Search in sources :

Example 11 with JavaStringObjectInspector

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

the class TestHiveAccumuloTypes method testUtf8Types.

@Test
public void testUtf8Types() 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)");
    conn.tableOperations().create(tableName);
    BatchWriterConfig writerConf = new BatchWriterConfig();
    BatchWriter writer = conn.createBatchWriter(tableName, writerConf);
    ByteArrayOutputStream baos = new ByteArrayOutputStream();
    String cf = "cf";
    byte[] cfBytes = cf.getBytes();
    ByteArrayRef byteRef = new ByteArrayRef();
    Mutation m = new Mutation("row1");
    // string
    String stringValue = "string";
    baos.reset();
    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.writePrimitiveUTF8(baos, booleanOI.create(booleanValue), booleanOI, false, (byte) 0, null);
    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.writePrimitiveUTF8(baos, tinyintValue, byteOI, false, (byte) 0, null);
    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.writePrimitiveUTF8(baos, smallintValue, shortOI, false, (byte) 0, null);
    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.writePrimitiveUTF8(baos, intValue, intOI, false, (byte) 0, null);
    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.writePrimitiveUTF8(baos, bigintValue, longOI, false, (byte) 0, null);
    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.writePrimitiveUTF8(baos, floatValue, floatOI, false, (byte) 0, null);
    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.writePrimitiveUTF8(baos, doubleValue, doubleOI, false, (byte) 0, null);
    m.put(cfBytes, "double".getBytes(), baos.toByteArray());
    // decimal
    HiveDecimal decimalValue = HiveDecimal.create("1.23");
    baos.reset();
    JavaHiveDecimalObjectInspector decimalOI = (JavaHiveDecimalObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(new DecimalTypeInfo(5, 2));
    LazyUtils.writePrimitiveUTF8(baos, decimalOI.create(decimalValue), decimalOI, false, (byte) 0, null);
    m.put(cfBytes, "decimal".getBytes(), baos.toByteArray());
    // date
    Date now = Date.ofEpochMilli(System.currentTimeMillis());
    DateWritableV2 dateWritable = new DateWritableV2(now);
    Date dateValue = dateWritable.get();
    baos.reset();
    JavaDateObjectInspector dateOI = (JavaDateObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.DATE_TYPE_NAME));
    LazyUtils.writePrimitiveUTF8(baos, dateOI.create(dateValue), dateOI, false, (byte) 0, null);
    m.put(cfBytes, "date".getBytes(), baos.toByteArray());
    // timestamp
    Timestamp timestampValue = Timestamp.valueOf(LocalDateTime.now().toString());
    baos.reset();
    JavaTimestampObjectInspector timestampOI = (JavaTimestampObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.TIMESTAMP_TYPE_NAME));
    LazyUtils.writePrimitiveUTF8(baos, timestampOI.create(timestampValue), timestampOI, false, (byte) 0, null);
    m.put(cfBytes, "timestamp".getBytes(), baos.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());
    // varchar
    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());
    // 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(new Text(stringValue), lazyString.getWritableObject());
    // 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.createLazyObject(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.createLazyObject(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.createLazyObject(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.createLazyObject(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.createLazyObject(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.createLazyObject(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.createLazyObject(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);
    LazyHiveDecimalObjectInspector lazyDecimalOI = (LazyHiveDecimalObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(new DecimalTypeInfo(5, 2));
    LazyHiveDecimal lazyDecimal = (LazyHiveDecimal) LazyFactory.createLazyObject(lazyDecimalOI);
    lazyDecimal.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(decimalValue, lazyDecimal.getWritableObject().getHiveDecimal());
    // date
    cqHolder.set("date");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyDateObjectInspector lazyDateOI = (LazyDateObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.DATE_TYPE_NAME));
    LazyDate lazyDate = (LazyDate) LazyFactory.createLazyObject(lazyDateOI);
    lazyDate.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(dateValue, lazyDate.getWritableObject().get());
    // timestamp
    cqHolder.set("timestamp");
    valueBytes = value.getValue(cfText, cqHolder);
    Assert.assertNotNull(valueBytes);
    byteRef.setData(valueBytes);
    LazyTimestampObjectInspector lazyTimestampOI = (LazyTimestampObjectInspector) LazyPrimitiveObjectInspectorFactory.getLazyObjectInspector(TypeInfoFactory.getPrimitiveTypeInfo(serdeConstants.TIMESTAMP_TYPE_NAME));
    LazyTimestamp lazyTimestamp = (LazyTimestamp) LazyFactory.createLazyObject(lazyTimestampOI);
