Search in sources :

Example 1 with LazyUnion

use of org.apache.hadoop.hive.serde2.lazy.LazyUnion in project hive by apache.

the class TestLazyArrayMapStruct method testNestedinArrayAtLevel.

/**
   * @param nestingLevel
   * @param dtype
   * @param tableProp
   * @throws SerDeException
   */
private void testNestedinArrayAtLevel(int nestingLevel, ObjectInspector.Category dtype, Properties tableProp) throws SerDeException {
    //create type with nestingLevel levels of nesting
    //set inner schema for dtype
    String inSchema = null;
    switch(dtype) {
        case LIST:
            inSchema = "array<tinyint>";
            break;
        case MAP:
            inSchema = "map<string,int>";
            break;
        case STRUCT:
            inSchema = "struct<s:string,i:tinyint>";
            break;
        case UNION:
            inSchema = "uniontype<string,tinyint>";
            break;
        default:
            fail("type not supported by test case");
    }
    StringBuilder schema = new StringBuilder(inSchema);
    for (int i = 0; i < nestingLevel - 1; i++) {
        schema.insert(0, "array<");
        schema.append(">");
    }
    System.err.println("Testing nesting level " + nestingLevel + ". Using schema " + schema);
    // Create the SerDe
    LazySimpleSerDe serDe = new LazySimpleSerDe();
    Configuration conf = new Configuration();
    tableProp.setProperty("columns", "narray");
    tableProp.setProperty("columns.types", schema.toString());
    SerDeUtils.initializeSerDe(serDe, conf, tableProp, null);
    LazySerDeParameters serdeParams = new LazySerDeParameters(conf, tableProp, LazySimpleSerDe.class.getName());
    //create the serialized string for type
    byte[] separators = serdeParams.getSeparators();
    System.err.println("Using separator " + (char) separators[nestingLevel]);
    byte[] serializedRow = null;
    switch(dtype) {
        case LIST:
            serializedRow = new byte[] { '8', separators[nestingLevel], '9' };
            break;
        case MAP:
            byte kvSep = separators[nestingLevel + 1];
            byte kvPairSep = separators[nestingLevel];
            serializedRow = new byte[] { '1', kvSep, '1', kvPairSep, '2', kvSep, '2' };
            break;
        case STRUCT:
            serializedRow = new byte[] { '8', separators[nestingLevel], '9' };
            break;
        case UNION:
            serializedRow = new byte[] { '0', separators[nestingLevel], '9' };
            break;
        default:
            fail("type not supported by test case");
    }
    //create LazyStruct with serialized string with expected separators
    StructObjectInspector oi = (StructObjectInspector) serDe.getObjectInspector();
    LazyStruct struct = (LazyStruct) LazyFactory.createLazyObject(oi);
    TestLazyPrimitive.initLazyObject(struct, serializedRow, 0, serializedRow.length);
    //Get fields out of the lazy struct and check if they match expected
    // results
    //Get first level array
    LazyArray array = (LazyArray) struct.getField(0);
    //Peel off the n-1 levels to get to the underlying array
    for (int i = 0; i < nestingLevel - 2; i++) {
        array = (LazyArray) array.getListElementObject(0);
    }
    //verify the serialized format for dtype
    switch(dtype) {
        case LIST:
            LazyArray array1 = (LazyArray) array.getListElementObject(0);
            //check elements of the innermost array
            assertEquals(2, array1.getListLength());
            assertEquals(new ByteWritable((byte) 8), ((LazyByte) array1.getListElementObject(0)).getWritableObject());
            assertEquals(new ByteWritable((byte) 9), ((LazyByte) array1.getListElementObject(1)).getWritableObject());
            break;
        case MAP:
            LazyMap lazyMap = (LazyMap) array.getListElementObject(0);
            Map map = lazyMap.getMap();
            System.err.println(map);
            assertEquals(2, map.size());
            Iterator<Map.Entry<LazyString, LazyInteger>> it = map.entrySet().iterator();
            Entry<LazyString, LazyInteger> e1 = it.next();
            assertEquals(e1.getKey().getWritableObject(), new Text(new byte[] { '1' }));
            assertEquals(e1.getValue().getWritableObject(), new IntWritable(1));
            Entry<LazyString, LazyInteger> e2 = it.next();
            assertEquals(e2.getKey().getWritableObject(), new Text(new byte[] { '2' }));
            assertEquals(e2.getValue().getWritableObject(), new IntWritable(2));
            break;
        case STRUCT:
            LazyStruct innerStruct = (LazyStruct) array.getListElementObject(0);
            //check elements of the innermost struct
            assertEquals(2, innerStruct.getFieldsAsList().size());
            assertEquals(new Text(new byte[] { '8' }), ((LazyString) innerStruct.getField(0)).getWritableObject());
            assertEquals(new ByteWritable((byte) 9), ((LazyByte) innerStruct.getField(1)).getWritableObject());
            break;
        case UNION:
            LazyUnion lazyUnion = (LazyUnion) array.getListElementObject(0);
            //check elements of the innermost union
            assertEquals(new Text(new byte[] { '9' }), ((LazyString) lazyUnion.getField()).getWritableObject());
            break;
        default:
            fail("type not supported by test case");
    }
    //test serialization
    Text serializedText = (Text) serDe.serialize(struct.getObject(), serDe.getObjectInspector());
    org.junit.Assert.assertArrayEquals(serializedRow, serializedText.getBytes());
}
Also used : Configuration(org.apache.hadoop.conf.Configuration) Text(org.apache.hadoop.io.Text) Entry(java.util.Map.Entry) ByteWritable(org.apache.hadoop.hive.serde2.io.ByteWritable) LazyBinaryMap(org.apache.hadoop.hive.serde2.lazybinary.LazyBinaryMap) Map(java.util.Map) IntWritable(org.apache.hadoop.io.IntWritable) StructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector)

