Search in sources :

Example 41 with TableReference

use of com.google.api.services.bigquery.model.TableReference in project beam by apache.

the class BigQueryInsertErrorCoderTest method testDecodeEncodeEqual.

@Test
public void testDecodeEncodeEqual() throws Exception {
    BigQueryInsertError value = new BigQueryInsertError(new TableRow().setF(Collections.singletonList(new TableCell().setV("Value"))), new TableDataInsertAllResponse.InsertErrors().setIndex(0L).setErrors(Collections.singletonList(new ErrorProto().setReason("a Reason").setLocation("A location").setMessage("A message").setDebugInfo("The debug info"))), new TableReference().setProjectId("dummy-project-id").setDatasetId("dummy-dataset-id").setTableId("dummy-table-id"));
    CoderProperties.coderDecodeEncodeEqual(TEST_CODER, value);
}
Also used : TableReference(com.google.api.services.bigquery.model.TableReference) TableCell(com.google.api.services.bigquery.model.TableCell) ErrorProto(com.google.api.services.bigquery.model.ErrorProto) TableRow(com.google.api.services.bigquery.model.TableRow) TableDataInsertAllResponse(com.google.api.services.bigquery.model.TableDataInsertAllResponse) Test(org.junit.Test)

Example 42 with TableReference

use of com.google.api.services.bigquery.model.TableReference in project beam by apache.

the class BigQueryIOReadTest method testBuildSourceWithTableReference.

@Test
public void testBuildSourceWithTableReference() {
    TableReference table = new TableReference().setProjectId("foo.com:project").setDatasetId("somedataset").setTableId("sometable");
    BigQueryIO.Read read = BigQueryIO.read().from(table);
    checkReadTableObject(read, "foo.com:project", "somedataset", "sometable");
}
Also used : TableReference(com.google.api.services.bigquery.model.TableReference) BigQueryResourceNaming.createTempTableReference(org.apache.beam.sdk.io.gcp.bigquery.BigQueryResourceNaming.createTempTableReference) Test(org.junit.Test)

Example 43 with TableReference

use of com.google.api.services.bigquery.model.TableReference in project beam by apache.

the class BigQueryIOReadTest method testBigQueryQuerySourceInitSplit.