    lazyTimestamp.init(byteRef, 0, valueBytes.length);
    Assert.assertEquals(timestampValue, lazyTimestamp.getWritableObject().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());
}
Also used : LazyHiveVarchar(org.apache.hadoop.hive.serde2.lazy.LazyHiveVarchar) VarcharTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.VarcharTypeInfo) LazyHiveDecimal(org.apache.hadoop.hive.serde2.lazy.LazyHiveDecimal) LazyIntObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyIntObjectInspector) LazyDoubleObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyDoubleObjectInspector) LazyString(org.apache.hadoop.hive.serde2.lazy.LazyString) AccumuloHiveRow(org.apache.hadoop.hive.accumulo.AccumuloHiveRow) LazyHiveChar(org.apache.hadoop.hive.serde2.lazy.LazyHiveChar) PasswordToken(org.apache.accumulo.core.client.security.tokens.PasswordToken) BatchWriterConfig(org.apache.accumulo.core.client.BatchWriterConfig) JobConf(org.apache.hadoop.mapred.JobConf) JavaHiveCharObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveCharObjectInspector) LazyShortObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyShortObjectInspector) Authorizations(org.apache.accumulo.core.security.Authorizations) LazyLong(org.apache.hadoop.hive.serde2.lazy.LazyLong) LazyStringObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyStringObjectInspector) JavaStringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaStringObjectInspector) LazyDateObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyDateObjectInspector) CharTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.CharTypeInfo) JavaLongObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaLongObjectInspector) DateWritableV2(org.apache.hadoop.hive.serde2.io.DateWritableV2) LazyHiveVarcharObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyHiveVarcharObjectInspector) JavaTimestampObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaTimestampObjectInspector) LazyBoolean(org.apache.hadoop.hive.serde2.lazy.LazyBoolean) LazyLongObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyLongObjectInspector) DecimalTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo) LazyByte(org.apache.hadoop.hive.serde2.lazy.LazyByte) JavaHiveVarcharObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveVarcharObjectInspector) ByteArrayRef(org.apache.hadoop.hive.serde2.lazy.ByteArrayRef) JavaFloatObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaFloatObjectInspector) Value(org.apache.accumulo.core.data.Value) LazyInteger(org.apache.hadoop.hive.serde2.lazy.LazyInteger) Mutation(org.apache.accumulo.core.data.Mutation) LazyDouble(org.apache.hadoop.hive.serde2.lazy.LazyDouble) Key(org.apache.accumulo.core.data.Key) Connector(org.apache.accumulo.core.client.Connector) LazyHiveCharObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyHiveCharObjectInspector) JavaIntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaIntObjectInspector) LazyTimestamp(org.apache.hadoop.hive.serde2.lazy.LazyTimestamp) HiveChar(org.apache.hadoop.hive.common.type.HiveChar) LazyHiveChar(org.apache.hadoop.hive.serde2.lazy.LazyHiveChar) LazyBooleanObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyBooleanObjectInspector) LazyFloat(org.apache.hadoop.hive.serde2.lazy.LazyFloat) LazyTimestamp(org.apache.hadoop.hive.serde2.lazy.LazyTimestamp) Timestamp(org.apache.hadoop.hive.common.type.Timestamp) JavaDoubleObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaDoubleObjectInspector) LazyShort(org.apache.hadoop.hive.serde2.lazy.LazyShort) LazyString(org.apache.hadoop.hive.serde2.lazy.LazyString) JavaHiveDecimalObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveDecimalObjectInspector) MockInstance(org.apache.accumulo.core.client.mock.MockInstance) JavaDateObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaDateObjectInspector) JavaByteObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaByteObjectInspector) LazyHiveDecimal(org.apache.hadoop.hive.serde2.lazy.LazyHiveDecimal) HiveDecimal(org.apache.hadoop.hive.common.type.HiveDecimal) JavaBooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaBooleanObjectInspector) InputSplit(org.apache.hadoop.mapred.InputSplit) Path(org.apache.hadoop.fs.Path) Text(org.apache.hadoop.io.Text) ByteArrayOutputStream(java.io.ByteArrayOutputStream) HiveVarchar(org.apache.hadoop.hive.common.type.HiveVarchar) LazyHiveVarchar(org.apache.hadoop.hive.serde2.lazy.LazyHiveVarchar) LazyFloatObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyFloatObjectInspector) LazyTimestampObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyTimestampObjectInspector) LazyDate(org.apache.hadoop.hive.serde2.lazy.LazyDate) Date(org.apache.hadoop.hive.common.type.Date) LazyDate(org.apache.hadoop.hive.serde2.lazy.LazyDate) LazyHiveDecimalObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyHiveDecimalObjectInspector) LazyByteObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.primitive.LazyByteObjectInspector) BatchWriter(org.apache.accumulo.core.client.BatchWriter) JavaShortObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaShortObjectInspector) Test(org.junit.Test)

