Examples with LazySimpleStructObjectInspector org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector used on opensource projects

Example 6 with LazySimpleStructObjectInspector

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

the class TestHiveAccumuloTableOutputFormat method testWriteToMockInstance.

@Test
public void testWriteToMockInstance() throws Exception {
    Instance inst = new MockInstance(test.getMethodName());
    Connector conn = inst.getConnector("root", new PasswordToken(""));
    HiveAccumuloTableOutputFormat outputFormat = new HiveAccumuloTableOutputFormat();
    String table = test.getMethodName();
    conn.tableOperations().create(table);
    JobConf conf = new JobConf();
    conf.set(AccumuloConnectionParameters.INSTANCE_NAME, inst.getInstanceName());
    conf.set(AccumuloConnectionParameters.USER_NAME, "root");
    conf.set(AccumuloConnectionParameters.USER_PASS, "");
    conf.setBoolean(AccumuloConnectionParameters.USE_MOCK_INSTANCE, true);
    conf.set(AccumuloConnectionParameters.TABLE_NAME, test.getMethodName());
    FileSystem local = FileSystem.getLocal(conf);
    outputFormat.checkOutputSpecs(local, conf);
    RecordWriter<Text, Mutation> recordWriter = outputFormat.getRecordWriter(local, conf, null, null);
    List<String> names = Arrays.asList("row", "col1", "col2");
    List<TypeInfo> types = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo);
    Properties tableProperties = new Properties();
    tableProperties.setProperty(AccumuloSerDeParameters.COLUMN_MAPPINGS, ":rowID,cf:cq1,cf:cq2");
    tableProperties.setProperty(serdeConstants.FIELD_DELIM, " ");
    tableProperties.setProperty(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(names));
    tableProperties.setProperty(serdeConstants.LIST_COLUMN_TYPES, Joiner.on(',').join(types));
    AccumuloSerDeParameters accumuloSerDeParams = new AccumuloSerDeParameters(new Configuration(), tableProperties, AccumuloSerDe.class.getSimpleName());
    LazySerDeParameters serDeParams = accumuloSerDeParams.getSerDeParameters();
    AccumuloRowSerializer serializer = new AccumuloRowSerializer(0, serDeParams, accumuloSerDeParams.getColumnMappings(), AccumuloSerDeParameters.DEFAULT_VISIBILITY_LABEL, accumuloSerDeParams.getRowIdFactory());
    TypeInfo stringTypeInfo = TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.STRING_TYPE_NAME);
    LazySimpleStructObjectInspector structOI = (LazySimpleStructObjectInspector) LazyFactory.createLazyStructInspector(Arrays.asList("row", "cq1", "cq2"), Arrays.asList(stringTypeInfo, stringTypeInfo, stringTypeInfo), serDeParams.getSeparators(), serDeParams.getNullSequence(), serDeParams.isLastColumnTakesRest(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
    LazyStruct struct = (LazyStruct) LazyFactory.createLazyObject(structOI);
    ByteArrayRef bytes = new ByteArrayRef();
    bytes.setData("row value1 value2".getBytes());
    struct.init(bytes, 0, bytes.getData().length);
    // Serialize the struct into a mutation
    Mutation m = serializer.serialize(struct, structOI);
    // Write the mutation
    recordWriter.write(new Text(table), m);
    // Close the writer
    recordWriter.close(null);
    Iterator<Entry<Key, Value>> iter = conn.createScanner(table, new Authorizations()).iterator();
    Assert.assertTrue("Iterator did not have an element as expected", iter.hasNext());
    Entry<Key, Value> entry = iter.next();
    Key k = entry.getKey();
    Value v = entry.getValue();
    Assert.assertEquals("row", k.getRow().toString());
    Assert.assertEquals("cf", k.getColumnFamily().toString());
    Assert.assertEquals("cq1", k.getColumnQualifier().toString());
    Assert.assertEquals("", k.getColumnVisibility().toString());
    Assert.assertEquals("value1", new String(v.get()));
    Assert.assertTrue("Iterator did not have an element as expected", iter.hasNext());
    entry = iter.next();
    k = entry.getKey();
    v = entry.getValue();
    Assert.assertEquals("row", k.getRow().toString());
    Assert.assertEquals("cf", k.getColumnFamily().toString());
    Assert.assertEquals("cq2", k.getColumnQualifier().toString());
    Assert.assertEquals("", k.getColumnVisibility().toString());
    Assert.assertEquals("value2", new String(v.get()));
    Assert.assertFalse("Iterator unexpectedly had more data", iter.hasNext());
}

Example 7 with LazySimpleStructObjectInspector

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

the class TestHiveAccumuloTableOutputFormat method testWriteToMockInstanceWithVisibility.