@Test
public void testBigQueryQuerySourceInitSplit() throws Exception {
    PipelineOptions options = PipelineOptionsFactory.create();
    BigQueryOptions bqOptions = options.as(BigQueryOptions.class);
    bqOptions.setProject("project");
    TableReference sourceTableRef = BigQueryHelpers.parseTableSpec("project:dataset.table");
    fakeDatasetService.createDataset(sourceTableRef.getProjectId(), sourceTableRef.getDatasetId(), "asia-northeast1", "Fake plastic tree^H^H^H^Htables", null);
    fakeDatasetService.createTable(new Table().setTableReference(sourceTableRef).setLocation("asia-northeast1"));
    Table queryResultTable = new Table().setSchema(new TableSchema().setFields(ImmutableList.of(new TableFieldSchema().setName("name").setType("STRING"), new TableFieldSchema().setName("number").setType("INTEGER"))));
    List<TableRow> expected = ImmutableList.of(new TableRow().set("name", "a").set("number", 1L), new TableRow().set("name", "b").set("number", 2L), new TableRow().set("name", "c").set("number", 3L), new TableRow().set("name", "d").set("number", 4L), new TableRow().set("name", "e").set("number", 5L), new TableRow().set("name", "f").set("number", 6L));
    String encodedQuery = FakeBigQueryServices.encodeQueryResult(queryResultTable, expected);
    String stepUuid = "testStepUuid";
    TableReference tempTableReference = createTempTableReference(bqOptions.getProject(), BigQueryResourceNaming.createJobIdPrefix(options.getJobName(), stepUuid, JobType.QUERY), Optional.empty());
    fakeJobService.expectDryRunQuery(bqOptions.getProject(), encodedQuery, new JobStatistics().setQuery(new JobStatistics2().setTotalBytesProcessed(100L).setReferencedTables(ImmutableList.of(sourceTableRef, tempTableReference))));
    BoundedSource<TableRow> bqSource = BigQueryQuerySourceDef.create(fakeBqServices, ValueProvider.StaticValueProvider.of(encodedQuery), true, /* flattenResults */
    true, /* useLegacySql */
    QueryPriority.BATCH, null, null, null).toSource(stepUuid, TableRowJsonCoder.of(), BigQueryIO.TableRowParser.INSTANCE, false);
    options.setTempLocation(testFolder.getRoot().getAbsolutePath());
    List<TableRow> read = convertStringsToLong(SourceTestUtils.readFromSplitsOfSource(bqSource, 0L, /* ignored */
    options));
    assertThat(read, containsInAnyOrder(Iterables.toArray(expected, TableRow.class)));
    List<? extends BoundedSource<TableRow>> sources = bqSource.split(100, options);
    assertEquals(2, sources.size());
}
Also used : JobStatistics(com.google.api.services.bigquery.model.JobStatistics) JobStatistics2(com.google.api.services.bigquery.model.JobStatistics2) Table(com.google.api.services.bigquery.model.Table) TableSchema(com.google.api.services.bigquery.model.TableSchema) ByteString(com.google.protobuf.ByteString) TableFieldSchema(com.google.api.services.bigquery.model.TableFieldSchema) TableReference(com.google.api.services.bigquery.model.TableReference) BigQueryResourceNaming.createTempTableReference(org.apache.beam.sdk.io.gcp.bigquery.BigQueryResourceNaming.createTempTableReference) PipelineOptions(org.apache.beam.sdk.options.PipelineOptions) TableRow(com.google.api.services.bigquery.model.TableRow) Test(org.junit.Test)

Example 44 with TableReference

use of com.google.api.services.bigquery.model.TableReference in project beam by apache.

