Search in sources :

Example 11 with ReadSession

use of com.google.cloud.bigquery.storage.v1.ReadSession in project beam by apache.

the class BigQueryIOStorageReadTest method testReadFromBigQueryIOArrow.

@Test
public void testReadFromBigQueryIOArrow() throws Exception {
    fakeDatasetService.createDataset("foo.com:project", "dataset", "", "", null);
    TableReference tableRef = BigQueryHelpers.parseTableSpec("foo.com:project:dataset.table");
    Table table = new Table().setTableReference(tableRef).setNumBytes(10L).setSchema(TABLE_SCHEMA);
    fakeDatasetService.createTable(table);
    CreateReadSessionRequest expectedCreateReadSessionRequest = CreateReadSessionRequest.newBuilder().setParent("projects/project-id").setReadSession(ReadSession.newBuilder().setTable("projects/foo.com:project/datasets/dataset/tables/table").setDataFormat(DataFormat.ARROW)).setMaxStreamCount(10).build();
    ReadSession readSession = ReadSession.newBuilder().setName("readSessionName").setArrowSchema(ArrowSchema.newBuilder().setSerializedSchema(serializeArrowSchema(ARROW_SCHEMA)).build()).addStreams(ReadStream.newBuilder().setName("streamName")).setDataFormat(DataFormat.ARROW).build();
    ReadRowsRequest expectedReadRowsRequest = ReadRowsRequest.newBuilder().setReadStream("streamName").build();
    List<String> names = Arrays.asList("A", "B", "C", "D");
    List<Long> values = Arrays.asList(1L, 2L, 3L, 4L);
    List<ReadRowsResponse> readRowsResponses = Lists.newArrayList(createResponseArrow(ARROW_SCHEMA, names.subList(0, 2), values.subList(0, 2), 0.0, 0.50), createResponseArrow(ARROW_SCHEMA, names.subList(2, 4), values.subList(2, 4), 0.5, 0.75));
    StorageClient fakeStorageClient = mock(StorageClient.class, withSettings().serializable());
    when(fakeStorageClient.createReadSession(expectedCreateReadSessionRequest)).thenReturn(readSession);
    when(fakeStorageClient.readRows(expectedReadRowsRequest, "")).thenReturn(new FakeBigQueryServerStream<>(readRowsResponses));
    PCollection<KV<String, Long>> output = p.apply(BigQueryIO.read(new ParseKeyValue()).from("foo.com:project:dataset.table").withMethod(Method.DIRECT_READ).withFormat(DataFormat.ARROW).withTestServices(new FakeBigQueryServices().withDatasetService(fakeDatasetService).withStorageClient(fakeStorageClient)));
    PAssert.that(output).containsInAnyOrder(ImmutableList.of(KV.of("A", 1L), KV.of("B", 2L), KV.of("C", 3L), KV.of("D", 4L)));
    p.run();
}
Also used : Table(com.google.api.services.bigquery.model.Table) ReadSession(com.google.cloud.bigquery.storage.v1.ReadSession) ReadRowsRequest(com.google.cloud.bigquery.storage.v1.ReadRowsRequest) StorageClient(org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient) ByteString(com.google.protobuf.ByteString) KV(org.apache.beam.sdk.values.KV) TableReference(com.google.api.services.bigquery.model.TableReference) ReadRowsResponse(com.google.cloud.bigquery.storage.v1.ReadRowsResponse) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) CreateReadSessionRequest(com.google.cloud.bigquery.storage.v1.CreateReadSessionRequest) Test(org.junit.Test)

Example 12 with ReadSession

use of com.google.cloud.bigquery.storage.v1.ReadSession in project beam by apache.

the class BigQueryIOStorageReadTest method testFractionConsumedWithSplit.

