperties(null));

    final String uuid = "00000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");

    for (int i = 0; i < 10; i++) {
        attributes.put("uuid", "00000000-0000-0000-0000-00000000000" + i);
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        repo.registerEvent(builder.build());
    }

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.FlowFileUUID, "00000*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.Filename, "file-*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "12?4"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.TransitURI, "nifi://*"));
    query.setMaxResults(100);

    final QuerySubmission submission = repo.submitQuery(query, createUser());
    while (!submission.getResult().isFinished()) {
        Thread.sleep(100L);
    }

    assertEquals(10, submission.getResult().getMatchingEvents().size());
    for (final ProvenanceEventRecord match : submission.getResult().getMatchingEvents()) {
        System.out.println(match);
    }
}
 

@Test
public void testIndexOnRolloverAndSubsequentSearch() throws IOException, InterruptedException, ParseException {
    assumeFalse(isWindowsEnvironment());
    final RepositoryConfiguration config = createConfiguration();
    config.setMaxEventFileLife(500, TimeUnit.MILLISECONDS);
    config.setSearchableFields(new ArrayList<>(SearchableFields.getStandardFields()));
    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    final String uuid = "00000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");

    for (int i = 0; i < 10; i++) {
        attributes.put("uuid", "00000000-0000-0000-0000-00000000000" + i);
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        repo.registerEvent(builder.build());
    }

    repo.waitForRollover();

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.FlowFileUUID, "000000*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.Filename, "file-*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "12?4"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.TransitURI, "nifi://*"));
    query.setMaxResults(100);

    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(10, result.getMatchingEvents().size());
    for (final ProvenanceEventRecord match : result.getMatchingEvents()) {
        System.out.println(match);
    }
}
 

@Test
public void testIndexOnRolloverWithImmenseAttribute() throws IOException {
    assumeFalse(isWindowsEnvironment());
    final RepositoryConfiguration config = createConfiguration();
    config.setMaxEventFileLife(500, TimeUnit.MILLISECONDS);
    config.setSearchableFields(new ArrayList<>(SearchableFields.getStandardFields()));
    config.setSearchableAttributes(SearchableFieldParser.extractSearchableFields("immense", false));
    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    int immenseAttrSize = 33000; // must be greater than 32766 for a meaningful test
    StringBuilder immenseBldr = new StringBuilder(immenseAttrSize);
    for (int i = 0; i < immenseAttrSize; i++) {
        immenseBldr.append('0');
    }
    final String uuid = "00000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);
    attributes.put("immense", immenseBldr.toString());

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");

    for (int i = 0; i < 10; i++) {
        attributes.put("uuid", "00000000-0000-0000-0000-00000000000" + i);
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        repo.registerEvent(builder.build());
    }

    repo.waitForRollover();

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.newSearchableAttribute("immense"), "000*"));
    query.setMaxResults(100);

    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(10, result.getMatchingEvents().size());
}
 

@Test
public void testAddToMultipleLogsAndRecover() throws IOException, InterruptedException {
    assumeFalse(isWindowsEnvironment());
    final List<SearchableField> searchableFields = new ArrayList<>();
    searchableFields.add(SearchableFields.ComponentID);

    final RepositoryConfiguration config = createConfiguration();
    config.setMaxEventFileCapacity(1024L * 1024L);
    config.setMaxEventFileLife(2, TimeUnit.SECONDS);
    config.setSearchableFields(searchableFields);
    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("uuid", UUID.randomUUID().toString());

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");
    final ProvenanceEventRecord record = builder.build();

    for (int i = 0; i < 10; i++) {
        repo.registerEvent(record);
    }

    builder.setComponentId("XXXX"); // create a different component id so that we can make sure we query this record.

    attributes.put("uuid", "11111111-1111-1111-1111-111111111111");

    builder.fromFlowFile(createFlowFile(11L, 11L, attributes));
    repo.registerEvent(builder.build());

    repo.waitForRollover();
    Thread.sleep(500L); // Give the repo time to shutdown (i.e., close all file handles, etc.)

