Examples with ShardNode com.bakdata.conquery.commands.ShardNode used on opensource projects

Example 1 with ShardNode

use of com.bakdata.conquery.commands.ShardNode in project conquery by bakdata.

the class ImportUpdateTest method execute.

@Override
public void execute(String name, TestConquery testConquery) throws Exception {
    final StandaloneSupport conquery = testConquery.getSupport(name);
    MetaStorage storage = conquery.getMetaStorage();
    String testJson = In.resource("/tests/query/UPDATE_IMPORT_TESTS/SIMPLE_TREECONCEPT_Query.json").withUTF8().readAll();
    final Dataset dataset = conquery.getDataset();
    final Namespace namespace = conquery.getNamespace();
    final ImportId importId1 = ImportId.Parser.INSTANCE.parse(dataset.getName(), "table1", "table1");
    final ImportId importId2 = ImportId.Parser.INSTANCE.parse(dataset.getName(), "table2", "table2");
    QueryTest test = (QueryTest) JsonIntegrationTest.readJson(dataset, testJson);
    final List<RequiredTable> tables = test.getContent().getTables();
    assertThat(tables.size()).isEqualTo(2);
    List<File> cqpps;
    // Manually import data, so we can do our own work.
    {
        ValidatorHelper.failOnError(log, conquery.getValidator().validate(test));
        importSecondaryIds(conquery, test.getContent().getSecondaryIds());
        conquery.waitUntilWorkDone();
        LoadingUtil.importTables(conquery, tables);
        conquery.waitUntilWorkDone();
        LoadingUtil.importConcepts(conquery, test.getRawConcepts());
        conquery.waitUntilWorkDone();
        cqpps = LoadingUtil.generateCqpp(conquery, tables);
        conquery.waitUntilWorkDone();
        assertThat(cqpps.size()).isEqualTo(tables.size());
        LoadingUtil.importCqppFiles(conquery, List.of(cqpps.get(0)));
        conquery.waitUntilWorkDone();
    }
    final Query query = IntegrationUtils.parseQuery(conquery, test.getRawQuery());
    // State before update.
    {
        log.info("Checking state before update");
        assertThat(namespace.getStorage().getAllImports()).hasSize(1);
        // Must contain the import.
        assertThat(namespace.getStorage().getAllImports()).filteredOn(imp -> imp.getId().equals(importId1)).isNotEmpty();
        assertThat(namespace.getStorage().getCentralRegistry().getOptional(importId1)).isNotEmpty();
        for (ShardNode node : conquery.getShardNodes()) {
            for (Worker worker : node.getWorkers().getWorkers().values()) {
                if (!worker.getInfo().getDataset().equals(dataset.getId())) {
                    continue;
                }
                final ModificationShieldedWorkerStorage workerStorage = worker.getStorage();
                assertThat(workerStorage.getAllCBlocks()).describedAs("CBlocks for Worker %s", worker.getInfo().getId()).filteredOn(block -> block.getBucket().getId().getDataset().equals(dataset.getId())).isNotEmpty();
                assertThat(workerStorage.getAllBuckets()).filteredOn(bucket -> bucket.getId().getDataset().equals(dataset.getId())).describedAs("Buckets for Worker %s", worker.getInfo().getId()).isNotEmpty();
                // Must contain the import.
                assertThat(workerStorage.getImport(importId1)).isNotNull();
            }
        }
        assertThat(namespace.getNumberOfEntities()).isEqualTo(4);
        // assert that the query can be executed after the import
        IntegrationUtils.assertQueryResult(conquery, query, 2L, ExecutionState.DONE, conquery.getTestUser(), 201);
    }
    // Try to update an import that does not exist should throw a Not-Found Webapplication Exception
    LoadingUtil.updateCqppFile(conquery, cqpps.get(1), Response.Status.Family.CLIENT_ERROR, "Not Found");
    conquery.waitUntilWorkDone();
    // Load manually new data for import and update the concerned import
