Java Code Examples for org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector#getStructFieldData()

@Test
public void TestDateParse() throws Exception {
	Configuration config = new Configuration();
	Text value = new Text();

	AbstractSerDe jserde = new GeoJsonSerDe();
	Properties proptab = new Properties();
	proptab.setProperty(HiveShims.serdeConstants.LIST_COLUMNS, "when");
	proptab.setProperty(HiveShims.serdeConstants.LIST_COLUMN_TYPES, "date");
	jserde.initialize(config, proptab);
       StructObjectInspector rowOI = (StructObjectInspector)jserde.getObjectInspector();

       value.set("{\"properties\":{\"when\":\"2020-02-20\"}}");
	Object row = jserde.deserialize(value);
	StructField f0 = rowOI.getStructFieldRef("when");
	Object fieldData = rowOI.getStructFieldData(row, f0);
	Assert.assertEquals("2020-02-20",
						((DateWritable)fieldData).get().toString());
       value.set("{\"properties\":{\"when\":\"2017-05-05\"}}");
       row = jserde.deserialize(value);
	fieldData = rowOI.getStructFieldData(row, f0);
	Assert.assertEquals("2017-05-05",
						((DateWritable)fieldData).get().toString());
}
 

@Test
public void TestDateParse() throws Exception {
	Configuration config = new Configuration();
	Text value = new Text();

	AbstractSerDe jserde = new EsriJsonSerDe();
	Properties proptab = new Properties();
	proptab.setProperty(HiveShims.serdeConstants.LIST_COLUMNS, "when");
	proptab.setProperty(HiveShims.serdeConstants.LIST_COLUMN_TYPES, "date");
	jserde.initialize(config, proptab);
       StructObjectInspector rowOI = (StructObjectInspector)jserde.getObjectInspector();

       value.set("{\"attributes\":{\"when\":\"2020-02-20\"}}");
	Object row = jserde.deserialize(value);
	StructField f0 = rowOI.getStructFieldRef("when");
	Object fieldData = rowOI.getStructFieldData(row, f0);
	Assert.assertEquals("2020-02-20",
						((DateWritable)fieldData).get().toString());
       value.set("{\"attributes\":{\"when\":\"2017-05-05\"}}");
       row = jserde.deserialize(value);
	fieldData = rowOI.getStructFieldData(row, f0);
	Assert.assertEquals("2017-05-05",
						((DateWritable)fieldData).get().toString());
}
 

@Test
public void TestIntParse() throws Exception {
	Configuration config = new Configuration();
	Text value = new Text();

	AbstractSerDe jserde = new EsriJsonSerDe();
	Properties proptab = new Properties();
	proptab.setProperty(HiveShims.serdeConstants.LIST_COLUMNS, "num");
	proptab.setProperty(HiveShims.serdeConstants.LIST_COLUMN_TYPES, "int");
	jserde.initialize(config, proptab);
       StructObjectInspector rowOI = (StructObjectInspector)jserde.getObjectInspector();

       //value.set("{\"attributes\":{\"num\":7},\"geometry\":null}");
       value.set("{\"attributes\":{\"num\":7}}");
	Object row = jserde.deserialize(value);
	StructField f0 = rowOI.getStructFieldRef("num");
	Object fieldData = rowOI.getStructFieldData(row, f0);
	Assert.assertEquals(7, ((IntWritable)fieldData).get());
       value.set("{\"attributes\":{\"num\":9}}");
       row = jserde.deserialize(value);
	f0 = rowOI.getStructFieldRef("num");
	fieldData = rowOI.getStructFieldData(row, f0);
	Assert.assertEquals(9, ((IntWritable)fieldData).get());
}
 

@Override
public Writable serialize(Object obj, ObjectInspector objectInspector) throws SerDeException {
    if (!objectInspector.getCategory().equals(ObjectInspector.Category.STRUCT)) {
        throw new SerDeException("Cannot serialize " + objectInspector.getCategory() + ". Can only serialize a struct");
    }

    StructObjectInspector inspector = (StructObjectInspector) objectInspector;
    List<? extends StructField> fields = inspector.getAllStructFieldRefs();
    Writable[] arr = new Writable[fields.size()];
    for (int i = 0; i < fields.size(); i++) {
        StructField field = fields.get(i);
        Object subObj = inspector.getStructFieldData(obj, field);
        ObjectInspector subInspector = field.getFieldObjectInspector();
        arr[i] = createPrimitive(subObj, (PrimitiveObjectInspector) subInspector);
    }
    serdeSize = arr.length;
    return new ArrayWritable(Writable.class, arr);
}
 

@Test
public void TestEpochParse() throws Exception {
	Configuration config = new Configuration();
	Text value = new Text();

	AbstractSerDe jserde = new EsriJsonSerDe();
	Properties proptab = new Properties();
	proptab.setProperty(HiveShims.serdeConstants.LIST_COLUMNS, "when");
	proptab.setProperty(HiveShims.serdeConstants.LIST_COLUMN_TYPES, "date");
	jserde.initialize(config, proptab);
       StructObjectInspector rowOI = (StructObjectInspector)jserde.getObjectInspector();

       value.set("{\"attributes\":{\"when\":147147147147}}");
	Object row = jserde.deserialize(value);
	StructField f0 = rowOI.getStructFieldRef("when");
	Object fieldData = rowOI.getStructFieldData(row, f0);
	//Assert.assertEquals(147147147147L, ((DateWritable)fieldData).get().getTime());
	Assert.assertEquals(new java.sql.Date(147147147147L).toString(),
						((DateWritable)fieldData).get().toString());
       value.set("{\"attributes\":{\"when\":142857142857}}");
       row = jserde.deserialize(value);
	fieldData = rowOI.getStructFieldData(row, f0);
	//Assert.assertEquals(142857142857L, ((DateWritable)fieldData).get());
	Assert.assertEquals(new java.sql.Date(142857142857L).toString(),
						((DateWritable)fieldData).get().toString());
}
 

