Search in sources :

Example 16 with FloatObjectInspector

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

the class TestOrcFile method testReadFormat_0_11.

@Test
public void testReadFormat_0_11() throws Exception {
    Path oldFilePath = new Path(HiveTestUtils.getFileFromClasspath("orc-file-11-format.orc"));
    Reader reader = OrcFile.createReader(oldFilePath, OrcFile.readerOptions(conf).filesystem(fs));
    int stripeCount = 0;
    int rowCount = 0;
    long currentOffset = -1;
    for (StripeInformation stripe : reader.getStripes()) {
        stripeCount += 1;
        rowCount += stripe.getNumberOfRows();
        if (currentOffset < 0) {
            currentOffset = stripe.getOffset() + stripe.getIndexLength() + stripe.getDataLength() + stripe.getFooterLength();
        } else {
            assertEquals(currentOffset, stripe.getOffset());
            currentOffset += stripe.getIndexLength() + stripe.getDataLength() + stripe.getFooterLength();
        }
    }
    assertEquals(reader.getNumberOfRows(), rowCount);
    assertEquals(2, stripeCount);
    // check the stats
    ColumnStatistics[] stats = reader.getStatistics();
    assertEquals(7500, stats[1].getNumberOfValues());
    assertEquals(3750, ((BooleanColumnStatistics) stats[1]).getFalseCount());
    assertEquals(3750, ((BooleanColumnStatistics) stats[1]).getTrueCount());
    assertEquals("count: 7500 hasNull: true true: 3750", stats[1].toString());
    assertEquals(2048, ((IntegerColumnStatistics) stats[3]).getMaximum());
    assertEquals(1024, ((IntegerColumnStatistics) stats[3]).getMinimum());
    assertEquals(true, ((IntegerColumnStatistics) stats[3]).isSumDefined());
    assertEquals(11520000, ((IntegerColumnStatistics) stats[3]).getSum());
    assertEquals("count: 7500 hasNull: true min: 1024 max: 2048 sum: 11520000", stats[3].toString());
    assertEquals(Long.MAX_VALUE, ((IntegerColumnStatistics) stats[5]).getMaximum());
    assertEquals(Long.MAX_VALUE, ((IntegerColumnStatistics) stats[5]).getMinimum());
    assertEquals(false, ((IntegerColumnStatistics) stats[5]).isSumDefined());
    assertEquals("count: 7500 hasNull: true min: 9223372036854775807 max: 9223372036854775807", stats[5].toString());
    assertEquals(-15.0, ((DoubleColumnStatistics) stats[7]).getMinimum());
    assertEquals(-5.0, ((DoubleColumnStatistics) stats[7]).getMaximum());
    assertEquals(-75000.0, ((DoubleColumnStatistics) stats[7]).getSum(), 0.00001);
    assertEquals("count: 7500 hasNull: true min: -15.0 max: -5.0 sum: -75000.0", stats[7].toString());
    assertEquals("count: 7500 hasNull: true min: bye max: hi sum: 0", stats[9].toString());
    // check the inspectors
    StructObjectInspector readerInspector = (StructObjectInspector) reader.getObjectInspector();
    assertEquals(ObjectInspector.Category.STRUCT, readerInspector.getCategory());
    assertEquals("struct<boolean1:boolean,byte1:tinyint,short1:smallint," + "int1:int,long1:bigint,float1:float,double1:double,bytes1:" + "binary,string1:string,middle:struct<list:array<struct<int1:int," + "string1:string>>>,list:array<struct<int1:int,string1:string>>," + "map:map<string,struct<int1:int,string1:string>>,ts:timestamp," + "decimal1:decimal(38,10)>", readerInspector.getTypeName());
    List<? extends StructField> fields = readerInspector.getAllStructFieldRefs();
    BooleanObjectInspector bo = (BooleanObjectInspector) readerInspector.getStructFieldRef("boolean1").getFieldObjectInspector();
    ByteObjectInspector by = (ByteObjectInspector) readerInspector.getStructFieldRef("byte1").getFieldObjectInspector();
    ShortObjectInspector sh = (ShortObjectInspector) readerInspector.getStructFieldRef("short1").getFieldObjectInspector();
    IntObjectInspector in = (IntObjectInspector) readerInspector.getStructFieldRef("int1").getFieldObjectInspector();
    LongObjectInspector lo = (LongObjectInspector) readerInspector.getStructFieldRef("long1").getFieldObjectInspector();
    FloatObjectInspector fl = (FloatObjectInspector) readerInspector.getStructFieldRef("float1").getFieldObjectInspector();
    DoubleObjectInspector dbl = (DoubleObjectInspector) readerInspector.getStructFieldRef("double1").getFieldObjectInspector();
    BinaryObjectInspector bi = (BinaryObjectInspector) readerInspector.getStructFieldRef("bytes1").getFieldObjectInspector();
    StringObjectInspector st = (StringObjectInspector) readerInspector.getStructFieldRef("string1").getFieldObjectInspector();
    StructObjectInspector mid = (StructObjectInspector) readerInspector.getStructFieldRef("middle").getFieldObjectInspector();
    List<? extends StructField> midFields = mid.getAllStructFieldRefs();
    ListObjectInspector midli = (ListObjectInspector) midFields.get(0).getFieldObjectInspector();
    StructObjectInspector inner = (StructObjectInspector) midli.getListElementObjectInspector();
    List<? extends StructField> inFields = inner.getAllStructFieldRefs();
    ListObjectInspector li = (ListObjectInspector) readerInspector.getStructFieldRef("list").getFieldObjectInspector();
    MapObjectInspector ma = (MapObjectInspector) readerInspector.getStructFieldRef("map").getFieldObjectInspector();
    TimestampObjectInspector tso = (TimestampObjectInspector) readerInspector.getStructFieldRef("ts").getFieldObjectInspector();
    HiveDecimalObjectInspector dco = (HiveDecimalObjectInspector) readerInspector.getStructFieldRef("decimal1").getFieldObjectInspector();
    StringObjectInspector mk = (StringObjectInspector) ma.getMapKeyObjectInspector();
    RecordReader rows = reader.rows();
    Object row = rows.next(null);
    assertNotNull(row);
    // check the contents of the first row
    assertEquals(false, bo.get(readerInspector.getStructFieldData(row, fields.get(0))));
    assertEquals(1, by.get(readerInspector.getStructFieldData(row, fields.get(1))));
    assertEquals(1024, sh.get(readerInspector.getStructFieldData(row, fields.get(2))));
    assertEquals(65536, in.get(readerInspector.getStructFieldData(row, fields.get(3))));
    assertEquals(Long.MAX_VALUE, lo.get(readerInspector.getStructFieldData(row, fields.get(4))));
    assertEquals(1.0, fl.get(readerInspector.getStructFieldData(row, fields.get(5))), 0.00001);
    assertEquals(-15.0, dbl.get(readerInspector.getStructFieldData(row, fields.get(6))), 0.00001);
    assertEquals(bytes(0, 1, 2, 3, 4), bi.getPrimitiveWritableObject(readerInspector.getStructFieldData(row, fields.get(7))));
    assertEquals("hi", st.getPrimitiveJavaObject(readerInspector.getStructFieldData(row, fields.get(8))));
    List<?> midRow = midli.getList(mid.getStructFieldData(readerInspector.getStructFieldData(row, fields.get(9)), midFields.get(0)));
    assertNotNull(midRow);
    assertEquals(2, midRow.size());
