Examples with TypeInfoUtils.getTypeInfosFromTypeString org.apache.hadoop.hive.serde2.typeinfo.TypeInfoUtils.getTypeInfosFromTypeString used on opensource projects

Example 26 with TypeInfoUtils.getTypeInfosFromTypeString

use of org.apache.hadoop.hive.serde2.typeinfo.TypeInfoUtils.getTypeInfosFromTypeString in project hive by apache.

the class TestLazyAccumuloMap method testIntMap.

@Test
public void testIntMap() throws SerDeException, IOException {
    AccumuloHiveRow row = new AccumuloHiveRow("row");
    row.add(new Text("cf1"), new Text("1"), "2".getBytes());
    row.add(new Text("cf1"), new Text("2"), "4".getBytes());
    row.add(new Text("cf1"), new Text("3"), "6".getBytes());
    HiveAccumuloMapColumnMapping mapping = new HiveAccumuloMapColumnMapping("cf1", null, ColumnEncoding.STRING, ColumnEncoding.STRING, "column", TypeInfoFactory.getMapTypeInfo(TypeInfoFactory.intTypeInfo, TypeInfoFactory.intTypeInfo).toString());
    // Map of Integer to Integer
    Text nullSequence = new Text("\\N");
    ObjectInspector oi = LazyFactory.createLazyObjectInspector(TypeInfoUtils.getTypeInfosFromTypeString("map<int,int>").get(0), new byte[] { (byte) 1, (byte) 2 }, 0, nullSequence, false, (byte) 0);
    LazyAccumuloMap map = new LazyAccumuloMap((LazyMapObjectInspector) oi);
    map.init(row, mapping);
    Assert.assertEquals(3, map.getMapSize());
    Object o = map.getMapValueElement(new IntWritable(1));
    Assert.assertNotNull(o);
    Assert.assertEquals(new IntWritable(2), ((LazyInteger) o).getWritableObject());
    o = map.getMapValueElement(new IntWritable(2));
    Assert.assertNotNull(o);
    Assert.assertEquals(new IntWritable(4), ((LazyInteger) o).getWritableObject());
    o = map.getMapValueElement(new IntWritable(3));
    Assert.assertNotNull(o);
    Assert.assertEquals(new IntWritable(6), ((LazyInteger) o).getWritableObject());
}

Example 27 with TypeInfoUtils.getTypeInfosFromTypeString

use of org.apache.hadoop.hive.serde2.typeinfo.TypeInfoUtils.getTypeInfosFromTypeString in project hadoop-pcap by RIPE-NCC.

the class PcapDeserializer method initialize.

@Override
public void initialize(Configuration cfg, Properties props) throws SerDeException {
    String columnNameProperty = props.getProperty(Constants.LIST_COLUMNS);
    columnNames = Arrays.asList(columnNameProperty.split(","));
    numColumns = columnNames.size();
    String columnTypeProperty = props.getProperty(Constants.LIST_COLUMN_TYPES);
    List<TypeInfo> columnTypes = TypeInfoUtils.getTypeInfosFromTypeString(columnTypeProperty);
    // Ensure we have the same number of column names and types
    assert numColumns == columnTypes.size();
    List<ObjectInspector> inspectors = new ArrayList<ObjectInspector>(numColumns);
    row = new ArrayList<Object>(numColumns);
    for (int c = 0; c < numColumns; c++) {
        ObjectInspector oi = TypeInfoUtils.getStandardJavaObjectInspectorFromTypeInfo(columnTypes.get(c));
        inspectors.add(oi);
        row.add(null);
    }
    inspector = ObjectInspectorFactory.getStandardStructObjectInspector(columnNames, inspectors);
}

