Search in sources :

Example 21 with StorageClient

use of org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient in project beam by apache.

the class BigQueryIOStorageReadTest method testFractionConsumedWithSplitArrow.

@Test
public void testFractionConsumedWithSplitArrow() 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("parentStream").build();
    List<String> names = Arrays.asList("A", "B", "C", "D", "E", "F", "G");
    List<Long> values = Arrays.asList(1L, 2L, 3L, 4L, 5L, 6L, 7L);
    List<ReadRowsResponse> parentResponse = Lists.newArrayList(createResponseArrow(ARROW_SCHEMA, names.subList(0, 2), values.subList(0, 2), 0.0, 0.25), createResponseArrow(ARROW_SCHEMA, names.subList(2, 4), values.subList(2, 4), 0.3, 0.5), createResponseArrow(ARROW_SCHEMA, names.subList(4, 7), values.subList(4, 7), 0.7, 0.875));
    StorageClient fakeStorageClient = mock(StorageClient.class);
    when(fakeStorageClient.readRows(expectedRequest, "")).thenReturn(new FakeBigQueryServerStream<>(parentResponse));
    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(createResponseArrow(ARROW_SCHEMA, names.subList(1, 3), values.subList(1, 3), 0.25, 0.75), createResponseArrow(ARROW_SCHEMA, names.subList(3, 4), values.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) 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 22 with StorageClient

use of org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient in project beam by apache.

the class BigQueryIOStorageReadTest method testStreamSourceSplitAtFractionRepeatedArrow.

@Test
public void testStreamSourceSplitAtFractionRepeatedArrow() throws Exception {
    List<ReadStream> readStreams = Lists.newArrayList(ReadStream.newBuilder().setName("stream1").build(), ReadStream.newBuilder().setName("stream2").build(), ReadStream.newBuilder().setName("stream3").build());
    StorageClient fakeStorageClient = mock(StorageClient.class);
    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, 4), values.subList(2, 4), 0.25, 0.5), createResponseArrow(ARROW_SCHEMA, names.subList(4, 6), values.subList(4, 6), 0.5, 0.75));
    // Mock the initial ReadRows call.
    when(fakeStorageClient.readRows(ReadRowsRequest.newBuilder().setReadStream(readStreams.get(0).getName()).build(), "")).thenReturn(new FakeBigQueryServerStream<>(parentResponses));
    // Mock the first SplitReadStream call.
    when(fakeStorageClient.splitReadStream(SplitReadStreamRequest.newBuilder().setName(readStreams.get(0).getName()).setFraction(0.83f).build())).thenReturn(SplitReadStreamResponse.newBuilder().setPrimaryStream(readStreams.get(1)).setRemainderStream(ReadStream.newBuilder().setName("ignored")).build());
    List<ReadRowsResponse> otherResponses = Lists.newArrayList(createResponseArrow(ARROW_SCHEMA, names.subList(1, 3), values.subList(1, 3), 0.0, 0.50), createResponseArrow(ARROW_SCHEMA, names.subList(3, 4), values.subList(3, 4), 0.5, 0.75));
    // Mock the second ReadRows call.
    when(fakeStorageClient.readRows(ReadRowsRequest.newBuilder().setReadStream(readStreams.get(1).getName()).setOffset(1).build(), "")).thenReturn(new FakeBigQueryServerStream<>(otherResponses));
    // Mock the second SplitReadStream call.
    when(fakeStorageClient.splitReadStream(SplitReadStreamRequest.newBuilder().setName(readStreams.get(1).getName()).setFraction(0.75f).build())).thenReturn(SplitReadStreamResponse.newBuilder().setPrimaryStream(readStreams.get(2)).setRemainderStream(ReadStream.newBuilder().setName("ignored")).build());
    List<ReadRowsResponse> lastResponses = Lists.newArrayList(createResponseArrow(ARROW_SCHEMA, names.subList(2, 4), values.subList(2, 4), 0.80, 0.90));
    // Mock the third ReadRows call.
    when(fakeStorageClient.readRows(ReadRowsRequest.newBuilder().setReadStream(readStreams.get(2).getName()).setOffset(2).build(), "")).thenReturn(new FakeBigQueryServerStream<>(lastResponses));
    BoundedSource<TableRow> source = BigQueryStorageStreamSource.create(ReadSession.newBuilder().setName("readSession").setArrowSchema(ArrowSchema.newBuilder().setSerializedSchema(serializeArrowSchema(ARROW_SCHEMA)).build()).setDataFormat(DataFormat.ARROW).build(), readStreams.get(0), TABLE_SCHEMA, new TableRowParser(), TableRowJsonCoder.of(), new FakeBigQueryServices().withStorageClient(fakeStorageClient));
    BoundedReader<TableRow> reader = source.createReader(options);
    assertTrue(reader.start());
    assertEquals("A", reader.getCurrent().get("name"));
    BoundedSource<TableRow> residualSource = reader.splitAtFraction(0.83f);
    assertNotNull(residualSource);
    assertEquals("A", reader.getCurrent().get("name"));
    assertTrue(reader.advance());
    assertEquals("B", reader.getCurrent().get("name"));
    residualSource = reader.splitAtFraction(0.75f);
    assertNotNull(residualSource);
    assertEquals("B", reader.getCurrent().get("name"));
    assertTrue(reader.advance());
    assertEquals("C", reader.getCurrent().get("name"));
    assertTrue(reader.advance());
    assertEquals("D", reader.getCurrent().get("name"));
    assertFalse(reader.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) ReadStream(com.google.cloud.bigquery.storage.v1.ReadStream) TableRow(com.google.api.services.bigquery.model.TableRow) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) Test(org.junit.Test)