Example 12 with JavaStringObjectInspector

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

the class TestAvroDeserializer method canDeserializeArrays.

@Test
public void canDeserializeArrays() throws SerDeException, IOException {
    Schema s = Schema.parse(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);
}
Also used : JavaStringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaStringObjectInspector) Schema(org.apache.avro.Schema) ArrayList(java.util.ArrayList) GenericData(org.apache.avro.generic.GenericData) ArrayList(java.util.ArrayList) List(java.util.List) Test(org.junit.Test)

Example 13 with JavaStringObjectInspector

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

the class OrcTester method getJavaObjectInspector.

private static ObjectInspector getJavaObjectInspector(Type type) {
    if (type.equals(BOOLEAN)) {
        return javaBooleanObjectInspector;
    }
    if (type.equals(BIGINT)) {
        return javaLongObjectInspector;
    }
    if (type.equals(INTEGER)) {
        return javaIntObjectInspector;
    }
    if (type.equals(SMALLINT)) {
        return javaShortObjectInspector;
    }
    if (type.equals(TINYINT)) {
        return javaByteObjectInspector;
    }
    if (type.equals(REAL)) {
        return javaFloatObjectInspector;
    }
    if (type.equals(DOUBLE)) {
        return javaDoubleObjectInspector;
    }
    if (type instanceof VarcharType) {
        return javaStringObjectInspector;
    }
    if (type instanceof CharType) {
        int charLength = ((CharType) type).getLength();
        return new JavaHiveCharObjectInspector(getCharTypeInfo(charLength));
    }
    if (type instanceof VarbinaryType) {
        return javaByteArrayObjectInspector;
    }
    if (type.equals(DATE)) {
        return javaDateObjectInspector;
    }
    if (type.equals(TIMESTAMP)) {
        return javaTimestampObjectInspector;
    }
    if (type instanceof DecimalType) {
        DecimalType decimalType = (DecimalType) type;
        return getPrimitiveJavaObjectInspector(new DecimalTypeInfo(decimalType.getPrecision(), decimalType.getScale()));
    }
    if (type.getTypeSignature().getBase().equals(StandardTypes.ARRAY)) {
        return getStandardListObjectInspector(getJavaObjectInspector(type.getTypeParameters().get(0)));
    }
    if (type.getTypeSignature().getBase().equals(StandardTypes.MAP)) {
        ObjectInspector keyObjectInspector = getJavaObjectInspector(type.getTypeParameters().get(0));
        ObjectInspector valueObjectInspector = getJavaObjectInspector(type.getTypeParameters().get(1));
        return getStandardMapObjectInspector(keyObjectInspector, valueObjectInspector);
    }
    if (type.getTypeSignature().getBase().equals(StandardTypes.ROW)) {
        return getStandardStructObjectInspector(type.getTypeSignature().getParameters().stream().map(parameter -> parameter.getNamedTypeSignature().getName().get()).collect(toList()), type.getTypeParameters().stream().map(OrcTester::getJavaObjectInspector).collect(toList()));
    }
    throw new IllegalArgumentException("unsupported type: " + type);
}
Also used : DecimalTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo) JavaHiveCharObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveCharObjectInspector) PrimitiveObjectInspectorFactory.javaByteObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaByteObjectInspector) PrimitiveObjectInspectorFactory.javaLongObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaLongObjectInspector) PrimitiveObjectInspectorFactory.javaTimestampObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaTimestampObjectInspector) PrimitiveObjectInspectorFactory.javaDateObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaDateObjectInspector) ObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector) PrimitiveObjectInspectorFactory.javaByteArrayObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaByteArrayObjectInspector) PrimitiveObjectInspectorFactory.javaFloatObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaFloatObjectInspector) PrimitiveObjectInspectorFactory.javaDoubleObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaDoubleObjectInspector) PrimitiveObjectInspectorFactory.javaIntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaIntObjectInspector) JavaHiveCharObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveCharObjectInspector) PrimitiveObjectInspectorFactory.javaShortObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaShortObjectInspector) ObjectInspectorFactory.getStandardStructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardStructObjectInspector) SettableStructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.SettableStructObjectInspector) StructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector) PrimitiveObjectInspectorFactory.javaBooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaBooleanObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector) ObjectInspectorFactory.getStandardMapObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardMapObjectInspector) ObjectInspectorFactory.getStandardListObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardListObjectInspector) PrimitiveObjectInspectorFactory.javaStringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaStringObjectInspector) VarbinaryType(com.facebook.presto.common.type.VarbinaryType) VarcharType(com.facebook.presto.common.type.VarcharType) DecimalType(com.facebook.presto.common.type.DecimalType) CharType(com.facebook.presto.common.type.CharType)