    {
        log.info("Manually loading new data for import");
        final RequiredTable importTable = test.getContent().getTables().stream().filter(table -> table.getName().equalsIgnoreCase(importId1.getTable().getTable())).findFirst().orElseThrow();
        final String csvName = importTable.getCsv().getName();
        final String path = importTable.getCsv().getPath();
        // copy new content of the importTable into the csv-File used by the preprocessor to avoid creating multiple files withe same names
        FileUtils.copyInputStreamToFile(In.resource(path.substring(0, path.lastIndexOf('/')) + "/" + csvName.replace(".csv", ".update.csv")).asStream(), new File(conquery.getTmpDir(), csvName));
        File descriptionFile = new File(conquery.getTmpDir(), importTable.getName() + ConqueryConstants.EXTENSION_DESCRIPTION);
        File newPreprocessedFile = new File(conquery.getTmpDir(), importTable.getName() + ConqueryConstants.EXTENSION_PREPROCESSED);
        // create import descriptor
        {
            TableImportDescriptor desc = new TableImportDescriptor();
            desc.setName(importTable.getName());
            desc.setTable(importTable.getName());
            TableInputDescriptor input = new TableInputDescriptor();
            {
                input.setPrimary(importTable.getPrimaryColumn().createOutput());
                input.setSourceFile(csvName);
                input.setOutput(new OutputDescription[importTable.getColumns().length]);
                for (int i = 0; i < importTable.getColumns().length; i++) {
                    input.getOutput()[i] = importTable.getColumns()[i].createOutput();
                }
            }
            desc.setInputs(new TableInputDescriptor[] { input });
            Jackson.MAPPER.writeValue(descriptionFile, desc);
        }
        // preprocess
        conquery.preprocessTmp(conquery.getTmpDir(), List.of(descriptionFile));
        log.info("updating import");
        // correct update of the import
        LoadingUtil.updateCqppFile(conquery, newPreprocessedFile, Response.Status.Family.SUCCESSFUL, "No Content");
        conquery.waitUntilWorkDone();
    }
    // State after update.
    {
        log.info("Checking state after update");
        assertThat(namespace.getStorage().getAllImports()).hasSize(1);
        // Must contain the import.
        assertThat(namespace.getStorage().getAllImports()).filteredOn(imp -> imp.getId().equals(importId1)).isNotEmpty();
        assertThat(namespace.getStorage().getCentralRegistry().getOptional(importId1)).isNotEmpty();
        for (ShardNode node : conquery.getShardNodes()) {
            for (Worker worker : node.getWorkers().getWorkers().values()) {
                if (!worker.getInfo().getDataset().equals(dataset.getId())) {
                    continue;
                }
                final ModificationShieldedWorkerStorage workerStorage = worker.getStorage();
                assertThat(workerStorage.getAllCBlocks()).describedAs("CBlocks for Worker %s", worker.getInfo().getId()).filteredOn(block -> block.getBucket().getId().getDataset().equals(dataset.getId())).isNotEmpty();
                assertThat(workerStorage.getAllBuckets()).filteredOn(bucket -> bucket.getId().getDataset().equals(dataset.getId())).describedAs("Buckets for Worker %s", worker.getInfo().getId()).isNotEmpty();
                // Must contain the import.
                assertThat(workerStorage.getImport(importId1)).isNotNull();
            }
        }
        assertThat(namespace.getNumberOfEntities()).isEqualTo(9);
        // Issue a query and assert that it has more content.
        IntegrationUtils.assertQueryResult(conquery, query, 4L, ExecutionState.DONE, conquery.getTestUser(), 201);
    }
}

