Examples with HiveVarcharWritable org.apache.hadoop.hive.serde2.io.HiveVarcharWritable used on opensource projects

Example 26 with HiveVarcharWritable

use of org.apache.hadoop.hive.serde2.io.HiveVarcharWritable in project hive by apache.

the class TestGenericUDFSortArrayByField method testSortPrimitiveTupleTwoField.

@Test
public void testSortPrimitiveTupleTwoField() throws HiveException {
    List<ObjectInspector> tuple = new ArrayList<ObjectInspector>();
    tuple.add(PrimitiveObjectInspectorFactory.writableStringObjectInspector);
    tuple.add(PrimitiveObjectInspectorFactory.writableStringObjectInspector);
    ObjectInspector[] inputOIs = { ObjectInspectorFactory.getStandardListObjectInspector(ObjectInspectorFactory.getStandardStructObjectInspector(asList("Company", "Department"), tuple)), PrimitiveObjectInspectorFactory.writableStringObjectInspector, PrimitiveObjectInspectorFactory.writableHiveVarcharObjectInspector };
    udf.initialize(inputOIs);
    Object i1 = asList(new Text("Linkedin"), new Text("HR"));
    Object i2 = asList(new Text("Linkedin"), new Text("IT"));
    Object i3 = asList(new Text("Linkedin"), new Text("Finance"));
    Object i4 = asList(new Text("Facebook"), new Text("IT"));
    Object i5 = asList(new Text("Facebook"), new Text("Finance"));
    Object i6 = asList(new Text("Facebook"), new Text("HR"));
    Object i7 = asList(new Text("Google"), new Text("Logistics"));
    Object i8 = asList(new Text("Google"), new Text("Finance"));
    Object i9 = asList(new Text("Google"), new Text("HR"));
    HiveVarchar vc = new HiveVarchar();
    vc.setValue("Department");
    GenericUDF.DeferredJavaObject[] argas = { new GenericUDF.DeferredJavaObject(asList(i1, i2, i3, i4, i5, i6, i7, i8, i9)), new GenericUDF.DeferredJavaObject(new Text("Company")), new GenericUDF.DeferredJavaObject(new HiveVarcharWritable(vc)) };
    runAndVerify(argas, asList(i5, i6, i4, i8, i9, i7, i3, i1, i2));
}

Example 27 with HiveVarcharWritable

use of org.apache.hadoop.hive.serde2.io.HiveVarcharWritable in project hive by apache.

the class TestGenericUDFSortArrayByField method testSortPrimitiveTupleOneFieldOrderASC.

@Test
public void testSortPrimitiveTupleOneFieldOrderASC() throws HiveException {
    List<ObjectInspector> tuple = new ArrayList<ObjectInspector>();
    tuple.add(PrimitiveObjectInspectorFactory.writableStringObjectInspector);
    tuple.add(PrimitiveObjectInspectorFactory.writableDoubleObjectInspector);
    ObjectInspector[] inputOIs = { ObjectInspectorFactory.getStandardListObjectInspector(ObjectInspectorFactory.getStandardStructObjectInspector(asList("Company", "Salary"), tuple)), PrimitiveObjectInspectorFactory.writableHiveVarcharObjectInspector, PrimitiveObjectInspectorFactory.writableHiveVarcharObjectInspector };
    udf.initialize(inputOIs);
    Object i1 = asList(new Text("Facebook"), new DoubleWritable(80223.25));
    Object i2 = asList(new Text("Facebook"), new DoubleWritable(50223.25));
    Object i3 = asList(new Text("Facebook"), new DoubleWritable(40223.25));
    Object i4 = asList(new Text("Facebook"), new DoubleWritable(60223.25));
    HiveVarchar vc = new HiveVarchar();
    vc.setValue("Salary");
    HiveVarchar order = new HiveVarchar();
    order.setValue("ASC");
    GenericUDF.DeferredJavaObject[] argas = { new GenericUDF.DeferredJavaObject(asList(i1, i2, i3, i4)), new GenericUDF.DeferredJavaObject(new HiveVarcharWritable(vc)), new GenericUDF.DeferredJavaObject(new HiveVarcharWritable(order)) };
    runAndVerify(argas, asList(i3, i2, i4, i1));
}

Example 28 with HiveVarcharWritable

use of org.apache.hadoop.hive.serde2.io.HiveVarcharWritable in project hive by apache.

the class TestGenericUDFSortArrayByField method testSortPrimitiveTupleOneField.