    assertEquals(1, in.get(inner.getStructFieldData(midRow.get(0), inFields.get(0))));
    assertEquals("bye", st.getPrimitiveJavaObject(inner.getStructFieldData(midRow.get(0), inFields.get(1))));
    assertEquals(2, in.get(inner.getStructFieldData(midRow.get(1), inFields.get(0))));
    assertEquals("sigh", st.getPrimitiveJavaObject(inner.getStructFieldData(midRow.get(1), inFields.get(1))));
    List<?> list = li.getList(readerInspector.getStructFieldData(row, fields.get(10)));
    assertEquals(2, list.size());
    assertEquals(3, in.get(inner.getStructFieldData(list.get(0), inFields.get(0))));
    assertEquals("good", st.getPrimitiveJavaObject(inner.getStructFieldData(list.get(0), inFields.get(1))));
    assertEquals(4, in.get(inner.getStructFieldData(list.get(1), inFields.get(0))));
    assertEquals("bad", st.getPrimitiveJavaObject(inner.getStructFieldData(list.get(1), inFields.get(1))));
    Map<?, ?> map = ma.getMap(readerInspector.getStructFieldData(row, fields.get(11)));
    assertEquals(0, map.size());
    assertEquals(Timestamp.valueOf("2000-03-12 15:00:00"), tso.getPrimitiveJavaObject(readerInspector.getStructFieldData(row, fields.get(12))));
    assertEquals(HiveDecimal.create("12345678.6547456"), dco.getPrimitiveJavaObject(readerInspector.getStructFieldData(row, fields.get(13))));
    // check the contents of second row
    assertEquals(true, rows.hasNext());
    rows.seekToRow(7499);
    row = rows.next(null);
    assertEquals(true, bo.get(readerInspector.getStructFieldData(row, fields.get(0))));
    assertEquals(100, by.get(readerInspector.getStructFieldData(row, fields.get(1))));
    assertEquals(2048, sh.get(readerInspector.getStructFieldData(row, fields.get(2))));
    assertEquals(65536, in.get(readerInspector.getStructFieldData(row, fields.get(3))));
    assertEquals(Long.MAX_VALUE, lo.get(readerInspector.getStructFieldData(row, fields.get(4))));
    assertEquals(2.0, fl.get(readerInspector.getStructFieldData(row, fields.get(5))), 0.00001);
    assertEquals(-5.0, dbl.get(readerInspector.getStructFieldData(row, fields.get(6))), 0.00001);
    assertEquals(bytes(), bi.getPrimitiveWritableObject(readerInspector.getStructFieldData(row, fields.get(7))));
    assertEquals("bye", st.getPrimitiveJavaObject(readerInspector.getStructFieldData(row, fields.get(8))));
    midRow = midli.getList(mid.getStructFieldData(readerInspector.getStructFieldData(row, fields.get(9)), midFields.get(0)));
    assertNotNull(midRow);
    assertEquals(2, midRow.size());
    assertEquals(1, in.get(inner.getStructFieldData(midRow.get(0), inFields.get(0))));
    assertEquals("bye", st.getPrimitiveJavaObject(inner.getStructFieldData(midRow.get(0), inFields.get(1))));
    assertEquals(2, in.get(inner.getStructFieldData(midRow.get(1), inFields.get(0))));
    assertEquals("sigh", st.getPrimitiveJavaObject(inner.getStructFieldData(midRow.get(1), inFields.get(1))));
    list = li.getList(readerInspector.getStructFieldData(row, fields.get(10)));
    assertEquals(3, list.size());
    assertEquals(100000000, in.get(inner.getStructFieldData(list.get(0), inFields.get(0))));
    assertEquals("cat", st.getPrimitiveJavaObject(inner.getStructFieldData(list.get(0), inFields.get(1))));
    assertEquals(-100000, in.get(inner.getStructFieldData(list.get(1), inFields.get(0))));
    assertEquals("in", st.getPrimitiveJavaObject(inner.getStructFieldData(list.get(1), inFields.get(1))));
    assertEquals(1234, in.get(inner.getStructFieldData(list.get(2), inFields.get(0))));
    assertEquals("hat", st.getPrimitiveJavaObject(inner.getStructFieldData(list.get(2), inFields.get(1))));
    map = ma.getMap(readerInspector.getStructFieldData(row, fields.get(11)));
    assertEquals(2, map.size());
    boolean[] found = new boolean[2];
    for (Object key : map.keySet()) {
        String str = mk.getPrimitiveJavaObject(key);
        if (str.equals("chani")) {
            assertEquals(false, found[0]);
            assertEquals(5, in.get(inner.getStructFieldData(map.get(key), inFields.get(0))));
            assertEquals(str, st.getPrimitiveJavaObject(inner.getStructFieldData(map.get(key), inFields.get(1))));
            found[0] = true;
        } else if (str.equals("mauddib")) {
            assertEquals(false, found[1]);
            assertEquals(1, in.get(inner.getStructFieldData(map.get(key), inFields.get(0))));
            assertEquals(str, st.getPrimitiveJavaObject(inner.getStructFieldData(map.get(key), inFields.get(1))));
            found[1] = true;
        } else {
            throw new IllegalArgumentException("Unknown key " + str);
        }
    }
    assertEquals(true, found[0]);
    assertEquals(true, found[1]);
    assertEquals(Timestamp.valueOf("2000-03-12 15:00:01"), tso.getPrimitiveJavaObject(readerInspector.getStructFieldData(row, fields.get(12))));
    assertEquals(HiveDecimal.create("12345678.6547457"), dco.getPrimitiveJavaObject(readerInspector.getStructFieldData(row, fields.get(13))));
    // handle the close up
    assertEquals(false, rows.hasNext());
    rows.close();
}
Also used : LongObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.LongObjectInspector) IntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.IntObjectInspector) BinaryObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BinaryObjectInspector) StringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.StringObjectInspector) FloatObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.FloatObjectInspector) ByteObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ByteObjectInspector) TimestampObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.TimestampObjectInspector) ShortObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ShortObjectInspector) MapObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector) ListObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ListObjectInspector) HiveDecimalObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveDecimalObjectInspector) Path(org.apache.hadoop.fs.Path) DecimalColumnStatistics(org.apache.orc.DecimalColumnStatistics) BooleanColumnStatistics(org.apache.orc.BooleanColumnStatistics) StringColumnStatistics(org.apache.orc.StringColumnStatistics) DoubleColumnStatistics(org.apache.orc.DoubleColumnStatistics) IntegerColumnStatistics(org.apache.orc.IntegerColumnStatistics) ColumnStatistics(org.apache.orc.ColumnStatistics) BinaryColumnStatistics(org.apache.orc.BinaryColumnStatistics) DoubleObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.DoubleObjectInspector) BooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BooleanObjectInspector) StripeInformation(org.apache.orc.StripeInformation) StructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector) Test(org.junit.Test)

