Examples with TimeAndDimsParseSpec org.apache.druid.data.input.impl.TimeAndDimsParseSpec used on opensource projects

Example 6 with TimeAndDimsParseSpec

use of org.apache.druid.data.input.impl.TimeAndDimsParseSpec in project hive by apache.

the class TestDruidRecordWriter method testWrite.

// Test is failing due to Guava dependency, Druid 0.13.0 should have less dependency on Guava
@Ignore
@Test
public void testWrite() throws IOException, SegmentLoadingException {
    final String dataSourceName = "testDataSource";
    final File segmentOutputDir = temporaryFolder.newFolder();
    final File workingDir = temporaryFolder.newFolder();
    Configuration config = new Configuration();
    final InputRowParser inputRowParser = new MapInputRowParser(new TimeAndDimsParseSpec(new TimestampSpec(DruidConstants.DEFAULT_TIMESTAMP_COLUMN, "auto", null), new DimensionsSpec(ImmutableList.of(new StringDimensionSchema("host")), null, null)));
    final Map<String, Object> parserMap = objectMapper.convertValue(inputRowParser, new TypeReference<Map<String, Object>>() {
    });
    DataSchema dataSchema = new DataSchema(dataSourceName, parserMap, new AggregatorFactory[] { new LongSumAggregatorFactory("visited_sum", "visited_sum"), new HyperUniquesAggregatorFactory("unique_hosts", "unique_hosts") }, new UniformGranularitySpec(Granularities.DAY, Granularities.NONE, ImmutableList.of(INTERVAL_FULL)), null, objectMapper);
    IndexSpec indexSpec = new IndexSpec(new RoaringBitmapSerdeFactory(true), null, null, null);
    RealtimeTuningConfig tuningConfig = new RealtimeTuningConfig(null, null, null, null, temporaryFolder.newFolder(), null, null, null, null, indexSpec, null, null, 0, 0, null, null, 0L, null, null);
    LocalFileSystem localFileSystem = FileSystem.getLocal(config);
    DataSegmentPusher dataSegmentPusher = new LocalDataSegmentPusher(new LocalDataSegmentPusherConfig() {

        @Override
        public File getStorageDirectory() {
            return segmentOutputDir;
        }
    });
    Path segmentDescriptorPath = new Path(workingDir.getAbsolutePath(), DruidStorageHandler.SEGMENTS_DESCRIPTOR_DIR_NAME);
    DruidRecordWriter druidRecordWriter = new DruidRecordWriter(dataSchema, tuningConfig, dataSegmentPusher, 20, segmentDescriptorPath, localFileSystem);
    List<DruidWritable> druidWritables = expectedRows.stream().map(input -> new DruidWritable(ImmutableMap.<String, Object>builder().putAll(input).put(Constants.DRUID_TIMESTAMP_GRANULARITY_COL_NAME, Granularities.DAY.bucketStart(new DateTime((long) input.get(DruidConstants.DEFAULT_TIMESTAMP_COLUMN))).getMillis()).build())).collect(Collectors.toList());
    for (DruidWritable druidWritable : druidWritables) {
        druidRecordWriter.write(druidWritable);
    }
    druidRecordWriter.close(false);
    List<DataSegment> dataSegmentList = DruidStorageHandlerUtils.getCreatedSegments(segmentDescriptorPath, config);
    Assert.assertEquals(1, dataSegmentList.size());
    File tmpUnzippedSegmentDir = temporaryFolder.newFolder();
    new LocalDataSegmentPuller().getSegmentFiles(dataSegmentList.get(0), tmpUnzippedSegmentDir);
    final QueryableIndex queryableIndex = DruidStorageHandlerUtils.INDEX_IO.loadIndex(tmpUnzippedSegmentDir);
    QueryableIndexStorageAdapter adapter = new QueryableIndexStorageAdapter(queryableIndex);
    Firehose firehose = new IngestSegmentFirehose(ImmutableList.of(new WindowedStorageAdapter(adapter, adapter.getInterval())), null, ImmutableList.of("host"), ImmutableList.of("visited_sum", "unique_hosts"), null);
    List<InputRow> rows = Lists.newArrayList();
    while (firehose.hasMore()) {
        rows.add(firehose.nextRow());
    }
    verifyRows(expectedRows, rows);
}

Example 7 with TimeAndDimsParseSpec

use of org.apache.druid.data.input.impl.TimeAndDimsParseSpec in project hive by apache.

the class DruidOutputFormat method getHiveRecordWriter.

