Search in sources :

Example 16 with LazyString

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

the class TestLazyAccumuloRow method testDeserializationOfBinaryEncoding.

@Test
public void testDeserializationOfBinaryEncoding() throws Exception {
    List<String> columns = Arrays.asList("row", "given_name", "surname", "age", "weight", "height");
    List<TypeInfo> types = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo, TypeInfoFactory.intTypeInfo, TypeInfoFactory.intTypeInfo, TypeInfoFactory.intTypeInfo);
    LazySimpleStructObjectInspector objectInspector = (LazySimpleStructObjectInspector) LazyFactory.createLazyStructInspector(columns, types, LazySerDeParameters.DefaultSeparators, new Text("\\N"), false, false, (byte) '\\');
    DefaultAccumuloRowIdFactory rowIdFactory = new DefaultAccumuloRowIdFactory();
    Properties props = new Properties();
    props.setProperty(AccumuloSerDeParameters.COLUMN_MAPPINGS, ":rowid#s,personal:given_name#s,personal:surname#s,personal:age,personal:weight,personal:height");
    props.setProperty(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(columns));
    props.setProperty(serdeConstants.LIST_COLUMN_TYPES, Joiner.on(',').join(types));
    props.setProperty(AccumuloSerDeParameters.DEFAULT_STORAGE_TYPE, ColumnEncoding.BINARY.getName());
    AccumuloSerDeParameters params = new AccumuloSerDeParameters(new Configuration(), props, AccumuloSerDe.class.getName());
    rowIdFactory.init(params, props);
    ByteArrayOutputStream baos = new ByteArrayOutputStream();
    DataOutputStream out = new DataOutputStream(baos);
    LazyAccumuloRow lazyRow = new LazyAccumuloRow(objectInspector);
    AccumuloHiveRow hiveRow = new AccumuloHiveRow("1");
    hiveRow.add("personal", "given_name", "Bob".getBytes());
    hiveRow.add("personal", "surname", "Stevens".getBytes());
    out.writeInt(30);
    hiveRow.add("personal", "age", baos.toByteArray());
    baos.reset();
    out.writeInt(200);
    hiveRow.add("personal", "weight", baos.toByteArray());
    baos.reset();
    out.writeInt(72);
    hiveRow.add("personal", "height", baos.toByteArray());
    ColumnMapper columnMapper = params.getColumnMapper();
    lazyRow.init(hiveRow, columnMapper.getColumnMappings(), rowIdFactory);
    Object o = lazyRow.getField(0);
    Assert.assertNotNull(o);
    Assert.assertEquals(LazyString.class, o.getClass());
    Assert.assertEquals("1", ((LazyString) o).toString());
    o = lazyRow.getField(1);
    Assert.assertNotNull(o);
    Assert.assertEquals(LazyString.class, o.getClass());
    Assert.assertEquals("Bob", ((LazyString) o).toString());
    o = lazyRow.getField(2);
    Assert.assertNotNull(o);
    Assert.assertEquals(LazyString.class, o.getClass());
    Assert.assertEquals("Stevens", ((LazyString) o).toString());
    o = lazyRow.getField(3);
    Assert.assertNotNull(o);
    Assert.assertEquals(LazyDioInteger.class, o.getClass());
    Assert.assertEquals("30", ((LazyDioInteger) o).toString());
    o = lazyRow.getField(4);
    Assert.assertNotNull(o);
    Assert.assertEquals(LazyDioInteger.class, o.getClass());
    Assert.assertEquals("200", ((LazyDioInteger) o).toString());
    o = lazyRow.getField(5);
    Assert.assertNotNull(o);
    Assert.assertEquals(LazyDioInteger.class, o.getClass());
    Assert.assertEquals("72", ((LazyDioInteger) o).toString());
}
Also used : LazySimpleStructObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector) Configuration(org.apache.hadoop.conf.Configuration) DataOutputStream(java.io.DataOutputStream) Text(org.apache.hadoop.io.Text) LazyString(org.apache.hadoop.hive.serde2.lazy.LazyString) DefaultAccumuloRowIdFactory(org.apache.hadoop.hive.accumulo.serde.DefaultAccumuloRowIdFactory) ByteArrayOutputStream(java.io.ByteArrayOutputStream) Properties(java.util.Properties) TypeInfo(org.apache.hadoop.hive.serde2.typeinfo.TypeInfo) AccumuloSerDe(org.apache.hadoop.hive.accumulo.serde.AccumuloSerDe) AccumuloSerDeParameters(org.apache.hadoop.hive.accumulo.serde.AccumuloSerDeParameters) ColumnMapper(org.apache.hadoop.hive.accumulo.columns.ColumnMapper) Test(org.junit.Test)

Example 17 with LazyString

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

the class TestLazyAccumuloRow method testExpectedDeserializationOfColumns.