@Test
public void TestEpochParse() throws Exception {
	Configuration config = new Configuration();
	Text value = new Text();

	AbstractSerDe jserde = new GeoJsonSerDe();
	Properties proptab = new Properties();
	proptab.setProperty(HiveShims.serdeConstants.LIST_COLUMNS, "when");
	proptab.setProperty(HiveShims.serdeConstants.LIST_COLUMN_TYPES, "date");
	jserde.initialize(config, proptab);
       StructObjectInspector rowOI = (StructObjectInspector)jserde.getObjectInspector();

       value.set("{\"properties\":{\"when\":147147147147}}");
	Object row = jserde.deserialize(value);
	StructField f0 = rowOI.getStructFieldRef("when");
	Object fieldData = rowOI.getStructFieldData(row, f0);
	//Assert.assertEquals(147147147147L, ((DateWritable)fieldData).get().getTime());
	Assert.assertEquals(new java.sql.Date(147147147147L).toString(),
						((DateWritable)fieldData).get().toString());
       value.set("{\"properties\":{\"when\":142857142857}}");
       row = jserde.deserialize(value);
	fieldData = rowOI.getStructFieldData(row, f0);
	//Assert.assertEquals(142857142857L, ((DateWritable)fieldData).get());
	Assert.assertEquals(new java.sql.Date(142857142857L).toString(),
						((DateWritable)fieldData).get().toString());
}
 

/**
 * This method takes an object representing a row of data from Hive, and uses
 * the ObjectInspector to get the data for each column and serialize.
 */
@Override
public DbRecordWritable serialize(Object row, ObjectInspector inspector) throws SerDeException {
    final StructObjectInspector structInspector = (StructObjectInspector) inspector;
    final List<? extends StructField> fields = structInspector.getAllStructFieldRefs();
    if(fields.size() != fieldCount) {
        throw new SerDeException(String.format("Required %d columns, received %d.", fieldCount, fields.size()));
    }

    cachedWritable.clear();

    for(int i = 0; i < fieldCount; i++) {
        StructField structField = fields.get(i);
        if(structField != null) {
            Object field = structInspector.getStructFieldData(row, structField);
            ObjectInspector fieldOI = structField.getFieldObjectInspector();
            Object javaObject = HiveJdbcBridgeUtils.deparseObject(field, fieldOI);
            cachedWritable.set(i, javaObject);
        }
    }

    return cachedWritable;
}
 

private static void assertFileContentsOrcHive(
        Type type,
        TempFile tempFile,
        Iterable<?> expectedValues)
        throws Exception
{
    JobConf configuration = new JobConf(new Configuration(false));
    configuration.set(READ_COLUMN_IDS_CONF_STR, "0");
    configuration.setBoolean(READ_ALL_COLUMNS, false);

    Reader reader = OrcFile.createReader(
            new Path(tempFile.getFile().getAbsolutePath()),
            new ReaderOptions(configuration));
    RecordReader recordReader = reader.rows();

    StructObjectInspector rowInspector = (StructObjectInspector) reader.getObjectInspector();
    StructField field = rowInspector.getStructFieldRef("test");

    Iterator<?> iterator = expectedValues.iterator();
    Object rowData = null;
    while (recordReader.hasNext()) {
        rowData = recordReader.next(rowData);
        Object expectedValue = iterator.next();

        Object actualValue = rowInspector.getStructFieldData(rowData, field);
        actualValue = decodeRecordReaderValue(type, actualValue);
        assertColumnValueEquals(type, actualValue, expectedValue);
    }
    assertFalse(iterator.hasNext());
}
 

private String getFieldData(String columnName, Object data, StructObjectInspector structObjectInspector,
    Map<String, StructField> allStructFieldRefs, String name) throws SerDeException {
  StructField structField = allStructFieldRefs.get(name);
  ObjectInspector fieldObjectInspector = structField.getFieldObjectInspector();
  Object structFieldData = structObjectInspector.getStructFieldData(data, structField);
  if (fieldObjectInspector instanceof PrimitiveObjectInspector) {
    return toString(columnName, structFieldData, (PrimitiveObjectInspector) fieldObjectInspector);
  } else {
    throw new SerDeException("Embedded non-primitive type is not supported columnName [" + columnName
        + "] objectInspector [" + fieldObjectInspector + "].");
  }
}
 

@Override
public HiveKuduWritable serialize(Object row, ObjectInspector inspector)
    throws SerDeException {

    final StructObjectInspector structInspector = (StructObjectInspector) inspector;
    final List<? extends StructField> fields = structInspector.getAllStructFieldRefs();
    if (fields.size() != fieldCount) {
        throw new SerDeException(String.format(
                "Required %d columns, received %d.", fieldCount,
                fields.size()));
    }

    cachedWritable.clear();

    for (int i = 0; i < fieldCount; i++) {
        StructField structField = fields.get(i);
        if (structField != null) {
            Object field = structInspector.getStructFieldData(row,
                    structField);
            ObjectInspector fieldOI = structField.getFieldObjectInspector();