Example 2 with ShardNode

use of com.bakdata.conquery.commands.ShardNode in project conquery by bakdata.

the class DatasetDeletionTest method execute.

@Override
public void execute(String name, TestConquery testConquery) throws Exception {
    final StandaloneSupport conquery = testConquery.getSupport(name);
    final MetaStorage storage = conquery.getMetaStorage();
    final Dataset dataset = conquery.getDataset();
    Namespace namespace = conquery.getNamespace();
    final String testJson = In.resource("/tests/query/DELETE_IMPORT_TESTS/SIMPLE_TREECONCEPT_Query.test.json").withUTF8().readAll();
    final QueryTest test = (QueryTest) JsonIntegrationTest.readJson(dataset, testJson);
    // Manually import data, so we can do our own work.
    {
        ValidatorHelper.failOnError(log, conquery.getValidator().validate(test));
        importSecondaryIds(conquery, test.getContent().getSecondaryIds());
        conquery.waitUntilWorkDone();
        LoadingUtil.importTables(conquery, test.getContent().getTables());
        conquery.waitUntilWorkDone();
        LoadingUtil.importConcepts(conquery, test.getRawConcepts());
        conquery.waitUntilWorkDone();
        LoadingUtil.importTableContents(conquery, test.getContent().getTables());
        conquery.waitUntilWorkDone();
    }
    final Query query = IntegrationUtils.parseQuery(conquery, test.getRawQuery());
    final int nImports = namespace.getStorage().getAllImports().size();
    log.info("Checking state before deletion");
    // Assert state before deletion.
    {
        // Must contain the import.
        assertThat(namespace.getStorage().getCentralRegistry().getOptional(dataset.getId())).isNotEmpty();
        for (ShardNode node : conquery.getShardNodes()) {
            for (Worker value : node.getWorkers().getWorkers().values()) {
                if (!value.getInfo().getDataset().equals(dataset.getId())) {
                    continue;
                }
                final ModificationShieldedWorkerStorage workerStorage = value.getStorage();
                assertThat(workerStorage.getAllCBlocks()).describedAs("CBlocks for Worker %s", value.getInfo().getId()).isNotEmpty();
                assertThat(workerStorage.getAllBuckets()).describedAs("Buckets for Worker %s", value.getInfo().getId()).isNotEmpty();
            }
        }
        log.info("Executing query before deletion");
        IntegrationUtils.assertQueryResult(conquery, query, 2L, ExecutionState.DONE, conquery.getTestUser(), 201);
    }
    // Delete Dataset.
    {
        log.info("Issuing deletion of import {}", dataset);
        // Delete the import.
        // But, we do not allow deletion of tables with associated connectors, so this should throw!
        assertThatThrownBy(() -> conquery.getDatasetsProcessor().deleteDataset(dataset)).isInstanceOf(WebApplicationException.class);
        // TODO use api
        conquery.getNamespace().getStorage().getTables().forEach(tableId -> conquery.getDatasetsProcessor().deleteTable(tableId, true));
        conquery.waitUntilWorkDone();
        // Finally delete dataset
        conquery.getDatasetsProcessor().deleteDataset(dataset);
        conquery.waitUntilWorkDone();
        assertThat(storage.getCentralRegistry().getOptional(dataset.getId())).isEmpty();
    }
    // State after deletion.
    {
        log.info("Checking state after deletion");
        // We have deleted an import now there should be two less!
        assertThat(namespace.getStorage().getAllImports().size()).isEqualTo(0);
        // The deleted import should not be found.
        assertThat(namespace.getStorage().getAllImports()).filteredOn(imp -> imp.getId().getTable().getDataset().equals(dataset.getId())).isEmpty();
        for (ShardNode node : conquery.getShardNodes()) {
            for (Worker value : node.getWorkers().getWorkers().values()) {
                if (!value.getInfo().getDataset().equals(dataset.getId())) {
                    continue;
                }
                final ModificationShieldedWorkerStorage workerStorage = value.getStorage();