@Test
public void testFractionConsumedWithSplit() throws Exception {
    ReadSession readSession = ReadSession.newBuilder().setName("readSession").setAvroSchema(AvroSchema.newBuilder().setSchema(AVRO_SCHEMA_STRING)).build();
    ReadRowsRequest expectedRequest = ReadRowsRequest.newBuilder().setReadStream("parentStream").build();
    List<GenericRecord> records = Lists.newArrayList(createRecord("A", 1, AVRO_SCHEMA), createRecord("B", 2, AVRO_SCHEMA), createRecord("C", 3, AVRO_SCHEMA), createRecord("D", 4, AVRO_SCHEMA), createRecord("E", 5, AVRO_SCHEMA), createRecord("F", 6, AVRO_SCHEMA), createRecord("G", 7, AVRO_SCHEMA));
    List<ReadRowsResponse> parentResponses = Lists.newArrayList(createResponse(AVRO_SCHEMA, records.subList(0, 2), 0.000, 0.250), createResponse(AVRO_SCHEMA, records.subList(2, 4), 0.300, 0.500), createResponse(AVRO_SCHEMA, records.subList(4, 7), 0.800, 0.875));
    StorageClient fakeStorageClient = mock(StorageClient.class);
    when(fakeStorageClient.readRows(expectedRequest, "")).thenReturn(new FakeBigQueryServerStream<>(parentResponses));
    when(fakeStorageClient.splitReadStream(SplitReadStreamRequest.newBuilder().setName("parentStream").setFraction(0.5f).build())).thenReturn(SplitReadStreamResponse.newBuilder().setPrimaryStream(ReadStream.newBuilder().setName("primaryStream")).setRemainderStream(ReadStream.newBuilder().setName("remainderStream")).build());
    List<ReadRowsResponse> primaryResponses = Lists.newArrayList(createResponse(AVRO_SCHEMA, records.subList(1, 3), 0.25, 0.75), createResponse(AVRO_SCHEMA, records.subList(3, 4), 0.8, 1.0));
    when(fakeStorageClient.readRows(ReadRowsRequest.newBuilder().setReadStream("primaryStream").setOffset(1).build(), "")).thenReturn(new FakeBigQueryServerStream<>(primaryResponses));
    BigQueryStorageStreamSource<TableRow> streamSource = BigQueryStorageStreamSource.create(readSession, ReadStream.newBuilder().setName("parentStream").build(), TABLE_SCHEMA, new TableRowParser(), TableRowJsonCoder.of(), new FakeBigQueryServices().withStorageClient(fakeStorageClient));
    BoundedReader<TableRow> reader = streamSource.createReader(options);
    // Before call to BoundedReader#start, fraction consumed must be zero.
    assertEquals(0.0, reader.getFractionConsumed(), DELTA);
    // Reads A.
    assertTrue(reader.start());
    assertEquals(0.125, reader.getFractionConsumed(), DELTA);
    reader.splitAtFraction(0.5);
    assertEquals(0.125, reader.getFractionConsumed(), DELTA);
    // Reads B.
    assertTrue(reader.advance());
    assertEquals(0.5, reader.getFractionConsumed(), DELTA);
    // Reads C.
    assertTrue(reader.advance());
    assertEquals(0.75, reader.getFractionConsumed(), DELTA);
    // Reads D.
    assertTrue(reader.advance());
    assertEquals(1.0, reader.getFractionConsumed(), DELTA);
    assertFalse(reader.advance());
    assertEquals(1.0, reader.getFractionConsumed(), DELTA);
}
Also used : ReadSession(com.google.cloud.bigquery.storage.v1.ReadSession) ReadRowsRequest(com.google.cloud.bigquery.storage.v1.ReadRowsRequest) StorageClient(org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient) TableRowParser(org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.TableRowParser) ReadRowsResponse(com.google.cloud.bigquery.storage.v1.ReadRowsResponse) TableRow(com.google.api.services.bigquery.model.TableRow) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) GenericRecord(org.apache.avro.generic.GenericRecord) Test(org.junit.Test)

Example 13 with ReadSession

use of com.google.cloud.bigquery.storage.v1.ReadSession in project beam by apache.

the class BigQueryIOStorageReadTest method testReadFromBigQueryIOWithTrimmedSchema.