@Test
public void testSortPrimitiveTupleOneField() throws HiveException {
    List<ObjectInspector> tuple = new ArrayList<ObjectInspector>();
    tuple.add(PrimitiveObjectInspectorFactory.writableStringObjectInspector);
    tuple.add(PrimitiveObjectInspectorFactory.writableDoubleObjectInspector);
    ObjectInspector[] inputOIs = { ObjectInspectorFactory.getStandardListObjectInspector(ObjectInspectorFactory.getStandardStructObjectInspector(asList("Company", "Salary"), tuple)), PrimitiveObjectInspectorFactory.writableHiveVarcharObjectInspector };
    udf.initialize(inputOIs);
    Object i1 = asList(new Text("Facebook"), new DoubleWritable(80223.25));
    Object i2 = asList(new Text("Facebook"), new DoubleWritable(50223.25));
    Object i3 = asList(new Text("Facebook"), new DoubleWritable(40223.25));
    Object i4 = asList(new Text("Facebook"), new DoubleWritable(60223.25));
    HiveVarchar vc = new HiveVarchar();
    vc.setValue("Salary");
    GenericUDF.DeferredJavaObject[] argas = { new GenericUDF.DeferredJavaObject(asList(i1, i2, i3, i4)), new GenericUDF.DeferredJavaObject(new HiveVarcharWritable(vc)) };
    runAndVerify(argas, asList(i3, i2, i4, i1));
}

Example 29 with HiveVarcharWritable

use of org.apache.hadoop.hive.serde2.io.HiveVarcharWritable in project hive by apache.

the class BinarySortableSerDe method serialize.

