Examples with LazyString org.apache.hadoop.hive.serde2.lazy.LazyString used on opensource projects

Example 11 with LazyString

use of org.apache.hadoop.hive.serde2.lazy.LazyString 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 = new Date(System.currentTimeMillis());
    DateWritable dateWritable = new DateWritable(now);
    Date dateValue = dateWritable.get();
    dateWritable.write(out);
    m.put(cfBytes, "date".getBytes(), baos.toByteArray());
    // tiemestamp
    baos.reset();
    Timestamp timestampValue = new Timestamp(now.getTime());
    ByteStream.Output output = new ByteStream.Output();
    TimestampWritable timestampWritable = new TimestampWritable(new Timestamp(now.getTime()));
    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 12 with LazyString

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

the class AccumuloRowSerializer method serializeRowId.

/**
   * Serialize an Accumulo rowid
   */
protected byte[] serializeRowId(Object rowId, StructField rowIdField, ColumnMapping rowIdMapping) throws IOException {
    if (rowId == null) {
        throw new IOException("Accumulo rowId cannot be NULL");
    }
    // Reset the buffer we're going to use
    output.reset();
    ObjectInspector rowIdFieldOI = rowIdField.getFieldObjectInspector();
    String rowIdMappingType = rowIdMapping.getColumnType();
    TypeInfo rowIdTypeInfo = TypeInfoUtils.getTypeInfoFromTypeString(rowIdMappingType);
    if (!rowIdFieldOI.getCategory().equals(ObjectInspector.Category.PRIMITIVE) && rowIdTypeInfo.getCategory() == ObjectInspector.Category.PRIMITIVE) {
        // we always serialize the String type using the escaped algorithm for LazyString
        writeString(output, SerDeUtils.getJSONString(rowId, rowIdFieldOI), PrimitiveObjectInspectorFactory.javaStringObjectInspector);
        return output.toByteArray();
    }
    // use the serialization option switch to write primitive values as either a variable
    // length UTF8 string or a fixed width bytes if serializing in binary format
    getSerializedValue(rowIdFieldOI, rowId, output, rowIdMapping);
    return output.toByteArray();
}

Example 13 with LazyString

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

the class TestLazyAccumuloMap method testStringMapWithProjection.

@Test
public void testStringMapWithProjection() throws SerDeException {
    AccumuloHiveRow row = new AccumuloHiveRow("row");
    row.add("cf1", "foo", "bar".getBytes());
    row.add("cf1", "bar", "foo".getBytes());
    row.add("cf2", "foo1", "bar1".getBytes());
    row.add("cf3", "bar1", "foo1".getBytes());
    HiveAccumuloMapColumnMapping mapping = new HiveAccumuloMapColumnMapping("cf1", null, ColumnEncoding.STRING, ColumnEncoding.STRING, "column", TypeInfoFactory.getMapTypeInfo(TypeInfoFactory.stringTypeInfo, TypeInfoFactory.stringTypeInfo).toString());
    // Map of Integer to String
    Text nullSequence = new Text("\\N");
    ObjectInspector oi = LazyFactory.createLazyObjectInspector(TypeInfoUtils.getTypeInfosFromTypeString("map<string,string>").get(0), new byte[] { (byte) 1, (byte) 2 }, 0, nullSequence, false, (byte) 0);
    LazyAccumuloMap map = new LazyAccumuloMap((LazyMapObjectInspector) oi);
    map.init(row, mapping);
    Assert.assertEquals(2, map.getMapSize());
    Object o = map.getMapValueElement(new Text("foo"));
    Assert.assertNotNull(o);
    Assert.assertEquals(new Text("bar"), ((LazyString) o).getWritableObject());
    o = map.getMapValueElement(new Text("bar"));
    Assert.assertNotNull(o);
    Assert.assertEquals(new Text("foo"), ((LazyString) o).getWritableObject());
}

Example 14 with LazyString

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

the class TestLazyHBaseObject method testLazyHBaseCellMap2.

/**
   * Test the LazyMap class with String-to-String.
   * @throws SerDeException
   */
public void testLazyHBaseCellMap2() throws SerDeException {
    // Map of String to String
    Text nullSequence = new Text("\\N");
    ObjectInspector oi = LazyFactory.createLazyObjectInspector(TypeInfoUtils.getTypeInfosFromTypeString("map<string,string>").get(0), new byte[] { (byte) '#', (byte) '\t' }, 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("d\tf")));
    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("d\tf"), ((LazyString) b.getMapValueElement(new Text("2"))).getWritableObject());
    assertNull(b.getMapValueElement(new Text("-1")));
    assertEquals(new Text("0"), ((LazyString) b.getMapValueElement(new Text("0"))).getWritableObject());
    assertEquals(new Text("abc"), ((LazyString) b.getMapValueElement(new Text("8"))).getWritableObject());
    assertNull(b.getMapValueElement(new Text("-")));
    assertEquals("{'0':'0','2':'d\\tf','8':'abc'}".replace('\'', '\"'), SerDeUtils.getJSONString(b, oi));
}

Example 15 with LazyString

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

the class TestLazyMap method lazyString.

private static LazyString lazyString(String string) {
    LazyString lazyString = new LazyString(LAZY_STRING_OBJECT_INSPECTOR);
    lazyString.init(newByteArrayRef(string), 0, string.length());
    return lazyString;
}