@Test
public void testReadFromBigQueryIOWithTrimmedSchema() throws Exception {
    fakeDatasetService.createDataset("foo.com:project", "dataset", "", "", null);
    TableReference tableRef = BigQueryHelpers.parseTableSpec("foo.com:project:dataset.table");
    Table table = new Table().setTableReference(tableRef).setNumBytes(10L).setSchema(TABLE_SCHEMA);
    fakeDatasetService.createTable(table);
    CreateReadSessionRequest expectedCreateReadSessionRequest = CreateReadSessionRequest.newBuilder().setParent("projects/project-id").setReadSession(ReadSession.newBuilder().setTable("projects/foo.com:project/datasets/dataset/tables/table").setReadOptions(ReadSession.TableReadOptions.newBuilder().addSelectedFields("name")).setDataFormat(DataFormat.AVRO)).setMaxStreamCount(10).build();
    ReadSession readSession = ReadSession.newBuilder().setName("readSessionName").setAvroSchema(AvroSchema.newBuilder().setSchema(TRIMMED_AVRO_SCHEMA_STRING)).addStreams(ReadStream.newBuilder().setName("streamName")).setDataFormat(DataFormat.AVRO).build();
    ReadRowsRequest expectedReadRowsRequest = ReadRowsRequest.newBuilder().setReadStream("streamName").build();
    List<GenericRecord> records = Lists.newArrayList(createRecord("A", TRIMMED_AVRO_SCHEMA), createRecord("B", TRIMMED_AVRO_SCHEMA), createRecord("C", TRIMMED_AVRO_SCHEMA), createRecord("D", TRIMMED_AVRO_SCHEMA));
    List<ReadRowsResponse> readRowsResponses = Lists.newArrayList(createResponse(TRIMMED_AVRO_SCHEMA, records.subList(0, 2), 0.0, 0.50), createResponse(TRIMMED_AVRO_SCHEMA, records.subList(2, 4), 0.5, 0.75));
    StorageClient fakeStorageClient = mock(StorageClient.class, withSettings().serializable());
    when(fakeStorageClient.createReadSession(expectedCreateReadSessionRequest)).thenReturn(readSession);
    when(fakeStorageClient.readRows(expectedReadRowsRequest, "")).thenReturn(new FakeBigQueryServerStream<>(readRowsResponses));
    PCollection<TableRow> output = p.apply(BigQueryIO.readTableRows().from("foo.com:project:dataset.table").withMethod(Method.DIRECT_READ).withSelectedFields(Lists.newArrayList("name")).withFormat(DataFormat.AVRO).withTestServices(new FakeBigQueryServices().withDatasetService(fakeDatasetService).withStorageClient(fakeStorageClient)));
    PAssert.that(output).containsInAnyOrder(ImmutableList.of(new TableRow().set("name", "A"), new TableRow().set("name", "B"), new TableRow().set("name", "C"), new TableRow().set("name", "D")));
    p.run();
}
Also used : Table(com.google.api.services.bigquery.model.Table) ReadSession(com.google.cloud.bigquery.storage.v1.ReadSession) ReadRowsRequest(com.google.cloud.bigquery.storage.v1.ReadRowsRequest) StorageClient(org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient) TableReference(com.google.api.services.bigquery.model.TableReference) ReadRowsResponse(com.google.cloud.bigquery.storage.v1.ReadRowsResponse) TableRow(com.google.api.services.bigquery.model.TableRow) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) GenericRecord(org.apache.avro.generic.GenericRecord) CreateReadSessionRequest(com.google.cloud.bigquery.storage.v1.CreateReadSessionRequest) Test(org.junit.Test)

Example 14 with ReadSession

use of com.google.cloud.bigquery.storage.v1.ReadSession in project beam by apache.

the class BigQueryIOStorageReadTest method testStreamSourceSplitAtFractionFailsWhenParentIsPastSplitPointArrow.