static void serialize(ByteStream.Output buffer, Object o, ObjectInspector oi, boolean invert, byte nullMarker, byte notNullMarker) throws SerDeException {
    // Is this field a null?
    if (o == null) {
        writeByte(buffer, nullMarker, invert);
        return;
    }
    // This field is not a null.
    writeByte(buffer, notNullMarker, invert);
    switch(oi.getCategory()) {
        case PRIMITIVE:
            {
                PrimitiveObjectInspector poi = (PrimitiveObjectInspector) oi;
                switch(poi.getPrimitiveCategory()) {
                    case VOID:
                        {
                            return;
                        }
                    case BOOLEAN:
                        {
                            boolean v = ((BooleanObjectInspector) poi).get(o);
                            writeByte(buffer, (byte) (v ? 2 : 1), invert);
                            return;
                        }
                    case BYTE:
                        {
                            ByteObjectInspector boi = (ByteObjectInspector) poi;
                            byte v = boi.get(o);
                            writeByte(buffer, (byte) (v ^ 0x80), invert);
                            return;
                        }
                    case SHORT:
                        {
                            ShortObjectInspector spoi = (ShortObjectInspector) poi;
                            short v = spoi.get(o);
                            serializeShort(buffer, v, invert);
                            return;
                        }
                    case INT:
                        {
                            IntObjectInspector ioi = (IntObjectInspector) poi;
                            int v = ioi.get(o);
                            serializeInt(buffer, v, invert);
                            return;
                        }
                    case LONG:
                        {
                            LongObjectInspector loi = (LongObjectInspector) poi;
                            long v = loi.get(o);
                            serializeLong(buffer, v, invert);
                            return;
                        }
                    case FLOAT:
                        {
                            FloatObjectInspector foi = (FloatObjectInspector) poi;
                            serializeFloat(buffer, foi.get(o), invert);
                            return;
                        }
                    case DOUBLE:
                        {
                            DoubleObjectInspector doi = (DoubleObjectInspector) poi;
                            serializeDouble(buffer, doi.get(o), invert);
                            return;
                        }
                    case STRING:
                        {
                            StringObjectInspector soi = (StringObjectInspector) poi;
                            Text t = soi.getPrimitiveWritableObject(o);
                            serializeBytes(buffer, t.getBytes(), t.getLength(), invert);
                            return;
                        }
                    case CHAR:
                        {
                            HiveCharObjectInspector hcoi = (HiveCharObjectInspector) poi;
                            HiveCharWritable hc = hcoi.getPrimitiveWritableObject(o);
                            // Trailing space should ignored for char comparisons.
                            // So write stripped values for this SerDe.
                            Text t = hc.getStrippedValue();
                            serializeBytes(buffer, t.getBytes(), t.getLength(), invert);
                            return;
                        }
                    case VARCHAR:
                        {
                            HiveVarcharObjectInspector hcoi = (HiveVarcharObjectInspector) poi;
                            HiveVarcharWritable hc = hcoi.getPrimitiveWritableObject(o);
                            // use varchar's text field directly
                            Text t = hc.getTextValue();
                            serializeBytes(buffer, t.getBytes(), t.getLength(), invert);
                            return;
                        }
                    case BINARY:
                        {
                            BinaryObjectInspector baoi = (BinaryObjectInspector) poi;
                            BytesWritable ba = baoi.getPrimitiveWritableObject(o);
                            byte[] toSer = new byte[ba.getLength()];
                            System.arraycopy(ba.getBytes(), 0, toSer, 0, ba.getLength());
                            serializeBytes(buffer, toSer, ba.getLength(), invert);