the class BigQueryIOStorageQueryTest method testQuerySourceInitialSplitWithBigQueryProject_EmptyResult.

@Test
@ProjectOverride
public void testQuerySourceInitialSplitWithBigQueryProject_EmptyResult() throws Exception {
    TableReference sourceTableRef = BigQueryHelpers.parseTableSpec("bigquery-project-id:dataset.table");
    fakeDatasetService.createDataset(sourceTableRef.getProjectId(), sourceTableRef.getDatasetId(), "asia-northeast1", "Fake plastic tree^H^H^H^Htables", null);
    fakeDatasetService.createTable(new Table().setTableReference(sourceTableRef).setLocation("asia-northeast1"));
    Table queryResultTable = new Table().setSchema(TABLE_SCHEMA).setNumBytes(0L);
    String encodedQuery = FakeBigQueryServices.encodeQueryResult(queryResultTable);
    fakeJobService.expectDryRunQuery(options.getBigQueryProject(), encodedQuery, new JobStatistics().setQuery(new JobStatistics2().setTotalBytesProcessed(1024L * 1024L).setReferencedTables(ImmutableList.of(sourceTableRef))));
    String stepUuid = "testStepUuid";
    TableReference tempTableReference = createTempTableReference(options.getBigQueryProject(), BigQueryResourceNaming.createJobIdPrefix(options.getJobName(), stepUuid, JobType.QUERY), Optional.empty());
    CreateReadSessionRequest expectedRequest = CreateReadSessionRequest.newBuilder().setParent("projects/" + options.getBigQueryProject()).setReadSession(ReadSession.newBuilder().setTable(BigQueryHelpers.toTableResourceName(tempTableReference)).setDataFormat(DataFormat.AVRO)).setMaxStreamCount(10).build();
    ReadSession emptyReadSession = ReadSession.newBuilder().build();
    StorageClient fakeStorageClient = mock(StorageClient.class);
    when(fakeStorageClient.createReadSession(expectedRequest)).thenReturn(emptyReadSession);
    BigQueryStorageQuerySource<TableRow> querySource = BigQueryStorageQuerySource.create(stepUuid, ValueProvider.StaticValueProvider.of(encodedQuery), /* flattenResults = */
    true, /* useLegacySql = */
    true, /* priority = */
    QueryPriority.BATCH, /* location = */
    null, /* queryTempDataset = */
    null, /* kmsKey = */
    null, DataFormat.AVRO, new TableRowParser(), TableRowJsonCoder.of(), new FakeBigQueryServices().withDatasetService(fakeDatasetService).withJobService(fakeJobService).withStorageClient(fakeStorageClient));
    List<? extends BoundedSource<TableRow>> sources = querySource.split(1024L, options);
    assertTrue(sources.isEmpty());
}
Also used : JobStatistics(com.google.api.services.bigquery.model.JobStatistics) JobStatistics2(com.google.api.services.bigquery.model.JobStatistics2) Table(com.google.api.services.bigquery.model.Table) ReadSession(com.google.cloud.bigquery.storage.v1.ReadSession) StorageClient(org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient) ByteString(com.google.protobuf.ByteString) TableRowParser(org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.TableRowParser) TableReference(com.google.api.services.bigquery.model.TableReference) BigQueryResourceNaming.createTempTableReference(org.apache.beam.sdk.io.gcp.bigquery.BigQueryResourceNaming.createTempTableReference) TableRow(com.google.api.services.bigquery.model.TableRow) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) CreateReadSessionRequest(com.google.cloud.bigquery.storage.v1.CreateReadSessionRequest) Test(org.junit.Test)