            Object javaObject = HiveKuduBridgeUtils.deparseObject(field,
                    fieldOI);
            LOG.warn("Column value of " + i + " is " + javaObject.toString());
            cachedWritable.set(i, javaObject);
        }
    }
    return cachedWritable;
}
 

@SuppressWarnings("serial")
public void testPropertyNotFound(boolean requireFieldExists) throws Exception {
	WritableEntity entity = new WritableEntity();
	entity.setProperties(new LinkedHashMap<String, EntityProperty>() {{
		put("a", new EntityProperty(7));
		put("b", new EntityProperty("hello"));
	}});
	AzureEntitySerDe serDe = new AzureEntitySerDe();
	Properties tbl = new Properties();
	tbl.put(LIST_COLUMNS, "a,b,c");
	tbl.put(LIST_COLUMN_TYPES, "int,string,int");
	Configuration conf = new Configuration();
	if (requireFieldExists) {
		conf.setBoolean(Keys.REQUIRE_FIELD_EXISTS.getKey(), true);
	}
	serDe.initialize(conf, tbl);
	StructObjectInspector inspector = (StructObjectInspector)serDe.getObjectInspector();
	assertEquals(7, inspector.getStructFieldData(entity,
			inspector.getStructFieldRef("a")));
	try {
		Object returned = inspector.getStructFieldData(entity,
				inspector.getStructFieldRef("c"));
		if (requireFieldExists) {
			fail("Should've thrown here.");
		} else {
			assertNull(returned);
		}
	} catch (IllegalArgumentException ex) {
		if (requireFieldExists) {
			assertEquals("No property found with name c. Properties found: a,b",
					ex.getMessage());
		} else {
			throw ex;
		}
	}
}
 

@Test
public void test_onTrigger_array_of_records() throws Exception {
    final Schema schema = new Schema.Parser().parse(new File("src/test/resources/array_of_records.avsc"));
    List<GenericRecord> innerRecords = new LinkedList<>();

    final GenericRecord outerRecord = new GenericData.Record(schema);

    Schema arraySchema = schema.getField("records").schema();
    Schema innerRecordSchema = arraySchema.getElementType();
    final GenericRecord innerRecord1 = new GenericData.Record(innerRecordSchema);
    innerRecord1.put("name", "Joe");
    innerRecord1.put("age", 42);

    innerRecords.add(innerRecord1);

    final GenericRecord innerRecord2 = new GenericData.Record(innerRecordSchema);
    innerRecord2.put("name", "Mary");
    innerRecord2.put("age", 28);

    innerRecords.add(innerRecord2);

    GenericData.Array<GenericRecord> array = new GenericData.Array<>(arraySchema, innerRecords);
    outerRecord.put("records", array);

    final DatumWriter<GenericRecord> datumWriter = new GenericDatumWriter<>(schema);
    ByteArrayOutputStream out = new ByteArrayOutputStream();
    try (DataFileWriter<GenericRecord> dataFileWriter = new DataFileWriter<>(datumWriter)) {
        dataFileWriter.create(schema, out);
        dataFileWriter.append(outerRecord);
    }
    out.close();

    // Build a flow file from the Avro record
    Map<String, String> attributes = new HashMap<String, String>() {{
        put(CoreAttributes.FILENAME.key(), "test");
    }};
    runner.enqueue(out.toByteArray(), attributes);
    runner.run();

    runner.assertAllFlowFilesTransferred(ConvertAvroToORC.REL_SUCCESS, 1);

    // Write the flow file out to disk, since the ORC Reader needs a path
    MockFlowFile resultFlowFile = runner.getFlowFilesForRelationship(ConvertAvroToORC.REL_SUCCESS).get(0);
    assertEquals("CREATE EXTERNAL TABLE IF NOT EXISTS org_apache_nifi_outer_record " +
            "(records ARRAY<STRUCT<name:STRING, age:INT>>)"
            + " STORED AS ORC", resultFlowFile.getAttribute(ConvertAvroToORC.HIVE_DDL_ATTRIBUTE));
    assertEquals("1", resultFlowFile.getAttribute(ConvertAvroToORC.RECORD_COUNT_ATTRIBUTE));
    assertEquals("test.orc", resultFlowFile.getAttribute(CoreAttributes.FILENAME.key()));
    byte[] resultContents = runner.getContentAsByteArray(resultFlowFile);
    FileOutputStream fos = new FileOutputStream("target/test1.orc");
    fos.write(resultContents);
    fos.flush();
    fos.close();

    Configuration conf = new Configuration();
    FileSystem fs = FileSystem.getLocal(conf);
    Reader reader = OrcFile.createReader(new Path("target/test1.orc"), OrcFile.readerOptions(conf).filesystem(fs));
    RecordReader rows = reader.rows();
    Object o = rows.next(null);
    assertNotNull(o);
    assertTrue(o instanceof OrcStruct);
    StructObjectInspector inspector = (StructObjectInspector) OrcStruct.createObjectInspector(NiFiOrcUtils.getOrcField(schema));

    // Verify the record contains an array
    Object arrayFieldObject = inspector.getStructFieldData(o, inspector.getStructFieldRef("records"));
    assertTrue(arrayFieldObject instanceof ArrayList);
    ArrayList<?> arrayField = (ArrayList<?>) arrayFieldObject;
    assertEquals(2, arrayField.size());