@Test
public void testExpectedDeserializationOfColumns() throws Exception {
    List<String> columns = Arrays.asList("row", "given_name", "surname", "age", "weight", "height");
    List<TypeInfo> types = Arrays.<TypeInfo>asList(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo, TypeInfoFactory.intTypeInfo, TypeInfoFactory.intTypeInfo, TypeInfoFactory.intTypeInfo);
    LazySimpleStructObjectInspector objectInspector = (LazySimpleStructObjectInspector) LazyFactory.createLazyStructInspector(columns, types, LazySerDeParameters.DefaultSeparators, new Text("\\N"), false, false, (byte) '\\');
    DefaultAccumuloRowIdFactory rowIdFactory = new DefaultAccumuloRowIdFactory();
    Properties props = new Properties();
    props.setProperty(AccumuloSerDeParameters.COLUMN_MAPPINGS, ":rowid,personal:given_name,personal:surname,personal:age,personal:weight,personal:height");
    props.setProperty(serdeConstants.LIST_COLUMNS, Joiner.on(',').join(columns));
    props.setProperty(serdeConstants.LIST_COLUMN_TYPES, Joiner.on(',').join(types));
    AccumuloSerDeParameters params = new AccumuloSerDeParameters(new Configuration(), props, AccumuloSerDe.class.getName());
    rowIdFactory.init(params, props);
    LazyAccumuloRow lazyRow = new LazyAccumuloRow(objectInspector);
    AccumuloHiveRow hiveRow = new AccumuloHiveRow("1");
    hiveRow.add("personal", "given_name", "Bob".getBytes());
    hiveRow.add("personal", "surname", "Stevens".getBytes());
    hiveRow.add("personal", "age", "30".getBytes());
    hiveRow.add("personal", "weight", "200".getBytes());
    hiveRow.add("personal", "height", "72".getBytes());
    ColumnMapper columnMapper = params.getColumnMapper();
    lazyRow.init(hiveRow, columnMapper.getColumnMappings(), rowIdFactory);
    Object o = lazyRow.getField(0);
    Assert.assertEquals(LazyString.class, o.getClass());
    Assert.assertEquals("1", ((LazyString) o).toString());
    o = lazyRow.getField(1);
    Assert.assertEquals(LazyString.class, o.getClass());
    Assert.assertEquals("Bob", ((LazyString) o).toString());
    o = lazyRow.getField(2);
    Assert.assertEquals(LazyString.class, o.getClass());
    Assert.assertEquals("Stevens", ((LazyString) o).toString());
    o = lazyRow.getField(3);
    Assert.assertEquals(LazyInteger.class, o.getClass());
    Assert.assertEquals("30", ((LazyInteger) o).toString());
    o = lazyRow.getField(4);
    Assert.assertEquals(LazyInteger.class, o.getClass());
    Assert.assertEquals("200", ((LazyInteger) o).toString());
    o = lazyRow.getField(5);
    Assert.assertEquals(LazyInteger.class, o.getClass());
    Assert.assertEquals("72", ((LazyInteger) o).toString());
}
Also used : LazySimpleStructObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector) Configuration(org.apache.hadoop.conf.Configuration) Text(org.apache.hadoop.io.Text) LazyString(org.apache.hadoop.hive.serde2.lazy.LazyString) DefaultAccumuloRowIdFactory(org.apache.hadoop.hive.accumulo.serde.DefaultAccumuloRowIdFactory) Properties(java.util.Properties) TypeInfo(org.apache.hadoop.hive.serde2.typeinfo.TypeInfo) AccumuloSerDe(org.apache.hadoop.hive.accumulo.serde.AccumuloSerDe) AccumuloSerDeParameters(org.apache.hadoop.hive.accumulo.serde.AccumuloSerDeParameters) ColumnMapper(org.apache.hadoop.hive.accumulo.columns.ColumnMapper) Test(org.junit.Test)

Example 18 with LazyString

use of org.apache.hadoop.hive.serde2.lazy.LazyString 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 = new Date(System.currentTimeMillis());
    DateWritable dateWritable = new DateWritable(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 = new Timestamp(now.getTime());
    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) 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) Timestamp(java.sql.Timestamp) LazyTimestamp(org.apache.hadoop.hive.serde2.lazy.LazyTimestamp) 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) DateWritable(org.apache.hadoop.hive.serde2.io.DateWritable) 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) LazyDate(org.apache.hadoop.hive.serde2.lazy.LazyDate) Date(java.sql.Date) 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 19 with LazyString

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

the class TestLazyHBaseObject method testLazyHBaseCellMap1.