Example 23 with StorageClient

use of org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient in project beam by apache.

the class BigQueryIOStorageReadTest method testStreamSourceSplitAtFractionSucceeds.

@Test
public void testStreamSourceSplitAtFractionSucceeds() throws Exception {
    List<ReadRowsResponse> parentResponses = Lists.newArrayList(createResponse(AVRO_SCHEMA, Lists.newArrayList(createRecord("A", 1, AVRO_SCHEMA), createRecord("B", 2, AVRO_SCHEMA)), 0.0, 0.25), createResponse(AVRO_SCHEMA, Lists.newArrayList(createRecord("C", 3, AVRO_SCHEMA)), 0.25, 0.50), createResponse(AVRO_SCHEMA, Lists.newArrayList(createRecord("D", 4, AVRO_SCHEMA), createRecord("E", 5, AVRO_SCHEMA)), 0.50, 0.75));
    StorageClient fakeStorageClient = mock(StorageClient.class);
    when(fakeStorageClient.readRows(ReadRowsRequest.newBuilder().setReadStream("parentStream").build(), "")).thenReturn(new FakeBigQueryServerStream<>(parentResponses));
    // 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(), "")).thenReturn(new FakeBigQueryServerStream<>(parentResponses.subList(1, 2)));
    when(fakeStorageClient.readRows(ReadRowsRequest.newBuilder().setReadStream("remainderStream").build(), "")).thenReturn(new FakeBigQueryServerStream<>(parentResponses.subList(2, parentResponses.size())));
    BigQueryStorageStreamSource<TableRow> streamSource = BigQueryStorageStreamSource.create(ReadSession.newBuilder().setName("readSession").setAvroSchema(AvroSchema.newBuilder().setSchema(AVRO_SCHEMA_STRING)).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"));
    // Now split the stream, and ensure that the "parent" reader has been replaced with the
    // primary stream and that the returned source points to the residual stream.
    BoundedReader<TableRow> primary = parent;
    BoundedSource<TableRow> residualSource = parent.splitAtFraction(0.5);
    assertNotNull(residualSource);
    BoundedReader<TableRow> residual = residualSource.createReader(options);
    assertTrue(primary.advance());
    assertEquals("C", primary.getCurrent().get("name"));
    assertFalse(primary.advance());
    assertTrue(residual.start());
    assertEquals("D", residual.getCurrent().get("name"));
    assertTrue(residual.advance());
    assertEquals("E", residual.getCurrent().get("name"));
    assertFalse(residual.advance());
}
Also used : ReadRowsResponse(com.google.cloud.bigquery.storage.v1.ReadRowsResponse) TableRow(com.google.api.services.bigquery.model.TableRow) StorageClient(org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) TableRowParser(org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.TableRowParser) Test(org.junit.Test)

Example 24 with StorageClient

use of org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient in project beam by apache.

the class BigQueryIOStorageReadTest method doTableSourceInitialSplitTest.

private void doTableSourceInitialSplitTest(long bundleSize, int streamCount) 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(1024L * 1024L).setSchema(TABLE_SCHEMA);
    fakeDatasetService.createTable(table);
    CreateReadSessionRequest expectedRequest = CreateReadSessionRequest.newBuilder().setParent("projects/project-id").setReadSession(ReadSession.newBuilder().setTable("projects/foo.com:project/datasets/dataset/tables/table")).setMaxStreamCount(streamCount).build();
    ReadSession.Builder builder = ReadSession.newBuilder().setAvroSchema(AvroSchema.newBuilder().setSchema(AVRO_SCHEMA_STRING)).setDataFormat(DataFormat.AVRO);
    for (int i = 0; i < streamCount; i++) {
        builder.addStreams(ReadStream.newBuilder().setName("stream-" + i));
    }
    StorageClient fakeStorageClient = mock(StorageClient.class);
    when(fakeStorageClient.createReadSession(expectedRequest)).thenReturn(builder.build());
    BigQueryStorageTableSource<TableRow> tableSource = BigQueryStorageTableSource.create(ValueProvider.StaticValueProvider.of(tableRef), null, null, new TableRowParser(), TableRowJsonCoder.of(), new FakeBigQueryServices().withDatasetService(fakeDatasetService).withStorageClient(fakeStorageClient));
    List<? extends BoundedSource<TableRow>> sources = tableSource.split(bundleSize, options);
    assertEquals(streamCount, sources.size());
}
Also used : TableReference(com.google.api.services.bigquery.model.TableReference) Table(com.google.api.services.bigquery.model.Table) ReadSession(com.google.cloud.bigquery.storage.v1.ReadSession) TableRow(com.google.api.services.bigquery.model.TableRow) StorageClient(org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) CreateReadSessionRequest(com.google.cloud.bigquery.storage.v1.CreateReadSessionRequest) TableRowParser(org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.TableRowParser)

