use of com.google.cloud.bigquery.storage.v1.ReadSession 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.ReadSession in project beam by apache.
the class BigQueryIOStorageReadTest method testTableSourceInitialSplit_EmptyTable.
@Test
public void testTableSourceInitialSplit_EmptyTable() 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(new TableSchema());
fakeDatasetService.createTable(table);
CreateReadSessionRequest expectedRequest = CreateReadSessionRequest.newBuilder().setParent("projects/project-id").setReadSession(ReadSession.newBuilder().setTable("projects/foo.com:project/datasets/dataset/tables/table")).setMaxStreamCount(1024).build();
ReadSession emptyReadSession = ReadSession.newBuilder().build();
StorageClient fakeStorageClient = mock(StorageClient.class);
when(fakeStorageClient.createReadSession(expectedRequest)).thenReturn(emptyReadSession);
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(1024L, options);
assertTrue(sources.isEmpty());
}
use of com.google.cloud.bigquery.storage.v1.ReadSession in project beam by apache.
the class BigQueryIOStorageQueryTest method testQuerySourceInitialSplitWithBigQueryProject_EmptyResult.
@Test
@ProjectOverride
public void testQuerySourceInitialSplitWithBigQueryProject_EmptyResult() throws Exception {
TableReference sourceTableRef = BigQueryHelpers.parseTableSpec("bigquery-project-id: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.getBigQueryProject(), encodedQuery, new JobStatistics().setQuery(new JobStatistics2().setTotalBytesProcessed(1024L * 1024L).setReferencedTables(ImmutableList.of(sourceTableRef))));
String stepUuid = "testStepUuid";
TableReference tempTableReference = createTempTableReference(options.getBigQueryProject(), BigQueryResourceNaming.createJobIdPrefix(options.getJobName(), stepUuid, JobType.QUERY), Optional.empty());
CreateReadSessionRequest expectedRequest = CreateReadSessionRequest.newBuilder().setParent("projects/" + options.getBigQueryProject()).setReadSession(ReadSession.newBuilder().setTable(BigQueryHelpers.toTableResourceName(tempTableReference)).setDataFormat(DataFormat.AVRO)).setMaxStreamCount(10).build();
ReadSession emptyReadSession = ReadSession.newBuilder().build();
StorageClient fakeStorageClient = mock(StorageClient.class);
when(fakeStorageClient.createReadSession(expectedRequest)).thenReturn(emptyReadSession);
BigQueryStorageQuerySource<TableRow> querySource = BigQueryStorageQuerySource.create(stepUuid, ValueProvider.StaticValueProvider.of(encodedQuery), /* flattenResults = */
true, /* useLegacySql = */
true, /* priority = */
QueryPriority.BATCH, /* location = */
null, /* queryTempDataset = */
null, /* kmsKey = */
null, DataFormat.AVRO, new TableRowParser(), TableRowJsonCoder.of(), new FakeBigQueryServices().withDatasetService(fakeDatasetService).withJobService(fakeJobService).withStorageClient(fakeStorageClient));
List<? extends BoundedSource<TableRow>> sources = querySource.split(1024L, options);
assertTrue(sources.isEmpty());
}
use of com.google.cloud.bigquery.storage.v1.ReadSession in project beam by apache.
the class BigQueryIOStorageQueryTest method testQuerySourceInitialSplit_EmptyResult.
@Test
public void testQuerySourceInitialSplit_EmptyResult() 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))));
String stepUuid = "testStepUuid";
TableReference tempTableReference = createTempTableReference(options.getProject(), BigQueryResourceNaming.createJobIdPrefix(options.getJobName(), stepUuid, JobType.QUERY), Optional.empty());
CreateReadSessionRequest expectedRequest = CreateReadSessionRequest.newBuilder().setParent("projects/" + options.getProject()).setReadSession(ReadSession.newBuilder().setTable(BigQueryHelpers.toTableResourceName(tempTableReference))).setMaxStreamCount(10).build();
ReadSession emptyReadSession = ReadSession.newBuilder().build();
StorageClient fakeStorageClient = mock(StorageClient.class);
when(fakeStorageClient.createReadSession(expectedRequest)).thenReturn(emptyReadSession);
BigQueryStorageQuerySource<TableRow> querySource = BigQueryStorageQuerySource.create(stepUuid, ValueProvider.StaticValueProvider.of(encodedQuery), /* flattenResults = */
true, /* useLegacySql = */
true, /* priority = */
QueryPriority.BATCH, /* location = */
null, /* queryTempDataset = */
null, /* kmsKey = */
null, null, new TableRowParser(), TableRowJsonCoder.of(), new FakeBigQueryServices().withDatasetService(fakeDatasetService).withJobService(fakeJobService).withStorageClient(fakeStorageClient));
List<? extends BoundedSource<TableRow>> sources = querySource.split(1024L, options);
assertTrue(sources.isEmpty());
}
use of com.google.cloud.bigquery.storage.v1.ReadSession 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());
}
Aggregations