Example 2 with LazyUnion

use of org.apache.hadoop.hive.serde2.lazy.LazyUnion in project hive by apache.

the class AvroLazyObjectInspector method toLazyUnionObject.

/**
   * Convert the given object to a lazy object using the given {@link ObjectInspector}
   *
   * @param obj Object to be converted to a {@link LazyObject}
   * @param oi ObjectInspector used for the conversion
   * @return the created {@link LazyObject lazy object}
   * */
private Object toLazyUnionObject(Object obj, ObjectInspector objectInspector) {
    if (obj == null) {
        return null;
    }
    if (!(objectInspector instanceof LazyUnionObjectInspector)) {
        throw new IllegalArgumentException("Invalid objectinspector found. Expected LazyUnionObjectInspector, Found " + objectInspector.getClass());
    }
    StandardUnion standardUnion = (StandardUnion) obj;
    LazyUnionObjectInspector lazyUnionOI = (LazyUnionObjectInspector) objectInspector;
    // Grab the tag and the field
    byte tag = standardUnion.getTag();
    Object field = standardUnion.getObject();
    ObjectInspector fieldOI = lazyUnionOI.getObjectInspectors().get(tag);
    // convert to lazy object
    Object convertedObj = null;
    if (field != null) {
        convertedObj = toLazyObject(field, fieldOI);
    }
    if (convertedObj == null) {
        return null;
    }
    return new LazyUnion(lazyUnionOI, tag, convertedObj);
}
Also used : LazyUnionObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazyUnionObjectInspector) LazyUnionObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazyUnionObjectInspector) ListObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ListObjectInspector) LazyListObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazyListObjectInspector) LazySimpleStructObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector) ObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector) LazyMapObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazyMapObjectInspector) MapObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector) StandardUnion(org.apache.hadoop.hive.serde2.objectinspector.StandardUnionObjectInspector.StandardUnion) LazyObject(org.apache.hadoop.hive.serde2.lazy.LazyObject) LazyUnion(org.apache.hadoop.hive.serde2.lazy.LazyUnion)

Aggregations

Map (java.util.Map)1 Entry (java.util.Map.Entry)1 Configuration (org.apache.hadoop.conf.Configuration)1 ByteWritable (org.apache.hadoop.hive.serde2.io.ByteWritable)1 LazyObject (org.apache.hadoop.hive.serde2.lazy.LazyObject)1 LazyUnion (org.apache.hadoop.hive.serde2.lazy.LazyUnion)1 LazyListObjectInspector (org.apache.hadoop.hive.serde2.lazy.objectinspector.LazyListObjectInspector)1 LazyMapObjectInspector (org.apache.hadoop.hive.serde2.lazy.objectinspector.LazyMapObjectInspector)1 LazySimpleStructObjectInspector (org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector)1 LazyUnionObjectInspector (org.apache.hadoop.hive.serde2.lazy.objectinspector.LazyUnionObjectInspector)1 LazyBinaryMap (org.apache.hadoop.hive.serde2.lazybinary.LazyBinaryMap)1 ListObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ListObjectInspector)1 MapObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector)1 ObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector)1 StandardUnion (org.apache.hadoop.hive.serde2.objectinspector.StandardUnionObjectInspector.StandardUnion)1 StructObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector)1 IntWritable (org.apache.hadoop.io.IntWritable)1 Text (org.apache.hadoop.io.Text)1