@Override
public FileSinkOperator.RecordWriter getHiveRecordWriter(JobConf jc, Path finalOutPath, Class<? extends Writable> valueClass, boolean isCompressed, Properties tableProperties, Progressable progress) throws IOException {
    final int targetNumShardsPerGranularity = Integer.parseUnsignedInt(tableProperties.getProperty(Constants.DRUID_TARGET_SHARDS_PER_GRANULARITY, "0"));
    final int maxPartitionSize = targetNumShardsPerGranularity > 0 ? -1 : HiveConf.getIntVar(jc, HiveConf.ConfVars.HIVE_DRUID_MAX_PARTITION_SIZE);
    // If datasource is in the table properties, it is an INSERT/INSERT OVERWRITE as the datasource
    // name was already persisted. Otherwise, it is a CT/CTAS and we need to get the name from the
    // job properties that are set by configureOutputJobProperties in the DruidStorageHandler
    final String dataSource = tableProperties.getProperty(Constants.DRUID_DATA_SOURCE) == null ? jc.get(Constants.DRUID_DATA_SOURCE) : tableProperties.getProperty(Constants.DRUID_DATA_SOURCE);
    final String segmentDirectory = jc.get(DruidConstants.DRUID_SEGMENT_INTERMEDIATE_DIRECTORY);
    final GranularitySpec granularitySpec = DruidStorageHandlerUtils.getGranularitySpec(jc, tableProperties);
    final String columnNameProperty = tableProperties.getProperty(serdeConstants.LIST_COLUMNS);
    final String columnTypeProperty = tableProperties.getProperty(serdeConstants.LIST_COLUMN_TYPES);
    if (StringUtils.isEmpty(columnNameProperty) || StringUtils.isEmpty(columnTypeProperty)) {
        throw new IllegalStateException(String.format("List of columns names [%s] or columns type [%s] is/are not present", columnNameProperty, columnTypeProperty));
    }
    ArrayList<String> columnNames = Lists.newArrayList(columnNameProperty.split(","));
    if (!columnNames.contains(DruidConstants.DEFAULT_TIMESTAMP_COLUMN)) {
        throw new IllegalStateException("Timestamp column (' " + DruidConstants.DEFAULT_TIMESTAMP_COLUMN + "') not specified in create table; list of columns is : " + tableProperties.getProperty(serdeConstants.LIST_COLUMNS));
    }
    ArrayList<TypeInfo> columnTypes = TypeInfoUtils.getTypeInfosFromTypeString(columnTypeProperty);
    Pair<List<DimensionSchema>, AggregatorFactory[]> dimensionsAndAggregates = DruidStorageHandlerUtils.getDimensionsAndAggregates(columnNames, columnTypes);
    final InputRowParser inputRowParser = new MapInputRowParser(new TimeAndDimsParseSpec(new TimestampSpec(DruidConstants.DEFAULT_TIMESTAMP_COLUMN, "auto", null), new DimensionsSpec(dimensionsAndAggregates.lhs, Lists.newArrayList(Constants.DRUID_TIMESTAMP_GRANULARITY_COL_NAME, Constants.DRUID_SHARD_KEY_COL_NAME), null)));
    Map<String, Object> inputParser = DruidStorageHandlerUtils.JSON_MAPPER.convertValue(inputRowParser, new TypeReference<Map<String, Object>>() {
    });
    final DataSchema dataSchema = new DataSchema(Preconditions.checkNotNull(dataSource, "Data source name is null"), inputParser, dimensionsAndAggregates.rhs, granularitySpec, null, DruidStorageHandlerUtils.JSON_MAPPER);
    final String workingPath = jc.get(DruidConstants.DRUID_JOB_WORKING_DIRECTORY);
    final String version = jc.get(DruidConstants.DRUID_SEGMENT_VERSION);
    String basePersistDirectory = HiveConf.getVar(jc, HiveConf.ConfVars.HIVE_DRUID_BASE_PERSIST_DIRECTORY);
    if (Strings.isNullOrEmpty(basePersistDirectory)) {
        basePersistDirectory = System.getProperty("java.io.tmpdir");
    }
    Integer maxRowInMemory = HiveConf.getIntVar(jc, HiveConf.ConfVars.HIVE_DRUID_MAX_ROW_IN_MEMORY);
    IndexSpec indexSpec = DruidStorageHandlerUtils.getIndexSpec(jc);
    RealtimeTuningConfig realtimeTuningConfig = new RealtimeTuningConfig(maxRowInMemory, null, null, null, new File(basePersistDirectory, dataSource), new CustomVersioningPolicy(version), null, null, null, indexSpec, null, true, 0, 0, true, null, 0L, null, null);
    LOG.debug(String.format("running with Data schema [%s] ", dataSchema));
    return new DruidRecordWriter(dataSchema, realtimeTuningConfig, DruidStorageHandlerUtils.createSegmentPusherForDirectory(segmentDirectory, jc), maxPartitionSize, new Path(workingPath, SEGMENTS_DESCRIPTOR_DIR_NAME), finalOutPath.getFileSystem(jc));
}

Example 8 with TimeAndDimsParseSpec

use of org.apache.druid.data.input.impl.TimeAndDimsParseSpec in project druid by druid-io.

