Examples with JavaStringObjectInspector org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaStringObjectInspector used on opensource projects

Example 6 with JavaStringObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaStringObjectInspector in project presto by prestodb.

the class AbstractTestParquetReader method testSchemaWithRequiredStruct.

@Test
public void testSchemaWithRequiredStruct() throws Exception {
    MessageType parquetSchema = parseMessageType("message hive_schema {" + "  required group a {" + "    required group b {" + "        required binary c (UTF8);" + "        required int32 d;" + "    }" + "    required binary e (UTF8);" + "  }" + "} ");
    Type bType = RowType.from(asList(field("c", VARCHAR), field("d", INTEGER)));
    Type aType = RowType.from(asList(field("b", bType), field("e", VARCHAR)));
    Iterable<String> cValues = limit(cycle(asList("c0", "c1", "c2", "c3", "c4", "c5", "c6", "c7")), 30000);
    Iterable<Integer> dValues = intsBetween(0, 30000);
    Iterable<String> eValues = limit(cycle(asList("e0", "e1", "e2", "e3", "e4", "e5", "e6", "e7")), 30000);
    List<List> bValues = createTestStructs(cValues, dValues);
    List<List> aValues = createTestStructs(bValues, eValues);
    ObjectInspector bInspector = getStandardStructObjectInspector(asList("c", "d"), asList(javaStringObjectInspector, javaIntObjectInspector));
    ObjectInspector aInspector = getStandardStructObjectInspector(asList("b", "e"), asList(bInspector, javaStringObjectInspector));
    tester.assertRoundTrip(singletonList(aInspector), new Iterable<?>[] { aValues }, new Iterable<?>[] { aValues }, singletonList("a"), singletonList(aType), Optional.of(parquetSchema));
}

Example 7 with JavaStringObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaStringObjectInspector in project presto by prestodb.

the class AbstractTestParquetReader method testSchemaWithRequiredOptionalRequiredFields.

@Test
public void testSchemaWithRequiredOptionalRequiredFields() throws Exception {
    MessageType parquetSchema = parseMessageType("message hive_schema {" + "  optional group a {" + "    required group b {" + "      optional group c {" + "        required binary d (UTF8);" + "      }" + "    }" + "  }" + "} ");
    Type cType = RowType.from(singletonList(field("d", VARCHAR)));
    Type bType = RowType.from(singletonList(field("c", cType)));
    Type aType = RowType.from(singletonList(field("b", bType)));
    Iterable<String> dValues = asList("d1", "d2", "d3", "d4", "d5", "d6", "d7");
    Iterable<List> cValues = createNullableTestStructs(dValues);
    List<List> bValues = createTestStructs(cValues);
    List<List> aValues = createTestStructs(bValues);
    ObjectInspector cInspector = getStandardStructObjectInspector(singletonList("d"), singletonList(javaStringObjectInspector));
    ObjectInspector bInspector = getStandardStructObjectInspector(singletonList("c"), singletonList(cInspector));
    ObjectInspector aInspector = getStandardStructObjectInspector(singletonList("b"), singletonList(bInspector));
    tester.testRoundTrip(aInspector, aValues, aValues, "a", aType, Optional.of(parquetSchema));
}

Example 8 with JavaStringObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaStringObjectInspector in project hive by apache.

the class TestHiveAccumuloTypes method testBinaryTypes.