Example 17 with FloatObjectInspector

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

the class TestOrcFile method test1.

@Test
public void test1() throws Exception {
    ObjectInspector inspector;
    synchronized (TestOrcFile.class) {
        inspector = ObjectInspectorFactory.getReflectionObjectInspector(BigRow.class, ObjectInspectorFactory.ObjectInspectorOptions.JAVA);
    }
    Writer writer = OrcFile.createWriter(testFilePath, OrcFile.writerOptions(conf).inspector(inspector).stripeSize(100000).bufferSize(10000));
    writer.addRow(new BigRow(false, (byte) 1, (short) 1024, 65536, Long.MAX_VALUE, (float) 1.0, -15.0, bytes(0, 1, 2, 3, 4), "hi", new MiddleStruct(inner(1, "bye"), inner(2, "sigh")), list(inner(3, "good"), inner(4, "bad")), map()));
    writer.addRow(new BigRow(true, (byte) 100, (short) 2048, 65536, Long.MAX_VALUE, (float) 2.0, -5.0, bytes(), "bye", new MiddleStruct(inner(1, "bye"), inner(2, "sigh")), list(inner(100000000, "cat"), inner(-100000, "in"), inner(1234, "hat")), map(inner(5, "chani"), inner(1, "mauddib"))));
    writer.close();
    Reader reader = OrcFile.createReader(testFilePath, OrcFile.readerOptions(conf).filesystem(fs));
    TypeDescription schema = writer.getSchema();
    assertEquals(23, schema.getMaximumId());
    boolean[] expected = new boolean[] { false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false, false };
    boolean[] included = OrcUtils.includeColumns("", schema);
    assertEquals(true, Arrays.equals(expected, included));
    expected = new boolean[] { false, true, false, false, false, false, false, false, false, true, true, true, true, true, true, false, false, false, false, true, true, true, true, true };
    included = OrcUtils.includeColumns("boolean1,string1,middle,map", schema);
    assertEquals(true, Arrays.equals(expected, included));
    expected = new boolean[] { false, true, false, false, false, false, false, false, false, true, true, true, true, true, true, false, false, false, false, true, true, true, true, true };
    included = OrcUtils.includeColumns("boolean1,string1,middle,map", schema);
    assertEquals(true, Arrays.equals(expected, included));
    expected = new boolean[] { false, true, true, true, true, true, true, true, true, true, true, true, true, true, true, true, true, true, true, true, true, true, true, true };
    included = OrcUtils.includeColumns("boolean1,byte1,short1,int1,long1,float1,double1,bytes1,string1,middle,list,map", schema);
    assertEquals(true, Arrays.equals(expected, included));
    // check the stats
    ColumnStatistics[] stats = reader.getStatistics();
    assertEquals(2, stats[1].getNumberOfValues());
    assertEquals(1, ((BooleanColumnStatistics) stats[1]).getFalseCount());
    assertEquals(1, ((BooleanColumnStatistics) stats[1]).getTrueCount());
    assertEquals("count: 2 hasNull: false bytesOnDisk: 5 true: 1", stats[1].toString());
    assertEquals(2048, ((IntegerColumnStatistics) stats[3]).getMaximum());
    assertEquals(1024, ((IntegerColumnStatistics) stats[3]).getMinimum());
    assertEquals(true, ((IntegerColumnStatistics) stats[3]).isSumDefined());
    assertEquals(3072, ((IntegerColumnStatistics) stats[3]).getSum());
    assertEquals("count: 2 hasNull: false bytesOnDisk: 9 min: 1024 max: 2048 sum: 3072", stats[3].toString());
    StripeStatistics ss = reader.getStripeStatistics().get(0);
    assertEquals(2, ss.getColumnStatistics()[0].getNumberOfValues());
    assertEquals(1, ((BooleanColumnStatistics) ss.getColumnStatistics()[1]).getTrueCount());
    assertEquals(1024, ((IntegerColumnStatistics) ss.getColumnStatistics()[3]).getMinimum());
    assertEquals(2048, ((IntegerColumnStatistics) ss.getColumnStatistics()[3]).getMaximum());
    assertEquals(3072, ((IntegerColumnStatistics) ss.getColumnStatistics()[3]).getSum());
    assertEquals(-15.0, ((DoubleColumnStatistics) stats[7]).getMinimum());
    assertEquals(-5.0, ((DoubleColumnStatistics) stats[7]).getMaximum());
    assertEquals(-20.0, ((DoubleColumnStatistics) stats[7]).getSum(), 0.00001);
    assertEquals("count: 2 hasNull: false bytesOnDisk: 15 min: -15.0 max: -5.0 sum: -20.0", stats[7].toString());
    assertEquals("count: 2 hasNull: false bytesOnDisk: 14 min: bye max: hi sum: 5", stats[9].toString());
    // check the inspectors
    StructObjectInspector readerInspector = (StructObjectInspector) reader.getObjectInspector();
    assertEquals(ObjectInspector.Category.STRUCT, readerInspector.getCategory());
    assertEquals("struct<boolean1:boolean,byte1:tinyint,short1:smallint," + "int1:int,long1:bigint,float1:float,double1:double,bytes1:" + "binary,string1:string,middle:struct<list:array<struct<int1:int," + "string1:string>>>,list:array<struct<int1:int,string1:string>>," + "map:map<string,struct<int1:int,string1:string>>>", readerInspector.getTypeName());
    List<? extends StructField> fields = readerInspector.getAllStructFieldRefs();
    BooleanObjectInspector bo = (BooleanObjectInspector) readerInspector.getStructFieldRef("boolean1").getFieldObjectInspector();