                            return;
                        }
                    case DATE:
                        {
                            DateObjectInspector doi = (DateObjectInspector) poi;
                            int v = doi.getPrimitiveWritableObject(o).getDays();
                            serializeInt(buffer, v, invert);
                            return;
                        }
                    case TIMESTAMP:
                        {
                            TimestampObjectInspector toi = (TimestampObjectInspector) poi;
                            TimestampWritable t = toi.getPrimitiveWritableObject(o);
                            serializeTimestampWritable(buffer, t, invert);
                            return;
                        }
                    case INTERVAL_YEAR_MONTH:
                        {
                            HiveIntervalYearMonthObjectInspector ioi = (HiveIntervalYearMonthObjectInspector) poi;
                            HiveIntervalYearMonth intervalYearMonth = ioi.getPrimitiveJavaObject(o);
                            serializeHiveIntervalYearMonth(buffer, intervalYearMonth, invert);
                            return;
                        }
                    case INTERVAL_DAY_TIME:
                        {
                            HiveIntervalDayTimeObjectInspector ioi = (HiveIntervalDayTimeObjectInspector) poi;
                            HiveIntervalDayTime intervalDayTime = ioi.getPrimitiveJavaObject(o);
                            serializeHiveIntervalDayTime(buffer, intervalDayTime, invert);
                            return;
                        }
                    case DECIMAL:
                        {
                            HiveDecimalObjectInspector boi = (HiveDecimalObjectInspector) poi;
                            HiveDecimal dec = boi.getPrimitiveJavaObject(o);
                            serializeHiveDecimal(buffer, dec, invert);
                            return;
                        }
                    default:
                        {
                            throw new RuntimeException("Unrecognized type: " + poi.getPrimitiveCategory());
                        }
                }
            }
        case LIST:
            {
                ListObjectInspector loi = (ListObjectInspector) oi;
                ObjectInspector eoi = loi.getListElementObjectInspector();
                // \1 followed by each element
                int size = loi.getListLength(o);
                for (int eid = 0; eid < size; eid++) {
                    writeByte(buffer, (byte) 1, invert);
                    serialize(buffer, loi.getListElement(o, eid), eoi, invert, nullMarker, notNullMarker);
                }
                // and \0 to terminate
                writeByte(buffer, (byte) 0, invert);
                return;
            }
        case MAP:
            {
                MapObjectInspector moi = (MapObjectInspector) oi;
                ObjectInspector koi = moi.getMapKeyObjectInspector();
                ObjectInspector voi = moi.getMapValueObjectInspector();
                // \1 followed by each key and then each value
                Map<?, ?> map = moi.getMap(o);
                for (Map.Entry<?, ?> entry : map.entrySet()) {
                    writeByte(buffer, (byte) 1, invert);
                    serialize(buffer, entry.getKey(), koi, invert, nullMarker, notNullMarker);
                    serialize(buffer, entry.getValue(), voi, invert, nullMarker, notNullMarker);
                }
                // and \0 to terminate
                writeByte(buffer, (byte) 0, invert);
                return;
            }
        case STRUCT:
            {
                StructObjectInspector soi = (StructObjectInspector) oi;
                List<? extends StructField> fields = soi.getAllStructFieldRefs();
                for (int i = 0; i < fields.size(); i++) {
                    serialize(buffer, soi.getStructFieldData(o, fields.get(i)), fields.get(i).getFieldObjectInspector(), invert, nullMarker, notNullMarker);
                }
                return;
            }
        case UNION:
            {
                UnionObjectInspector uoi = (UnionObjectInspector) oi;
                byte tag = uoi.getTag(o);
                writeByte(buffer, tag, invert);
                serialize(buffer, uoi.getField(o), uoi.getObjectInspectors().get(tag), invert, nullMarker, notNullMarker);
                return;
            }
        default:
            {
                throw new RuntimeException("Unrecognized type: " + oi.getCategory());
            }
    }
}