Example 28 with TypeInfoUtils.getTypeInfosFromTypeString

use of org.apache.hadoop.hive.serde2.typeinfo.TypeInfoUtils.getTypeInfosFromTypeString 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 String segmentGranularity = tableProperties.getProperty(Constants.DRUID_SEGMENT_GRANULARITY) != null ? tableProperties.getProperty(Constants.DRUID_SEGMENT_GRANULARITY) : HiveConf.getVar(jc, HiveConf.ConfVars.HIVE_DRUID_INDEXING_GRANULARITY);
    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(Constants.DRUID_SEGMENT_INTERMEDIATE_DIRECTORY);
    final GranularitySpec granularitySpec = new UniformGranularitySpec(Granularity.fromString(segmentGranularity), Granularity.fromString(tableProperties.getProperty(Constants.DRUID_QUERY_GRANULARITY) == null ? "NONE" : tableProperties.getProperty(Constants.DRUID_QUERY_GRANULARITY)), null);
    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 = new ArrayList<String>();
    for (String name : columnNameProperty.split(",")) {
        columnNames.add(name);
    }
    if (!columnNames.contains(DruidStorageHandlerUtils.DEFAULT_TIMESTAMP_COLUMN)) {
        throw new IllegalStateException("Timestamp column (' " + DruidStorageHandlerUtils.DEFAULT_TIMESTAMP_COLUMN + "') not specified in create table; list of columns is : " + tableProperties.getProperty(serdeConstants.LIST_COLUMNS));
    }
    ArrayList<TypeInfo> columnTypes = TypeInfoUtils.getTypeInfosFromTypeString(columnTypeProperty);
    final boolean approximationAllowed = HiveConf.getBoolVar(jc, HiveConf.ConfVars.HIVE_DRUID_APPROX_RESULT);
    // Default, all columns that are not metrics or timestamp, are treated as dimensions
    final List<DimensionSchema> dimensions = new ArrayList<>();
    ImmutableList.Builder<AggregatorFactory> aggregatorFactoryBuilder = ImmutableList.builder();
    for (int i = 0; i < columnTypes.size(); i++) {
        final PrimitiveObjectInspector.PrimitiveCategory primitiveCategory = ((PrimitiveTypeInfo) columnTypes.get(i)).getPrimitiveCategory();
        AggregatorFactory af;
        switch(primitiveCategory) {
            case BYTE:
            case SHORT:
            case INT:
            case LONG:
                af = new LongSumAggregatorFactory(columnNames.get(i), columnNames.get(i));
                break;
            case FLOAT:
            case DOUBLE:
                af = new DoubleSumAggregatorFactory(columnNames.get(i), columnNames.get(i));
                break;
            case DECIMAL:
                if (approximationAllowed) {
                    af = new DoubleSumAggregatorFactory(columnNames.get(i), columnNames.get(i));
                } else {
                    throw new UnsupportedOperationException(String.format("Druid does not support decimal column type." + "Either cast column [%s] to double or Enable Approximate Result for Druid by setting property [%s] to true", columnNames.get(i), HiveConf.ConfVars.HIVE_DRUID_APPROX_RESULT.varname));
                }
                break;
            case TIMESTAMP:
                // Granularity column
                String tColumnName = columnNames.get(i);
                if (!tColumnName.equals(Constants.DRUID_TIMESTAMP_GRANULARITY_COL_NAME)) {
                    throw new IOException("Dimension " + tColumnName + " does not have STRING type: " + primitiveCategory);
                }
                continue;
            case TIMESTAMPLOCALTZ:
                // Druid timestamp column
                String tLocalTZColumnName = columnNames.get(i);
                if (!tLocalTZColumnName.equals(DruidStorageHandlerUtils.DEFAULT_TIMESTAMP_COLUMN)) {
                    throw new IOException("Dimension " + tLocalTZColumnName + " does not have STRING type: " + primitiveCategory);
                }
                continue;
            default:
                // Dimension
                String dColumnName = columnNames.get(i);
                if (PrimitiveObjectInspectorUtils.getPrimitiveGrouping(primitiveCategory) != PrimitiveGrouping.STRING_GROUP && primitiveCategory != PrimitiveObjectInspector.PrimitiveCategory.BOOLEAN) {
                    throw new IOException("Dimension " + dColumnName + " does not have STRING type: " + primitiveCategory);
                }
                dimensions.add(new StringDimensionSchema(dColumnName));
                continue;
        }
        aggregatorFactoryBuilder.add(af);
    }
    List<AggregatorFactory> aggregatorFactories = aggregatorFactoryBuilder.build();
    final InputRowParser inputRowParser = new MapInputRowParser(new TimeAndDimsParseSpec(new TimestampSpec(DruidStorageHandlerUtils.DEFAULT_TIMESTAMP_COLUMN, "auto", null), new DimensionsSpec(dimensions, Lists.newArrayList(Constants.DRUID_TIMESTAMP_GRANULARITY_COL_NAME, Constants.DRUID_SHARD_KEY_COL_NAME), null)));
    Map<String, Object> inputParser = DruidStorageHandlerUtils.JSON_MAPPER.convertValue(inputRowParser, Map.class);
    final DataSchema dataSchema = new DataSchema(Preconditions.checkNotNull(dataSource, "Data source name is null"), inputParser, aggregatorFactories.toArray(new AggregatorFactory[aggregatorFactories.size()]), granularitySpec, DruidStorageHandlerUtils.JSON_MAPPER);
    final String workingPath = jc.get(Constants.DRUID_JOB_WORKING_DIRECTORY);
    final String version = jc.get(Constants.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;
    if ("concise".equals(HiveConf.getVar(jc, HiveConf.ConfVars.HIVE_DRUID_BITMAP_FACTORY_TYPE))) {
        indexSpec = new IndexSpec(new ConciseBitmapSerdeFactory(), null, null, null);
    } else {
        indexSpec = new IndexSpec(new RoaringBitmapSerdeFactory(true), null, null, null);
    }
    RealtimeTuningConfig realtimeTuningConfig = new RealtimeTuningConfig(maxRowInMemory, null, null, new File(basePersistDirectory, dataSource), new CustomVersioningPolicy(version), null, null, null, indexSpec, true, 0, 0, true, null, 0L);
    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 29 with TypeInfoUtils.getTypeInfosFromTypeString