    ByteObjectInspector by = (ByteObjectInspector) readerInspector.getStructFieldRef("byte1").getFieldObjectInspector();
    ShortObjectInspector sh = (ShortObjectInspector) readerInspector.getStructFieldRef("short1").getFieldObjectInspector();
    IntObjectInspector in = (IntObjectInspector) readerInspector.getStructFieldRef("int1").getFieldObjectInspector();
    LongObjectInspector lo = (LongObjectInspector) readerInspector.getStructFieldRef("long1").getFieldObjectInspector();
    FloatObjectInspector fl = (FloatObjectInspector) readerInspector.getStructFieldRef("float1").getFieldObjectInspector();
    DoubleObjectInspector dbl = (DoubleObjectInspector) readerInspector.getStructFieldRef("double1").getFieldObjectInspector();
    BinaryObjectInspector bi = (BinaryObjectInspector) readerInspector.getStructFieldRef("bytes1").getFieldObjectInspector();
    StringObjectInspector st = (StringObjectInspector) readerInspector.getStructFieldRef("string1").getFieldObjectInspector();
    StructObjectInspector mid = (StructObjectInspector) readerInspector.getStructFieldRef("middle").getFieldObjectInspector();
    List<? extends StructField> midFields = mid.getAllStructFieldRefs();
    ListObjectInspector midli = (ListObjectInspector) midFields.get(0).getFieldObjectInspector();
    StructObjectInspector inner = (StructObjectInspector) midli.getListElementObjectInspector();
    List<? extends StructField> inFields = inner.getAllStructFieldRefs();
    ListObjectInspector li = (ListObjectInspector) readerInspector.getStructFieldRef("list").getFieldObjectInspector();
    MapObjectInspector ma = (MapObjectInspector) readerInspector.getStructFieldRef("map").getFieldObjectInspector();
    StringObjectInspector mk = (StringObjectInspector) ma.getMapKeyObjectInspector();
    RecordReader rows = reader.rows();
    Object row = rows.next(null);
    assertNotNull(row);
    // check the contents of the first row
    assertEquals(false, bo.get(readerInspector.getStructFieldData(row, fields.get(0))));
    assertEquals(1, by.get(readerInspector.getStructFieldData(row, fields.get(1))));
    assertEquals(1024, sh.get(readerInspector.getStructFieldData(row, fields.get(2))));
    assertEquals(65536, in.get(readerInspector.getStructFieldData(row, fields.get(3))));
    assertEquals(Long.MAX_VALUE, lo.get(readerInspector.getStructFieldData(row, fields.get(4))));
    assertEquals(1.0, fl.get(readerInspector.getStructFieldData(row, fields.get(5))), 0.00001);
    assertEquals(-15.0, dbl.get(readerInspector.getStructFieldData(row, fields.get(6))), 0.00001);
    assertEquals(bytes(0, 1, 2, 3, 4), bi.getPrimitiveWritableObject(readerInspector.getStructFieldData(row, fields.get(7))));
    assertEquals("hi", st.getPrimitiveJavaObject(readerInspector.getStructFieldData(row, fields.get(8))));
    List<?> midRow = midli.getList(mid.getStructFieldData(readerInspector.getStructFieldData(row, fields.get(9)), midFields.get(0)));
    assertNotNull(midRow);
    assertEquals(2, midRow.size());
    assertEquals(1, in.get(inner.getStructFieldData(midRow.get(0), inFields.get(0))));
    assertEquals("bye", st.getPrimitiveJavaObject(inner.getStructFieldData(midRow.get(0), inFields.get(1))));
    assertEquals(2, in.get(inner.getStructFieldData(midRow.get(1), inFields.get(0))));
    assertEquals("sigh", st.getPrimitiveJavaObject(inner.getStructFieldData(midRow.get(1), inFields.get(1))));
    List<?> list = li.getList(readerInspector.getStructFieldData(row, fields.get(10)));
    assertEquals(2, list.size());
    assertEquals(3, in.get(inner.getStructFieldData(list.get(0), inFields.get(0))));
    assertEquals("good", st.getPrimitiveJavaObject(inner.getStructFieldData(list.get(0), inFields.get(1))));
    assertEquals(4, in.get(inner.getStructFieldData(list.get(1), inFields.get(0))));
    assertEquals("bad", st.getPrimitiveJavaObject(inner.getStructFieldData(list.get(1), inFields.get(1))));
    Map<?, ?> map = ma.getMap(readerInspector.getStructFieldData(row, fields.get(11)));
    assertEquals(0, map.size());
    // check the contents of second row
    assertEquals(true, rows.hasNext());
    row = rows.next(row);
    assertEquals(true, bo.get(readerInspector.getStructFieldData(row, fields.get(0))));
    assertEquals(100, by.get(readerInspector.getStructFieldData(row, fields.get(1))));
    assertEquals(2048, sh.get(readerInspector.getStructFieldData(row, fields.get(2))));
    assertEquals(65536, in.get(readerInspector.getStructFieldData(row, fields.get(3))));
    assertEquals(Long.MAX_VALUE, lo.get(readerInspector.getStructFieldData(row, fields.get(4))));
    assertEquals(2.0, fl.get(readerInspector.getStructFieldData(row, fields.get(5))), 0.00001);
    assertEquals(-5.0, dbl.get(readerInspector.getStructFieldData(row, fields.get(6))), 0.00001);
    assertEquals(bytes(), bi.getPrimitiveWritableObject(readerInspector.getStructFieldData(row, fields.get(7))));
    assertEquals("bye", st.getPrimitiveJavaObject(readerInspector.getStructFieldData(row, fields.get(8))));
    midRow = midli.getList(mid.getStructFieldData(readerInspector.getStructFieldData(row, fields.get(9)), midFields.get(0)));