    // Verify the first element. Should be a record with two fields "name" and "age"
    Object element = arrayField.get(0);
    assertTrue(element instanceof OrcStruct);
    StructObjectInspector elementInspector = (StructObjectInspector) OrcStruct.createObjectInspector(NiFiOrcUtils.getOrcField(innerRecordSchema));
    Object nameObject = elementInspector.getStructFieldData(element, elementInspector.getStructFieldRef("name"));
    assertTrue(nameObject instanceof Text);
    assertEquals("Joe", nameObject.toString());
    Object ageObject = elementInspector.getStructFieldData(element, elementInspector.getStructFieldRef("age"));
    assertTrue(ageObject instanceof IntWritable);
    assertEquals(42, ((IntWritable) ageObject).get());

    // Verify the first element. Should be a record with two fields "name" and "age"
    element = arrayField.get(1);
    assertTrue(element instanceof OrcStruct);
    nameObject = elementInspector.getStructFieldData(element, elementInspector.getStructFieldRef("name"));
    assertTrue(nameObject instanceof Text);
    assertEquals("Mary", nameObject.toString());
    ageObject = elementInspector.getStructFieldData(element, elementInspector.getStructFieldRef("age"));
    assertTrue(ageObject instanceof IntWritable);
    assertEquals(28, ((IntWritable) ageObject).get());
}
 

@Test
public void test_onTrigger_complex_record() throws Exception {

    Map<String, Double> mapData1 = new TreeMap<String, Double>() {{
        put("key1", 1.0);
        put("key2", 2.0);
    }};

    GenericData.Record record = TestNiFiOrcUtils.buildComplexAvroRecord(10, mapData1, "DEF", 3.0f, Arrays.asList(10, 20));

    DatumWriter<GenericData.Record> writer = new GenericDatumWriter<>(record.getSchema());
    DataFileWriter<GenericData.Record> fileWriter = new DataFileWriter<>(writer);
    ByteArrayOutputStream out = new ByteArrayOutputStream();
    fileWriter.create(record.getSchema(), out);
    fileWriter.append(record);

    // Put another record in
    Map<String, Double> mapData2 = new TreeMap<String, Double>() {{
        put("key1", 3.0);
        put("key2", 4.0);
    }};

    record = TestNiFiOrcUtils.buildComplexAvroRecord(null, mapData2, "XYZ", 4L, Arrays.asList(100, 200));
    fileWriter.append(record);

    fileWriter.flush();
    fileWriter.close();
    out.close();

    Map<String, String> attributes = new HashMap<String, String>() {{
        put(CoreAttributes.FILENAME.key(), "test");
    }};
    runner.enqueue(out.toByteArray(), attributes);
    runner.run();

    runner.assertAllFlowFilesTransferred(ConvertAvroToORC.REL_SUCCESS, 1);

    // Write the flow file out to disk, since the ORC Reader needs a path
    MockFlowFile resultFlowFile = runner.getFlowFilesForRelationship(ConvertAvroToORC.REL_SUCCESS).get(0);
    assertEquals("CREATE EXTERNAL TABLE IF NOT EXISTS complex_record " +
            "(myInt INT, myMap MAP<STRING, DOUBLE>, myEnum STRING, myLongOrFloat UNIONTYPE<BIGINT, FLOAT>, myIntList ARRAY<INT>)"
            + " STORED AS ORC", resultFlowFile.getAttribute(ConvertAvroToORC.HIVE_DDL_ATTRIBUTE));
    assertEquals("2", resultFlowFile.getAttribute(ConvertAvroToORC.RECORD_COUNT_ATTRIBUTE));
    assertEquals("test.orc", resultFlowFile.getAttribute(CoreAttributes.FILENAME.key()));
    byte[] resultContents = runner.getContentAsByteArray(resultFlowFile);
    FileOutputStream fos = new FileOutputStream("target/test1.orc");
    fos.write(resultContents);
    fos.flush();
    fos.close();

    Configuration conf = new Configuration();
    FileSystem fs = FileSystem.getLocal(conf);
    Reader reader = OrcFile.createReader(new Path("target/test1.orc"), OrcFile.readerOptions(conf).filesystem(fs));
    RecordReader rows = reader.rows();
    Object o = rows.next(null);
    assertNotNull(o);
    assertTrue(o instanceof OrcStruct);
    TypeInfo resultSchema = TestNiFiOrcUtils.buildComplexOrcSchema();
    StructObjectInspector inspector = (StructObjectInspector) OrcStruct.createObjectInspector(resultSchema);

    // Check some fields in the first row
    Object intFieldObject = inspector.getStructFieldData(o, inspector.getStructFieldRef("myInt"));
    assertTrue(intFieldObject instanceof IntWritable);
    assertEquals(10, ((IntWritable) intFieldObject).get());

    Object mapFieldObject = inspector.getStructFieldData(o, inspector.getStructFieldRef("myMap"));
    assertTrue(mapFieldObject instanceof Map);
    Map map = (Map) mapFieldObject;
    Object mapValue = map.get(new Text("key1"));
    assertNotNull(mapValue);
    assertTrue(mapValue instanceof DoubleWritable);
    assertEquals(1.0, ((DoubleWritable) mapValue).get(), Double.MIN_VALUE);

    mapValue = map.get(new Text("key2"));
    assertNotNull(mapValue);
    assertTrue(mapValue instanceof DoubleWritable);
    assertEquals(2.0, ((DoubleWritable) mapValue).get(), Double.MIN_VALUE);
}
 

protected Object getField(String col, Object row, StructObjectInspector rowOI) {
	StructField f0 = rowOI.getStructFieldRef(col);
	return rowOI.getStructFieldData(row, f0);
}
 

