Search in sources :

Example 21 with ReadRowsResponse

use of com.google.cloud.bigquery.storage.v1.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();
}
Also used : JobStatistics(com.google.api.services.bigquery.model.JobStatistics) JobStatistics2(com.google.api.services.bigquery.model.JobStatistics2) 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) BigQueryResourceNaming.createTempTableReference(org.apache.beam.sdk.io.gcp.bigquery.BigQueryResourceNaming.createTempTableReference) ReadRowsResponse(com.google.cloud.bigquery.storage.v1.ReadRowsResponse) FakeBigQueryServices(org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices) GenericRecord(org.apache.avro.generic.GenericRecord)

Aggregations

ReadRowsResponse (com.google.cloud.bigquery.storage.v1.ReadRowsResponse)18 StorageClient (org.apache.beam.sdk.io.gcp.bigquery.BigQueryServices.StorageClient)17 FakeBigQueryServices (org.apache.beam.sdk.io.gcp.testing.FakeBigQueryServices)17 Test (org.junit.Test)17 TableRow (com.google.api.services.bigquery.model.TableRow)14 TableRowParser (org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.TableRowParser)13 ReadRowsRequest (com.google.cloud.bigquery.storage.v1.ReadRowsRequest)11 ReadSession (com.google.cloud.bigquery.storage.v1.ReadSession)10 ByteString (com.google.protobuf.ByteString)9 GenericRecord (org.apache.avro.generic.GenericRecord)6 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 StatusRuntimeException (io.grpc.StatusRuntimeException)3 KV (org.apache.beam.sdk.values.KV)3 FailedPreconditionException (com.google.api.gax.rpc.FailedPreconditionException)2 ArrayList (java.util.ArrayList)2 JobStatistics (com.google.api.services.bigquery.model.JobStatistics)1 JobStatistics2 (com.google.api.services.bigquery.model.JobStatistics2)1 ArrowRecordBatch (com.google.cloud.bigquery.storage.v1.ArrowRecordBatch)1