    assertNotNull(midRow);
    assertEquals(2, midRow.size());
    assertEquals(1, in.get(inner.getStructFieldData(midRow.get(0), inFields.get(0))));
    assertEquals("bye", st.getPrimitiveJavaObject(inner.getStructFieldData(midRow.get(0), inFields.get(1))));
    assertEquals(2, in.get(inner.getStructFieldData(midRow.get(1), inFields.get(0))));
    assertEquals("sigh", st.getPrimitiveJavaObject(inner.getStructFieldData(midRow.get(1), inFields.get(1))));
    list = li.getList(readerInspector.getStructFieldData(row, fields.get(10)));
    assertEquals(3, list.size());
    assertEquals(100000000, in.get(inner.getStructFieldData(list.get(0), inFields.get(0))));
    assertEquals("cat", st.getPrimitiveJavaObject(inner.getStructFieldData(list.get(0), inFields.get(1))));
    assertEquals(-100000, in.get(inner.getStructFieldData(list.get(1), inFields.get(0))));
    assertEquals("in", st.getPrimitiveJavaObject(inner.getStructFieldData(list.get(1), inFields.get(1))));
    assertEquals(1234, in.get(inner.getStructFieldData(list.get(2), inFields.get(0))));
    assertEquals("hat", st.getPrimitiveJavaObject(inner.getStructFieldData(list.get(2), inFields.get(1))));
    map = ma.getMap(readerInspector.getStructFieldData(row, fields.get(11)));
    assertEquals(2, map.size());
    boolean[] found = new boolean[2];
    for (Object key : map.keySet()) {
        String str = mk.getPrimitiveJavaObject(key);
        if (str.equals("chani")) {
            assertEquals(false, found[0]);
            assertEquals(5, in.get(inner.getStructFieldData(map.get(key), inFields.get(0))));
            assertEquals(str, st.getPrimitiveJavaObject(inner.getStructFieldData(map.get(key), inFields.get(1))));
            found[0] = true;
        } else if (str.equals("mauddib")) {
            assertEquals(false, found[1]);
            assertEquals(1, in.get(inner.getStructFieldData(map.get(key), inFields.get(0))));
            assertEquals(str, st.getPrimitiveJavaObject(inner.getStructFieldData(map.get(key), inFields.get(1))));
            found[1] = true;
        } else {
            throw new IllegalArgumentException("Unknown key " + str);
        }
    }
    assertEquals(true, found[0]);
    assertEquals(true, found[1]);
    // handle the close up
    assertEquals(false, rows.hasNext());
    rows.close();
}
Also used : LongObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.LongObjectInspector) IntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.IntObjectInspector) BinaryObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BinaryObjectInspector) StringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.StringObjectInspector) FloatObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.FloatObjectInspector) ByteObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ByteObjectInspector) ShortObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ShortObjectInspector) MapObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector) ListObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ListObjectInspector) TypeDescription(org.apache.orc.TypeDescription) DecimalColumnStatistics(org.apache.orc.DecimalColumnStatistics) BooleanColumnStatistics(org.apache.orc.BooleanColumnStatistics) StringColumnStatistics(org.apache.orc.StringColumnStatistics) DoubleColumnStatistics(org.apache.orc.DoubleColumnStatistics) IntegerColumnStatistics(org.apache.orc.IntegerColumnStatistics) ColumnStatistics(org.apache.orc.ColumnStatistics) BinaryColumnStatistics(org.apache.orc.BinaryColumnStatistics) HiveDecimalObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveDecimalObjectInspector) BooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BooleanObjectInspector) ShortObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ShortObjectInspector) ObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ObjectInspector) MapObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector) StructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector) FloatObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.FloatObjectInspector) StringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.StringObjectInspector) ListObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.ListObjectInspector) IntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.IntObjectInspector) LongObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.LongObjectInspector) BinaryObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BinaryObjectInspector) ByteObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ByteObjectInspector) DoubleObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.DoubleObjectInspector) TimestampObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.TimestampObjectInspector) StripeStatistics(org.apache.orc.StripeStatistics) DoubleObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.DoubleObjectInspector) BooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BooleanObjectInspector) StructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector) Test(org.junit.Test)