Example 14 with JavaStringObjectInspector

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

the class AbstractTestParquetReader method testSingleLevelSchemaArrayOfStucts.

@Test
public void testSingleLevelSchemaArrayOfStucts() throws Exception {
    Iterable<Long> aValues = limit(cycle(asList(1L, null, 3L, 5L, null, null, null, 7L, 11L, null, 13L, 17L)), 30_000);
    Iterable<Boolean> bValues = limit(cycle(asList(null, true, false, null, null, true, false)), 30_000);
    Iterable<String> cValues = transform(intsBetween(0, 31_234), Object::toString);
    Iterable<List> structs = createTestStructs(aValues, bValues, cValues);
    Iterable<List<List>> values = createTestArrays(structs);
    List<String> structFieldNames = asList("a", "b", "c");
    Type structType = RowType.from(asList(field("a", BIGINT), field("b", BOOLEAN), field("c", VARCHAR)));
    ObjectInspector objectInspector = getStandardListObjectInspector(getStandardStructObjectInspector(structFieldNames, asList(javaLongObjectInspector, javaBooleanObjectInspector, javaStringObjectInspector)));
    tester.testSingleLevelArraySchemaRoundTrip(objectInspector, values, values, new ArrayType(structType));
}
Also used : JavaHiveDecimalObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveDecimalObjectInspector) PrimitiveObjectInspectorFactory.javaByteObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaByteObjectInspector) PrimitiveObjectInspectorFactory.javaLongObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaLongObjectInspector) PrimitiveObjectInspectorFactory.javaTimestampObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaTimestampObjectInspector) PrimitiveObjectInspectorFactory.javaDateObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaDateObjectInspector) ObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector) PrimitiveObjectInspectorFactory.javaByteArrayObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaByteArrayObjectInspector) PrimitiveObjectInspectorFactory.javaFloatObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaFloatObjectInspector) PrimitiveObjectInspectorFactory.javaDoubleObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaDoubleObjectInspector) PrimitiveObjectInspectorFactory.javaIntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaIntObjectInspector) PrimitiveObjectInspectorFactory.javaShortObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaShortObjectInspector) ObjectInspectorFactory.getStandardStructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardStructObjectInspector) PrimitiveObjectInspectorFactory.javaBooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaBooleanObjectInspector) ObjectInspectorFactory.getStandardMapObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardMapObjectInspector) ObjectInspectorFactory.getStandardListObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardListObjectInspector) PrimitiveObjectInspectorFactory.javaStringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaStringObjectInspector) ArrayType(com.facebook.presto.common.type.ArrayType) VarcharType.createUnboundedVarcharType(com.facebook.presto.common.type.VarcharType.createUnboundedVarcharType) StructuralTestUtil.mapType(com.facebook.presto.tests.StructuralTestUtil.mapType) ArrayType(com.facebook.presto.common.type.ArrayType) RowType(com.facebook.presto.common.type.RowType) PrimitiveType(org.apache.parquet.schema.PrimitiveType) MessageType(org.apache.parquet.schema.MessageType) Type(com.facebook.presto.common.type.Type) MessageTypeParser.parseMessageType(org.apache.parquet.schema.MessageTypeParser.parseMessageType) DecimalType.createDecimalType(com.facebook.presto.common.type.DecimalType.createDecimalType) Collections.singletonList(java.util.Collections.singletonList) Arrays.asList(java.util.Arrays.asList) ArrayList(java.util.ArrayList) List(java.util.List) ImmutableList(com.google.common.collect.ImmutableList) Test(org.testng.annotations.Test)

Example 15 with JavaStringObjectInspector

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

the class AbstractTestParquetReader method testSingleLevelSchemaArrayOfMaps.