use of org.apache.hadoop.hive.serde2.typeinfo.TypeInfoUtils.getTypeInfosFromTypeString in project hive by apache.

the class HiveHBaseTableInputFormat method createFilterScan.

/**
 * Converts a filter (which has been pushed down from Hive's optimizer)
 * into corresponding restrictions on the HBase scan.  The
 * filter should already be in a form which can be fully converted.
 *
 * @param jobConf configuration for the scan
 *
 * @param iKey 0-based offset of key column within Hive table
 *
 * @return converted table split if any
 */
private Scan createFilterScan(JobConf jobConf, int iKey, int iTimestamp, boolean isKeyBinary) throws IOException {
    // TODO: assert iKey is HBaseSerDe#HBASE_KEY_COL
    Scan scan = new Scan();
    String filterObjectSerialized = jobConf.get(TableScanDesc.FILTER_OBJECT_CONF_STR);
    if (filterObjectSerialized != null) {
        HiveHBaseInputFormatUtil.setupScanRange(scan, filterObjectSerialized, jobConf, false);
        return scan;
    }
    String filterExprSerialized = jobConf.get(TableScanDesc.FILTER_EXPR_CONF_STR);
    if (filterExprSerialized == null) {
        return scan;
    }
    ExprNodeGenericFuncDesc filterExpr = SerializationUtilities.deserializeExpression(filterExprSerialized);
    String keyColName = jobConf.get(serdeConstants.LIST_COLUMNS).split(",")[iKey];
    ArrayList<TypeInfo> cols = TypeInfoUtils.getTypeInfosFromTypeString(jobConf.get(serdeConstants.LIST_COLUMN_TYPES));
    String colType = cols.get(iKey).getTypeName();
    boolean isKeyComparable = isKeyBinary || colType.equalsIgnoreCase("string");
    String tsColName = null;
    if (iTimestamp >= 0) {
        tsColName = jobConf.get(serdeConstants.LIST_COLUMNS).split(",")[iTimestamp];
    }
    IndexPredicateAnalyzer analyzer = newIndexPredicateAnalyzer(keyColName, isKeyComparable, tsColName);
    List<IndexSearchCondition> conditions = new ArrayList<IndexSearchCondition>();
    ExprNodeDesc residualPredicate = analyzer.analyzePredicate(filterExpr, conditions);
    // THIS IGNORES RESIDUAL PARSING FROM HBaseStorageHandler#decomposePredicate
    if (residualPredicate != null) {
        LOG.debug("Ignoring residual predicate " + residualPredicate.getExprString());
    }
    Map<String, List<IndexSearchCondition>> split = HiveHBaseInputFormatUtil.decompose(conditions);
    List<IndexSearchCondition> keyConditions = split.get(keyColName);
    if (keyConditions != null && !keyConditions.isEmpty()) {
        HiveHBaseInputFormatUtil.setupKeyRange(scan, keyConditions, isKeyBinary);
    }
    List<IndexSearchCondition> tsConditions = split.get(tsColName);
    if (tsConditions != null && !tsConditions.isEmpty()) {
        HiveHBaseInputFormatUtil.setupTimeRange(scan, tsConditions);
    }
    return scan;
}

Example 30 with TypeInfoUtils.getTypeInfosFromTypeString

use of org.apache.hadoop.hive.serde2.typeinfo.TypeInfoUtils.getTypeInfosFromTypeString in project hive by apache.

the class TestLazyHBaseObject method testLazyHBaseRow2.