                // No bucket should be found referencing the import.
                assertThat(workerStorage.getAllBuckets()).describedAs("Buckets for Worker %s", value.getInfo().getId()).filteredOn(bucket -> bucket.getTable().getDataset().getId().equals(dataset.getId())).isEmpty();
                // No CBlock associated with import may exist
                assertThat(workerStorage.getAllCBlocks()).describedAs("CBlocks for Worker %s", value.getInfo().getId()).filteredOn(cBlock -> cBlock.getBucket().getTable().getDataset().getId().equals(dataset.getId())).isEmpty();
            }
        }
        // It's not exactly possible to issue a query for a non-existant dataset, so we assert that parsing the fails.
        assertThatThrownBy(() -> {
            IntegrationUtils.parseQuery(conquery, test.getRawQuery());
        }).isNotNull();
        IntegrationUtils.assertQueryResult(conquery, query, 0, ExecutionState.FAILED, conquery.getTestUser(), 404);
    }
    // Reload the dataset and assert the state.
    // We have to do some weird trix with StandaloneSupport to open it with another Dataset
    final StandaloneSupport conqueryReimport = testConquery.getSupport(namespace.getDataset().getName());
    {
        // only import the deleted import/table
        LoadingUtil.importTables(conqueryReimport, test.getContent().getTables());
        assertThat(conqueryReimport.getNamespace().getStorage().getTables()).isNotEmpty();
        conqueryReimport.waitUntilWorkDone();
        LoadingUtil.importTableContents(conqueryReimport, test.getContent().getTables());
        conqueryReimport.waitUntilWorkDone();
        LoadingUtil.importConcepts(conqueryReimport, test.getRawConcepts());
        conqueryReimport.waitUntilWorkDone();
        assertThat(conqueryReimport.getDatasetsProcessor().getDatasetRegistry().get(conqueryReimport.getDataset().getId())).describedAs("Dataset after re-import.").isNotNull();
        assertThat(conqueryReimport.getNamespace().getStorage().getAllImports().size()).isEqualTo(nImports);
        for (ShardNode node : conqueryReimport.getShardNodes()) {
            assertThat(node.getWorkers().getWorkers().values()).filteredOn(w -> w.getInfo().getDataset().equals(conqueryReimport.getDataset().getId())).describedAs("Workers for node {}", node.getName()).isNotEmpty();
        }
        log.info("Executing query after re-import");
        final Query query2 = IntegrationUtils.parseQuery(conqueryReimport, test.getRawQuery());
        // Issue a query and assert that it has the same content as the first time around.
        IntegrationUtils.assertQueryResult(conqueryReimport, query2, 2L, ExecutionState.DONE, conqueryReimport.getTestUser(), 201);
    }
    // Finally, restart conquery and assert again, that the data is correct.
    {
        testConquery.shutdown();
        // restart
        testConquery.beforeAll();
        final StandaloneSupport conqueryRestart = testConquery.openDataset(conqueryReimport.getDataset().getId());
        log.info("Checking state after re-start");
        assertThat(conqueryRestart.getNamespace().getStorage().getAllImports().size()).isEqualTo(2);
        for (ShardNode node : conqueryRestart.getShardNodes()) {
            for (Worker value : node.getWorkers().getWorkers().values()) {
                if (!value.getInfo().getDataset().equals(dataset.getId())) {
                    continue;
                }
                final ModificationShieldedWorkerStorage workerStorage = value.getStorage();
                assertThat(workerStorage.getAllBuckets().stream().filter(bucket -> bucket.getTable().getDataset().getId().equals(dataset.getId()))).describedAs("Buckets for Worker %s", value.getInfo().getId()).isNotEmpty();
            }
        }
        log.info("Executing query after restart");
        final Query query3 = IntegrationUtils.parseQuery(conqueryRestart, test.getRawQuery());
        // Issue a query and assert that it has the same content as the first time around.
        IntegrationUtils.assertQueryResult(conqueryRestart, query3, 2L, ExecutionState.DONE, conquery.getTestUser(), 201);
    }
}