@Test
public void testWriteToMockInstanceWithVisibility() throws Exception {
    Instance inst = new MockInstance(test.getMethodName());
    Connector conn = inst.getConnector("root", new PasswordToken(""));
    Authorizations auths = new Authorizations("foo");
    conn.securityOperations().changeUserAuthorizations("root", auths);
    HiveAccumuloTableOutputFormat outputFormat = new HiveAccumuloTableOutputFormat();
    String table = test.getMethodName();
    conn.tableOperations().create(table);
    JobConf conf = new JobConf();
    conf.set(AccumuloConnectionParameters.INSTANCE_NAME, inst.getInstanceName());
    conf.set(AccumuloConnectionParameters.USER_NAME, "root");
    conf.set(AccumuloConnectionParameters.USER_PASS, "");
    conf.setBoolean(AccumuloConnectionParameters.USE_MOCK_INSTANCE, true);
    conf.set(AccumuloConnectionParameters.TABLE_NAME, test.getMethodName());
    FileSystem local = FileSystem.getLocal(conf);
    outputFormat.checkOutputSpecs(local, conf);
    RecordWriter<Text, Mutation> recordWriter = outputFormat.getRecordWriter(local, conf, null, null);
    List<String> names = Arrays.asList("row", "col1", "col2");
    List<TypeInfo> types = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo);
    Properties tableProperties = new Properties();
    tableProperties.setProperty(AccumuloSerDeParameters.COLUMN_MAPPINGS, ":rowID,cf:cq1,cf:cq2");
    tableProperties.setProperty(serdeConstants.FIELD_DELIM, " ");
    tableProperties.setProperty(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(names));
    tableProperties.setProperty(serdeConstants.LIST_COLUMN_TYPES, Joiner.on(',').join(types));
    AccumuloSerDeParameters accumuloSerDeParams = new AccumuloSerDeParameters(new Configuration(), tableProperties, AccumuloSerDe.class.getSimpleName());
    LazySerDeParameters serDeParams = accumuloSerDeParams.getSerDeParameters();
    AccumuloRowSerializer serializer = new AccumuloRowSerializer(0, serDeParams, accumuloSerDeParams.getColumnMappings(), new ColumnVisibility("foo"), accumuloSerDeParams.getRowIdFactory());
    LazySimpleStructObjectInspector structOI = (LazySimpleStructObjectInspector) LazyFactory.createLazyStructInspector(Arrays.asList("row", "cq1", "cq2"), Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo), serDeParams.getSeparators(), serDeParams.getNullSequence(), serDeParams.isLastColumnTakesRest(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
    LazyStruct struct = (LazyStruct) LazyFactory.createLazyObject(structOI);
    ByteArrayRef bytes = new ByteArrayRef();
    bytes.setData("row value1 value2".getBytes());
    struct.init(bytes, 0, bytes.getData().length);
    // Serialize the struct into a mutation
    Mutation m = serializer.serialize(struct, structOI);
    // Write the mutation
    recordWriter.write(new Text(table), m);
    // Close the writer
    recordWriter.close(null);
    Iterator<Entry<Key, Value>> iter = conn.createScanner(table, auths).iterator();
    Assert.assertTrue("Iterator did not have an element as expected", iter.hasNext());
    Entry<Key, Value> entry = iter.next();
    Key k = entry.getKey();
    Value v = entry.getValue();
    Assert.assertEquals("row", k.getRow().toString());
    Assert.assertEquals("cf", k.getColumnFamily().toString());
    Assert.assertEquals("cq1", k.getColumnQualifier().toString());
    Assert.assertEquals("foo", k.getColumnVisibility().toString());
    Assert.assertEquals("value1", new String(v.get()));
    Assert.assertTrue("Iterator did not have an element as expected", iter.hasNext());
    entry = iter.next();
    k = entry.getKey();
    v = entry.getValue();
    Assert.assertEquals("row", k.getRow().toString());
    Assert.assertEquals("cf", k.getColumnFamily().toString());
    Assert.assertEquals("cq2", k.getColumnQualifier().toString());
    Assert.assertEquals("foo", k.getColumnVisibility().toString());
    Assert.assertEquals("value2", new String(v.get()));
    Assert.assertFalse("Iterator unexpectedly had more data", iter.hasNext());
}