@Test
public void testBinaryTypes() throws Exception {
    final String tableName = test.getMethodName(), user = "root", pass = "";
    MockInstance mockInstance = new MockInstance(test.getMethodName());
    Connector conn = mockInstance.getConnector(user, new PasswordToken(pass));
    HiveAccumuloTableInputFormat inputformat = new HiveAccumuloTableInputFormat();
    JobConf conf = new JobConf();
    conf.set(AccumuloSerDeParameters.TABLE_NAME, tableName);
    conf.set(AccumuloSerDeParameters.USE_MOCK_INSTANCE, "true");
    conf.set(AccumuloSerDeParameters.INSTANCE_NAME, test.getMethodName());
    conf.set(AccumuloSerDeParameters.USER_NAME, user);
    conf.set(AccumuloSerDeParameters.USER_PASS, pass);
    // not used for mock, but
    conf.set(AccumuloSerDeParameters.ZOOKEEPERS, "localhost:2181");
    // required by input format.
    conf.set(AccumuloSerDeParameters.COLUMN_MAPPINGS, AccumuloHiveConstants.ROWID + ",cf:string,cf:boolean,cf:tinyint,cf:smallint,cf:int,cf:bigint" + ",cf:float,cf:double,cf:decimal,cf:date,cf:timestamp,cf:char,cf:varchar");
    conf.set(serdeConstants.LIST_COLUMNS, "string,string,boolean,tinyint,smallint,int,bigint,float,double,decimal,date,timestamp,char(4),varchar(7)");
    conf.set(serdeConstants.LIST_COLUMN_TYPES, "string,string,boolean,tinyint,smallint,int,bigint,float,double,decimal,date,timestamp,char(4),varchar(7)");
    conf.set(AccumuloSerDeParameters.DEFAULT_STORAGE_TYPE, "binary");
    conn.tableOperations().create(tableName);
    BatchWriterConfig writerConf = new BatchWriterConfig();
    BatchWriter writer = conn.createBatchWriter(tableName, writerConf);
    ByteArrayOutputStream baos = new ByteArrayOutputStream();
    DataOutputStream out = new DataOutputStream(baos);
    String cf = "cf";
    byte[] cfBytes = cf.getBytes();
    Mutation m = new Mutation("row1");
    // string
    String stringValue = "string";
    JavaStringObjectInspector stringOI = (JavaStringObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.STRING_TYPE_NAME));
    LazyUtils.writePrimitiveUTF8(baos, stringOI.create(stringValue), stringOI, false, (byte) 0, null);
    m.put(cfBytes, "string".getBytes(), baos.toByteArray());
    // boolean
    boolean booleanValue = true;
    baos.reset();
    JavaBooleanObjectInspector booleanOI = (JavaBooleanObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.BOOLEAN_TYPE_NAME));
    LazyUtils.writePrimitive(baos, booleanOI.create(booleanValue), booleanOI);
    m.put(cfBytes, "boolean".getBytes(), baos.toByteArray());
    // tinyint
    byte tinyintValue = -127;
    baos.reset();
    JavaByteObjectInspector byteOI = (JavaByteObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.TINYINT_TYPE_NAME));
    LazyUtils.writePrimitive(baos, tinyintValue, byteOI);
    m.put(cfBytes, "tinyint".getBytes(), baos.toByteArray());
    // smallint
    short smallintValue = Short.MAX_VALUE;
    baos.reset();
    JavaShortObjectInspector shortOI = (JavaShortObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.SMALLINT_TYPE_NAME));
    LazyUtils.writePrimitive(baos, smallintValue, shortOI);
    m.put(cfBytes, "smallint".getBytes(), baos.toByteArray());
    // int
    int intValue = Integer.MAX_VALUE;
    baos.reset();
    JavaIntObjectInspector intOI = (JavaIntObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.INT_TYPE_NAME));
    LazyUtils.writePrimitive(baos, intValue, intOI);
    m.put(cfBytes, "int".getBytes(), baos.toByteArray());
    // bigint
    long bigintValue = Long.MAX_VALUE;
    baos.reset();
    JavaLongObjectInspector longOI = (JavaLongObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.BIGINT_TYPE_NAME));
    LazyUtils.writePrimitive(baos, bigintValue, longOI);
    m.put(cfBytes, "bigint".getBytes(), baos.toByteArray());
    // float
    float floatValue = Float.MAX_VALUE;
    baos.reset();
    JavaFloatObjectInspector floatOI = (JavaFloatObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.FLOAT_TYPE_NAME));
    LazyUtils.writePrimitive(baos, floatValue, floatOI);