@Test
public void testStreamSourceSplitAtFractionFailsWhenParentIsPastSplitPointArrow() throws Exception {
    List<String> names = Arrays.asList("A", "B", "C", "D", "E", "F");
    List<Long> values = Arrays.asList(1L, 2L, 3L, 4L, 5L, 6L);
    List<ReadRowsResponse> parentResponses = Lists.newArrayList(createResponseArrow(ARROW_SCHEMA, names.subList(0, 2), values.subList(0, 2), 0.0, 0.25), createResponseArrow(ARROW_SCHEMA, names.subList(2, 3), values.subList(2, 3), 0.25, 0.5), createResponseArrow(ARROW_SCHEMA, names.subList(3, 5), values.subList(3, 5), 0.5, 0.75));
    StorageClient fakeStorageClient = mock(StorageClient.class);
    when(fakeStorageClient.readRows(ReadRowsRequest.newBuilder().setReadStream("parentStream").build(), "")).thenReturn(new FakeBigQueryServerStream<>(parentResponses));
    // Mocks the split call. A response without a primary_stream and remainder_stream means
    // that the split is not possible.
    // Mocks the split call.
    when(fakeStorageClient.splitReadStream(SplitReadStreamRequest.newBuilder().setName("parentStream").setFraction(0.5f).build())).thenReturn(SplitReadStreamResponse.newBuilder().setPrimaryStream(ReadStream.newBuilder().setName("primaryStream")).setRemainderStream(ReadStream.newBuilder().setName("remainderStream")).build());
    // Mocks the ReadRows calls expected on the primary and residual streams.
    when(fakeStorageClient.readRows(ReadRowsRequest.newBuilder().setReadStream("primaryStream").setOffset(2).build(), "")).thenThrow(new FailedPreconditionException("Given row offset is invalid for stream.", new StatusRuntimeException(Status.FAILED_PRECONDITION), GrpcStatusCode.of(Code.FAILED_PRECONDITION), /* retryable = */
    false));
    BigQueryStorageStreamSource<TableRow> streamSource = BigQueryStorageStreamSource.create(ReadSession.newBuilder().setName("readSession").setArrowSchema(ArrowSchema.newBuilder().setSerializedSchema(serializeArrowSchema(ARROW_SCHEMA)).build()).setDataFormat(DataFormat.ARROW).build(), ReadStream.newBuilder().setName("parentStream").build(), TABLE_SCHEMA, new TableRowParser(), TableRowJsonCoder.of(), new FakeBigQueryServices().withStorageClient(fakeStorageClient));
    // Read a few records from the parent stream and ensure that records are returned in the
    // prescribed order.
    BoundedReader<TableRow> parent = streamSource.createReader(options);
    assertTrue(parent.start());
    assertEquals("A", parent.getCurrent().get("name"));
    assertTrue(parent.advance());
    assertEquals("B", parent.getCurrent().get("name"));
    assertNull(parent.splitAtFraction(0.5));
    // Verify that the parent source still works okay even after an unsuccessful split attempt.
    assertTrue(parent.advance());
    assertEquals("C", parent.getCurrent().get("name"));
    assertTrue(parent.advance());
    assertEquals("D", parent.getCurrent().get("name"));
    assertTrue(parent.advance());
    assertEquals("E", parent.getCurrent().get("name"));
    assertFalse(parent.advance());
}
Also used : 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) ReadRowsResponse(com.google.cloud.bigquery.storage.v1.ReadRowsResponse) TableRow(com.google.api.services.bigquery.model.TableRow) FailedPreconditionException(com.google.api.gax.rpc.FailedPreconditionException) StatusRuntimeException(io.grpc.StatusRuntimeException) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) Test(org.junit.Test)

Example 15 with ReadSession

use of com.google.cloud.bigquery.storage.v1.ReadSession in project beam by apache.

the class BigQueryIOStorageReadTest method testStreamSourceSplitAtFractionFailsWhenSplitIsNotPossibleArrow.