Example 8 with LazySimpleStructObjectInspector

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

the class TestHiveAccumuloTableOutputFormat method testBinarySerializationOnStringFallsBackToUtf8.

@Test
public void testBinarySerializationOnStringFallsBackToUtf8() throws Exception {
    Instance inst = new MockInstance(test.getMethodName());
    Connector conn = inst.getConnector("root", new PasswordToken(""));
    HiveAccumuloTableOutputFormat outputFormat = new HiveAccumuloTableOutputFormat();
    String table = test.getMethodName();
    conn.tableOperations().create(table);
    JobConf conf = new JobConf();
    conf.set(AccumuloConnectionParameters.INSTANCE_NAME, inst.getInstanceName());
    conf.set(AccumuloConnectionParameters.USER_NAME, "root");
    conf.set(AccumuloConnectionParameters.USER_PASS, "");
    conf.setBoolean(AccumuloConnectionParameters.USE_MOCK_INSTANCE, true);
    conf.set(AccumuloConnectionParameters.TABLE_NAME, test.getMethodName());
    FileSystem local = FileSystem.getLocal(conf);
    outputFormat.checkOutputSpecs(local, conf);
    RecordWriter<Text, Mutation> recordWriter = outputFormat.getRecordWriter(local, conf, null, null);
    List<String> names = Arrays.asList("row", "col1", "col2");
    List<TypeInfo> types = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo);
    Properties tableProperties = new Properties();
    tableProperties.setProperty(AccumuloSerDeParameters.COLUMN_MAPPINGS, ":rowID,cf:cq1,cf:cq2");
    tableProperties.setProperty(serdeConstants.FIELD_DELIM, " ");
    tableProperties.setProperty(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(names));
    tableProperties.setProperty(serdeConstants.LIST_COLUMN_TYPES, Joiner.on(',').join(types));
    tableProperties.setProperty(AccumuloSerDeParameters.DEFAULT_STORAGE_TYPE, ColumnEncoding.BINARY.getName());
    AccumuloSerDeParameters accumuloSerDeParams = new AccumuloSerDeParameters(new Configuration(), tableProperties, AccumuloSerDe.class.getSimpleName());
    LazySerDeParameters serDeParams = accumuloSerDeParams.getSerDeParameters();
    AccumuloRowSerializer serializer = new AccumuloRowSerializer(0, serDeParams, accumuloSerDeParams.getColumnMappings(), AccumuloSerDeParameters.DEFAULT_VISIBILITY_LABEL, accumuloSerDeParams.getRowIdFactory());
    TypeInfo stringTypeInfo = TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.STRING_TYPE_NAME);
    LazySimpleStructObjectInspector structOI = (LazySimpleStructObjectInspector) LazyFactory.createLazyStructInspector(Arrays.asList("row", "cq1", "cq2"), Arrays.asList(stringTypeInfo, stringTypeInfo, stringTypeInfo), serDeParams.getSeparators(), serDeParams.getNullSequence(), serDeParams.isLastColumnTakesRest(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
    LazyStruct struct = (LazyStruct) LazyFactory.createLazyObject(structOI);
    ByteArrayRef bytes = new ByteArrayRef();
    bytes.setData("row value1 value2".getBytes());
    struct.init(bytes, 0, bytes.getData().length);
    // Serialize the struct into a mutation
    Mutation m = serializer.serialize(struct, structOI);
    // Write the mutation
    recordWriter.write(new Text(table), m);
    // Close the writer
    recordWriter.close(null);
    Iterator<Entry<Key, Value>> iter = conn.createScanner(table, new Authorizations()).iterator();
    Assert.assertTrue("Iterator did not have an element as expected", iter.hasNext());
    Entry<Key, Value> entry = iter.next();
    Key k = entry.getKey();
    Value v = entry.getValue();
    Assert.assertEquals("row", k.getRow().toString());
    Assert.assertEquals("cf", k.getColumnFamily().toString());
    Assert.assertEquals("cq1", k.getColumnQualifier().toString());
    Assert.assertEquals("", k.getColumnVisibility().toString());
    Assert.assertEquals("value1", new String(v.get()));
    Assert.assertTrue("Iterator did not have an element as expected", iter.hasNext());
    entry = iter.next();
    k = entry.getKey();
    v = entry.getValue();
    Assert.assertEquals("row", k.getRow().toString());
    Assert.assertEquals("cf", k.getColumnFamily().toString());
    Assert.assertEquals("cq2", k.getColumnQualifier().toString());
    Assert.assertEquals("", k.getColumnVisibility().toString());
    Assert.assertEquals("value2", new String(v.get()));
    Assert.assertFalse("Iterator unexpectedly had more data", iter.hasNext());
}