    // Create a new repo and add another record with component id XXXX so that we can ensure that it's added to a different
    // log file than the previous one.
    attributes.put("uuid", "22222222-2222-2222-2222-222222222222");
    builder.fromFlowFile(createFlowFile(11L, 11L, attributes));
    repo.registerEvent(builder.build());
    repo.waitForRollover();

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "XXXX"));
    query.setMaxResults(100);

    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(2, result.getMatchingEvents().size());
    for (final ProvenanceEventRecord match : result.getMatchingEvents()) {
        System.out.println(match);
    }
}
 

/**
 * Submits a provenance query.
 *
 * @param provenanceDto dto
 * @return provenance info
 */
public ProvenanceDTO submitProvenance(ProvenanceDTO provenanceDto) {
    final ProvenanceRequestDTO requestDto = provenanceDto.getRequest();

    // create the query
    final Query query = new Query(provenanceDto.getId());

    // if the request was specified
    if (requestDto != null) {
        // add each search term specified
        final Map<String, String> searchTerms = requestDto.getSearchTerms();
        if (searchTerms != null) {
            for (final Map.Entry<String, String> searchTerm : searchTerms.entrySet()) {
                SearchableField field;

                field = SearchableFields.getSearchableField(searchTerm.getKey());
                if (field == null) {
                    field = SearchableFields.newSearchableAttribute(searchTerm.getKey());
                }
                query.addSearchTerm(SearchTerms.newSearchTerm(field, searchTerm.getValue()));
            }
        }

        // specify the start date if specified
        if (requestDto.getStartDate() != null) {
            query.setStartDate(requestDto.getStartDate());
        }

        // ensure an end date is populated
        if (requestDto.getEndDate() != null) {
            query.setEndDate(requestDto.getEndDate());
        }

        // set the min/max file size
        query.setMinFileSize(requestDto.getMinimumFileSize());
        query.setMaxFileSize(requestDto.getMaximumFileSize());

        // set the max results desired
        query.setMaxResults(requestDto.getMaxResults());
    }

    // submit the query to the provenance repository
    final ProvenanceRepository provenanceRepository = flowController.getProvenanceRepository();
    final QuerySubmission querySubmission = provenanceRepository.submitQuery(query, NiFiUserUtils.getNiFiUser());

    // return the query with the results populated at this point
    return getProvenanceQuery(querySubmission.getQueryIdentifier(), requestDto.getSummarize(), requestDto.getIncrementalResults());
}
 

@Test
public void testAddToMultipleLogsAndRecover() throws IOException, InterruptedException {
    final List<SearchableField> searchableFields = new ArrayList<>();
    searchableFields.add(SearchableFields.ComponentID);

    final RepositoryConfiguration config = createConfiguration();
    config.setMaxEventFileCapacity(1024L * 1024L);
    config.setMaxEventFileLife(2, TimeUnit.SECONDS);
    config.setSearchableFields(searchableFields);
    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("uuid", UUID.randomUUID().toString());

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");
    final ProvenanceEventRecord record = builder.build();

    for (int i = 0; i < 10; i++) {
        repo.registerEvent(record);
    }

    builder.setComponentId("XXXX"); // create a different component id so that we can make sure we query this record.

    attributes.put("uuid", "11111111-1111-1111-1111-111111111111");

    builder.fromFlowFile(createFlowFile(11L, 11L, attributes));
    repo.registerEvent(builder.build());

    repo.waitForRollover();
    Thread.sleep(500L); // Give the repo time to shutdown (i.e., close all file handles, etc.)

    // Create a new repo and add another record with component id XXXX so that we can ensure that it's added to a different
    // log file than the previous one.
    attributes.put("uuid", "22222222-2222-2222-2222-222222222222");
    builder.fromFlowFile(createFlowFile(11L, 11L, attributes));
    repo.registerEvent(builder.build());
    repo.waitForRollover();

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "XXXX"));
    query.setMaxResults(100);

    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(2, result.getMatchingEvents().size());
    for (final ProvenanceEventRecord match : result.getMatchingEvents()) {
        System.out.println(match);
    }
}
 

@Test
@Ignore("This test relies too much on timing of background events by using Thread.sleep().")
public void testIndexDirectoryRemoved() throws InterruptedException, IOException, ParseException {
    final RepositoryConfiguration config = createConfiguration();
    config.setMaxRecordLife(5, TimeUnit.MINUTES);
    config.setMaxStorageCapacity(1024L * 1024L);
    config.setMaxEventFileLife(500, TimeUnit.MILLISECONDS);
    config.setMaxEventFileCapacity(1024L * 1024L);
    config.setSearchableFields(new ArrayList<>(SearchableFields.getStandardFields()));
    config.setDesiredIndexSize(10); // force new index to be created for each rollover

