Examples with RowLevelPolicyCheckResults org.apache.gobblin.qualitychecker.row.RowLevelPolicyCheckResults used on opensource projects

Example 1 with RowLevelPolicyCheckResults

use of org.apache.gobblin.qualitychecker.row.RowLevelPolicyCheckResults in project incubator-gobblin by apache.

the class Task method runSynchronousModel.

@Deprecated
private void runSynchronousModel() throws Exception {
    // Get the fork operator. By default IdentityForkOperator is used with a single branch.
    ForkOperator forkOperator = closer.register(this.taskContext.getForkOperator());
    forkOperator.init(this.taskState);
    int branches = forkOperator.getBranches(this.taskState);
    // Set fork.branches explicitly here so the rest task flow can pick it up
    this.taskState.setProp(ConfigurationKeys.FORK_BRANCHES_KEY, branches);
    // Extract, convert, and fork the source schema.
    Object schema = converter.convertSchema(extractor.getSchema(), this.taskState);
    List<Boolean> forkedSchemas = forkOperator.forkSchema(this.taskState, schema);
    if (forkedSchemas.size() != branches) {
        throw new ForkBranchMismatchException(String.format("Number of forked schemas [%d] is not equal to number of branches [%d]", forkedSchemas.size(), branches));
    }
    if (inMultipleBranches(forkedSchemas) && !(CopyHelper.isCopyable(schema))) {
        throw new CopyNotSupportedException(schema + " is not copyable");
    }
    RowLevelPolicyCheckResults rowResults = new RowLevelPolicyCheckResults();
    if (!areSingleBranchTasksSynchronous(this.taskContext) || branches > 1) {
        // Create one fork for each forked branch
        for (int i = 0; i < branches; i++) {
            if (forkedSchemas.get(i)) {
                AsynchronousFork fork = closer.register(new AsynchronousFork(this.taskContext, schema instanceof Copyable ? ((Copyable) schema).copy() : schema, branches, i, this.taskMode));
                configureStreamingFork(fork, watermarkingStrategy);
                // Run the Fork
                this.forks.put(Optional.<Fork>of(fork), Optional.<Future<?>>of(this.taskExecutor.submit(fork)));
            } else {
                this.forks.put(Optional.<Fork>absent(), Optional.<Future<?>>absent());
            }
        }
    } else {
        SynchronousFork fork = closer.register(new SynchronousFork(this.taskContext, schema instanceof Copyable ? ((Copyable) schema).copy() : schema, branches, 0, this.taskMode));
        configureStreamingFork(fork, watermarkingStrategy);
        this.forks.put(Optional.<Fork>of(fork), Optional.<Future<?>>of(this.taskExecutor.submit(fork)));
    }
    if (isStreamingTask()) {
        // Start watermark manager and tracker
        if (this.watermarkTracker.isPresent()) {
            this.watermarkTracker.get().start();
        }
        this.watermarkManager.get().start();
        ((StreamingExtractor) this.taskContext.getRawSourceExtractor()).start(this.watermarkStorage.get());
        RecordEnvelope recordEnvelope;
        // Extract, convert, and fork one source record at a time.
        while (!shutdownRequested() && (recordEnvelope = extractor.readRecordEnvelope()) != null) {
            onRecordExtract();
            AcknowledgableWatermark ackableWatermark = new AcknowledgableWatermark(recordEnvelope.getWatermark());
            if (watermarkTracker.isPresent()) {
                watermarkTracker.get().track(ackableWatermark);
            }
            for (Object convertedRecord : converter.convertRecord(schema, recordEnvelope, this.taskState)) {
                processRecord(convertedRecord, forkOperator, rowChecker, rowResults, branches, ackableWatermark.incrementAck());
            }
            ackableWatermark.ack();
        }
    } else {
        RecordEnvelope record;
        // Extract, convert, and fork one source record at a time.
        long errRecords = 0;
        while ((record = extractor.readRecordEnvelope()) != null) {
            onRecordExtract();
            try {
                for (Object convertedRecord : converter.convertRecord(schema, record.getRecord(), this.taskState)) {
                    processRecord(convertedRecord, forkOperator, rowChecker, rowResults, branches, null);
                }
            } catch (Exception e) {
                if (!(e instanceof DataConversionException) && !(e.getCause() instanceof DataConversionException)) {
                    LOG.error("Processing record incurs an unexpected exception: ", e);
                    throw new RuntimeException(e.getCause());
                }
                errRecords++;
                if (errRecords > this.taskState.getPropAsLong(TaskConfigurationKeys.TASK_SKIP_ERROR_RECORDS, TaskConfigurationKeys.DEFAULT_TASK_SKIP_ERROR_RECORDS)) {
                    throw new RuntimeException(e);
                }
            }
        }
    }
    LOG.info("Extracted " + this.recordsPulled + " data records");
    LOG.info("Row quality checker finished with results: " + rowResults.getResults());
    this.taskState.setProp(ConfigurationKeys.EXTRACTOR_ROWS_EXTRACTED, this.recordsPulled);
    this.taskState.setProp(ConfigurationKeys.EXTRACTOR_ROWS_EXPECTED, extractor.getExpectedRecordCount());
    for (Optional<Fork> fork : this.forks.keySet()) {
        if (fork.isPresent()) {
            // Tell the fork that the main branch is completed and no new incoming data records should be expected
            fork.get().markParentTaskDone();
        }
    }
    for (Optional<Future<?>> forkFuture : this.forks.values()) {
        if (forkFuture.isPresent()) {
            try {
                long forkFutureStartTime = System.nanoTime();
                forkFuture.get().get();
                long forkDuration = System.nanoTime() - forkFutureStartTime;
                LOG.info("Task shutdown: Fork future reaped in {} millis", forkDuration / 1000000);
            } catch (InterruptedException ie) {
                Thread.currentThread().interrupt();
            }
        }
    }
}

