Search in sources :

Example 1 with JobStatus

use of com.google.api.services.bigquery.model.JobStatus in project beam by apache.

the class FakeJobService method startCopyJob.

@Override
public void startCopyJob(JobReference jobRef, JobConfigurationTableCopy copyConfig) throws IOException, InterruptedException {
    synchronized (allJobs) {
        Job job = new Job();
        job.setJobReference(jobRef);
        job.setConfiguration(new JobConfiguration().setCopy(copyConfig));
        job.setKind(" bigquery#job");
        job.setStatus(new JobStatus().setState("PENDING"));
        allJobs.put(jobRef.getProjectId(), jobRef.getJobId(), new JobInfo(job));
    }
}
Also used : JobStatus(com.google.api.services.bigquery.model.JobStatus) Job(com.google.api.services.bigquery.model.Job) JobConfiguration(com.google.api.services.bigquery.model.JobConfiguration)

Example 2 with JobStatus

use of com.google.api.services.bigquery.model.JobStatus in project beam by apache.

the class BigQueryTableRowIteratorTest method testReadFromQuery.

/**
   * Verifies that when the query runs, the correct data is returned and the temporary dataset and
   * table are both cleaned up.
   */
@Test
public void testReadFromQuery() throws IOException, InterruptedException {
    // Mock job inserting.
    Job dryRunJob = new Job().setStatistics(new JobStatistics().setQuery(new JobStatistics2().setReferencedTables(ImmutableList.of(new TableReference()))));
    Job insertedJob = new Job().setJobReference(new JobReference());
    when(mockJobsInsert.execute()).thenReturn(dryRunJob, insertedJob);
    // Mock job polling.
    JobStatus status = new JobStatus().setState("DONE");
    JobConfigurationQuery resultQueryConfig = new JobConfigurationQuery().setDestinationTable(new TableReference().setProjectId("project").setDatasetId("tempdataset").setTableId("temptable"));
    Job getJob = new Job().setJobReference(new JobReference()).setStatus(status).setConfiguration(new JobConfiguration().setQuery(resultQueryConfig));
    when(mockJobsGet.execute()).thenReturn(getJob);
    // Mock table schema fetch.
    when(mockTablesGet.execute()).thenReturn(tableWithLocation(), tableWithBasicSchema());
    byte[] photoBytes = "photograph".getBytes();
    String photoBytesEncoded = BaseEncoding.base64().encode(photoBytes);
    // Mock table data fetch.
    when(mockTabledataList.execute()).thenReturn(rawDataList(rawRow("Arthur", 42, photoBytesEncoded, "2000-01-01", "2000-01-01 00:00:00.000005", "00:00:00.000005")));
    // Run query and verify
    String query = "SELECT name, count, photo, anniversary_date, " + "anniversary_datetime, anniversary_time from table";
    JobConfigurationQuery queryConfig = new JobConfigurationQuery().setQuery(query);
    try (BigQueryTableRowIterator iterator = BigQueryTableRowIterator.fromQuery(queryConfig, "project", mockClient)) {
        iterator.open();
        assertTrue(iterator.advance());
        TableRow row = iterator.getCurrent();
        assertTrue(row.containsKey("name"));
        assertTrue(row.containsKey("answer"));
        assertTrue(row.containsKey("photo"));
        assertTrue(row.containsKey("anniversary_date"));
        assertTrue(row.containsKey("anniversary_datetime"));
        assertTrue(row.containsKey("anniversary_time"));
        assertEquals("Arthur", row.get("name"));
        assertEquals(42, row.get("answer"));
        assertEquals(photoBytesEncoded, row.get("photo"));
        assertEquals("2000-01-01", row.get("anniversary_date"));
        assertEquals("2000-01-01 00:00:00.000005", row.get("anniversary_datetime"));
        assertEquals("00:00:00.000005", row.get("anniversary_time"));
        assertFalse(iterator.advance());
    }
    // Temp dataset created and later deleted.
    verify(mockClient, times(2)).datasets();
    verify(mockDatasets).insert(anyString(), any(Dataset.class));
    verify(mockDatasetsInsert).execute();
    verify(mockDatasets).delete(anyString(), anyString());
    verify(mockDatasetsDelete).execute();
    // Job inserted to run the query, polled once.
    verify(mockClient, times(3)).jobs();
    verify(mockJobs, times(2)).insert(anyString(), any(Job.class));
    verify(mockJobsInsert, times(2)).execute();
    verify(mockJobs).get(anyString(), anyString());
    verify(mockJobsGet).execute();
    // Temp table get after query finish, deleted after reading.
    verify(mockClient, times(3)).tables();
    verify(mockTables, times(2)).get(anyString(), anyString(), anyString());
    verify(mockTablesGet, times(2)).execute();
    verify(mockTables).delete(anyString(), anyString(), anyString());
    verify(mockTablesDelete).execute();
    // Table data read.
    verify(mockClient).tabledata();
    verify(mockTabledata).list("project", "tempdataset", "temptable");
    verify(mockTabledataList).execute();
}
Also used : JobStatistics(com.google.api.services.bigquery.model.JobStatistics) JobStatistics2(com.google.api.services.bigquery.model.JobStatistics2) JobReference(com.google.api.services.bigquery.model.JobReference) JobConfigurationQuery(com.google.api.services.bigquery.model.JobConfigurationQuery) Dataset(com.google.api.services.bigquery.model.Dataset) Matchers.anyString(org.mockito.Matchers.anyString) Matchers.containsString(org.hamcrest.Matchers.containsString) JobStatus(com.google.api.services.bigquery.model.JobStatus) TableReference(com.google.api.services.bigquery.model.TableReference) TableRow(com.google.api.services.bigquery.model.TableRow) Job(com.google.api.services.bigquery.model.Job) JobConfiguration(com.google.api.services.bigquery.model.JobConfiguration) Test(org.junit.Test)