@Test
public void testSingleLevelSchemaArrayOfMaps() throws Exception {
    Iterable<Map<String, Long>> maps = createTestMaps(transform(intsBetween(0, 10), Object::toString), longsBetween(0, 10));
    List<List<Map<String, Long>>> values = createTestArrays(maps);
    ObjectInspector objectInspector = getStandardListObjectInspector(getStandardMapObjectInspector(javaStringObjectInspector, javaLongObjectInspector));
    tester.testSingleLevelArraySchemaRoundTrip(objectInspector, values, values, new ArrayType(mapType(VARCHAR, BIGINT)));
}
Also used : ArrayType(com.facebook.presto.common.type.ArrayType) JavaHiveDecimalObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveDecimalObjectInspector) PrimitiveObjectInspectorFactory.javaByteObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaByteObjectInspector) PrimitiveObjectInspectorFactory.javaLongObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaLongObjectInspector) PrimitiveObjectInspectorFactory.javaTimestampObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaTimestampObjectInspector) PrimitiveObjectInspectorFactory.javaDateObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaDateObjectInspector) ObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector) PrimitiveObjectInspectorFactory.javaByteArrayObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaByteArrayObjectInspector) PrimitiveObjectInspectorFactory.javaFloatObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaFloatObjectInspector) PrimitiveObjectInspectorFactory.javaDoubleObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaDoubleObjectInspector) PrimitiveObjectInspectorFactory.javaIntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaIntObjectInspector) PrimitiveObjectInspectorFactory.javaShortObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaShortObjectInspector) ObjectInspectorFactory.getStandardStructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardStructObjectInspector) PrimitiveObjectInspectorFactory.javaBooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaBooleanObjectInspector) ObjectInspectorFactory.getStandardMapObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardMapObjectInspector) ObjectInspectorFactory.getStandardListObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardListObjectInspector) PrimitiveObjectInspectorFactory.javaStringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaStringObjectInspector) Collections.singletonList(java.util.Collections.singletonList) Arrays.asList(java.util.Arrays.asList) ArrayList(java.util.ArrayList) List(java.util.List) ImmutableList(com.google.common.collect.ImmutableList) Map(java.util.Map) HashMap(java.util.HashMap) Test(org.testng.annotations.Test)

Aggregations

ArrayList (java.util.ArrayList)11 List (java.util.List)11 ObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector)11 ObjectInspectorFactory.getStandardListObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardListObjectInspector)11 ObjectInspectorFactory.getStandardMapObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardMapObjectInspector)11 ObjectInspectorFactory.getStandardStructObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ObjectInspectorFactory.getStandardStructObjectInspector)11 JavaHiveDecimalObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveDecimalObjectInspector)11 PrimitiveObjectInspectorFactory.javaBooleanObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaBooleanObjectInspector)11 PrimitiveObjectInspectorFactory.javaByteArrayObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaByteArrayObjectInspector)11 PrimitiveObjectInspectorFactory.javaByteObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaByteObjectInspector)11 PrimitiveObjectInspectorFactory.javaDateObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaDateObjectInspector)11 PrimitiveObjectInspectorFactory.javaDoubleObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaDoubleObjectInspector)11 PrimitiveObjectInspectorFactory.javaFloatObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaFloatObjectInspector)11 PrimitiveObjectInspectorFactory.javaIntObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaIntObjectInspector)11 PrimitiveObjectInspectorFactory.javaLongObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaLongObjectInspector)11 PrimitiveObjectInspectorFactory.javaShortObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaShortObjectInspector)11 PrimitiveObjectInspectorFactory.javaStringObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaStringObjectInspector)11 PrimitiveObjectInspectorFactory.javaTimestampObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.PrimitiveObjectInspectorFactory.javaTimestampObjectInspector)11 ArrayType (com.facebook.presto.common.type.ArrayType)10 Test (org.testng.annotations.Test)10