@Test
public void testStreamSourceSplitAtFractionFailsWhenSplitIsNotPossibleArrow() throws Exception {
    List<String> names = Arrays.asList("A", "B", "C", "D", "E", "F");
    List<Long> values = Arrays.asList(1L, 2L, 3L, 4L, 5L, 6L);
    List<ReadRowsResponse> parentResponses = Lists.newArrayList(createResponseArrow(ARROW_SCHEMA, names.subList(0, 2), values.subList(0, 2), 0.0, 0.25), createResponseArrow(ARROW_SCHEMA, names.subList(2, 3), values.subList(2, 3), 0.25, 0.5), createResponseArrow(ARROW_SCHEMA, names.subList(3, 5), values.subList(3, 5), 0.5, 0.75));
    StorageClient fakeStorageClient = mock(StorageClient.class);
    when(fakeStorageClient.readRows(ReadRowsRequest.newBuilder().setReadStream("parentStream").build(), "")).thenReturn(new FakeBigQueryServerStream<>(parentResponses));
    // Mocks the split call. A response without a primary_stream and remainder_stream means
    // that the split is not possible.
    when(fakeStorageClient.splitReadStream(SplitReadStreamRequest.newBuilder().setName("parentStream").setFraction(0.5f).build())).thenReturn(SplitReadStreamResponse.getDefaultInstance());
    BigQueryStorageStreamSource<TableRow> streamSource = BigQueryStorageStreamSource.create(ReadSession.newBuilder().setName("readSession").setArrowSchema(ArrowSchema.newBuilder().setSerializedSchema(serializeArrowSchema(ARROW_SCHEMA)).build()).setDataFormat(DataFormat.ARROW).build(), ReadStream.newBuilder().setName("parentStream").build(), TABLE_SCHEMA, new TableRowParser(), TableRowJsonCoder.of(), new FakeBigQueryServices().withStorageClient(fakeStorageClient));
    // Read a few records from the parent stream and ensure that records are returned in the
    // prescribed order.
    BoundedReader<TableRow> parent = streamSource.createReader(options);
    assertTrue(parent.start());
    assertEquals("A", parent.getCurrent().get("name"));
    assertTrue(parent.advance());
    assertEquals("B", parent.getCurrent().get("name"));
    assertNull(parent.splitAtFraction(0.5));
    verify(fakeStorageClient, times(1)).splitReadStream(ArgumentMatchers.any());
    // Verify that subsequent splitAtFraction() calls after a failed splitAtFraction() attempt
    // do NOT invoke SplitReadStream.
    assertNull(parent.splitAtFraction(0.5));
    verify(fakeStorageClient, times(1)).splitReadStream(ArgumentMatchers.any());
    // Verify that the parent source still works okay even after an unsuccessful split attempt.
    assertTrue(parent.advance());
    assertEquals("C", parent.getCurrent().get("name"));
    assertTrue(parent.advance());
    assertEquals("D", parent.getCurrent().get("name"));
    assertTrue(parent.advance());
    assertEquals("E", parent.getCurrent().get("name"));
    assertFalse(parent.advance());
}
Also used : 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) ReadRowsResponse(com.google.cloud.bigquery.storage.v1.ReadRowsResponse) TableRow(com.google.api.services.bigquery.model.TableRow) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) Test(org.junit.Test)

Aggregations

StorageClient (org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient)22 FakeBigQueryServices (org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices)21 Test (org.junit.Test)20 TableRow (com.google.api.services.bigquery.model.TableRow)18 ReadRowsResponse (com.google.cloud.bigquery.storage.v1.ReadRowsResponse)17 TableRowParser (org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.TableRowParser)17 ReadSession (com.google.cloud.bigquery.storage.v1.ReadSession)14 ByteString (com.google.protobuf.ByteString)11 ReadRowsRequest (com.google.cloud.bigquery.storage.v1.ReadRowsRequest)10 Table (com.google.api.services.bigquery.model.Table)8 TableReference (com.google.api.services.bigquery.model.TableReference)7 CreateReadSessionRequest (com.google.cloud.bigquery.storage.v1.CreateReadSessionRequest)7 GenericRecord (org.apache.avro.generic.GenericRecord)6 JobStatistics (com.google.api.services.bigquery.model.JobStatistics)3 JobStatistics2 (com.google.api.services.bigquery.model.JobStatistics2)3 ReadStream (com.google.cloud.bigquery.storage.v1.ReadStream)3 BigQueryResourceNaming.createTempTableReference (org.apache.beam.sdk.io.gcp.bigquery.BigQueryResourceNaming.createTempTableReference)3 KV (org.apache.beam.sdk.values.KV)3 FailedPreconditionException (com.google.api.gax.rpc.FailedPreconditionException)2 TableSchema (com.google.api.services.bigquery.model.TableSchema)2