Examples with AccumuloRowSerializer org.apache.hadoop.hive.accumulo.serde.AccumuloRowSerializer used on opensource projects

Example 1 with AccumuloRowSerializer

use of org.apache.hadoop.hive.accumulo.serde.AccumuloRowSerializer 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 2 with AccumuloRowSerializer

use of org.apache.hadoop.hive.accumulo.serde.AccumuloRowSerializer 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 3 with AccumuloRowSerializer

use of org.apache.hadoop.hive.accumulo.serde.AccumuloRowSerializer 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 4 with AccumuloRowSerializer

use of org.apache.hadoop.hive.accumulo.serde.AccumuloRowSerializer in project hive by apache.

the class TestHiveAccumuloTableOutputFormat method testWriteMap.

@Test
public void testWriteMap() 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");
    List<TypeInfo> types = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo);
    Properties tableProperties = new Properties();
    tableProperties.setProperty(AccumuloSerDeParameters.COLUMN_MAPPINGS, ":rowID,cf:*");
    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);
    LazyStringObjectInspector stringOI = (LazyStringObjectInspector) LazyFactory.createLazyObjectInspector(stringTypeInfo, new byte[] { 0 }, 0, serDeParams.getNullSequence(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
    LazyMapObjectInspector mapOI = LazyObjectInspectorFactory.getLazySimpleMapObjectInspector(stringOI, stringOI, (byte) ',', (byte) ':', serDeParams.getNullSequence(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
    LazySimpleStructObjectInspector structOI = (LazySimpleStructObjectInspector) LazyObjectInspectorFactory.getLazySimpleStructObjectInspector(Arrays.asList("row", "data"), Arrays.asList(stringOI, mapOI), (byte) ' ', serDeParams.getNullSequence(), serDeParams.isLastColumnTakesRest(), serDeParams.isEscaped(), serDeParams.getEscapeChar());
    LazyStruct struct = (LazyStruct) LazyFactory.createLazyObject(structOI);
    ByteArrayRef bytes = new ByteArrayRef();
    bytes.setData("row cq1:value1,cq2: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(AccumuloSerDeParameters.DEFAULT_VISIBILITY_LABEL, k.getColumnVisibilityParsed());
    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(AccumuloSerDeParameters.DEFAULT_VISIBILITY_LABEL, k.getColumnVisibilityParsed());
    Assert.assertEquals("value2", new String(v.get()));
    Assert.assertFalse("Iterator unexpectedly had more data", iter.hasNext());
}