Search in sources :

Example 1 with JavaHiveDecimalObjectInspector

use of org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveDecimalObjectInspector 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 2 with JavaHiveDecimalObjectInspector

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

the class AvroDeserializer method deserializePrimitive.

private Object deserializePrimitive(Object datum, Schema fileSchema, Schema recordSchema, PrimitiveTypeInfo columnType) throws AvroSerdeException {
    switch(columnType.getPrimitiveCategory()) {
        case STRING:
            // To workaround AvroUTF8
            return datum.toString();
        // and convert it to a string. Yay!
        case BINARY:
            if (recordSchema.getType() == Type.FIXED) {
                Fixed fixed = (Fixed) datum;
                return fixed.bytes();
            } else if (recordSchema.getType() == Type.BYTES) {
                return AvroSerdeUtils.getBytesFromByteBuffer((ByteBuffer) datum);
            } else {
                throw new AvroSerdeException("Unexpected Avro schema for Binary TypeInfo: " + recordSchema.getType());
            }
        case DECIMAL:
            if (fileSchema == null) {
                throw new AvroSerdeException("File schema is missing for decimal field. Reader schema is " + columnType);
            }
            int scale = 0;
            try {
                scale = AvroSerdeUtils.getIntFromSchema(fileSchema, AvroSerDe.AVRO_PROP_SCALE);
            } catch (Exception ex) {
                throw new AvroSerdeException("Failed to obtain scale value from file schema: " + fileSchema, ex);
            }
            HiveDecimal dec = AvroSerdeUtils.getHiveDecimalFromByteBuffer((ByteBuffer) datum, scale);
            JavaHiveDecimalObjectInspector oi = (JavaHiveDecimalObjectInspector) PrimitiveObjectInspectorFactory.getPrimitiveJavaObjectInspector((DecimalTypeInfo) columnType);
            return oi.set(null, dec);
        case CHAR:
            if (fileSchema == null) {
                throw new AvroSerdeException("File schema is missing for char field. Reader schema is " + columnType);
            }
            int maxLength = 0;
            try {
                maxLength = AvroSerdeUtils.getIntFromSchema(fileSchema, AvroSerDe.AVRO_PROP_MAX_LENGTH);
            } catch (Exception ex) {
                throw new AvroSerdeException("Failed to obtain maxLength value for char field from file schema: " + fileSchema, ex);
            }
            String str = datum.toString();
            HiveChar hc = new HiveChar(str, maxLength);
            return hc;
        case VARCHAR:
            if (fileSchema == null) {
                throw new AvroSerdeException("File schema is missing for varchar field. Reader schema is " + columnType);
            }
            maxLength = 0;
            try {
                maxLength = AvroSerdeUtils.getIntFromSchema(fileSchema, AvroSerDe.AVRO_PROP_MAX_LENGTH);
            } catch (Exception ex) {
                throw new AvroSerdeException("Failed to obtain maxLength value for varchar field from file schema: " + fileSchema, ex);
            }
            str = datum.toString();
            HiveVarchar hvc = new HiveVarchar(str, maxLength);
            return hvc;
        case DATE:
            {
                if (recordSchema.getType() != Type.INT) {
                    throw new AvroSerdeException("Unexpected Avro schema for Date TypeInfo: " + recordSchema.getType());
                }
                final boolean skipProlepticConversion;
                if (writerProleptic != null) {
                    skipProlepticConversion = writerProleptic;
                } else {
                    if (configuration != null) {
                        skipProlepticConversion = HiveConf.getBoolVar(configuration, HiveConf.ConfVars.HIVE_AVRO_PROLEPTIC_GREGORIAN_DEFAULT);
                    } else {
                        skipProlepticConversion = HiveConf.ConfVars.HIVE_AVRO_PROLEPTIC_GREGORIAN_DEFAULT.defaultBoolVal;
                    }
                }
                return Date.ofEpochMilli(DateWritableV2.daysToMillis(skipProlepticConversion ? (Integer) datum : CalendarUtils.convertDateToProleptic((Integer) datum)));
            }
        case TIMESTAMP:
            {
                if (recordSchema.getType() != Type.LONG) {
                    throw new AvroSerdeException("Unexpected Avro schema for Date TypeInfo: " + recordSchema.getType());
                }
                // If a time zone is found in file metadata (property name: writer.time.zone), convert the
                // timestamp to that (writer) time zone in order to emulate time zone agnostic behavior.
                // If not, then the file was written by an older version of hive, so we convert the timestamp
                // to the server's (reader) time zone for backwards compatibility reasons - unless the
                // session level configuration hive.avro.timestamp.skip.conversion is set to true, in which
                // case we assume it was written by a time zone agnostic writer, so we don't convert it.
                final boolean skipUTCConversion;
                if (configuration != null) {
                    skipUTCConversion = HiveConf.getBoolVar(configuration, HiveConf.ConfVars.HIVE_AVRO_TIMESTAMP_SKIP_CONVERSION);
                } else {
                    skipUTCConversion = HiveConf.ConfVars.HIVE_AVRO_TIMESTAMP_SKIP_CONVERSION.defaultBoolVal;
                }
                final boolean legacyConversion;
                if (writerZoneConversionLegacy != null) {
                    legacyConversion = writerZoneConversionLegacy;
                } else if (writerTimezone != null) {
                    legacyConversion = false;
                } else if (configuration != null) {
                    legacyConversion = HiveConf.getBoolVar(configuration, HiveConf.ConfVars.HIVE_AVRO_TIMESTAMP_LEGACY_CONVERSION_ENABLED);
                } else {
                    legacyConversion = HiveConf.ConfVars.HIVE_AVRO_TIMESTAMP_LEGACY_CONVERSION_ENABLED.defaultBoolVal;
                }
                ZoneId convertToTimeZone;
                if (writerTimezone != null) {
                    convertToTimeZone = writerTimezone;
                } else if (skipUTCConversion) {
                    convertToTimeZone = ZoneOffset.UTC;
                } else {
                    convertToTimeZone = TimeZone.getDefault().toZoneId();
                }
                final boolean skipProlepticConversion;
                if (writerProleptic != null) {
                    skipProlepticConversion = writerProleptic;
                } else {
                    if (configuration != null) {
                        skipProlepticConversion = HiveConf.getBoolVar(configuration, HiveConf.ConfVars.HIVE_AVRO_PROLEPTIC_GREGORIAN_DEFAULT);
                    } else {
                        skipProlepticConversion = HiveConf.ConfVars.HIVE_AVRO_PROLEPTIC_GREGORIAN_DEFAULT.defaultBoolVal;
                    }
                }
                Timestamp timestamp = TimestampTZUtil.convertTimestampToZone(Timestamp.ofEpochMilli((Long) datum), ZoneOffset.UTC, convertToTimeZone, legacyConversion);
                if (!skipProlepticConversion) {
                    timestamp = Timestamp.ofEpochMilli(CalendarUtils.convertTimeToProleptic(timestamp.toEpochMilli()));
                }
                return timestamp;
            }
        default:
            return datum;
    }
}
Also used : ZoneId(java.time.ZoneId) HiveChar(org.apache.hadoop.hive.common.type.HiveChar) HiveVarchar(org.apache.hadoop.hive.common.type.HiveVarchar) ByteBuffer(java.nio.ByteBuffer) Timestamp(org.apache.hadoop.hive.common.type.Timestamp) IOException(java.io.IOException) DecimalTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo) JavaHiveDecimalObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveDecimalObjectInspector) HiveDecimal(org.apache.hadoop.hive.common.type.HiveDecimal) Fixed(org.apache.avro.generic.GenericData.Fixed)