    m.put(cfBytes, "float".getBytes(), baos.toByteArray());
    // double
    double doubleValue = Double.MAX_VALUE;
    baos.reset();
    JavaDoubleObjectInspector doubleOI = (JavaDoubleObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.DOUBLE_TYPE_NAME));
    LazyUtils.writePrimitive(baos, doubleValue, doubleOI);
    m.put(cfBytes, "double".getBytes(), baos.toByteArray());
    // decimal
    baos.reset();
    HiveDecimal decimalValue = HiveDecimal.create(65536l);
    HiveDecimalWritable decimalWritable = new HiveDecimalWritable(decimalValue);
    decimalWritable.write(out);
    m.put(cfBytes, "decimal".getBytes(), baos.toByteArray());
    // date
    baos.reset();
    Date now = Date.ofEpochMilli(System.currentTimeMillis());
    DateWritableV2 dateWritable = new DateWritableV2(now);
    Date dateValue = dateWritable.get();
    dateWritable.write(out);
    m.put(cfBytes, "date".getBytes(), baos.toByteArray());
    // tiemestamp
    baos.reset();
    Timestamp timestampValue = Timestamp.ofEpochMilli(System.currentTimeMillis());
    ByteStream.Output output = new ByteStream.Output();
    TimestampWritableV2 timestampWritable = new TimestampWritableV2(timestampValue);
    timestampWritable.write(new DataOutputStream(output));
    output.close();
    m.put(cfBytes, "timestamp".getBytes(), output.toByteArray());
    // char
    baos.reset();
    HiveChar charValue = new HiveChar("char", 4);
    JavaHiveCharObjectInspector charOI = (JavaHiveCharObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(new CharTypeInfo(4));
    LazyUtils.writePrimitiveUTF8(baos, charOI.create(charValue), charOI, false, (byte) 0, null);
    m.put(cfBytes, "char".getBytes(), baos.toByteArray());
    baos.reset();
    HiveVarchar varcharValue = new HiveVarchar("varchar", 7);
    JavaHiveVarcharObjectInspector varcharOI = (JavaHiveVarcharObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(new VarcharTypeInfo(7));
    LazyUtils.writePrimitiveUTF8(baos, varcharOI.create(varcharValue), varcharOI, false, (byte) 0, null);
    m.put(cfBytes, "varchar".getBytes(), baos.toByteArray());
    writer.addMutation(m);
    writer.close();
    for (Entry<Key, Value> e : conn.createScanner(tableName, new Authorizations())) {
        System.out.println(e);
    }
    // Create the RecordReader
    FileInputFormat.addInputPath(conf, new Path("unused"));
    InputSplit[] splits = inputformat.getSplits(conf, 0);
    assertEquals(splits.length, 1);
    RecordReader<Text, AccumuloHiveRow> reader = inputformat.getRecordReader(splits[0], conf, null);
    Text key = reader.createKey();
    AccumuloHiveRow value = reader.createValue();
    reader.next(key, value);
    Assert.assertEquals(13, value.getTuples().size());
    ByteArrayRef byteRef = new ByteArrayRef();
    // string
    Text cfText = new Text(cf), cqHolder = new Text();
    cqHolder.set("string");
    byte[] valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyStringObjectInspector lazyStringOI = LazyPrimitiveObjectInspectorFactory.getLazyStringObjectInspector(false, (byte) 0);
    LazyString lazyString = (LazyString) LazyFactory.createLazyObject(lazyStringOI);
    lazyString.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(stringValue, lazyString.getWritableObject().toString());
    // boolean
    cqHolder.set("boolean");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyBooleanObjectInspector lazyBooleanOI = (LazyBooleanObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.BOOLEAN_TYPE_NAME));
    LazyBoolean lazyBoolean = (LazyBoolean) LazyFactory.createLazyPrimitiveBinaryClass(lazyBooleanOI);
    lazyBoolean.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(booleanValue, lazyBoolean.getWritableObject().get());
    // tinyint
    cqHolder.set("tinyint");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyByteObjectInspector lazyByteOI = (LazyByteObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.TINYINT_TYPE_NAME));
    LazyByte lazyByte = (LazyByte) LazyFactory.createLazyPrimitiveBinaryClass(lazyByteOI);