Example 45 with TableReference

use of com.google.api.services.bigquery.model.TableReference in project beam by apache.

the class BigQueryIOStorageQueryTest method doQuerySourceInitialSplit.

private void doQuerySourceInitialSplit(long bundleSize, int requestedStreamCount, int expectedStreamCount) throws Exception {
    TableReference sourceTableRef = BigQueryHelpers.parseTableSpec("project:dataset.table");
    fakeDatasetService.createDataset(sourceTableRef.getProjectId(), sourceTableRef.getDatasetId(), "asia-northeast1", "Fake plastic tree^H^H^H^Htables", null);
    fakeDatasetService.createTable(new Table().setTableReference(sourceTableRef).setLocation("asia-northeast1"));
    Table queryResultTable = new Table().setSchema(new TableSchema().setFields(ImmutableList.of(new TableFieldSchema().setName("name").setType("STRING"), new TableFieldSchema().setName("number").setType("INTEGER")))).setNumBytes(1024L * 1024L);
    String encodedQuery = FakeBigQueryServices.encodeQueryResult(queryResultTable);
    fakeJobService.expectDryRunQuery(options.getProject(), encodedQuery, new JobStatistics().setQuery(new JobStatistics2().setTotalBytesProcessed(1024L * 1024L).setReferencedTables(ImmutableList.of(sourceTableRef))));
    String stepUuid = "testStepUuid";
    TableReference tempTableReference = createTempTableReference(options.getProject(), BigQueryResourceNaming.createJobIdPrefix(options.getJobName(), stepUuid, JobType.QUERY), Optional.empty());
    CreateReadSessionRequest expectedRequest = CreateReadSessionRequest.newBuilder().setParent("projects/" + options.getProject()).setReadSession(ReadSession.newBuilder().setTable(BigQueryHelpers.toTableResourceName(tempTableReference))).setMaxStreamCount(requestedStreamCount).build();
    Schema sessionSchema = SchemaBuilder.record("__root__").fields().name("name").type().nullable().stringType().noDefault().name("number").type().nullable().longType().noDefault().endRecord();
    ReadSession.Builder builder = ReadSession.newBuilder().setAvroSchema(AvroSchema.newBuilder().setSchema(sessionSchema.toString())).setDataFormat(DataFormat.AVRO);
    for (int i = 0; i < expectedStreamCount; i++) {
        builder.addStreams(ReadStream.newBuilder().setName("stream-" + i));
    }
    StorageClient fakeStorageClient = mock(StorageClient.class);
    when(fakeStorageClient.createReadSession(expectedRequest)).thenReturn(builder.build());
    BigQueryStorageQuerySource<TableRow> querySource = BigQueryStorageQuerySource.create(stepUuid, ValueProvider.StaticValueProvider.of(encodedQuery), /* flattenResults = */
    true, /* useLegacySql = */
    true, /* priority = */
    QueryPriority.BATCH, /* location = */
    null, /* queryTempDataset = */
    null, /* kmsKey = */
    null, null, new TableRowParser(), TableRowJsonCoder.of(), new FakeBigQueryServices().withDatasetService(fakeDatasetService).withJobService(fakeJobService).withStorageClient(fakeStorageClient));
    List<? extends BoundedSource<TableRow>> sources = querySource.split(bundleSize, options);
    assertEquals(expectedStreamCount, sources.size());
}
Also used : JobStatistics(com.google.api.services.bigquery.model.JobStatistics) JobStatistics2(com.google.api.services.bigquery.model.JobStatistics2) Table(com.google.api.services.bigquery.model.Table) TableSchema(com.google.api.services.bigquery.model.TableSchema) AvroSchema(com.google.cloud.bigquery.storage.v1.AvroSchema) TableSchema(com.google.api.services.bigquery.model.TableSchema) Schema(org.apache.avro.Schema) TableFieldSchema(com.google.api.services.bigquery.model.TableFieldSchema) ReadSession(com.google.cloud.bigquery.storage.v1.ReadSession) StorageClient(org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient) ByteString(com.google.protobuf.ByteString) TableFieldSchema(com.google.api.services.bigquery.model.TableFieldSchema) TableRowParser(org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.TableRowParser) TableReference(com.google.api.services.bigquery.model.TableReference) BigQueryResourceNaming.createTempTableReference(org.apache.beam.sdk.io.gcp.bigquery.BigQueryResourceNaming.createTempTableReference) TableRow(com.google.api.services.bigquery.model.TableRow) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) CreateReadSessionRequest(com.google.cloud.bigquery.storage.v1.CreateReadSessionRequest)

Aggregations

TableReference (com.google.api.services.bigquery.model.TableReference)139 Test (org.junit.Test)75 TableRow (com.google.api.services.bigquery.model.TableRow)68 Table (com.google.api.services.bigquery.model.Table)61 TableSchema (com.google.api.services.bigquery.model.TableSchema)36 DatasetServiceImpl (org.apache.beam.sdk.io.gcp.bigquery.BigQueryServicesImpl.DatasetServiceImpl)29 TableFieldSchema (com.google.api.services.bigquery.model.TableFieldSchema)22 TableDataInsertAllResponse (com.google.api.services.bigquery.model.TableDataInsertAllResponse)19 FakeBigQueryServices (org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices)19 MockSleeper (com.google.api.client.testing.util.MockSleeper)17 BigQueryHelpers.createTempTableReference (org.apache.beam.sdk.io.gcp.bigquery.BigQueryHelpers.createTempTableReference)17 FailsafeValueInSingleWindow (org.apache.beam.sdk.values.FailsafeValueInSingleWindow)16 JobStatus (com.google.api.services.bigquery.model.JobStatus)15 ReadSession (com.google.cloud.bigquery.storage.v1.ReadSession)15 BigQueryResourceNaming.createTempTableReference (org.apache.beam.sdk.io.gcp.bigquery.BigQueryResourceNaming.createTempTableReference)15 ErrorProto (com.google.api.services.bigquery.model.ErrorProto)14 JobStatistics (com.google.api.services.bigquery.model.JobStatistics)14 CreateReadSessionRequest (com.google.cloud.bigquery.storage.v1.CreateReadSessionRequest)14 ByteString (com.google.protobuf.ByteString)14 IOException (java.io.IOException)14