/**
   * This method takes an object representing a row of data from Hive, and
   * uses the ObjectInspector to get the data for each column and serialize
   * it.
   */
  //@Override
  public Writable serialize(Object obj, ObjectInspector oi)
          throws SerDeException {
  	ExecRow row = null;
  	int[] execRowFormatIds = null;
      try {
	List<NameType> nameTypes = sqlUtil.getTableStructure(tableName);
	execRowFormatIds = sqlUtil.getExecRowFormatIds(colNames, nameTypes);
	row = sqlUtil.getExecRow(execRowFormatIds);
	if (row == null)
		throw new SerDeException("ExecRow Cannot be Null");
} catch (SQLException | StandardException | IOException e1) {
	throw new SerDeException(e1);
}    	
  	if (Log.isTraceEnabled())
  		SpliceLogUtils.trace(Log, "serialize with obj=%s, oi=%s",obj,oi);
      if (oi.getCategory() != ObjectInspector.Category.STRUCT) {
          throw new SerDeException(getClass().toString()
                  + " can only serialize struct types, but we got: "
                  + oi.getTypeName());
      }

      StructObjectInspector soi = (StructObjectInspector) oi;
      List<? extends StructField> fields = soi.getAllStructFieldRefs();

      try {

      	DataValueDescriptor dvd;
          for (int i = 0; i < fields.size(); i++) {
              StructField field = fields.get(i);
              dvd = row.getColumn(i+1);
              ObjectInspector fieldOI = field.getFieldObjectInspector();
              Object fieldObj = soi.getStructFieldData(obj, field);
              PrimitiveObjectInspector primOI = (PrimitiveObjectInspector) fieldOI;
              Object data = primOI.getPrimitiveJavaObject(fieldObj);

              PrimitiveCategory primitiveCategory = primOI.getPrimitiveCategory();
              switch (primitiveCategory) {
                  case BYTE:
                      dvd.setValue(((Byte) data).byteValue());
                      break;
                  case INT:
                      dvd.setValue(((Integer) data).intValue());
                      break;
                  case VARCHAR:
                      dvd.setValue(((HiveVarchar)data).getValue());
                      break;
                  case CHAR:
                      dvd.setValue(((HiveChar)data).getValue());
                      break;
                  case STRING:
                      dvd.setValue((String) data);
                      break;
                  case BINARY:
                      dvd.setValue((SerializationUtils.serialize((Serializable) data))); // is this right?  Should just be a byte[]
                      break;
                  case BOOLEAN:
                      dvd.setValue(((Boolean) data).booleanValue());
                      break;
                  case DECIMAL:
                      dvd.setValue(((HiveDecimal) data).doubleValue());
                      break;
                  case DOUBLE:
                      dvd.setValue(((Double) data).doubleValue());
                      break;
                  case FLOAT:
                      dvd.setValue(((Float) data).floatValue());
                      break;
                  case LONG:
                      dvd.setValue(((Long) data).longValue());
                      break;
                  case SHORT:
                      dvd.setValue(((Short) data).shortValue());
                      break;
                  case TIMESTAMP:
                      dvd.setValue((Timestamp) data);
                      break;
                  case DATE:
                      dvd.setValue((java.sql.Date) data);
                      break;
                  default:
                      throw new SerDeException(String.format("Hive Type %s Not Supported Yet",primOI.getPrimitiveCategory()));
              }
          }

      } catch (StandardException e) {
          // TODO Auto-generated catch block
          throw new RuntimeException("Serialized Object To Java Type Error");
      }
      ExecRowWritable rowWritable = new ExecRowWritable(WriteReadUtils.getExecRowFromTypeFormatIds(execRowFormatIds));
      rowWritable.set(row);
      return rowWritable;
  }
 

/** 
 * Writes a row in Hive containing exactly 5 elements (String) to a SpreadSheetCellDAO. Order: "formattedValue","comment","formula","address","sheetName"
 * 
 */
@Override
public Writable serialize(Object arg0, ObjectInspector arg1) throws SerDeException {
	if (!(arg1 instanceof StructObjectInspector)) {
		throw new SerDeException("Expect a row of Strings for serialization");
	}
	final StructObjectInspector outputOI = (StructObjectInspector) arg1;
	final List<? extends StructField> outputFields = outputOI.getAllStructFieldRefs();
	if (outputFields.size()!=ExcelSpreadSheetCellDAOSerde.columnNames.length) {
		throw new SerDeException("Expected "+ExcelSpreadSheetCellDAOSerde.columnNames.length+" fields characterizing a cell: \"formattedValue\",\"comment\",\"formula\",\"address\",\"sheetName\", but found "+outputFields.size()+" fields");
	}
	if (arg0==null) {
		return null;
	}
	// get field data
	// formattedValue
	int columnNum=0;
	final Object foFormattedValue = outputOI.getStructFieldData(arg0, outputFields.get(columnNum));
	final ObjectInspector oiFormattedValue = outputFields.get(columnNum).getFieldObjectInspector();
	String formattedValue = String.valueOf(((PrimitiveObjectInspector) oiFormattedValue).getPrimitiveJavaObject(foFormattedValue));
	// comment
	columnNum=1;
	final Object foComment= outputOI.getStructFieldData(arg0, outputFields.get(columnNum));
	final ObjectInspector oiComment = outputFields.get(columnNum).getFieldObjectInspector();
	String comment = String.valueOf(((PrimitiveObjectInspector) oiComment).getPrimitiveJavaObject(foComment));
	// formula
	columnNum=2;
	final Object foFormula= outputOI.getStructFieldData(arg0, outputFields.get(columnNum));
	final ObjectInspector oiFormula = outputFields.get(columnNum).getFieldObjectInspector();
	String formula = String.valueOf(((PrimitiveObjectInspector) oiFormula).getPrimitiveJavaObject(foFormula));
	// address
	columnNum=3;
	final Object foAddress= outputOI.getStructFieldData(arg0, outputFields.get(columnNum));
	final ObjectInspector oiAddress = outputFields.get(columnNum).getFieldObjectInspector();
	String address = String.valueOf(((PrimitiveObjectInspector) oiAddress).getPrimitiveJavaObject(foAddress));
	// sheetName
	columnNum=4;
	final Object foSheetName= outputOI.getStructFieldData(arg0, outputFields.get(columnNum));
	final ObjectInspector oiSheetName = outputFields.get(columnNum).getFieldObjectInspector();
	String sheetName = String.valueOf(((PrimitiveObjectInspector) oiSheetName).getPrimitiveJavaObject(foSheetName));
	return new SpreadSheetCellDAO(formattedValue,comment,formula,address,sheetName);
}
 