Example 30 with HiveVarcharWritable

use of org.apache.hadoop.hive.serde2.io.HiveVarcharWritable in project hive by apache.

the class BinarySortableSerDe method deserialize.

static Object deserialize(InputByteBuffer buffer, TypeInfo type, boolean invert, byte nullMarker, byte notNullMarker, Object reuse) throws IOException {
    // Is this field a null?
    byte isNull = buffer.read(invert);
    if (isNull == nullMarker) {
        return null;
    }
    assert (isNull == notNullMarker);
    switch(type.getCategory()) {
        case PRIMITIVE:
            {
                PrimitiveTypeInfo ptype = (PrimitiveTypeInfo) type;
                switch(ptype.getPrimitiveCategory()) {
                    case VOID:
                        {
                            return null;
                        }
                    case BOOLEAN:
                        {
                            BooleanWritable r = reuse == null ? new BooleanWritable() : (BooleanWritable) reuse;
                            byte b = buffer.read(invert);
                            assert (b == 1 || b == 2);
                            r.set(b == 2);
                            return r;
                        }
                    case BYTE:
                        {
                            ByteWritable r = reuse == null ? new ByteWritable() : (ByteWritable) reuse;
                            r.set((byte) (buffer.read(invert) ^ 0x80));
                            return r;
                        }
                    case SHORT:
                        {
                            ShortWritable r = reuse == null ? new ShortWritable() : (ShortWritable) reuse;
                            int v = buffer.read(invert) ^ 0x80;
                            v = (v << 8) + (buffer.read(invert) & 0xff);
                            r.set((short) v);
                            return r;
                        }
                    case INT:
                        {
                            IntWritable r = reuse == null ? new IntWritable() : (IntWritable) reuse;
                            r.set(deserializeInt(buffer, invert));
                            return r;
                        }
                    case LONG:
                        {
                            LongWritable r = reuse == null ? new LongWritable() : (LongWritable) reuse;
                            r.set(deserializeLong(buffer, invert));
                            return r;
                        }
                    case FLOAT:
                        {
                            FloatWritable r = reuse == null ? new FloatWritable() : (FloatWritable) reuse;
                            int v = 0;
                            for (int i = 0; i < 4; i++) {
                                v = (v << 8) + (buffer.read(invert) & 0xff);
                            }
                            if ((v & (1 << 31)) == 0) {
                                // negative number, flip all bits
                                v = ~v;
                            } else {
                                // positive number, flip the first bit
                                v = v ^ (1 << 31);
                            }
                            r.set(Float.intBitsToFloat(v));
                            return r;
                        }
                    case DOUBLE:
                        {
                            DoubleWritable r = reuse == null ? new DoubleWritable() : (DoubleWritable) reuse;
                            long v = 0;
                            for (int i = 0; i < 8; i++) {
                                v = (v << 8) + (buffer.read(invert) & 0xff);
                            }
                            if ((v & (1L << 63)) == 0) {
                                // negative number, flip all bits
                                v = ~v;
                            } else {
                                // positive number, flip the first bit
                                v = v ^ (1L << 63);
                            }
                            r.set(Double.longBitsToDouble(v));
                            return r;
                        }
                    case STRING:
                        {
                            Text r = reuse == null ? new Text() : (Text) reuse;
                            return deserializeText(buffer, invert, r);
                        }
                    case CHAR:
                        {
                            HiveCharWritable r = reuse == null ? new HiveCharWritable() : (HiveCharWritable) reuse;
                            // Use internal text member to read value
                            deserializeText(buffer, invert, r.getTextValue());
                            r.enforceMaxLength(getCharacterMaxLength(type));
                            return r;
                        }
                    case VARCHAR:
                        {
                            HiveVarcharWritable r = reuse == null ? new HiveVarcharWritable() : (HiveVarcharWritable) reuse;
                            // Use HiveVarchar's internal Text member to read the value.
                            deserializeText(buffer, invert, r.getTextValue());
                            // If we cache helper data for deserialization we could avoid having
                            // to call getVarcharMaxLength() on every deserialize call.
                            r.enforceMaxLength(getCharacterMaxLength(type));
                            return r;
                        }
                    case BINARY:
                        {
                            BytesWritable bw = new BytesWritable();
                            // Get the actual length first
                            int start = buffer.tell();
                            int length = 0;
                            do {
                                byte b = buffer.read(invert);
                                if (b == 0) {
                                    // end of string
                                    break;
                                }
                                if (b == 1) {
                                    // the last char is an escape char. read the actual char
                                    buffer.read(invert);
                                }
                                length++;
                            } while (true);
                            if (length == buffer.tell() - start) {
                                // No escaping happened, so we are already done.
                                bw.set(buffer.getData(), start, length);
                            } else {
                                // Escaping happened, we need to copy byte-by-byte.
                                // 1. Set the length first.
                                bw.set(buffer.getData(), start, length);
                                // 2. Reset the pointer.
                                buffer.seek(start);
                                // 3. Copy the data.
                                byte[] rdata = bw.getBytes();
                                for (int i = 0; i < length; i++) {
                                    byte b = buffer.read(invert);
                                    if (b == 1) {
                                        // The last char is an escape char, read the actual char.
                                        // The serialization format escape \0 to \1, and \1 to \2,
                                        // to make sure the string is null-terminated.
                                        b = (byte) (buffer.read(invert) - 1);
                                    }
                                    rdata[i] = b;
                                }
                                // 4. Read the null terminator.
                                byte b = buffer.read(invert);
                                assert (b == 0);