/**
   * Test the LazyMap class with Integer-to-String.
   * @throws SerDeException
   */
public void testLazyHBaseCellMap1() throws SerDeException {
    // Map of Integer to String
    Text nullSequence = new Text("\\N");
    ObjectInspector oi = LazyFactory.createLazyObjectInspector(TypeInfoUtils.getTypeInfosFromTypeString("map<int,string>").get(0), new byte[] { (byte) 1, (byte) 2 }, 0, nullSequence, false, (byte) 0);
    LazyHBaseCellMap b = new LazyHBaseCellMap((LazyMapObjectInspector) oi);
    // Initialize a result
    List<KeyValue> kvs = new ArrayList<KeyValue>();
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfa"), Bytes.toBytes("col1"), Bytes.toBytes("cfacol1")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfa"), Bytes.toBytes("col2"), Bytes.toBytes("cfacol2")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfb"), Bytes.toBytes("2"), Bytes.toBytes("def")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfb"), Bytes.toBytes("-1"), Bytes.toBytes("")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfb"), Bytes.toBytes("0"), Bytes.toBytes("0")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfb"), Bytes.toBytes("8"), Bytes.toBytes("abc")));
    kvs.add(new KeyValue(Bytes.toBytes("test-row"), Bytes.toBytes("cfc"), Bytes.toBytes("col3"), Bytes.toBytes("cfccol3")));
    Result r = new Result(kvs);
    List<Boolean> mapBinaryStorage = new ArrayList<Boolean>();
    mapBinaryStorage.add(false);
    mapBinaryStorage.add(false);
    b.init(r, "cfb".getBytes(), mapBinaryStorage);
    assertEquals(new Text("def"), ((LazyString) b.getMapValueElement(new IntWritable(2))).getWritableObject());
    assertNull(b.getMapValueElement(new IntWritable(-1)));
    assertEquals(new Text("0"), ((LazyString) b.getMapValueElement(new IntWritable(0))).getWritableObject());
    assertEquals(new Text("abc"), ((LazyString) b.getMapValueElement(new IntWritable(8))).getWritableObject());
    assertNull(b.getMapValueElement(new IntWritable(12345)));
    assertEquals("{0:'0',2:'def',8:'abc'}".replace('\'', '\"'), SerDeUtils.getJSONString(b, oi));
}
Also used : LazySimpleStructObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector) ObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector) LazyMapObjectInspector(org.apache.hadoop.hive.serde2.lazy.objectinspector.LazyMapObjectInspector) StructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector) KeyValue(org.apache.hadoop.hbase.KeyValue) ArrayList(java.util.ArrayList) Text(org.apache.hadoop.io.Text) IntWritable(org.apache.hadoop.io.IntWritable) Result(org.apache.hadoop.hbase.client.Result)

Aggregations

LazyString (org.apache.hadoop.hive.serde2.lazy.LazyString)10 Text (org.apache.hadoop.io.Text)10 Test (org.junit.Test)9 Configuration (org.apache.hadoop.conf.Configuration)8 Properties (java.util.Properties)6 AccumuloHiveRow (org.apache.hadoop.hive.accumulo.AccumuloHiveRow)6 ObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector)6 StructObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector)6 LazyAccumuloRow (org.apache.hadoop.hive.accumulo.LazyAccumuloRow)4 ByteArrayOutputStream (java.io.ByteArrayOutputStream)3 Path (org.apache.hadoop.fs.Path)3 ByteArrayRef (org.apache.hadoop.hive.serde2.lazy.ByteArrayRef)3 LazySimpleStructObjectInspector (org.apache.hadoop.hive.serde2.lazy.objectinspector.LazySimpleStructObjectInspector)3 ListObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ListObjectInspector)3 MapObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector)3 PrimitiveObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector)3 DataOutputStream (java.io.DataOutputStream)2 IOException (java.io.IOException)2 Date (java.sql.Date)2 Timestamp (java.sql.Timestamp)2