@Override
public int populateData() throws IOException, SerDeException {
  final SkipRecordsInspector skipRecordsInspector = this.skipRecordsInspector;
  final RecordReader<Object, Object> reader = this.reader;
  final Converter partTblObjectInspectorConverter = this.partTblObjectInspectorConverter;
  final Object key = this.key;

  final int numRowsPerBatch = (int) this.numRowsPerBatch;

  final StructField[] selectedStructFieldRefs = this.selectedStructFieldRefs;
  final AbstractSerDe partitionSerDe = this.partitionSerDe;
  final StructObjectInspector finalOI = this.finalOI;
  final ObjectInspector[] selectedColumnObjInspectors = this.selectedColumnObjInspectors;
  final HiveFieldConverter[] selectedColumnFieldConverters = this.selectedColumnFieldConverters;
  final ValueVector[] vectors = this.vectors;

  skipRecordsInspector.reset();
  Object value;

  int recordCount = 0;

  while (recordCount < numRowsPerBatch) {
    try (OperatorStats.WaitRecorder recorder = OperatorStats.getWaitRecorder(this.context.getStats())) {
      boolean hasNext = reader.next(key, value = skipRecordsInspector.getNextValue());
      if (!hasNext) {
        break;
      }
    }
    catch(FSError e) {
      throw HadoopFileSystemWrapper.propagateFSError(e);
    }
    if (skipRecordsInspector.doSkipHeader(recordCount++)) {
      continue;
    }
    Object bufferedValue = skipRecordsInspector.bufferAdd(value);
    if (bufferedValue != null) {
      Object deSerializedValue = partitionSerDe.deserialize((Writable) bufferedValue);
      if (partTblObjectInspectorConverter != null) {
        deSerializedValue = partTblObjectInspectorConverter.convert(deSerializedValue);
      }

      for (int i = 0; i < selectedStructFieldRefs.length; i++) {
        Object hiveValue = finalOI.getStructFieldData(deSerializedValue, selectedStructFieldRefs[i]);
        if (hiveValue != null) {
          selectedColumnFieldConverters[i].setSafeValue(selectedColumnObjInspectors[i], hiveValue, vectors[i], skipRecordsInspector.getActualCount());
        }
      }
      skipRecordsInspector.incrementActualCount();
    }
    skipRecordsInspector.incrementTempCount();
  }
  for (int i = 0; i < selectedStructFieldRefs.length; i++) {
    vectors[i].setValueCount(skipRecordsInspector.getActualCount());
  }

  skipRecordsInspector.updateContinuance();
  return skipRecordsInspector.getActualCount();
}
 

@Test
public void test_onTrigger_array_of_records() throws Exception {
    final Schema schema = new Schema.Parser().parse(new File("src/test/resources/array_of_records.avsc"));
    List<GenericRecord> innerRecords = new LinkedList<>();

    final GenericRecord outerRecord = new GenericData.Record(schema);

    Schema arraySchema = schema.getField("records").schema();
    Schema innerRecordSchema = arraySchema.getElementType();
    final GenericRecord innerRecord1 = new GenericData.Record(innerRecordSchema);
    innerRecord1.put("name", "Joe");
    innerRecord1.put("age", 42);

    innerRecords.add(innerRecord1);

    final GenericRecord innerRecord2 = new GenericData.Record(innerRecordSchema);
    innerRecord2.put("name", "Mary");
    innerRecord2.put("age", 28);

    innerRecords.add(innerRecord2);

    GenericData.Array<GenericRecord> array = new GenericData.Array<>(arraySchema, innerRecords);
    outerRecord.put("records", array);

    final DatumWriter<GenericRecord> datumWriter = new GenericDatumWriter<>(schema);
    ByteArrayOutputStream out = new ByteArrayOutputStream();
    try (DataFileWriter<GenericRecord> dataFileWriter = new DataFileWriter<>(datumWriter)) {
        dataFileWriter.create(schema, out);
        dataFileWriter.append(outerRecord);
    }
    out.close();

    // Build a flow file from the Avro record
    Map<String, String> attributes = new HashMap<String, String>() {{
        put(CoreAttributes.FILENAME.key(), "test");
    }};
    runner.enqueue(out.toByteArray(), attributes);
    runner.run();

    runner.assertAllFlowFilesTransferred(ConvertAvroToORC.REL_SUCCESS, 1);