Example 9 with LazySimpleStructObjectInspector

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

the class TestLazyAccumuloRow method testNullInit.

@Test
public void testNullInit() throws SerDeException {
    List<String> columns = Arrays.asList("row", "1", "2", "3");
    List<TypeInfo> types = Arrays.<TypeInfo>asList(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.STRING_TYPE_NAME), TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.STRING_TYPE_NAME), TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.STRING_TYPE_NAME), TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.STRING_TYPE_NAME));
    LazySimpleStructObjectInspector objectInspector = (LazySimpleStructObjectInspector) LazyFactory.createLazyStructInspector(columns, types, LazySerDeParameters.DefaultSeparators, new Text("\\N"), false, false, (byte) '\\');
    DefaultAccumuloRowIdFactory rowIdFactory = new DefaultAccumuloRowIdFactory();
    Properties props = new Properties();
    props.setProperty(AccumuloSerDeParameters.COLUMN_MAPPINGS, ":rowid,cf:cq1,cf:cq2,cf:cq3");
    props.setProperty(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(columns));
    props.setProperty(serdeConstants.LIST_COLUMN_TYPES, Joiner.on(',').join(types));
    AccumuloSerDeParameters params = new AccumuloSerDeParameters(new Configuration(), props, AccumuloSerDe.class.getName());
    rowIdFactory.init(params, props);
    ColumnMapper columnMapper = params.getColumnMapper();
    LazyAccumuloRow lazyRow = new LazyAccumuloRow(objectInspector);
    AccumuloHiveRow hiveRow = new AccumuloHiveRow("1");
    hiveRow.add("cf", "cq1", "foo".getBytes());
    hiveRow.add("cf", "cq3", "bar".getBytes());
    lazyRow.init(hiveRow, columnMapper.getColumnMappings(), rowIdFactory);
    // Noticed that we also suffer from the same issue as HIVE-3179
    // Only want to call a field init'ed when it's non-NULL
    // Check it twice, make sure we get null both times
    Assert.assertEquals("{'row':'1','1':'foo','2':null,'3':'bar'}".replace('\'', '"'), SerDeUtils.getJSONString(lazyRow, objectInspector));
    Assert.assertEquals("{'row':'1','1':'foo','2':null,'3':'bar'}".replace('\'', '"'), SerDeUtils.getJSONString(lazyRow, objectInspector));
}

Example 10 with LazySimpleStructObjectInspector

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

the class TestLazyHBaseObject method testLazyHBaseRow3.