Example 18 with FloatObjectInspector

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

the class GenericUDFFormatNumber method evaluate.

@Override
public Object evaluate(DeferredObject[] arguments) throws HiveException {
    Object arg0;
    Object arg1;
    if ((arg0 = arguments[0].get()) == null || (arg1 = arguments[1].get()) == null) {
        return null;
    }
    if (!dType.equals(PrimitiveCategory.STRING)) {
        int dValue = ((IntObjectInspector) argumentOIs[1]).get(arg1);
        if (dValue < 0) {
            throw new HiveException("Argument 2 of function FORMAT_NUMBER must be >= 0, but \"" + dValue + "\" was found");
        }
        if (dValue != lastDValue) {
            // construct a new DecimalFormat only if a new dValue
            pattern.delete(0, pattern.length());
            pattern.append("#,###,###,###,###,###,##0");
            // decimal place
            if (dValue > 0) {
                pattern.append(".");
                for (int i = 0; i < dValue; i++) {
                    pattern.append("0");
                }
            }
            DecimalFormat dFormat = new DecimalFormat(pattern.toString());
            lastDValue = dValue;
            numberFormat.applyPattern(dFormat.toPattern());
        }
    }
    double xDoubleValue = 0.0;
    float xFloatValue = 0.0f;
    HiveDecimal xDecimalValue = null;
    int xIntValue = 0;
    long xLongValue = 0L;
    PrimitiveObjectInspector xObjectInspector = (PrimitiveObjectInspector) argumentOIs[0];
    switch(xObjectInspector.getPrimitiveCategory()) {
        case VOID:
        case DOUBLE:
            xDoubleValue = ((DoubleObjectInspector) argumentOIs[0]).get(arg0);
            resultText.set(numberFormat.format(xDoubleValue));
            break;
        case FLOAT:
            xFloatValue = ((FloatObjectInspector) argumentOIs[0]).get(arg0);
            resultText.set(numberFormat.format(xFloatValue));
            break;
        case DECIMAL:
            xDecimalValue = ((HiveDecimalObjectInspector) argumentOIs[0]).getPrimitiveJavaObject(arg0);
            resultText.set(numberFormat.format(xDecimalValue.bigDecimalValue()));
            break;
        case BYTE:
        case SHORT:
        case INT:
            xIntValue = ((IntObjectInspector) argumentOIs[0]).get(arg0);
            resultText.set(numberFormat.format(xIntValue));
            break;
        case LONG:
            xLongValue = ((LongObjectInspector) argumentOIs[0]).get(arg0);
            resultText.set(numberFormat.format(xLongValue));
            break;
        default:
            throw new HiveException("Argument 1 of function FORMAT_NUMBER must be " + serdeConstants.TINYINT_TYPE_NAME + "\"" + " or \"" + serdeConstants.SMALLINT_TYPE_NAME + "\"" + " or \"" + serdeConstants.INT_TYPE_NAME + "\"" + " or \"" + serdeConstants.BIGINT_TYPE_NAME + "\"" + " or \"" + serdeConstants.DOUBLE_TYPE_NAME + "\"" + " or \"" + serdeConstants.FLOAT_TYPE_NAME + "\"" + " or \"" + serdeConstants.DECIMAL_TYPE_NAME + "\", but \"" + argumentOIs[0].getTypeName() + "\" was found.");
    }
    return resultText;
}
Also used : IntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.IntObjectInspector) HiveException(org.apache.hadoop.hive.ql.metadata.HiveException) DecimalFormat(java.text.DecimalFormat) HiveDecimal(org.apache.hadoop.hive.common.type.HiveDecimal) PrimitiveObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector)