    lazyByte.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(tinyintValue, lazyByte.getWritableObject().get());
    // smallint
    cqHolder.set("smallint");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyShortObjectInspector lazyShortOI = (LazyShortObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.SMALLINT_TYPE_NAME));
    LazyShort lazyShort = (LazyShort) LazyFactory.createLazyPrimitiveBinaryClass(lazyShortOI);
    lazyShort.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(smallintValue, lazyShort.getWritableObject().get());
    // int
    cqHolder.set("int");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyIntObjectInspector lazyIntOI = (LazyIntObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.INT_TYPE_NAME));
    LazyInteger lazyInt = (LazyInteger) LazyFactory.createLazyPrimitiveBinaryClass(lazyIntOI);
    lazyInt.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(intValue, lazyInt.getWritableObject().get());
    // bigint
    cqHolder.set("bigint");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyLongObjectInspector lazyLongOI = (LazyLongObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.BIGINT_TYPE_NAME));
    LazyLong lazyLong = (LazyLong) LazyFactory.createLazyPrimitiveBinaryClass(lazyLongOI);
    lazyLong.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(bigintValue, lazyLong.getWritableObject().get());
    // float
    cqHolder.set("float");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyFloatObjectInspector lazyFloatOI = (LazyFloatObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.FLOAT_TYPE_NAME));
    LazyFloat lazyFloat = (LazyFloat) LazyFactory.createLazyPrimitiveBinaryClass(lazyFloatOI);
    lazyFloat.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(floatValue, lazyFloat.getWritableObject().get(), 0);
    // double
    cqHolder.set("double");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyDoubleObjectInspector lazyDoubleOI = (LazyDoubleObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.DOUBLE_TYPE_NAME));
    LazyDouble lazyDouble = (LazyDouble) LazyFactory.createLazyPrimitiveBinaryClass(lazyDoubleOI);
    lazyDouble.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(doubleValue, lazyDouble.getWritableObject().get(), 0);
    // decimal
    cqHolder.set("decimal");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    ByteArrayInputStream bais = new ByteArrayInputStream(valueBytes);
    DataInputStream in = new DataInputStream(bais);
    decimalWritable.readFields(in);
    Assert.assertEquals(decimalValue, decimalWritable.getHiveDecimal());
    // date
    cqHolder.set("date");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    bais = new ByteArrayInputStream(valueBytes);
    in = new DataInputStream(bais);
    dateWritable.readFields(in);
    Assert.assertEquals(dateValue, dateWritable.get());
    // timestamp
    cqHolder.set("timestamp");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    bais = new ByteArrayInputStream(valueBytes);
    in = new DataInputStream(bais);
    timestampWritable.readFields(in);
    Assert.assertEquals(timestampValue, timestampWritable.getTimestamp());
    // char
    cqHolder.set("char");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyHiveCharObjectInspector lazyCharOI = (LazyHiveCharObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(new CharTypeInfo(4));
    LazyHiveChar lazyChar = (LazyHiveChar) LazyFactory.createLazyObject(lazyCharOI);
    lazyChar.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(charValue, lazyChar.getWritableObject().getHiveChar());
    // varchar
    cqHolder.set("varchar");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyHiveVarcharObjectInspector lazyVarcharOI = (LazyHiveVarcharObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(new VarcharTypeInfo(7));
    LazyHiveVarchar lazyVarchar = (LazyHiveVarchar) LazyFactory.createLazyObject(lazyVarcharOI);
    lazyVarchar.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(varcharValue.toString(), lazyVarchar.getWritableObject().getHiveVarchar().toString());
}