    // Write the flow file out to disk, since the ORC Reader needs a path
    MockFlowFile resultFlowFile = runner.getFlowFilesForRelationship(ConvertAvroToORC.REL_SUCCESS).get(0);
    assertEquals("CREATE EXTERNAL TABLE IF NOT EXISTS org_apache_nifi_outer_record " +
            "(records ARRAY<STRUCT<name:STRING, age:INT>>)"
            + " STORED AS ORC", resultFlowFile.getAttribute(ConvertAvroToORC.HIVE_DDL_ATTRIBUTE));
    assertEquals("1", resultFlowFile.getAttribute(ConvertAvroToORC.RECORD_COUNT_ATTRIBUTE));
    assertEquals("test.orc", resultFlowFile.getAttribute(CoreAttributes.FILENAME.key()));
    byte[] resultContents = runner.getContentAsByteArray(resultFlowFile);
    FileOutputStream fos = new FileOutputStream("target/test1.orc");
    fos.write(resultContents);
    fos.flush();
    fos.close();

    Configuration conf = new Configuration();
    FileSystem fs = FileSystem.getLocal(conf);
    Reader reader = OrcFile.createReader(new Path("target/test1.orc"), OrcFile.readerOptions(conf).filesystem(fs));
    RecordReader rows = reader.rows();
    Object o = rows.next(null);
    assertNotNull(o);
    assertTrue(o instanceof OrcStruct);
    StructObjectInspector inspector = (StructObjectInspector) OrcStruct.createObjectInspector(NiFiOrcUtils.getOrcField(schema));

    // Verify the record contains an array
    Object arrayFieldObject = inspector.getStructFieldData(o, inspector.getStructFieldRef("records"));
    assertTrue(arrayFieldObject instanceof ArrayList);
    ArrayList<?> arrayField = (ArrayList<?>) arrayFieldObject;
    assertEquals(2, arrayField.size());

    // Verify the first element. Should be a record with two fields "name" and "age"
    Object element = arrayField.get(0);
    assertTrue(element instanceof OrcStruct);
    StructObjectInspector elementInspector = (StructObjectInspector) OrcStruct.createObjectInspector(NiFiOrcUtils.getOrcField(innerRecordSchema));
    Object nameObject = elementInspector.getStructFieldData(element, elementInspector.getStructFieldRef("name"));
    assertTrue(nameObject instanceof Text);
    assertEquals("Joe", nameObject.toString());
    Object ageObject = elementInspector.getStructFieldData(element, elementInspector.getStructFieldRef("age"));
    assertTrue(ageObject instanceof IntWritable);
    assertEquals(42, ((IntWritable) ageObject).get());

    // Verify the first element. Should be a record with two fields "name" and "age"
    element = arrayField.get(1);
    assertTrue(element instanceof OrcStruct);
    nameObject = elementInspector.getStructFieldData(element, elementInspector.getStructFieldRef("name"));
    assertTrue(nameObject instanceof Text);
    assertEquals("Mary", nameObject.toString());
    ageObject = elementInspector.getStructFieldData(element, elementInspector.getStructFieldRef("age"));
    assertTrue(ageObject instanceof IntWritable);
    assertEquals(28, ((IntWritable) ageObject).get());
}
 

@Test
public void test_onTrigger_complex_record() throws Exception {

    Map<String, Double> mapData1 = new TreeMap<String, Double>() {{
        put("key1", 1.0);
        put("key2", 2.0);
    }};

    GenericData.Record record = TestNiFiOrcUtils.buildComplexAvroRecord(10, mapData1, "DEF", 3.0f, Arrays.asList(10, 20));

    DatumWriter<GenericData.Record> writer = new GenericDatumWriter<>(record.getSchema());
    DataFileWriter<GenericData.Record> fileWriter = new DataFileWriter<>(writer);
    ByteArrayOutputStream out = new ByteArrayOutputStream();
    fileWriter.create(record.getSchema(), out);
    fileWriter.append(record);

    // Put another record in
    Map<String, Double> mapData2 = new TreeMap<String, Double>() {{
        put("key1", 3.0);
        put("key2", 4.0);
    }};

    record = TestNiFiOrcUtils.buildComplexAvroRecord(null, mapData2, "XYZ", 4L, Arrays.asList(100, 200));
    fileWriter.append(record);

    fileWriter.flush();
    fileWriter.close();
    out.close();

    Map<String, String> attributes = new HashMap<String, String>() {{
        put(CoreAttributes.FILENAME.key(), "test");
    }};
    runner.enqueue(out.toByteArray(), attributes);
    runner.run();

    runner.assertAllFlowFilesTransferred(ConvertAvroToORC.REL_SUCCESS, 1);

    // Write the flow file out to disk, since the ORC Reader needs a path
    MockFlowFile resultFlowFile = runner.getFlowFilesForRelationship(ConvertAvroToORC.REL_SUCCESS).get(0);
    assertEquals("CREATE EXTERNAL TABLE IF NOT EXISTS complex_record " +
            "(myInt INT, myMap MAP<STRING, DOUBLE>, myEnum STRING, myLongOrFloat UNIONTYPE<BIGINT, FLOAT>, myIntList ARRAY<INT>)"
            + " STORED AS ORC", resultFlowFile.getAttribute(ConvertAvroToORC.HIVE_DDL_ATTRIBUTE));
    assertEquals("2", resultFlowFile.getAttribute(ConvertAvroToORC.RECORD_COUNT_ATTRIBUTE));
    assertEquals("test.orc", resultFlowFile.getAttribute(CoreAttributes.FILENAME.key()));
    byte[] resultContents = runner.getContentAsByteArray(resultFlowFile);
    FileOutputStream fos = new FileOutputStream("target/test1.orc");
    fos.write(resultContents);
    fos.flush();
    fos.close();