Example 3 with JavaHiveDecimalObjectInspector

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

the class AbstractTestParquetReader method testDecimalBackedByFixedLenByteArray.

@Test
public void testDecimalBackedByFixedLenByteArray() throws Exception {
    for (int precision = MAX_PRECISION_INT64 + 1; precision < MAX_PRECISION; precision++) {
        int scale = ThreadLocalRandom.current().nextInt(precision);
        ContiguousSet<BigInteger> values = bigIntegersBetween(BigDecimal.valueOf(Math.pow(10, precision - 1)).toBigInteger(), BigDecimal.valueOf(Math.pow(10, precision)).toBigInteger());
        ImmutableList.Builder<SqlDecimal> expectedValues = new ImmutableList.Builder<>();
        ImmutableList.Builder<HiveDecimal> writeValues = new ImmutableList.Builder<>();
        for (BigInteger value : limit(values, 1_000)) {
            writeValues.add(HiveDecimal.create(value, scale));
            expectedValues.add(new SqlDecimal(value, precision, scale));
        }
        tester.testRoundTrip(new JavaHiveDecimalObjectInspector(new DecimalTypeInfo(precision, scale)), writeValues.build(), expectedValues.build(), createDecimalType(precision, scale));
    }
}
Also used : DecimalTypeInfo(org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo) JavaHiveDecimalObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveDecimalObjectInspector) ImmutableList(com.google.common.collect.ImmutableList) CacheBuilder(com.google.common.cache.CacheBuilder) HiveDecimal(org.apache.hadoop.hive.common.type.HiveDecimal) BigInteger(java.math.BigInteger) SqlDecimal(com.facebook.presto.common.type.SqlDecimal) Test(org.testng.annotations.Test)

Aggregations

HiveDecimal (org.apache.hadoop.hive.common.type.HiveDecimal)3 JavaHiveDecimalObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.JavaHiveDecimalObjectInspector)3 DecimalTypeInfo (org.apache.hadoop.hive.serde2.typeinfo.DecimalTypeInfo)3 HiveChar (org.apache.hadoop.hive.common.type.HiveChar)2 HiveVarchar (org.apache.hadoop.hive.common.type.HiveVarchar)2 Timestamp (org.apache.hadoop.hive.common.type.Timestamp)2 SqlDecimal (com.facebook.presto.common.type.SqlDecimal)1 CacheBuilder (com.google.common.cache.CacheBuilder)1 ImmutableList (com.google.common.collect.ImmutableList)1 ByteArrayOutputStream (java.io.ByteArrayOutputStream)1 IOException (java.io.IOException)1 BigInteger (java.math.BigInteger)1 ByteBuffer (java.nio.ByteBuffer)1 ZoneId (java.time.ZoneId)1 BatchWriter (org.apache.accumulo.core.client.BatchWriter)1 BatchWriterConfig (org.apache.accumulo.core.client.BatchWriterConfig)1 Connector (org.apache.accumulo.core.client.Connector)1 MockInstance (org.apache.accumulo.core.client.mock.MockInstance)1 PasswordToken (org.apache.accumulo.core.client.security.tokens.PasswordToken)1 Key (org.apache.accumulo.core.data.Key)1