    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    final String uuid = "00000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");

    for (int i = 0; i < 10; i++) {
        attributes.put("uuid", "00000000-0000-0000-0000-00000000000" + i);
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        builder.setEventTime(10L); // make sure the events are destroyed when we call purge
        repo.registerEvent(builder.build());
    }

    repo.waitForRollover();

    Thread.sleep(2000L);

    final FileFilter indexFileFilter = file -> file.getName().startsWith("index");
    final int numIndexDirs = config.getStorageDirectories().values().iterator().next().listFiles(indexFileFilter).length;
    assertEquals(1, numIndexDirs);

    // add more records so that we will create a new index
    final long secondBatchStartTime = System.currentTimeMillis();
    for (int i = 0; i < 10; i++) {
        attributes.put("uuid", "00000000-0000-0000-0000-00000000001" + i);
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        builder.setEventTime(System.currentTimeMillis());
        repo.registerEvent(builder.build());
    }

    // wait for indexing to happen
    repo.waitForRollover();

    // verify we get the results expected
    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.Filename, "file-*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "12?4"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.TransitURI, "nifi://*"));
    query.setMaxResults(100);

    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(20, result.getMatchingEvents().size());

    // Ensure index directories exists
    File[] indexDirs = config.getStorageDirectories().values().iterator().next().listFiles(indexFileFilter);
    assertEquals(2, indexDirs.length);

    // expire old events and indexes
    final long timeSinceSecondBatch = System.currentTimeMillis() - secondBatchStartTime;
    config.setMaxRecordLife(timeSinceSecondBatch + 1000L, TimeUnit.MILLISECONDS);
    repo.purgeOldEvents();
    Thread.sleep(2000L);

    final QueryResult newRecordSet = repo.queryEvents(query, createUser());
    assertEquals(10, newRecordSet.getMatchingEvents().size());

    // Ensure that one index directory is gone
    indexDirs = config.getStorageDirectories().values().iterator().next().listFiles(indexFileFilter);
    assertEquals(1, indexDirs.length);
}
 

@Test
@Ignore("This test relies too much on timing of background events by using Thread.sleep().")
public void testIndexDirectoryRemoved() throws InterruptedException, IOException, ParseException {
    final RepositoryConfiguration config = createConfiguration();
    config.setMaxRecordLife(5, TimeUnit.MINUTES);
    config.setMaxStorageCapacity(1024L * 1024L);
    config.setMaxEventFileLife(500, TimeUnit.MILLISECONDS);
    config.setMaxEventFileCapacity(1024L * 1024L);
    config.setSearchableFields(new ArrayList<>(SearchableFields.getStandardFields()));
    config.setDesiredIndexSize(10); // force new index to be created for each rollover

    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    final String uuid = "00000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");

    for (int i = 0; i < 10; i++) {
        attributes.put("uuid", "00000000-0000-0000-0000-00000000000" + i);
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        builder.setEventTime(10L); // make sure the events are destroyed when we call purge
        repo.registerEvent(builder.build());
    }

    repo.waitForRollover();

    Thread.sleep(2000L);

    final FileFilter indexFileFilter = file -> file.getName().startsWith("index");
    final int numIndexDirs = config.getStorageDirectories().values().iterator().next().listFiles(indexFileFilter).length;
    assertEquals(1, numIndexDirs);

    // add more records so that we will create a new index
    final long secondBatchStartTime = System.currentTimeMillis();
    for (int i = 0; i < 10; i++) {
        attributes.put("uuid", "00000000-0000-0000-0000-00000000001" + i);
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        builder.setEventTime(System.currentTimeMillis());
        repo.registerEvent(builder.build());
    }

    // wait for indexing to happen
    repo.waitForRollover();

    // verify we get the results expected
    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.Filename, "file-*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "12?4"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.TransitURI, "nifi://*"));
    query.setMaxResults(100);

    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(20, result.getMatchingEvents().size());

    // Ensure index directories exists
    File[] indexDirs = config.getStorageDirectories().values().iterator().next().listFiles(indexFileFilter);
    assertEquals(2, indexDirs.length);

