Search in sources :

Example 6 with ReadRowsRequest

use of com.google.cloud.bigquery.storage.v1.ReadRowsRequest 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 7 with ReadRowsRequest

use of com.google.cloud.bigquery.storage.v1.ReadRowsRequest 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 8 with ReadRowsRequest

use of com.google.cloud.bigquery.storage.v1.ReadRowsRequest 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 9 with ReadRowsRequest

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

the class BigQueryIOStorageReadTest method testReadFromStreamSourceArrow.

@Test
public void testReadFromStreamSourceArrow() throws Exception {
    ReadSession readSession = ReadSession.newBuilder().setName("readSession").setArrowSchema(ArrowSchema.newBuilder().setSerializedSchema(serializeArrowSchema(ARROW_SCHEMA)).build()).setDataFormat(DataFormat.ARROW).build();
    ReadRowsRequest expectedRequest = ReadRowsRequest.newBuilder().setReadStream("readStream").build();
    List<String> names = Arrays.asList("A", "B", "C");
    List<Long> values = Arrays.asList(1L, 2L, 3L);
    List<ReadRowsResponse> responses = Lists.newArrayList(createResponseArrow(ARROW_SCHEMA, names.subList(0, 2), values.subList(0, 2), 0.0, 0.50), createResponseArrow(ARROW_SCHEMA, names.subList(2, 3), values.subList(2, 3), 0.5, 0.75));
    StorageClient fakeStorageClient = mock(StorageClient.class);
    when(fakeStorageClient.readRows(expectedRequest, "")).thenReturn(new FakeBigQueryServerStream<>(responses));
    BigQueryStorageStreamSource<TableRow> streamSource = BigQueryStorageStreamSource.create(readSession, ReadStream.newBuilder().setName("readStream").build(), TABLE_SCHEMA, new TableRowParser(), TableRowJsonCoder.of(), new FakeBigQueryServices().withStorageClient(fakeStorageClient));
    List<TableRow> rows = new ArrayList<>();
    BoundedReader<TableRow> reader = streamSource.createReader(options);
    for (boolean hasNext = reader.start(); hasNext; hasNext = reader.advance()) {
        rows.add(reader.getCurrent());
    }
    System.out.println("Rows: " + rows);
    assertEquals(3, rows.size());
}
Also used : ReadSession(com.google.cloud.bigquery.storage.v1.ReadSession) ArrayList(java.util.ArrayList) ReadRowsRequest(com.google.cloud.bigquery.storage.v1.ReadRowsRequest) 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)

Example 10 with ReadRowsRequest

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

the class BigQueryServicesImplTest method testReadRowsSetsRequestCountMetricOnError.

@Test
public void testReadRowsSetsRequestCountMetricOnError() throws InterruptedException, IOException {
    BigQueryServices.StorageClient client = mock(BigQueryServicesImpl.StorageClientImpl.class);
    ReadRowsRequest request = null;
    StatusCode statusCode = new StatusCode() {

        @Override
        public Code getCode() {
            return Code.INTERNAL;
        }

        @Override
        public Object getTransportCode() {
            return null;
        }
    };
    when(client.readRows(request)).thenThrow(// Mock implementation.
    new ApiException("Internal", null, statusCode, false));
    // Real implementation.
    when(client.readRows(any(), any())).thenCallRealMethod();
    thrown.expect(ApiException.class);
    thrown.expectMessage("Internal");
    client.readRows(request, "myproject:mydataset.mytable");
    verifyReadMetricWasSet("myproject", "mydataset", "mytable", "internal", 1);
}
Also used : ReadRowsRequest(com.google.cloud.bigquery.storage.v1.ReadRowsRequest) StatusCode(com.google.api.gax.rpc.StatusCode) ApiException(com.google.api.gax.rpc.ApiException) Test(org.junit.Test)

Aggregations

ReadRowsRequest (com.google.cloud.bigquery.storage.v1.ReadRowsRequest)12 ReadRowsResponse (com.google.cloud.bigquery.storage.v1.ReadRowsResponse)11 Test (org.junit.Test)11 ReadSession (com.google.cloud.bigquery.storage.v1.ReadSession)10 StorageClient (org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient)10 FakeBigQueryServices (org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices)10 TableRow (com.google.api.services.bigquery.model.TableRow)7 GenericRecord (org.apache.avro.generic.GenericRecord)6 TableRowParser (org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.TableRowParser)6 ByteString (com.google.protobuf.ByteString)5 Table (com.google.api.services.bigquery.model.Table)4 TableReference (com.google.api.services.bigquery.model.TableReference)4 CreateReadSessionRequest (com.google.cloud.bigquery.storage.v1.CreateReadSessionRequest)3 KV (org.apache.beam.sdk.values.KV)3 ArrayList (java.util.ArrayList)2 ApiException (com.google.api.gax.rpc.ApiException)1 StatusCode (com.google.api.gax.rpc.StatusCode)1 JobStatistics (com.google.api.services.bigquery.model.JobStatistics)1 JobStatistics2 (com.google.api.services.bigquery.model.JobStatistics2)1 BigQueryResourceNaming.createTempTableReference (org.apache.beam.sdk.io.gcp.bigquery.BigQueryResourceNaming.createTempTableReference)1