use of com.google.cloud.bigquery.storage.v1.ReadRowsResponse in project beam by apache.
the class BigQueryIOStorageReadTest method testStreamSourceSplitAtFractionSucceedsArrow.
@Test
public void testStreamSourceSplitAtFractionSucceedsArrow() throws Exception {
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> 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.
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").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"));
// 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());
}
use of com.google.cloud.bigquery.storage.v1.ReadRowsResponse in project beam by apache.
the class BigQueryIOStorageReadTest method createResponseArrow.
private ReadRowsResponse createResponseArrow(org.apache.arrow.vector.types.pojo.Schema arrowSchema, List<String> name, List<Long> number, double progressAtResponseStart, double progressAtResponseEnd) {
ArrowRecordBatch serializedRecord;
try (VectorSchemaRoot schemaRoot = VectorSchemaRoot.create(arrowSchema, allocator)) {
schemaRoot.allocateNew();
schemaRoot.setRowCount(name.size());
VarCharVector strVector = (VarCharVector) schemaRoot.getFieldVectors().get(0);
BigIntVector bigIntVector = (BigIntVector) schemaRoot.getFieldVectors().get(1);
for (int i = 0; i < name.size(); i++) {
bigIntVector.set(i, number.get(i));
strVector.set(i, new Text(name.get(i)));
}
VectorUnloader unLoader = new VectorUnloader(schemaRoot);
try (org.apache.arrow.vector.ipc.message.ArrowRecordBatch records = unLoader.getRecordBatch()) {
try (ByteArrayOutputStream os = new ByteArrayOutputStream()) {
MessageSerializer.serialize(new WriteChannel(Channels.newChannel(os)), records);
serializedRecord = ArrowRecordBatch.newBuilder().setRowCount(records.getLength()).setSerializedRecordBatch(ByteString.copyFrom(os.toByteArray())).build();
} catch (IOException e) {
throw new RuntimeException("Error writing to byte array output stream", e);
}
}
}
return ReadRowsResponse.newBuilder().setArrowRecordBatch(serializedRecord).setRowCount(name.size()).setStats(StreamStats.newBuilder().setProgress(Progress.newBuilder().setAtResponseStart(progressAtResponseStart).setAtResponseEnd(progressAtResponseEnd))).build();
}
use of com.google.cloud.bigquery.storage.v1.ReadRowsResponse in project beam by apache.
the class BigQueryIOStorageReadTest method testReadFromStreamSource.
@Test
public void testReadFromStreamSource() throws Exception {
ReadSession readSession = ReadSession.newBuilder().setName("readSession").setAvroSchema(AvroSchema.newBuilder().setSchema(AVRO_SCHEMA_STRING)).build();
ReadRowsRequest expectedRequest = ReadRowsRequest.newBuilder().setReadStream("readStream").build();
List<GenericRecord> records = Lists.newArrayList(createRecord("A", 1, AVRO_SCHEMA), createRecord("B", 2, AVRO_SCHEMA), createRecord("C", 3, AVRO_SCHEMA));
List<ReadRowsResponse> responses = Lists.newArrayList(createResponse(AVRO_SCHEMA, records.subList(0, 2), 0.0, 0.50), createResponse(AVRO_SCHEMA, records.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.v1.ReadRowsResponse in project beam by apache.
the class BigQueryStorageArrowReader method processReadRowsResponse.
@Override
public void processReadRowsResponse(ReadRowsResponse readRowsResponse) throws IOException {
com.google.cloud.bigquery.storage.v1.ArrowRecordBatch recordBatch = readRowsResponse.getArrowRecordBatch();
rowCount = recordBatch.getRowCount();
this.alloc = new RootAllocator(Long.MAX_VALUE);
InputStream input = protoSchema.getSerializedSchema().newInput();
Schema arrowSchema = ArrowConversion.arrowSchemaFromInput(input);
this.recordBatchIterator = ArrowConversion.rowsFromSerializedRecordBatch(arrowSchema, recordBatch.getSerializedRecordBatch().newInput(), this.alloc);
}
use of com.google.cloud.bigquery.storage.v1.ReadRowsResponse in project beam by apache.
the class BigQueryStorageAvroReader method processReadRowsResponse.
@Override
public void processReadRowsResponse(ReadRowsResponse readRowsResponse) {
AvroRows avroRows = readRowsResponse.getAvroRows();
rowCount = avroRows.getRowCount();
decoder = DecoderFactory.get().binaryDecoder(avroRows.getSerializedBinaryRows().toByteArray(), decoder);
}
Aggregations