/**
 * Test the LazyHBaseRow class with a mapping from a Hive field to
 * an HBase column family.
 * @throws SerDeException
 */
public void testLazyHBaseRow2() throws SerDeException {
    // column family is mapped to Map<string,string>
    List<TypeInfo> fieldTypeInfos = TypeInfoUtils.getTypeInfosFromTypeString("string,int,array<string>,map<string,string>,string");
    List<String> fieldNames = Arrays.asList(new String[] { "key", "a", "b", "c", "d" });
    Text nullSequence = new Text("\\N");
    String hbaseColsMapping = ":key,cfa:a,cfa:b,cfb:,cfc:d";
    ColumnMappings columnMappings = null;
    try {
        columnMappings = HBaseSerDe.parseColumnsMapping(hbaseColsMapping);
    } catch (SerDeException e) {
        fail(e.toString());
    }
    for (ColumnMapping colMap : columnMappings) {
        if (!colMap.hbaseRowKey && colMap.qualifierName == null) {
            colMap.binaryStorage.add(false);
            colMap.binaryStorage.add(false);
        } else {
            colMap.binaryStorage.add(false);
        }
    }
    ObjectInspector oi = LazyFactory.createLazyStructInspector(fieldNames, fieldTypeInfos, new byte[] { ' ', ':', '=' }, nullSequence, false, false, (byte) 0);
    LazyHBaseRow o = new LazyHBaseRow((LazySimpleStructObjectInspector) oi, columnMappings);
    List<Cell> kvs = new ArrayList<Cell>();
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfa"), Bytes.toBytes("a"), Bytes.toBytes("123")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfa"), Bytes.toBytes("b"), Bytes.toBytes("a:b:c")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfb"), Bytes.toBytes("d"), Bytes.toBytes("e")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfb"), Bytes.toBytes("f"), Bytes.toBytes("g")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfc"), Bytes.toBytes("d"), Bytes.toBytes("hi")));
    Result r = Result.create(kvs);
    o.init(r);
    assertEquals(("{'key':'test-row','a':123,'b':['a','b','c']," + "'c':{'d':'e','f':'g'},'d':'hi'}").replace("'", "\""), SerDeUtils.getJSONString(o, oi));
    kvs.clear();
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfa"), Bytes.toBytes("a"), Bytes.toBytes("123")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfb"), Bytes.toBytes("d"), Bytes.toBytes("e")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfb"), Bytes.toBytes("f"), Bytes.toBytes("g")));
    r = Result.create(kvs);
    o.init(r);
    assertEquals(("{'key':'test-row','a':123,'b':null," + "'c':{'d':'e','f':'g'},'d':null}").replace("'", "\""), SerDeUtils.getJSONString(o, oi));
    kvs.clear();
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfa"), Bytes.toBytes("b"), Bytes.toBytes("a")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfb"), Bytes.toBytes("f"), Bytes.toBytes("g")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfc"), Bytes.toBytes("d"), Bytes.toBytes("no")));
    r = Result.create(kvs);
    o.init(r);
    assertEquals(("{'key':'test-row','a':null,'b':['a']," + "'c':{'f':'g'},'d':'no'}").replace("'", "\""), SerDeUtils.getJSONString(o, oi));
    kvs.clear();
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfa"), Bytes.toBytes("b"), Bytes.toBytes(":a::")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfc"), Bytes.toBytes("d"), Bytes.toBytes("no")));
    r = Result.create(kvs);
    o.init(r);
    assertEquals(("{'key':'test-row','a':null,'b':['','a','','']," + "'c':{},'d':'no'}").replace("'", "\""), SerDeUtils.getJSONString(o, oi));
    kvs.clear();
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfa"), Bytes.toBytes("a"), Bytes.toBytes("123")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfa"), Bytes.toBytes("b"), Bytes.toBytes("")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfc"), Bytes.toBytes("d"), Bytes.toBytes("")));
    r = Result.create(kvs);
    o.init(r);
    assertEquals("{'key':'test-row','a':123,'b':[],'c':{},'d':''}".replace("'", "\""), SerDeUtils.getJSONString(o, oi));
}