Example 3 with ShardNode

use of com.bakdata.conquery.commands.ShardNode in project conquery by bakdata.

the class TestConquery method isBusy.

private boolean isBusy() {
    boolean busy;
    busy = standaloneCommand.getManager().getJobManager().isSlowWorkerBusy();
    busy |= standaloneCommand.getManager().getStorage().getAllExecutions().stream().map(ManagedExecution::getState).anyMatch(ExecutionState.RUNNING::equals);
    for (Namespace namespace : standaloneCommand.getManager().getDatasetRegistry().getDatasets()) {
        busy |= namespace.getJobManager().isSlowWorkerBusy();
    }
    for (ShardNode slave : standaloneCommand.getShardNodes()) {
        busy |= slave.isBusy();
    }
    return busy;
}

Example 4 with ShardNode

use of com.bakdata.conquery.commands.ShardNode in project conquery by bakdata.

the class ConceptUpdateAndDeletionTest method execute.

@Override
public void execute(String name, TestConquery testConquery) throws Exception {
    StandaloneSupport conquery = testConquery.getSupport(name);
    // Read two JSONs with different Trees
    final String testJson = In.resource("/tests/query/UPDATE_CONCEPT_TESTS/SIMPLE_TREECONCEPT_Query.json").withUTF8().readAll();
    final String testJson2 = In.resource("/tests/query/UPDATE_CONCEPT_TESTS/SIMPLE_TREECONCEPT_2_Query.json").withUTF8().readAll();
    final Dataset dataset = conquery.getDataset();
    final Namespace namespace = conquery.getNamespace();
    final ConceptId conceptId = ConceptId.Parser.INSTANCE.parse(dataset.getName(), "test_tree");
    final Concept<?> concept;
    final QueryTest test = (QueryTest) JsonIntegrationTest.readJson(dataset, testJson);
    final QueryTest test2 = (QueryTest) JsonIntegrationTest.readJson(dataset, testJson2);
    // Manually import data, so we can do our own work.
    {
        ValidatorHelper.failOnError(log, conquery.getValidator().validate(test));
        importSecondaryIds(conquery, test.getContent().getSecondaryIds());
        conquery.waitUntilWorkDone();
        LoadingUtil.importTables(conquery, test.getContent().getTables());
        conquery.waitUntilWorkDone();
        LoadingUtil.importConcepts(conquery, test.getRawConcepts());
        conquery.waitUntilWorkDone();
        assertThat(namespace.getStorage().getConcept(conceptId)).isNotNull();
        LoadingUtil.importTableContents(conquery, test.getContent().getTables());
        conquery.waitUntilWorkDone();
    }
    final Query query = IntegrationUtils.parseQuery(conquery, test.getRawQuery());
    // State before update.
    {
        log.info("Checking state before update");
        // Must contain the concept.
        assertThat(namespace.getStorage().getAllConcepts()).filteredOn(con -> con.getId().equals(conceptId)).isNotEmpty();
        assertThat(namespace.getStorage().getCentralRegistry().getOptional(conceptId)).isNotEmpty();
        for (ShardNode node : conquery.getShardNodes()) {
            for (Worker value : node.getWorkers().getWorkers().values()) {
                if (!value.getInfo().getDataset().equals(dataset.getId())) {
                    continue;
                }
                final ModificationShieldedWorkerStorage workerStorage = value.getStorage();
                assertThat(workerStorage.getCentralRegistry().getOptional(conceptId)).isNotEmpty();
                assertThat(workerStorage.getAllCBlocks()).describedAs("CBlocks for Worker %s", value.getInfo().getId()).filteredOn(cBlock -> cBlock.getConnector().getConcept().getId().equals(conceptId)).isNotEmpty();
            }
        }
        log.info("Executing query before update");
        IntegrationUtils.assertQueryResult(conquery, query, 1L, ExecutionState.DONE, conquery.getTestUser(), 201);
        conquery.waitUntilWorkDone();
        log.info("Query before update executed");
    }
    // Load a different concept with the same id (it has different children "C1" that are more than "A1")
    // To perform the update, the old concept will be deleted first and the new concept will be added. That means the deletion of concept is also covered here
    {
        log.info("Executing  update");
        LoadingUtil.updateConcepts(conquery, test2.getRawConcepts(), Response.Status.Family.SUCCESSFUL);
        conquery.waitUntilWorkDone();
        log.info("Update executed");
    }
    // Check state after update.
    {
        log.info("Checking state after update");
        // Must contain the concept now.
        assertThat(namespace.getStorage().getAllConcepts()).filteredOn(con -> con.getId().equals(conceptId)).isNotEmpty();
        assertThat(namespace.getStorage().getCentralRegistry().getOptional(conceptId)).isNotEmpty();
        for (ShardNode node : conquery.getShardNodes()) {
            for (Worker value : node.getWorkers().getWorkers().values()) {
                if (!value.getInfo().getDataset().equals(dataset.getId())) {
                    continue;
                }
                final ModificationShieldedWorkerStorage workerStorage = value.getStorage();
                assertThat(workerStorage.getCentralRegistry().getOptional(conceptId)).isNotEmpty();