Example 19 with FloatObjectInspector

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

the class TypedBytesSerDe method serializeField.

private void serializeField(Object o, ObjectInspector oi, Object reuse) throws IOException {
    switch(oi.getCategory()) {
        case PRIMITIVE:
            {
                PrimitiveObjectInspector poi = (PrimitiveObjectInspector) oi;
                switch(poi.getPrimitiveCategory()) {
                    case VOID:
                        {
                            return;
                        }
                    case BOOLEAN:
                        {
                            BooleanObjectInspector boi = (BooleanObjectInspector) poi;
                            BooleanWritable r = reuse == null ? new BooleanWritable() : (BooleanWritable) reuse;
                            r.set(boi.get(o));
                            tbOut.write(r);
                            return;
                        }
                    case BYTE:
                        {
                            ByteObjectInspector boi = (ByteObjectInspector) poi;
                            ByteWritable r = reuse == null ? new ByteWritable() : (ByteWritable) reuse;
                            r.set(boi.get(o));
                            tbOut.write(r);
                            return;
                        }
                    case SHORT:
                        {
                            ShortObjectInspector spoi = (ShortObjectInspector) poi;
                            ShortWritable r = reuse == null ? new ShortWritable() : (ShortWritable) reuse;
                            r.set(spoi.get(o));
                            tbOut.write(r);
                            return;
                        }
                    case INT:
                        {
                            IntObjectInspector ioi = (IntObjectInspector) poi;
                            IntWritable r = reuse == null ? new IntWritable() : (IntWritable) reuse;
                            r.set(ioi.get(o));
                            tbOut.write(r);
                            return;
                        }
                    case LONG:
                        {
                            LongObjectInspector loi = (LongObjectInspector) poi;
                            LongWritable r = reuse == null ? new LongWritable() : (LongWritable) reuse;
                            r.set(loi.get(o));
                            tbOut.write(r);
                            return;
                        }
                    case FLOAT:
                        {
                            FloatObjectInspector foi = (FloatObjectInspector) poi;
                            FloatWritable r = reuse == null ? new FloatWritable() : (FloatWritable) reuse;
                            r.set(foi.get(o));
                            tbOut.write(r);
                            return;
                        }
                    case DOUBLE:
                        {
                            DoubleObjectInspector doi = (DoubleObjectInspector) poi;
                            DoubleWritable r = reuse == null ? new DoubleWritable() : (DoubleWritable) reuse;
                            r.set(doi.get(o));
                            tbOut.write(r);
                            return;
                        }
                    case STRING:
                        {
                            StringObjectInspector soi = (StringObjectInspector) poi;
                            Text t = soi.getPrimitiveWritableObject(o);
                            tbOut.write(t);
                            return;
                        }
                    default:
                        {
                            throw new RuntimeException("Unrecognized type: " + poi.getPrimitiveCategory());
                        }
                }
            }
        case LIST:
        case MAP:
        case STRUCT:
            {
                // For complex object, serialize to JSON format
                String s = SerDeUtils.getJSONString(o, oi);
                Text t = reuse == null ? new Text() : (Text) reuse;
                // convert to Text and write it
                t.set(s);
                tbOut.write(t);
            }
        default:
            {
                throw new RuntimeException("Unrecognized type: " + oi.getCategory());
            }
    }
}
Also used : LongObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.LongObjectInspector) IntObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.IntObjectInspector) DoubleWritable(org.apache.hadoop.hive.serde2.io.DoubleWritable) Text(org.apache.hadoop.io.Text) ShortWritable(org.apache.hadoop.hive.serde2.io.ShortWritable) StringObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.StringObjectInspector) FloatObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.FloatObjectInspector) ByteObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ByteObjectInspector) FloatWritable(org.apache.hadoop.io.FloatWritable) DoubleObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.DoubleObjectInspector) BooleanWritable(org.apache.hadoop.io.BooleanWritable) ShortObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.ShortObjectInspector) PrimitiveObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector) LongWritable(org.apache.hadoop.io.LongWritable) BooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BooleanObjectInspector) ByteWritable(org.apache.hadoop.hive.serde2.io.ByteWritable) IntWritable(org.apache.hadoop.io.IntWritable)

Example 20 with FloatObjectInspector

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

the class DruidSerDe method serialize.