the class RealtimeIndexTaskTest method makeRealtimeTask.

private RealtimeIndexTask makeRealtimeTask(final String taskId, final TransformSpec transformSpec, final boolean reportParseExceptions, final long handoffTimeout) {
    ObjectMapper objectMapper = new DefaultObjectMapper();
    DataSchema dataSchema = new DataSchema("test_ds", TestHelper.makeJsonMapper().convertValue(new MapInputRowParser(new TimeAndDimsParseSpec(new TimestampSpec("t", "auto", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.of("dim1", "dim2", "dim1t"))))), JacksonUtils.TYPE_REFERENCE_MAP_STRING_OBJECT), new AggregatorFactory[] { new CountAggregatorFactory("rows"), new LongSumAggregatorFactory("met1", "met1") }, new UniformGranularitySpec(Granularities.DAY, Granularities.NONE, null), transformSpec, objectMapper);
    RealtimeIOConfig realtimeIOConfig = new RealtimeIOConfig(new TestFirehose.TestFirehoseFactory(), null);
    RealtimeTuningConfig realtimeTuningConfig = new RealtimeTuningConfig(null, 1000, null, null, new Period("P1Y"), new Period("PT10M"), null, null, new ServerTimeRejectionPolicyFactory(), null, null, null, null, 0, 0, reportParseExceptions, handoffTimeout, null, null, null);
    return new RealtimeIndexTask(taskId, null, new FireDepartment(dataSchema, realtimeIOConfig, realtimeTuningConfig), null) {

        @Override
        protected boolean isFirehoseDrainableByClosing(FirehoseFactory firehoseFactory) {
            return true;
        }
    };
}

Example 9 with TimeAndDimsParseSpec

use of org.apache.druid.data.input.impl.TimeAndDimsParseSpec in project druid by druid-io.

the class BloomFilterSqlAggregatorTest method createQuerySegmentWalker.

@Override
public SpecificSegmentsQuerySegmentWalker createQuerySegmentWalker() throws IOException {
    InputRowParser parser = new MapInputRowParser(new TimeAndDimsParseSpec(new TimestampSpec("t", "iso", null), new DimensionsSpec(ImmutableList.<DimensionSchema>builder().addAll(DimensionsSpec.getDefaultSchemas(ImmutableList.of("dim1", "dim2", "dim3"))).add(new DoubleDimensionSchema("d1")).add(new FloatDimensionSchema("f1")).add(new LongDimensionSchema("l1")).build())));
    final QueryableIndex index = IndexBuilder.create().tmpDir(temporaryFolder.newFolder()).segmentWriteOutMediumFactory(OffHeapMemorySegmentWriteOutMediumFactory.instance()).schema(new IncrementalIndexSchema.Builder().withMetrics(new CountAggregatorFactory("cnt"), new DoubleSumAggregatorFactory("m1", "m1")).withDimensionsSpec(parser).withRollup(false).build()).rows(CalciteTests.ROWS1_WITH_NUMERIC_DIMS).buildMMappedIndex();
    return new SpecificSegmentsQuerySegmentWalker(conglomerate).add(DataSegment.builder().dataSource(DATA_SOURCE).interval(index.getDataInterval()).version("1").shardSpec(new LinearShardSpec(0)).size(0).build(), index);
}

Example 10 with TimeAndDimsParseSpec

use of org.apache.druid.data.input.impl.TimeAndDimsParseSpec in project druid by druid-io.

the class SqlFirehoseTest method testFirehoseTransformingParser.

@Test
public void testFirehoseTransformingParser() throws Exception {
    final TestCloseable closeable = new TestCloseable();
    List<Object> expectedResults = new ArrayList<>();
    for (Map<String, Object> map : inputs) {
        expectedResults.add(map.get("x") + "foo");
    }
    final List<JsonIterator<Map<String, Object>>> lineIterators = fileList.stream().map(s -> new JsonIterator<Map<String, Object>>(TYPE_REF, s, closeable, objectMapper)).collect(Collectors.toList());
    final InputRowParser stringParser = new TransformingStringInputRowParser(new TimeAndDimsParseSpec(new TimestampSpec("timestamp", "auto", null), new DimensionsSpec(DimensionsSpec.getDefaultSchemas(ImmutableList.of("x")))), Charset.defaultCharset().name(), new TransformSpec(null, ImmutableList.of(new ExpressionTransform("xfoo", "concat(x,'foo')", ExprMacroTable.nil()))));
    try (final SqlFirehose firehose = new SqlFirehose(lineIterators.iterator(), stringParser, closeable)) {
        final List<Object> results = new ArrayList<>();
        while (firehose.hasMore()) {
            final InputRow inputRow = firehose.nextRow();
            if (inputRow == null) {
                results.add(null);
            } else {
                results.add(inputRow.getDimension("xfoo").get(0));
            }
        }
        Assert.assertEquals(expectedResults, results);
    }
}