Example 3 with JobStatus

use of com.google.api.services.bigquery.model.JobStatus in project beam by apache.

the class FakeJobService method runLoadJob.

private JobStatus runLoadJob(JobReference jobRef, JobConfigurationLoad load) throws InterruptedException, IOException {
    TableReference destination = load.getDestinationTable();
    TableSchema schema = load.getSchema();
    List<ResourceId> sourceFiles = filesForLoadJobs.get(jobRef.getProjectId(), jobRef.getJobId());
    WriteDisposition writeDisposition = WriteDisposition.valueOf(load.getWriteDisposition());
    CreateDisposition createDisposition = CreateDisposition.valueOf(load.getCreateDisposition());
    checkArgument(load.getSourceFormat().equals("NEWLINE_DELIMITED_JSON"));
    Table existingTable = datasetService.getTable(destination);
    if (!validateDispositions(existingTable, createDisposition, writeDisposition)) {
        return new JobStatus().setState("FAILED").setErrorResult(new ErrorProto());
    }
    datasetService.createTable(new Table().setTableReference(destination).setSchema(schema));
    List<TableRow> rows = Lists.newArrayList();
    for (ResourceId filename : sourceFiles) {
        rows.addAll(readRows(filename.toString()));
    }
    datasetService.insertAll(destination, rows, null);
    return new JobStatus().setState("DONE");
}
Also used : JobStatus(com.google.api.services.bigquery.model.JobStatus) TableReference(com.google.api.services.bigquery.model.TableReference) CreateDisposition(org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.Write.CreateDisposition) HashBasedTable(com.google.common.collect.HashBasedTable) Table(com.google.api.services.bigquery.model.Table) ErrorProto(com.google.api.services.bigquery.model.ErrorProto) TableSchema(com.google.api.services.bigquery.model.TableSchema) ResourceId(org.apache.beam.sdk.io.fs.ResourceId) TableRow(com.google.api.services.bigquery.model.TableRow) WriteDisposition(org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.Write.WriteDisposition)

Example 4 with JobStatus

use of com.google.api.services.bigquery.model.JobStatus in project beam by apache.

the class FakeJobService method pollJob.