/**
   * Test the LazyHBaseRow class with a one-to-one/onto mapping between Hive columns and
   * HBase column family/column qualifier pairs. The column types are primitive and fields
   * are stored in binary format in HBase.
   * @throws SerDeException
   */
public void testLazyHBaseRow3() throws SerDeException {
    List<TypeInfo> fieldTypeInfos = TypeInfoUtils.getTypeInfosFromTypeString("string,int,tinyint,smallint,bigint,float,double,string,boolean");
    List<String> fieldNames = Arrays.asList(new String[] { "key", "c_int", "c_byte", "c_short", "c_long", "c_float", "c_double", "c_string", "c_bool" });
    Text nullSequence = new Text("\\N");
    String hbaseColumnsMapping = ":key#str,cf-int:cq-int#bin,cf-byte:cq-byte#bin," + "cf-short:cq-short#bin,cf-long:cq-long#bin,cf-float:cq-float#bin,cf-double:cq-double#bin," + "cf-string:cq-string#str,cf-bool:cq-bool#bin";
    ColumnMappings columnMappings = null;
    try {
        columnMappings = HBaseSerDe.parseColumnsMapping(hbaseColumnsMapping);
    } catch (SerDeException e) {
        fail(e.toString());
    }
    ColumnMapping[] columnsMapping = columnMappings.getColumnsMapping();
    for (int i = 0; i < columnsMapping.length; i++) {
        ColumnMapping colMap = columnsMapping[i];
        if (i == 0 || i == 7) {
            colMap.binaryStorage.add(false);
        } else {
            colMap.binaryStorage.add(true);
        }
    }
    ObjectInspector oi = LazyFactory.createLazyStructInspector(fieldNames, fieldTypeInfos, new byte[] { ' ', ':', '=' }, nullSequence, false, false, (byte) 0);
    LazyHBaseRow o = new LazyHBaseRow((LazySimpleStructObjectInspector) oi, columnMappings);
    byte[] rowKey = "row-key".getBytes();
    List<KeyValue> kvs = new ArrayList<KeyValue>();
    byte[] value;
    for (int i = 1; i < columnsMapping.length; i++) {
        switch(i) {
            case 1:
                value = Bytes.toBytes(1);
                break;
            case 2:
                value = new byte[] { (byte) 1 };
                break;
            case 3:
                value = Bytes.toBytes((short) 1);
                break;
            case 4:
                value = Bytes.toBytes((long) 1);
                break;
            case 5:
                value = Bytes.toBytes((float) 1.0F);
                break;
            case 6:
                value = Bytes.toBytes((double) 1.0);
                break;
            case 7:
                value = "Hadoop, Hive, with HBase storage handler.".getBytes();
                break;
            case 8:
                value = Bytes.toBytes(true);
                break;
            default:
                throw new RuntimeException("Not expected: " + i);
        }
        ColumnMapping colMap = columnsMapping[i];
        kvs.add(new KeyValue(rowKey, colMap.familyNameBytes, colMap.qualifierNameBytes, value));
    }
    Collections.sort(kvs, KeyValue.COMPARATOR);
    Result result = new Result(kvs);
    o.init(result);
    List<? extends StructField> fieldRefs = ((StructObjectInspector) oi).getAllStructFieldRefs();
    for (int i = 0; i < fieldRefs.size(); i++) {
        Object fieldData = ((StructObjectInspector) oi).getStructFieldData(o, fieldRefs.get(i));
        assert (fieldData != null);
        assert (fieldData instanceof LazyPrimitive<?, ?>);
        Writable writable = ((LazyPrimitive<?, ?>) fieldData).getWritableObject();
        switch(i) {
            case 0:
                Text text = new Text("row-key");
                assertEquals(text, writable);
                break;
            case 1:
                IntWritable iw = new IntWritable(1);
                assertEquals(iw, writable);
                break;
            case 2:
                ByteWritable bw = new ByteWritable((byte) 1);
                assertEquals(bw, writable);
                break;
            case 3:
                ShortWritable sw = new ShortWritable((short) 1);
                assertEquals(sw, writable);
                break;
            case 4:
                LongWritable lw = new LongWritable(1);
                assertEquals(lw, writable);
                break;
            case 5:
                FloatWritable fw = new FloatWritable(1.0F);
                assertEquals(fw, writable);
                break;
            case 6:
                DoubleWritable dw = new DoubleWritable(1.0);
                assertEquals(dw, writable);
                break;
            case 7:
                Text t = new Text("Hadoop, Hive, with HBase storage handler.");
                assertEquals(t, writable);
                break;
            case 8:
                BooleanWritable boolWritable = new BooleanWritable(true);
                assertEquals(boolWritable, writable);
                break;
            default:
                fail("Error: Unanticipated value in deserializing fields for HBaseSerDe.");
                break;
        }
    }
}