Example 9 with JavaStringObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaStringObjectInspector in project hive by apache.

the class TestAvroDeserializer method canDeserializeArrays.

@Test
public void canDeserializeArrays() throws SerDeException, IOException {
    Schema s = AvroSerdeUtils.getSchemaFor(TestAvroObjectInspectorGenerator.ARRAY_WITH_PRIMITIVE_ELEMENT_TYPE);
    GenericData.Record record = new GenericData.Record(s);
    List<String> list = new ArrayList<String>();
    list.add("Eccleston");
    list.add("Tennant");
    list.add("Smith");
    record.put("anArray", list);
    assertTrue(GENERIC_DATA.validate(s, record));
    System.out.println("Array-backed record = " + record);
    AvroGenericRecordWritable garw = Utils.serializeAndDeserializeRecord(record);
    AvroObjectInspectorGenerator aoig = new AvroObjectInspectorGenerator(s);
    AvroDeserializer de = new AvroDeserializer();
    ArrayList<Object> row = (ArrayList<Object>) de.deserialize(aoig.getColumnNames(), aoig.getColumnTypes(), garw, s);
    assertEquals(1, row.size());
    Object theArrayObject = row.get(0);
    assertTrue(theArrayObject instanceof List);
    List theList = (List) theArrayObject;
    // Verify the raw object that's been created
    assertEquals("Eccleston", theList.get(0));
    assertEquals("Tennant", theList.get(1));
    assertEquals("Smith", theList.get(2));
    // Now go the correct way, through objectinspectors
    StandardStructObjectInspector oi = (StandardStructObjectInspector) aoig.getObjectInspector();
    StructField fieldRefToArray = oi.getStructFieldRef("anArray");
    Object anArrayData = oi.getStructFieldData(row, fieldRefToArray);
    StandardListObjectInspector anArrayOI = (StandardListObjectInspector) fieldRefToArray.getFieldObjectInspector();
    assertEquals(3, anArrayOI.getListLength(anArrayData));
    JavaStringObjectInspector elementOI = (JavaStringObjectInspector) anArrayOI.getListElementObjectInspector();
    Object firstElement = anArrayOI.getListElement(anArrayData, 0);
    assertEquals("Eccleston", elementOI.getPrimitiveJavaObject(firstElement));
    assertTrue(firstElement instanceof String);
    Object secondElement = anArrayOI.getListElement(anArrayData, 1);
    assertEquals("Tennant", elementOI.getPrimitiveJavaObject(secondElement));
    assertTrue(secondElement instanceof String);
    Object thirdElement = anArrayOI.getListElement(anArrayData, 2);
    assertEquals("Smith", elementOI.getPrimitiveJavaObject(thirdElement));
    assertTrue(thirdElement instanceof String);
}

Example 10 with JavaStringObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaStringObjectInspector in project presto by prestodb.

the class HiveBucketing method getJavaObjectInspector.

private static ObjectInspector getJavaObjectInspector(ObjectInspector objectInspector) {
    checkArgument(objectInspector.getCategory() == Category.PRIMITIVE, "Unsupported object inspector category %s", objectInspector.getCategory());
    PrimitiveObjectInspector poi = ((PrimitiveObjectInspector) objectInspector);
    switch(poi.getPrimitiveCategory()) {
        case BOOLEAN:
            return javaBooleanObjectInspector;
        case BYTE:
            return javaByteObjectInspector;
        case SHORT:
            return javaShortObjectInspector;
        case INT:
            return javaIntObjectInspector;
        case LONG:
            return javaLongObjectInspector;
        case STRING:
            return javaStringObjectInspector;
    }
    throw new RuntimeException("Unsupported type: " + poi.getPrimitiveCategory());
}