@Override
public Job pollJob(JobReference jobRef, int maxAttempts) throws InterruptedException {
    BackOff backoff = BackOffAdapter.toGcpBackOff(FluentBackoff.DEFAULT.withMaxRetries(maxAttempts).withInitialBackoff(Duration.millis(10)).withMaxBackoff(Duration.standardSeconds(1)).backoff());
    Sleeper sleeper = Sleeper.DEFAULT;
    try {
        do {
            Job job = getJob(jobRef);
            if (job != null) {
                JobStatus status = job.getStatus();
                if (status != null && status.getState() != null && (status.getState().equals("DONE") || status.getState().equals("FAILED"))) {
                    return job;
                }
            }
        } while (BackOffUtils.next(sleeper, backoff));
    } catch (IOException e) {
        return null;
    }
    return null;
}
Also used : JobStatus(com.google.api.services.bigquery.model.JobStatus) Sleeper(com.google.api.client.util.Sleeper) IOException(java.io.IOException) Job(com.google.api.services.bigquery.model.Job) BackOff(com.google.api.client.util.BackOff)

Example 5 with JobStatus

use of com.google.api.services.bigquery.model.JobStatus in project beam by apache.

the class BigQueryIOTest method testBigQueryQuerySourceInitSplit.

@Test
public void testBigQueryQuerySourceInitSplit() throws Exception {
    TableReference dryRunTable = new TableReference();
    Job queryJob = new Job();
    JobStatistics queryJobStats = new JobStatistics();
    JobStatistics2 queryStats = new JobStatistics2();
    queryStats.setReferencedTables(ImmutableList.of(dryRunTable));
    queryJobStats.setQuery(queryStats);
    queryJob.setStatus(new JobStatus()).setStatistics(queryJobStats);
    Job extractJob = new Job();
    JobStatistics extractJobStats = new JobStatistics();
    JobStatistics4 extractStats = new JobStatistics4();
    extractStats.setDestinationUriFileCounts(ImmutableList.of(1L));
    extractJobStats.setExtract(extractStats);
    extractJob.setStatus(new JobStatus()).setStatistics(extractJobStats);
    FakeJobService fakeJobService = new FakeJobService();
    FakeDatasetService fakeDatasetService = new FakeDatasetService();
    FakeBigQueryServices fakeBqServices = new FakeBigQueryServices().withJobService(fakeJobService).withDatasetService(fakeDatasetService);
    List<TableRow> expected = ImmutableList.of(new TableRow().set("name", "a").set("number", 1L), new TableRow().set("name", "b").set("number", 2L), new TableRow().set("name", "c").set("number", 3L), new TableRow().set("name", "d").set("number", 4L), new TableRow().set("name", "e").set("number", 5L), new TableRow().set("name", "f").set("number", 6L));
    PipelineOptions options = PipelineOptionsFactory.create();
    BigQueryOptions bqOptions = options.as(BigQueryOptions.class);
    bqOptions.setProject("project");
    String stepUuid = "testStepUuid";
    TableReference tempTableReference = createTempTableReference(bqOptions.getProject(), createJobIdToken(bqOptions.getJobName(), stepUuid));
    fakeDatasetService.createDataset(bqOptions.getProject(), tempTableReference.getDatasetId(), "", "");
    fakeDatasetService.createTable(new Table().setTableReference(tempTableReference).setSchema(new TableSchema().setFields(ImmutableList.of(new TableFieldSchema().setName("name").setType("STRING"), new TableFieldSchema().setName("number").setType("INTEGER")))));
    Path baseDir = Files.createTempDirectory(tempFolder, "testBigQueryQuerySourceInitSplit");
    String query = FakeBigQueryServices.encodeQuery(expected);
    BoundedSource<TableRow> bqSource = BigQueryQuerySource.create(stepUuid, StaticValueProvider.of(query), true, /* flattenResults */
    true, /* useLegacySql */
    fakeBqServices);
    options.setTempLocation(baseDir.toString());
    TableReference queryTable = new TableReference().setProjectId(bqOptions.getProject()).setDatasetId(tempTableReference.getDatasetId()).setTableId(tempTableReference.getTableId());
    fakeJobService.expectDryRunQuery(bqOptions.getProject(), query, new JobStatistics().setQuery(new JobStatistics2().setTotalBytesProcessed(100L).setReferencedTables(ImmutableList.of(queryTable))));
    List<TableRow> read = SourceTestUtils.readFromSource(bqSource, options);
    assertThat(read, containsInAnyOrder(Iterables.toArray(expected, TableRow.class)));
    SourceTestUtils.assertSplitAtFractionBehavior(bqSource, 2, 0.3, ExpectedSplitOutcome.MUST_BE_CONSISTENT_IF_SUCCEEDS, options);
    List<? extends BoundedSource<TableRow>> sources = bqSource.split(100, options);
    assertEquals(2, sources.size());
    BoundedSource<TableRow> actual = sources.get(0);
    assertThat(actual, CoreMatchers.instanceOf(TransformingSource.class));
}
Also used : Path(java.nio.file.Path) JobStatistics(com.google.api.services.bigquery.model.JobStatistics) JobStatistics2(com.google.api.services.bigquery.model.JobStatistics2) HashBasedTable(com.google.common.collect.HashBasedTable) Table(com.google.api.services.bigquery.model.Table) JobStatistics4(com.google.api.services.bigquery.model.JobStatistics4) TableSchema(com.google.api.services.bigquery.model.TableSchema) JsonSchemaToTableSchema(org.apache.beam.sdk.io.gcp.bigquery.BigQueryHelpers.JsonSchemaToTableSchema) BigQueryHelpers.toJsonString(org.apache.beam.sdk.io.gcp.bigquery.BigQueryHelpers.toJsonString) TableFieldSchema(com.google.api.services.bigquery.model.TableFieldSchema) JobStatus(com.google.api.services.bigquery.model.JobStatus) BigQueryHelpers.createTempTableReference(org.apache.beam.sdk.io.gcp.bigquery.BigQueryHelpers.createTempTableReference) TableReference(com.google.api.services.bigquery.model.TableReference) PipelineOptions(org.apache.beam.sdk.options.PipelineOptions) TableRow(com.google.api.services.bigquery.model.TableRow) Job(com.google.api.services.bigquery.model.Job) Test(org.junit.Test)