                assertThat(workerStorage.getAllCBlocks()).describedAs("CBlocks for Worker %s", value.getInfo().getId()).filteredOn(cBlock -> cBlock.getConnector().getConcept().getId().equals(conceptId)).isNotEmpty();
            }
        }
        log.info("Executing query after update");
        // Assert that it now contains 2 instead of 1.
        IntegrationUtils.assertQueryResult(conquery, query, 2L, ExecutionState.DONE, conquery.getTestUser(), 201);
        conquery.waitUntilWorkDone();
        log.info("Query after update executed");
    }
    // new Conquery generated after restarting
    // StandaloneSupport conquery;
    // Restart conquery and assert again, that the data is correct.
    {
        testConquery.shutdown();
        // restart
        testConquery.beforeAll();
        conquery = testConquery.openDataset(dataset.getId());
        log.info("Checking state after re-start");
        {
            // Must contain the concept.
            assertThat(conquery.getNamespace().getStorage().getAllConcepts()).filteredOn(con -> con.getId().equals(conceptId)).isNotEmpty();
            assertThat(conquery.getNamespace().getStorage().getCentralRegistry().getOptional(conceptId)).isNotEmpty();
            for (ShardNode node : conquery.getShardNodes()) {
                for (Worker value : node.getWorkers().getWorkers().values()) {
                    if (!value.getInfo().getDataset().equals(dataset.getId())) {
                        continue;
                    }
                    final ModificationShieldedWorkerStorage workerStorage = value.getStorage();
                    assertThat(workerStorage.getCentralRegistry().getOptional(conceptId)).isNotEmpty();
                    assertThat(workerStorage.getAllCBlocks()).describedAs("CBlocks for Worker %s", value.getInfo().getId()).filteredOn(cBlock -> cBlock.getConnector().getConcept().getId().equals(conceptId)).isNotEmpty();
                }
            }
            log.info("Executing query after restart.");
            // Re-assert state.
            IntegrationUtils.assertQueryResult(conquery, query, 2L, ExecutionState.DONE, conquery.getTestUser(), 201);
            conquery.waitUntilWorkDone();
        }
    }
    // Delete the Concept.
    {
        log.info("Issuing deletion of import {}", conceptId);
        concept = Objects.requireNonNull(conquery.getNamespace().getStorage().getConcept(conceptId));
        conquery.getDatasetsProcessor().deleteConcept(concept);
        conquery.waitUntilWorkDone();
    }
    // Check state after deletion.
    {
        log.info("Checking state after deletion");
        // We've deleted the concept so it and it's associated cblock should be gone.
        assertThat(conquery.getNamespace().getStorage().getAllConcepts()).filteredOn(con -> con.getId().equals(conceptId)).isEmpty();
        assertThat(conquery.getNamespace().getStorage().getCentralRegistry().getOptional(conceptId)).isEmpty();
        assertThat(conquery.getShardNodes().stream().flatMap(node -> node.getWorkers().getWorkers().values().stream()).filter(worker -> worker.getInfo().getDataset().equals(dataset.getId())).map(Worker::getStorage)).noneMatch(workerStorage -> workerStorage.getConcept(conceptId) != null).noneMatch(workerStorage -> workerStorage.getAllCBlocks().stream().anyMatch(cBlock -> cBlock.getConnector().getConcept().getId().equals(conceptId)));
        log.info("Executing query after deletion (EXPECTING AN EXCEPTION IN THE LOGS!)");
        // Issue a query and assert that it is failing.
        IntegrationUtils.assertQueryResult(conquery, query, 0L, ExecutionState.FAILED, conquery.getTestUser(), 400);
    }
    // Restart conquery and assert again, that the state after deletion was maintained.
    {
        {
            testConquery.shutdown();
            // restart
            testConquery.beforeAll();
            conquery = testConquery.openDataset(dataset.getId());
        }
        // Check state after restart.
        {
            log.info("Checking state after restart");
            // We've deleted the concept so it and it's associated cblock should be gone.
            assertThat(conquery.getNamespace().getStorage().getAllConcepts()).filteredOn(con -> con.getId().equals(conceptId)).isEmpty();
            assertThat(conquery.getNamespace().getStorage().getCentralRegistry().getOptional(conceptId)).isEmpty();
            assertThat(conquery.getShardNodes().stream().flatMap(node -> node.getWorkers().getWorkers().values().stream()).filter(worker -> worker.getInfo().getDataset().equals(dataset.getId())).map(Worker::getStorage)).noneMatch(workerStorage -> workerStorage.getConcept(conceptId) != null).noneMatch(workerStorage -> workerStorage.getAllCBlocks().stream().anyMatch(cBlock -> cBlock.getConnector().getConcept().getId().equals(conceptId)));
            log.info("Executing query after restart (EXPECTING AN EXCEPTION IN THE LOGS!)");
            // Issue a query and assert that it is failing.
            IntegrationUtils.assertQueryResult(conquery, query, 0L, ExecutionState.FAILED, conquery.getTestUser(), 400);
        }
    }
}