    // expire old events and indexes
    final long timeSinceSecondBatch = System.currentTimeMillis() - secondBatchStartTime;
    config.setMaxRecordLife(timeSinceSecondBatch + 1000L, TimeUnit.MILLISECONDS);
    repo.purgeOldEvents();
    Thread.sleep(2000L);

    final QueryResult newRecordSet = repo.queryEvents(query, createUser());
    assertEquals(10, newRecordSet.getMatchingEvents().size());

    // Ensure that one index directory is gone
    indexDirs = config.getStorageDirectories().values().iterator().next().listFiles(indexFileFilter);
    assertEquals(1, indexDirs.length);
}
 

@Test
public void testIndexAndCompressOnRolloverAndSubsequentSearchMultipleStorageDirs() throws IOException, InterruptedException, ParseException {
    final RepositoryConfiguration config = createConfiguration();
    config.addStorageDirectory("2", new File("target/storage/" + UUID.randomUUID().toString()));
    config.setMaxRecordLife(30, TimeUnit.SECONDS);
    config.setMaxStorageCapacity(1024L * 1024L);
    config.setMaxEventFileLife(1, TimeUnit.SECONDS);
    config.setMaxEventFileCapacity(1024L * 1024L);
    config.setSearchableFields(new ArrayList<>(SearchableFields.getStandardFields()));

    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    final String uuid = "00000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    for (int j = 0; j < 3; j++) {
        attributes.put("iteration", String.valueOf(j));

        builder.setEventTime(System.currentTimeMillis());
        builder.setEventType(ProvenanceEventType.RECEIVE);
        builder.setTransitUri("nifi://unit-test");
        builder.setComponentId("1234");
        builder.setComponentType("dummy processor");
        builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));

        for (int i = 0; i < 10; i++) {
            String uuidSuffix = String.valueOf(i + j * 10);
            if (uuidSuffix.length() < 2) {
                uuidSuffix = "0" + uuidSuffix;
            }

            attributes.put("uuid", "00000000-0000-0000-0000-0000000000" + uuidSuffix);
            builder.fromFlowFile(createFlowFile(i + j * 10, 3000L, attributes));
            repo.registerEvent(builder.build());
        }

        repo.waitForRollover();
    }

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.Filename, "file-*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "12?4"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.TransitURI, "nifi://*"));
    query.setMaxResults(100);

    final QuerySubmission submission = repo.submitQuery(query, createUser());
    while (!submission.getResult().isFinished()) {
        Thread.sleep(100L);
    }

    assertEquals(30, submission.getResult().getMatchingEvents().size());
    final Map<String, Integer> counts = new HashMap<>();
    for (final ProvenanceEventRecord match : submission.getResult().getMatchingEvents()) {
        System.out.println(match);

        final String index = match.getAttributes().get("iteration");
        Integer count = counts.get(index);
        if (count == null) {
            count = 0;
        }
        counts.put(index, count + 1);
    }

    assertEquals(3, counts.size());
    assertEquals(10, counts.get("0").intValue());
    assertEquals(10, counts.get("1").intValue());
    assertEquals(10, counts.get("2").intValue());

    config.setMaxRecordLife(1, TimeUnit.MILLISECONDS);

    repo.purgeOldEvents();

    Thread.sleep(2000L); // purge is async. Give it time to do its job.

    query.setMaxResults(100);
    final QuerySubmission noResultSubmission = repo.submitQuery(query, createUser());
    while (!noResultSubmission.getResult().isFinished()) {
        Thread.sleep(10L);
    }

    assertEquals(0, noResultSubmission.getResult().getTotalHitCount());
}
 

@Test
public void testIndexAndCompressOnRolloverAndSubsequentSearch() throws IOException, InterruptedException, ParseException {
    final RepositoryConfiguration config = createConfiguration();
    config.setMaxRecordLife(30, TimeUnit.SECONDS);
    config.setMaxStorageCapacity(1024L * 1024L * 10);
    config.setMaxEventFileLife(500, TimeUnit.MILLISECONDS);
    config.setMaxEventFileCapacity(1024L * 1024L * 10);
    config.setSearchableFields(new ArrayList<>(SearchableFields.getStandardFields()));