Example 25 with StorageClient

use of org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient in project beam by apache.

the class BigQueryIOStorageReadTest method testTableSourceInitialSplit_WithSelectedFieldsAndRowRestriction.

@Test
public void testTableSourceInitialSplit_WithSelectedFieldsAndRowRestriction() 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(100L).setSchema(TABLE_SCHEMA);
    fakeDatasetService.createTable(table);
    CreateReadSessionRequest expectedRequest = CreateReadSessionRequest.newBuilder().setParent("projects/project-id").setReadSession(ReadSession.newBuilder().setTable("projects/foo.com:project/datasets/dataset/tables/table").setReadOptions(ReadSession.TableReadOptions.newBuilder().addSelectedFields("name").setRowRestriction("number > 5"))).setMaxStreamCount(10).build();
    ReadSession.Builder builder = ReadSession.newBuilder().setAvroSchema(AvroSchema.newBuilder().setSchema(TRIMMED_AVRO_SCHEMA_STRING)).setDataFormat(DataFormat.AVRO);
    for (int i = 0; i < 10; i++) {
        builder.addStreams(ReadStream.newBuilder().setName("stream-" + i));
    }
    StorageClient fakeStorageClient = mock(StorageClient.class);
    when(fakeStorageClient.createReadSession(expectedRequest)).thenReturn(builder.build());
    BigQueryStorageTableSource<TableRow> tableSource = BigQueryStorageTableSource.create(ValueProvider.StaticValueProvider.of(tableRef), StaticValueProvider.of(Lists.newArrayList("name")), StaticValueProvider.of("number > 5"), new TableRowParser(), TableRowJsonCoder.of(), new FakeBigQueryServices().withDatasetService(fakeDatasetService).withStorageClient(fakeStorageClient));
    List<? extends BoundedSource<TableRow>> sources = tableSource.split(10L, options);
    assertEquals(10L, sources.size());
}
Also used : TableReference(com.google.api.services.bigquery.model.TableReference) Table(com.google.api.services.bigquery.model.Table) ReadSession(com.google.cloud.bigquery.storage.v1.ReadSession) TableRow(com.google.api.services.bigquery.model.TableRow) StorageClient(org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) CreateReadSessionRequest(com.google.cloud.bigquery.storage.v1.CreateReadSessionRequest) TableRowParser(org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.TableRowParser) Test(org.junit.Test)

Aggregations

StorageClient (org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient)27 FakeBigQueryServices (org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices)26 TableRow (com.google.api.services.bigquery.model.TableRow)23 Test (org.junit.Test)23 TableRowParser (org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.TableRowParser)22 ReadSession (com.google.cloud.bigquery.storage.v1.ReadSession)19 ReadRowsResponse (com.google.cloud.bigquery.storage.v1.ReadRowsResponse)17 Table (com.google.api.services.bigquery.model.Table)13 ByteString (com.google.protobuf.ByteString)13 TableReference (com.google.api.services.bigquery.model.TableReference)12 CreateReadSessionRequest (com.google.cloud.bigquery.storage.v1.CreateReadSessionRequest)12 ReadRowsRequest (com.google.cloud.bigquery.storage.v1.ReadRowsRequest)10 GenericRecord (org.apache.avro.generic.GenericRecord)6 JobStatistics (com.google.api.services.bigquery.model.JobStatistics)5 JobStatistics2 (com.google.api.services.bigquery.model.JobStatistics2)5 BigQueryResourceNaming.createTempTableReference (org.apache.beam.sdk.io.gcp.bigquery.BigQueryResourceNaming.createTempTableReference)5 TableSchema (com.google.api.services.bigquery.model.TableSchema)4 ReadStream (com.google.cloud.bigquery.storage.v1.ReadStream)3 Schema (org.apache.avro.Schema)3 KV (org.apache.beam.sdk.values.KV)3