Example 5 with ShardNode

use of com.bakdata.conquery.commands.ShardNode in project conquery by bakdata.

the class ImportDeletionTest method execute.

@Override
public void execute(String name, TestConquery testConquery) throws Exception {
    final StandaloneSupport conquery = testConquery.getSupport(name);
    MetaStorage storage = conquery.getMetaStorage();
    final String testJson = In.resource("/tests/query/DELETE_IMPORT_TESTS/SIMPLE_TREECONCEPT_Query.test.json").withUTF8().readAll();
    final Dataset dataset = conquery.getDataset();
    final Namespace namespace = conquery.getNamespace();
    final ImportId importId = ImportId.Parser.INSTANCE.parse(dataset.getName(), "test_table2", "test_table2");
    final QueryTest test = (QueryTest) JsonIntegrationTest.readJson(dataset, testJson);
    // Manually import data, so we can do our own work.
    {
        ValidatorHelper.failOnError(log, conquery.getValidator().validate(test));
        importSecondaryIds(conquery, test.getContent().getSecondaryIds());
        conquery.waitUntilWorkDone();
        LoadingUtil.importTables(conquery, test.getContent().getTables());
        conquery.waitUntilWorkDone();
        LoadingUtil.importConcepts(conquery, test.getRawConcepts());
        conquery.waitUntilWorkDone();
        LoadingUtil.importTableContents(conquery, test.getContent().getTables());
        conquery.waitUntilWorkDone();
    }
    final Query query = IntegrationUtils.parseQuery(conquery, test.getRawQuery());
    final int nImports = namespace.getStorage().getAllImports().size();
    // State before deletion.
    {
        log.info("Checking state before deletion");
        // Must contain the import.
        assertThat(namespace.getStorage().getAllImports()).filteredOn(imp -> imp.getId().equals(importId)).isNotEmpty();
        assertThat(namespace.getStorage().getCentralRegistry().getOptional(importId)).isNotEmpty();
        for (ShardNode node : conquery.getShardNodes()) {
            for (Worker worker : node.getWorkers().getWorkers().values()) {
                if (!worker.getInfo().getDataset().equals(dataset.getId())) {
                    continue;
                }
                final ModificationShieldedWorkerStorage workerStorage = worker.getStorage();
                assertThat(workerStorage.getAllCBlocks()).describedAs("CBlocks for Worker %s", worker.getInfo().getId()).filteredOn(block -> block.getBucket().getId().getDataset().equals(dataset.getId())).isNotEmpty();
                assertThat(workerStorage.getAllBuckets()).filteredOn(bucket -> bucket.getId().getDataset().equals(dataset.getId())).describedAs("Buckets for Worker %s", worker.getInfo().getId()).isNotEmpty();
                // Must contain the import.
                assertThat(workerStorage.getImport(importId)).isNotNull();
            }
        }
        log.info("Executing query before deletion");
        IntegrationUtils.assertQueryResult(conquery, query, 2L, ExecutionState.DONE, conquery.getTestUser(), 201);
    }
    // Delete the import.
    {
        log.info("Issuing deletion of import {}", importId);
        final URI deleteImportUri = HierarchyHelper.hierarchicalPath(conquery.defaultAdminURIBuilder(), AdminTablesResource.class, "deleteImport").buildFromMap(Map.of(ResourceConstants.DATASET, conquery.getDataset().getId(), ResourceConstants.TABLE, importId.getTable(), ResourceConstants.IMPORT_ID, importId));
        final Response delete = conquery.getClient().target(deleteImportUri).request(MediaType.APPLICATION_JSON).delete();
        assertThat(delete.getStatusInfo().getFamily()).isEqualTo(Response.Status.Family.SUCCESSFUL);
        conquery.waitUntilWorkDone();
    }
    // State after deletion.
    {
        log.info("Checking state after deletion");
        // We have deleted an import now there should be one less!
        assertThat(namespace.getStorage().getAllImports().size()).isEqualTo(nImports - 1);
        // The deleted import should not be found.
        assertThat(namespace.getStorage().getAllImports()).filteredOn(imp -> imp.getId().equals(importId)).isEmpty();
        for (ShardNode node : conquery.getShardNodes()) {
            for (Worker worker : node.getWorkers().getWorkers().values()) {
                if (!worker.getInfo().getDataset().equals(dataset.getId())) {
                    continue;
                }
                final ModificationShieldedWorkerStorage workerStorage = worker.getStorage();
                // No bucket should be found referencing the import.
                assertThat(workerStorage.getAllBuckets()).describedAs("Buckets for Worker %s", worker.getInfo().getId()).filteredOn(bucket -> bucket.getImp().getId().equals(importId)).isEmpty();
                // No CBlock associated with import may exist
                assertThat(workerStorage.getAllCBlocks()).describedAs("CBlocks for Worker %s", worker.getInfo().getId()).filteredOn(cBlock -> cBlock.getBucket().getId().getImp().equals(importId)).isEmpty();
                // Import should not exists anymore