@Override
public Writable serialize(Object o, ObjectInspector objectInspector) throws SerDeException {
    if (objectInspector.getCategory() != ObjectInspector.Category.STRUCT) {
        throw new SerDeException(getClass().toString() + " can only serialize struct types, but we got: " + objectInspector.getTypeName());
    }
    // Prepare the field ObjectInspectors
    StructObjectInspector soi = (StructObjectInspector) objectInspector;
    List<? extends StructField> fields = soi.getAllStructFieldRefs();
    List<Object> values = soi.getStructFieldsDataAsList(o);
    // We deserialize the result
    final Map<String, Object> value = new HashMap<>();
    for (int i = 0; i < columns.length; i++) {
        if (values.get(i) == null) {
            // null, we just add it
            value.put(columns[i], null);
            continue;
        }
        final Object res;
        switch(types[i].getPrimitiveCategory()) {
            case TIMESTAMP:
                res = ((TimestampObjectInspector) fields.get(i).getFieldObjectInspector()).getPrimitiveJavaObject(values.get(i)).toEpochMilli();
                break;
            case TIMESTAMPLOCALTZ:
                res = ((TimestampLocalTZObjectInspector) fields.get(i).getFieldObjectInspector()).getPrimitiveJavaObject(values.get(i)).getZonedDateTime().toInstant().toEpochMilli();
                break;
            case BYTE:
                res = ((ByteObjectInspector) fields.get(i).getFieldObjectInspector()).get(values.get(i));
                break;
            case SHORT:
                res = ((ShortObjectInspector) fields.get(i).getFieldObjectInspector()).get(values.get(i));
                break;
            case INT:
                res = ((IntObjectInspector) fields.get(i).getFieldObjectInspector()).get(values.get(i));
                break;
            case LONG:
                res = ((LongObjectInspector) fields.get(i).getFieldObjectInspector()).get(values.get(i));
                break;
            case FLOAT:
                res = ((FloatObjectInspector) fields.get(i).getFieldObjectInspector()).get(values.get(i));
                break;
            case DOUBLE:
                res = ((DoubleObjectInspector) fields.get(i).getFieldObjectInspector()).get(values.get(i));
                break;
            case CHAR:
                res = ((HiveCharObjectInspector) fields.get(i).getFieldObjectInspector()).getPrimitiveJavaObject(values.get(i)).getValue();
                break;
            case VARCHAR:
                res = ((HiveVarcharObjectInspector) fields.get(i).getFieldObjectInspector()).getPrimitiveJavaObject(values.get(i)).getValue();
                break;
            case STRING:
                res = ((StringObjectInspector) fields.get(i).getFieldObjectInspector()).getPrimitiveJavaObject(values.get(i));
                break;
            case BOOLEAN:
                res = ((BooleanObjectInspector) fields.get(i).getFieldObjectInspector()).get(values.get(i)) ? 1L : 0L;
                break;
            default:
                throw new SerDeException("Unsupported type: " + types[i].getPrimitiveCategory());
        }
        value.put(columns[i], res);
    }
    // Extract the partitions keys segments granularity and partition key if any
    // First Segment Granularity has to be here.
    final int granularityFieldIndex = columns.length;
    assert values.size() > granularityFieldIndex;
    Preconditions.checkArgument(fields.get(granularityFieldIndex).getFieldName().equals(Constants.DRUID_TIMESTAMP_GRANULARITY_COL_NAME));
    Timestamp timestamp = ((TimestampObjectInspector) fields.get(granularityFieldIndex).getFieldObjectInspector()).getPrimitiveJavaObject(values.get(granularityFieldIndex));
    Preconditions.checkNotNull(timestamp, "Timestamp column cannot have null value");
    value.put(Constants.DRUID_TIMESTAMP_GRANULARITY_COL_NAME, timestamp.toEpochMilli());
    if (values.size() == columns.length + 2) {
        // Then partition number if any.
        final int partitionNumPos = granularityFieldIndex + 1;
        Preconditions.checkArgument(fields.get(partitionNumPos).getFieldName().equals(Constants.DRUID_SHARD_KEY_COL_NAME), String.format("expecting to encounter %s but was %s", Constants.DRUID_SHARD_KEY_COL_NAME, fields.get(partitionNumPos).getFieldName()));
        value.put(Constants.DRUID_SHARD_KEY_COL_NAME, ((LongObjectInspector) fields.get(partitionNumPos).getFieldObjectInspector()).get(values.get(partitionNumPos)));
    }
    return new DruidWritable(value);
}
Also used : HashMap(java.util.HashMap) TimestampLocalTZObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.TimestampLocalTZObjectInspector) Timestamp(org.apache.hadoop.hive.common.type.Timestamp) TimestampObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.TimestampObjectInspector) HiveVarcharObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveVarcharObjectInspector) HiveCharObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveCharObjectInspector) BooleanObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.primitive.BooleanObjectInspector) SerDeException(org.apache.hadoop.hive.serde2.SerDeException) StructObjectInspector(org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector)

Aggregations

FloatObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.FloatObjectInspector)17 IntObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.IntObjectInspector)17 BooleanObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.BooleanObjectInspector)16 ByteObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.ByteObjectInspector)16 DoubleObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.DoubleObjectInspector)16 LongObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.LongObjectInspector)16 ShortObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.ShortObjectInspector)16 StringObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.StringObjectInspector)16 TimestampObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.TimestampObjectInspector)16 BinaryObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.BinaryObjectInspector)14 HiveDecimalObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveDecimalObjectInspector)14 StructObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.StructObjectInspector)12 DateObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.DateObjectInspector)12 HiveCharObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveCharObjectInspector)12 PrimitiveObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.PrimitiveObjectInspector)11 HiveVarcharObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.primitive.HiveVarcharObjectInspector)11 MapObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.MapObjectInspector)10 Text (org.apache.hadoop.io.Text)10 Map (java.util.Map)9 ListObjectInspector (org.apache.hadoop.hive.serde2.objectinspector.ListObjectInspector)9