use of com.google.cloud.bigquery.storage.v1beta2.ReadRowsResponse in project java-bigtable by googleapis.
the class SimpleGceRawRead method main.
public static void main(String[] args) {
configureLogging();
String projectId = System.getProperty("bigtable.project", "gcloud-devel");
String instanceId = System.getProperty("bigtable.instance", "google-cloud-bigtable");
String tableId = System.getProperty("bigtable.table", "integration-tests");
String endpoint = System.getProperty("bigtable.data-endpoint", "bigtable.googleapis.com");
String tableName = String.format("projects/%s/instances/%s/tables/%s", projectId, instanceId, tableId);
System.out.printf(">>>>>>>>> Trying to connect to: %s, to read %s%n%n%n", endpoint, tableName);
ManagedChannel channel = ComputeEngineChannelBuilder.forAddress(endpoint, 443).disableServiceConfigLookUp().defaultServiceConfig(newServiceConfig()).build();
try {
BigtableGrpc.BigtableBlockingStub stub = BigtableGrpc.newBlockingStub(channel);
Iterator<ReadRowsResponse> iter = stub.readRows(ReadRowsRequest.newBuilder().setTableName(tableName).setRowsLimit(1).build());
System.out.printf("%n%n>>>>>>>>> Success Rows Read: %d%n%n", Lists.newArrayList(iter).size());
} finally {
channel.shutdown();
}
}
use of com.google.cloud.bigquery.storage.v1beta2.ReadRowsResponse in project beam by apache.
the class BigQueryIOStorageQueryTest method doReadFromBigQueryIO.
private void doReadFromBigQueryIO(boolean templateCompatibility) 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(TABLE_SCHEMA).setNumBytes(0L);
String encodedQuery = FakeBigQueryServices.encodeQueryResult(queryResultTable);
fakeJobService.expectDryRunQuery(options.getProject(), encodedQuery, new JobStatistics().setQuery(new JobStatistics2().setTotalBytesProcessed(1024L * 1024L).setReferencedTables(ImmutableList.of(sourceTableRef))));
ReadSession readSession = ReadSession.newBuilder().setName("readSessionName").setAvroSchema(AvroSchema.newBuilder().setSchema(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", 1, AVRO_SCHEMA), createRecord("B", 2, AVRO_SCHEMA), createRecord("C", 3, AVRO_SCHEMA), createRecord("D", 4, AVRO_SCHEMA));
List<ReadRowsResponse> readRowsResponses = Lists.newArrayList(createResponse(AVRO_SCHEMA, records.subList(0, 2), 0.0, 0.500), createResponse(AVRO_SCHEMA, records.subList(2, 4), 0.5, 0.875));
//
// Note that since the temporary table name is generated by the pipeline, we can't match the
// expected create read session request exactly. For now, match against any appropriately typed
// proto object.
//
StorageClient fakeStorageClient = mock(StorageClient.class, withSettings().serializable());
when(fakeStorageClient.createReadSession(any())).thenReturn(readSession);
when(fakeStorageClient.readRows(expectedReadRowsRequest, "")).thenReturn(new FakeBigQueryServerStream<>(readRowsResponses));
BigQueryIO.TypedRead<KV<String, Long>> typedRead = BigQueryIO.read(new ParseKeyValue()).fromQuery(encodedQuery).withMethod(Method.DIRECT_READ).withTestServices(new FakeBigQueryServices().withDatasetService(fakeDatasetService).withJobService(fakeJobService).withStorageClient(fakeStorageClient));
if (templateCompatibility) {
typedRead = typedRead.withTemplateCompatibility();
}
PCollection<KV<String, Long>> output = p.apply(typedRead);
PAssert.that(output).containsInAnyOrder(ImmutableList.of(KV.of("A", 1L), KV.of("B", 2L), KV.of("C", 3L), KV.of("D", 4L)));
p.run();
}
use of com.google.cloud.bigquery.storage.v1beta2.ReadRowsResponse 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();
}
use of com.google.cloud.bigquery.storage.v1beta2.ReadRowsResponse 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());
}
use of com.google.cloud.bigquery.storage.v1beta2.ReadRowsResponse 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());
}
Aggregations