    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    final String uuid = "10000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    attributes.put("uuid", uuid);
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");

    for (int i = 0; i < 10; i++) {
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        attributes.put("uuid", "00000000-0000-0000-0000-00000000000" + i);
        repo.registerEvent(builder.build());
    }

    repo.waitForRollover();

    final Query query = new Query(UUID.randomUUID().toString());
    // query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.FlowFileUUID, "00000000-0000-0000-0000*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.Filename, "file-*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "12?4"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.TransitURI, "nifi://*"));
    query.setMaxResults(100);

    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(10, result.getMatchingEvents().size());
    for (final ProvenanceEventRecord match : result.getMatchingEvents()) {
        System.out.println(match);
    }

    Thread.sleep(2000L);

    config.setMaxStorageCapacity(100L);
    config.setMaxRecordLife(500, TimeUnit.MILLISECONDS);
    repo.purgeOldEvents();
    Thread.sleep(2000L);

    final QueryResult newRecordSet = repo.queryEvents(query, createUser());
    assertTrue(newRecordSet.getMatchingEvents().isEmpty());
}
 

@Test
public void testIndexOnRolloverAndSubsequentSearch() throws IOException, InterruptedException, ParseException {
    final RepositoryConfiguration config = createConfiguration();
    config.setMaxEventFileLife(500, TimeUnit.MILLISECONDS);
    config.setSearchableFields(new ArrayList<>(SearchableFields.getStandardFields()));
    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    final String uuid = "00000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");

    for (int i = 0; i < 10; i++) {
        attributes.put("uuid", "00000000-0000-0000-0000-00000000000" + i);
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        repo.registerEvent(builder.build());
    }

    repo.waitForRollover();

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.FlowFileUUID, "000000*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.Filename, "file-*"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.ComponentID, "12?4"));
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.TransitURI, "nifi://*"));
    query.setMaxResults(100);

    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(10, result.getMatchingEvents().size());
    for (final ProvenanceEventRecord match : result.getMatchingEvents()) {
        System.out.println(match);
    }
}
 

@Test
public void testIndexOnRolloverWithImmenseAttribute() throws IOException {
    final RepositoryConfiguration config = createConfiguration();
    config.setMaxEventFileLife(500, TimeUnit.MILLISECONDS);
    config.setSearchableFields(new ArrayList<>(SearchableFields.getStandardFields()));
    config.setSearchableAttributes(SearchableFieldParser.extractSearchableFields("immense", false));
    repo = new PersistentProvenanceRepository(config, DEFAULT_ROLLOVER_MILLIS);
    repo.initialize(getEventReporter(), null, null, IdentifierLookup.EMPTY);

    int immenseAttrSize = 33000; // must be greater than 32766 for a meaningful test
    StringBuilder immenseBldr = new StringBuilder(immenseAttrSize);
    for (int i=0; i < immenseAttrSize; i++) {
        immenseBldr.append('0');
    }
    final String uuid = "00000000-0000-0000-0000-000000000000";
    final Map<String, String> attributes = new HashMap<>();
    attributes.put("abc", "xyz");
    attributes.put("xyz", "abc");
    attributes.put("filename", "file-" + uuid);
    attributes.put("immense", immenseBldr.toString());

    final ProvenanceEventBuilder builder = new StandardProvenanceEventRecord.Builder();
    builder.setEventTime(System.currentTimeMillis());
    builder.setEventType(ProvenanceEventType.RECEIVE);
    builder.setTransitUri("nifi://unit-test");
    builder.fromFlowFile(createFlowFile(3L, 3000L, attributes));
    builder.setComponentId("1234");
    builder.setComponentType("dummy processor");

    for (int i = 0; i < 10; i++) {
        attributes.put("uuid", "00000000-0000-0000-0000-00000000000" + i);
        builder.fromFlowFile(createFlowFile(i, 3000L, attributes));
        repo.registerEvent(builder.build());
    }

    repo.waitForRollover();

    final Query query = new Query(UUID.randomUUID().toString());
    query.addSearchTerm(SearchTerms.newSearchTerm(SearchableFields.newSearchableAttribute("immense"), "000*"));
    query.setMaxResults(100);

    final QueryResult result = repo.queryEvents(query, createUser());
    assertEquals(10, result.getMatchingEvents().size());
}