                            }
                            return bw;
                        }
                    case DATE:
                        {
                            DateWritable d = reuse == null ? new DateWritable() : (DateWritable) reuse;
                            d.set(deserializeInt(buffer, invert));
                            return d;
                        }
                    case TIMESTAMP:
                        TimestampWritable t = (reuse == null ? new TimestampWritable() : (TimestampWritable) reuse);
                        byte[] bytes = new byte[TimestampWritable.BINARY_SORTABLE_LENGTH];
                        for (int i = 0; i < bytes.length; i++) {
                            bytes[i] = buffer.read(invert);
                        }
                        t.setBinarySortable(bytes, 0);
                        return t;
                    case INTERVAL_YEAR_MONTH:
                        {
                            HiveIntervalYearMonthWritable i = reuse == null ? new HiveIntervalYearMonthWritable() : (HiveIntervalYearMonthWritable) reuse;
                            i.set(deserializeInt(buffer, invert));
                            return i;
                        }
                    case INTERVAL_DAY_TIME:
                        {
                            HiveIntervalDayTimeWritable i = reuse == null ? new HiveIntervalDayTimeWritable() : (HiveIntervalDayTimeWritable) reuse;
                            long totalSecs = deserializeLong(buffer, invert);
                            int nanos = deserializeInt(buffer, invert);
                            i.set(totalSecs, nanos);
                            return i;
                        }
                    case DECIMAL:
                        {
                            // See serialization of decimal for explanation (below)
                            HiveDecimalWritable bdw = (reuse == null ? new HiveDecimalWritable() : (HiveDecimalWritable) reuse);
                            int b = buffer.read(invert) - 1;
                            assert (b == 1 || b == -1 || b == 0);
                            boolean positive = b != -1;
                            int factor = buffer.read(invert) ^ 0x80;
                            for (int i = 0; i < 3; i++) {
                                factor = (factor << 8) + (buffer.read(invert) & 0xff);
                            }
                            if (!positive) {
                                factor = -factor;
                            }
                            int start = buffer.tell();
                            int length = 0;
                            do {
                                b = buffer.read(positive ? invert : !invert);
                                assert (b != 1);
                                if (b == 0) {
                                    // end of digits
                                    break;
                                }
                                length++;
                            } while (true);
                            final byte[] decimalBuffer = new byte[length];
                            buffer.seek(start);
                            for (int i = 0; i < length; ++i) {
                                decimalBuffer[i] = buffer.read(positive ? invert : !invert);
                            }
                            // read the null byte again
                            buffer.read(positive ? invert : !invert);
                            String digits = new String(decimalBuffer, 0, length, decimalCharSet);
                            BigInteger bi = new BigInteger(digits);
                            HiveDecimal bd = HiveDecimal.create(bi).scaleByPowerOfTen(factor - length);
                            if (!positive) {
                                bd = bd.negate();
                            }
                            bdw.set(bd);
                            return bdw;
                        }
                    default:
                        {
                            throw new RuntimeException("Unrecognized type: " + ptype.getPrimitiveCategory());
                        }
                }
            }
        case LIST:
            {
                ListTypeInfo ltype = (ListTypeInfo) type;
                TypeInfo etype = ltype.getListElementTypeInfo();
                // Create the list if needed
                ArrayList<Object> r = reuse == null ? new ArrayList<Object>() : (ArrayList<Object>) reuse;
                // Read the list
                int size = 0;
                while (true) {
                    int more = buffer.read(invert);
                    if (more == 0) {
                        // \0 to terminate
                        break;
                    }
                    // \1 followed by each element
                    assert (more == 1);
                    if (size == r.size()) {
                        r.add(null);
                    }
                    r.set(size, deserialize(buffer, etype, invert, nullMarker, notNullMarker, r.get(size)));
                    size++;
                }
                // Remove additional elements if the list is reused
                while (r.size() > size) {
                    r.remove(r.size() - 1);
                }
                return r;
            }
        case MAP:
            {
                MapTypeInfo mtype = (MapTypeInfo) type;
                TypeInfo ktype = mtype.getMapKeyTypeInfo();
                TypeInfo vtype = mtype.getMapValueTypeInfo();
                // Create the map if needed
                Map<Object, Object> r;
                if (reuse == null) {
                    r = new HashMap<Object, Object>();
                } else {
                    r = (HashMap<Object, Object>) reuse;
                    r.clear();
                }
                while (true) {
                    int more = buffer.read(invert);
                    if (more == 0) {
                        // \0 to terminate
                        break;
                    }
                    // \1 followed by each key and then each value
                    assert (more == 1);
                    Object k = deserialize(buffer, ktype, invert, nullMarker, notNullMarker, null);
                    Object v = deserialize(buffer, vtype, invert, nullMarker, notNullMarker, null);
                    r.put(k, v);
                }
                return r;
            }
        case STRUCT:
            {
                StructTypeInfo stype = (StructTypeInfo) type;
                List<TypeInfo> fieldTypes = stype.getAllStructFieldTypeInfos();
                int size = fieldTypes.size();
                // Create the struct if needed
                ArrayList<Object> r = reuse == null ? new ArrayList<Object>(size) : (ArrayList<Object>) reuse;
                assert (r.size() <= size);
                // Set the size of the struct
                while (r.size() < size) {
                    r.add(null);
                }
                // Read one field by one field
                for (int eid = 0; eid < size; eid++) {
                    r.set(eid, deserialize(buffer, fieldTypes.get(eid), invert, nullMarker, notNullMarker, r.get(eid)));
                }
                return r;
            }
        case UNION:
            {
                UnionTypeInfo utype = (UnionTypeInfo) type;
                StandardUnion r = reuse == null ? new StandardUnion() : (StandardUnion) reuse;
                // Read the tag
                byte tag = buffer.read(invert);
                r.setTag(tag);
                r.setObject(deserialize(buffer, utype.getAllUnionObjectTypeInfos().get(tag), invert, nullMarker, notNullMarker, null));
                return r;
            }
        default:
            {
                throw new RuntimeException("Unrecognized type: " + type.getCategory());
            }
    }
}