Aggregations

JobStatus (com.google.api.services.bigquery.model.JobStatus)31 Job (com.google.api.services.bigquery.model.Job)23 TableReference (com.google.api.services.bigquery.model.TableReference)12 TableRow (com.google.api.services.bigquery.model.TableRow)12 JobConfiguration (com.google.api.services.bigquery.model.JobConfiguration)11 Table (com.google.api.services.bigquery.model.Table)10 Test (org.junit.Test)10 JobReference (com.google.api.services.bigquery.model.JobReference)8 JobStatistics (com.google.api.services.bigquery.model.JobStatistics)8 TableSchema (com.google.api.services.bigquery.model.TableSchema)8 ErrorProto (com.google.api.services.bigquery.model.ErrorProto)7 HashBasedTable (com.google.common.collect.HashBasedTable)6 JobStatistics4 (com.google.api.services.bigquery.model.JobStatistics4)5 JobStatistics2 (com.google.api.services.bigquery.model.JobStatistics2)4 IOException (java.io.IOException)4 ResourceId (org.apache.beam.sdk.io.fs.ResourceId)4 CreateDisposition (org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.Write.CreateDisposition)4 WriteDisposition (org.apache.beam.sdk.io.gcp.bigquery.BigQueryIO.Write.WriteDisposition)4 JobServiceImpl (org.apache.beam.sdk.io.gcp.bigquery.BigQueryServicesImpl.JobServiceImpl)4 TableFieldSchema (com.google.api.services.bigquery.model.TableFieldSchema)3