                assertThat(workerStorage.getImport(importId)).describedAs("Import for Worker %s", worker.getInfo().getId()).isNull();
            }
        }
        log.info("Executing query after deletion");
        // Issue a query and assert that it has less content.
        IntegrationUtils.assertQueryResult(conquery, query, 1L, ExecutionState.DONE, conquery.getTestUser(), 201);
    }
    conquery.waitUntilWorkDone();
    // Load more data under the same name into the same table, with only the deleted import/table
    {
        // only import the deleted import/table
        final RequiredTable import2Table = test.getContent().getTables().stream().filter(table -> table.getName().equalsIgnoreCase(importId.getTable().getTable())).findFirst().orElseThrow();
        final ResourceFile csv = import2Table.getCsv();
        final String path = csv.getPath();
        // copy csv to tmp folder
        // Content 2.2 contains an extra entry of a value that hasn't been seen before.
        FileUtils.copyInputStreamToFile(In.resource(path.substring(0, path.lastIndexOf('/')) + "/" + "content2.2.csv").asStream(), new File(conquery.getTmpDir(), csv.getName()));
        File descriptionFile = new File(conquery.getTmpDir(), import2Table.getName() + ConqueryConstants.EXTENSION_DESCRIPTION);
        File preprocessedFile = new File(conquery.getTmpDir(), import2Table.getName() + ConqueryConstants.EXTENSION_PREPROCESSED);
        // create import descriptor
        TableImportDescriptor desc = new TableImportDescriptor();
        desc.setName(import2Table.getName());
        desc.setTable(import2Table.getName());
        TableInputDescriptor input = new TableInputDescriptor();
        {
            input.setPrimary(import2Table.getPrimaryColumn().createOutput());
            input.setSourceFile(import2Table.getCsv().getName());
            input.setOutput(new OutputDescription[import2Table.getColumns().length]);
            for (int i = 0; i < import2Table.getColumns().length; i++) {
                input.getOutput()[i] = import2Table.getColumns()[i].createOutput();
            }
        }
        desc.setInputs(new TableInputDescriptor[] { input });
        Jackson.MAPPER.writeValue(descriptionFile, desc);
        // preprocess
        conquery.preprocessTmp(conquery.getTmpDir(), List.of(descriptionFile));
        // import preprocessedFiles
        conquery.getDatasetsProcessor().addImport(conquery.getNamespace(), new GZIPInputStream(new FileInputStream(preprocessedFile)));
        conquery.waitUntilWorkDone();
    }
    // State after reimport.
    {
        log.info("Checking state after re-import");
        assertThat(namespace.getStorage().getAllImports().size()).isEqualTo(nImports);
        for (ShardNode node : conquery.getShardNodes()) {
            for (Worker worker : node.getWorkers().getWorkers().values()) {
                if (!worker.getInfo().getDataset().equals(dataset.getId())) {
                    continue;
                }
                final ModificationShieldedWorkerStorage workerStorage = worker.getStorage();
                assertThat(workerStorage.getAllBuckets()).describedAs("Buckets for Worker %s", worker.getInfo().getId()).filteredOn(bucket -> bucket.getImp().getId().equals(importId)).filteredOn(bucket -> bucket.getId().getDataset().equals(dataset.getId())).isNotEmpty();
            }
        }
        log.info("Executing query after re-import");
        // Issue a query and assert that it has the same content as the first time around.
        IntegrationUtils.assertQueryResult(conquery, query, 2L, ExecutionState.DONE, conquery.getTestUser(), 201);
    }
    // Finally, restart conquery and assert again, that the data is correct.
    {
        testConquery.shutdown();
        // restart
        testConquery.beforeAll();
        StandaloneSupport conquery2 = testConquery.openDataset(dataset.getId());
        log.info("Checking state after re-start");
        {
            assertThat(namespace.getStorage().getAllImports().size()).isEqualTo(2);
            for (ShardNode node : conquery2.getShardNodes()) {
                for (Worker worker : node.getWorkers().getWorkers().values()) {
                    if (!worker.getInfo().getDataset().equals(dataset.getId()))
                        continue;
                    final ModificationShieldedWorkerStorage workerStorage = worker.getStorage();
                    assertThat(workerStorage.getAllBuckets()).describedAs("Buckets for Worker %s", worker.getInfo().getId()).filteredOn(bucket -> bucket.getId().getDataset().equals(dataset.getId())).filteredOn(bucket -> bucket.getImp().getId().equals(importId)).isNotEmpty();
                }
            }
            log.info("Executing query after re-import");
            // Issue a query and assert that it has the same content as the first time around.
            IntegrationUtils.assertQueryResult(conquery2, query, 2L, ExecutionState.DONE, conquery.getTestUser(), 201);
        }
    }
}