Example 2 with RowLevelPolicyCheckResults

use of org.apache.gobblin.qualitychecker.row.RowLevelPolicyCheckResults in project incubator-gobblin by apache.

the class RowLevelQualityCheckerTest method testRowLevelPolicy.

@Test(groups = { "ignore" })
public void testRowLevelPolicy() throws Exception {
    State state = new State();
    state.setProp(ConfigurationKeys.ROW_LEVEL_POLICY_LIST, "org.apache.gobblin.qualitychecker.TestRowLevelPolicy");
    state.setProp(ConfigurationKeys.ROW_LEVEL_POLICY_LIST_TYPE, "FAIL");
    RowLevelPolicyChecker checker = new RowLevelPolicyCheckerBuilderFactory().newPolicyCheckerBuilder(state, -1).build();
    RowLevelPolicyCheckResults results = new RowLevelPolicyCheckResults();
    FileReader<GenericRecord> fileReader = openFile(state);
    for (GenericRecord datum : fileReader) {
        Assert.assertTrue(checker.executePolicies(datum, results));
    }
}

Example 3 with RowLevelPolicyCheckResults

use of org.apache.gobblin.qualitychecker.row.RowLevelPolicyCheckResults in project incubator-gobblin by apache.

the class RowLevelQualityCheckerTest method testWriteToErrFile.

@Test(groups = { "ignore" })
public void testWriteToErrFile() throws Exception {
    State state = new State();
    state.setProp(ConfigurationKeys.ROW_LEVEL_POLICY_LIST, "org.apache.gobblin.qualitychecker.TestRowLevelPolicyFail");
    state.setProp(ConfigurationKeys.ROW_LEVEL_POLICY_LIST_TYPE, "ERR_FILE");
    state.setProp(ConfigurationKeys.ROW_LEVEL_ERR_FILE, TestConstants.TEST_ERR_FILE);
    state.setProp(ConfigurationKeys.WRITER_FILE_SYSTEM_URI, TestConstants.TEST_FS_URI);
    RowLevelPolicyChecker checker = new RowLevelPolicyCheckerBuilderFactory().newPolicyCheckerBuilder(state, -1).build();
    RowLevelPolicyCheckResults results = new RowLevelPolicyCheckResults();
    FileReader<GenericRecord> fileReader = openFile(state);
    for (GenericRecord datum : fileReader) {
        Assert.assertFalse(checker.executePolicies(datum, results));
    }
    FileSystem fs = FileSystem.get(new URI(TestConstants.TEST_FS_URI), new Configuration());
    Path outputPath = new Path(TestConstants.TEST_ERR_FILE, state.getProp(ConfigurationKeys.ROW_LEVEL_POLICY_LIST).replaceAll("\\.", "-") + ".err");
    Assert.assertTrue(fs.exists(outputPath));
    fs.delete(new Path(TestConstants.TEST_ERR_FILE), true);
}