    Configuration conf = new Configuration();
    FileSystem fs = FileSystem.getLocal(conf);
    Reader reader = OrcFile.createReader(new Path("target/test1.orc"), OrcFile.readerOptions(conf).filesystem(fs));
    RecordReader rows = reader.rows();
    Object o = rows.next(null);
    assertNotNull(o);
    assertTrue(o instanceof OrcStruct);
    TypeInfo resultSchema = TestNiFiOrcUtils.buildComplexOrcSchema();
    StructObjectInspector inspector = (StructObjectInspector) OrcStruct.createObjectInspector(resultSchema);

    // Check some fields in the first row
    Object intFieldObject = inspector.getStructFieldData(o, inspector.getStructFieldRef("myInt"));
    assertTrue(intFieldObject instanceof IntWritable);
    assertEquals(10, ((IntWritable) intFieldObject).get());

    // This is pretty awkward and messy. The map object is a Map (not a MapWritable) but the keys are writables (in this case Text)
    // and so are the values (DoubleWritables in this case).
    Object mapFieldObject = inspector.getStructFieldData(o, inspector.getStructFieldRef("myMap"));
    assertTrue(mapFieldObject instanceof Map);
    Map map = (Map) mapFieldObject;
    Object mapValue = map.get(new Text("key1"));
    assertNotNull(mapValue);
    assertTrue(mapValue instanceof DoubleWritable);
    assertEquals(1.0, ((DoubleWritable) mapValue).get(), Double.MIN_VALUE);

    mapValue = map.get(new Text("key2"));
    assertNotNull(mapValue);
    assertTrue(mapValue instanceof DoubleWritable);
    assertEquals(2.0, ((DoubleWritable) mapValue).get(), Double.MIN_VALUE);
}
 

@Test
public void test_onTrigger_primitive_record() throws Exception {
    GenericData.Record record = TestNiFiOrcUtils.buildPrimitiveAvroRecord(10, 20L, true, 30.0f, 40, StandardCharsets.UTF_8.encode("Hello"), "World");

    DatumWriter<GenericData.Record> writer = new GenericDatumWriter<>(record.getSchema());
    DataFileWriter<GenericData.Record> fileWriter = new DataFileWriter<>(writer);
    ByteArrayOutputStream out = new ByteArrayOutputStream();
    fileWriter.create(record.getSchema(), out);
    fileWriter.append(record);
    // Put another record in
    record = TestNiFiOrcUtils.buildPrimitiveAvroRecord(1, 2L, false, 3.0f, 4L, StandardCharsets.UTF_8.encode("I am"), "another record");
    fileWriter.append(record);
    // And one more
    record = TestNiFiOrcUtils.buildPrimitiveAvroRecord(100, 200L, true, 300.0f, 400L, StandardCharsets.UTF_8.encode("Me"), "too!");
    fileWriter.append(record);
    fileWriter.flush();
    fileWriter.close();
    out.close();
    Map<String, String> attributes = new HashMap<String, String>() {{
        put(CoreAttributes.FILENAME.key(), "test.avro");
    }};
    runner.enqueue(out.toByteArray(), attributes);
    runner.run();

    runner.assertAllFlowFilesTransferred(ConvertAvroToORC.REL_SUCCESS, 1);

    // Write the flow file out to disk, since the ORC Reader needs a path
    MockFlowFile resultFlowFile = runner.getFlowFilesForRelationship(ConvertAvroToORC.REL_SUCCESS).get(0);
    assertEquals("CREATE EXTERNAL TABLE IF NOT EXISTS test_record (int INT, long BIGINT, boolean BOOLEAN, float FLOAT, double DOUBLE, bytes BINARY, string STRING)"
            + " STORED AS ORC", resultFlowFile.getAttribute(ConvertAvroToORC.HIVE_DDL_ATTRIBUTE));
    assertEquals("3", resultFlowFile.getAttribute(ConvertAvroToORC.RECORD_COUNT_ATTRIBUTE));
    assertEquals("test.orc", resultFlowFile.getAttribute(CoreAttributes.FILENAME.key()));
    byte[] resultContents = runner.getContentAsByteArray(resultFlowFile);
    FileOutputStream fos = new FileOutputStream("target/test1.orc");
    fos.write(resultContents);
    fos.flush();
    fos.close();

    Configuration conf = new Configuration();
    FileSystem fs = FileSystem.getLocal(conf);
    Reader reader = OrcFile.createReader(new Path("target/test1.orc"), OrcFile.readerOptions(conf).filesystem(fs));
    RecordReader rows = reader.rows();
    Object o = rows.next(null);
    assertNotNull(o);
    assertTrue(o instanceof OrcStruct);
    TypeInfo resultSchema = TestNiFiOrcUtils.buildPrimitiveOrcSchema();
    StructObjectInspector inspector = (StructObjectInspector) OrcStruct.createObjectInspector(resultSchema);

    // Check some fields in the first row
    Object intFieldObject = inspector.getStructFieldData(o, inspector.getStructFieldRef("int"));
    assertTrue(intFieldObject instanceof IntWritable);
    assertEquals(10, ((IntWritable) intFieldObject).get());
    Object stringFieldObject = inspector.getStructFieldData(o, inspector.getStructFieldRef("string"));
    assertTrue(stringFieldObject instanceof Text);
    assertEquals("World", stringFieldObject.toString());

}