Search in sources :

Example 16 with LazySerDeParameters

use of org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters 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 17 with LazySerDeParameters

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

the class TestLazySimpleSerDe method serializeAndDeserialize.

private Object serializeAndDeserialize(Object row, StructObjectInspector rowOI, LazySimpleSerDe serde, LazySerDeParameters serdeParams) throws IOException, SerDeException {
    ByteStream.Output serializeStream = new ByteStream.Output();
    LazySimpleSerDe.serialize(serializeStream, row, rowOI, serdeParams.getSeparators(), 0, serdeParams.getNullSequence(), serdeParams.isEscaped(), serdeParams.getEscapeChar(), serdeParams.getNeedsEscape());
    Text t = new Text(serializeStream.toByteArray());
    return serde.deserialize(t);
}
Also used : ByteStream(org.apache.hadoop.hive.serde2.ByteStream) Text(org.apache.hadoop.io.Text)

Example 18 with LazySerDeParameters

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

the class TestCrossMapEqualComparer method serializeAndDeserialize.

Object serializeAndDeserialize(TextStringMapHolder o1, StructObjectInspector oi1, LazySimpleSerDe serde, LazySerDeParameters serdeParams) throws IOException, SerDeException {
    ByteStream.Output serializeStream = new ByteStream.Output();
    LazySimpleSerDe.serialize(serializeStream, o1, oi1, serdeParams.getSeparators(), 0, serdeParams.getNullSequence(), serdeParams.isEscaped(), serdeParams.getEscapeChar(), serdeParams.getNeedsEscape());
    Text t = new Text(serializeStream.toByteArray());
    return serde.deserialize(t);
}
Also used : ByteStream(org.apache.hadoop.hive.serde2.ByteStream) Text(org.apache.hadoop.io.Text)

Example 19 with LazySerDeParameters

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

the class TestSimpleMapEqualComparer method serializeAndDeserialize.

Object serializeAndDeserialize(TextStringMapHolder o1, StructObjectInspector oi1, LazySimpleSerDe serde, LazySerDeParameters serdeParams) throws IOException, SerDeException {
    ByteStream.Output serializeStream = new ByteStream.Output();
    LazySimpleSerDe.serialize(serializeStream, o1, oi1, serdeParams.getSeparators(), 0, serdeParams.getNullSequence(), serdeParams.isEscaped(), serdeParams.getEscapeChar(), serdeParams.getNeedsEscape());
    Text t = new Text(serializeStream.toByteArray());
    return serde.deserialize(t);
}
Also used : ByteStream(org.apache.hadoop.hive.serde2.ByteStream) Text(org.apache.hadoop.io.Text)

Example 20 with LazySerDeParameters

use of org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters in project cdap by caskdata.

the class SimpleMapEqualComparerTest method serializeAndDeserialize.

Object serializeAndDeserialize(StringTextMapHolder o1, StructObjectInspector oi1, LazySimpleSerDe serde, LazySerDeParameters serdeParams) throws IOException, SerDeException {
    ByteStream.Output serializeStream = new ByteStream.Output();
    LazySimpleSerDe.serialize(serializeStream, o1, oi1, serdeParams.getSeparators(), 0, serdeParams.getNullSequence(), serdeParams.isEscaped(), serdeParams.getEscapeChar(), serdeParams.getNeedsEscape());
    Text t = new Text(serializeStream.toByteArray());
    return serde.deserialize(t);
}
Also used : ByteStream(org.apache.hadoop.hive.serde2.ByteStream) Text(org.apache.hadoop.io.Text)

Aggregations

LazySerDeParameters (org.apache.hadoop.hive.serde2.lazy.LazySerDeParameters)26 Text (org.apache.hadoop.io.Text)20 Configuration (org.apache.hadoop.conf.Configuration)19 Properties (java.util.Properties)18 TypeInfo (org.apache.hadoop.hive.serde2.typeinfo.TypeInfo)16 LazySimpleStructObjectInspector (org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector)14 Test (org.junit.Test)12 ObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector)10 StructObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector)10 Mutation (org.apache.accumulo.core.data.Mutation)8 ByteArrayRef (org.apache.hadoop.hive.serde2.lazy.ByteArrayRef)8 LazyStruct (org.apache.hadoop.hive.serde2.lazy.LazyStruct)8 ArrayList (java.util.ArrayList)7 ByteStream (org.apache.hadoop.hive.serde2.ByteStream)7 LazySimpleSerDe (org.apache.hadoop.hive.serde2.lazy.LazySimpleSerDe)7 LazySimpleDeserializeRead (org.apache.hadoop.hive.serde2.lazy.fast.LazySimpleDeserializeRead)6 LazySimpleSerializeWrite (org.apache.hadoop.hive.serde2.lazy.fast.LazySimpleSerializeWrite)6 Entry (java.util.Map.Entry)5 PrimitiveTypeInfo (org.apache.hadoop.hive.serde2.typeinfo.PrimitiveTypeInfo)5 Connector (org.apache.accumulo.core.client.Connector)4