public void NextBigIntegerTest()
{
Random rand = new Random();
byte[] aaa = new byte[] { 10, 20, 30, 40, 50 };
byte[] bbb = aaa.Reverse().ToArray();
bool isSqeEq = bbb.SequenceEqual(new byte[] { 50, 40, 30, 20, 10 });
Assert.IsTrue(isSqeEq);
BigInteger b0 = rand.NextBigInteger(0);
BigInteger b1 = rand.NextBigInteger(1);
BigInteger b2 = rand.NextBigInteger(2);
BigInteger b3 = rand.NextBigInteger(100);
BigInteger b4 = rand.NextBigInteger(BigInteger.Pow(2, 127));
Assert.IsTrue(b0 >= 0);
Assert.IsTrue(b0 == 0);
Assert.IsTrue(b1 >= 0);
Assert.IsTrue(b1 < 1);
Assert.IsTrue(b2 >= 0);
Assert.IsTrue(b2 < 2);
Assert.IsTrue(b3 >= 0);
Assert.IsTrue(b3 < 100);
Assert.IsTrue(b4 >= 0);
Assert.IsTrue(b4 < BigInteger.Pow(2, 130));
try
{
BigInteger b5 = rand.NextBigInteger(-BigInteger.Pow(2, 127));
Assert.Fail();
}
catch (Exception ex) { }
}
public void testUnionAndTimestamp()
{
List<OrcProto.Type> types = new List<OrcProto.Type>();
types.Add(OrcProto.Type.CreateBuilder().SetKind(OrcProto.Type.Types.Kind.STRUCT).
AddFieldNames("time").AddFieldNames("union").AddFieldNames("decimal").
AddSubtypes(1).AddSubtypes(2).AddSubtypes(5).Build());
types.Add(OrcProto.Type.CreateBuilder().SetKind(OrcProto.Type.Types.Kind.TIMESTAMP).
Build());
types.Add(OrcProto.Type.CreateBuilder().SetKind(OrcProto.Type.Types.Kind.UNION).
AddSubtypes(3).AddSubtypes(4).Build());
types.Add(OrcProto.Type.CreateBuilder().SetKind(OrcProto.Type.Types.Kind.INT).
Build());
types.Add(OrcProto.Type.CreateBuilder().SetKind(OrcProto.Type.Types.Kind.STRING).
Build());
types.Add(OrcProto.Type.CreateBuilder().SetKind(OrcProto.Type.Types.Kind.DECIMAL).
Build());
ObjectInspector inspector = OrcStruct.createObjectInspector(0, types);
HiveDecimal maxValue = HiveDecimal.Parse("10000000000000000000");
OrcStruct row = new OrcStruct(3);
OrcUnion union = new OrcUnion();
Random rand;
using (Stream file = FileOpenWrite(TestFilePath))
using (Writer writer = OrcFile.createWriter(TestFilePath, file, OrcFile.writerOptions(conf)
.inspector(inspector)
.stripeSize(1000)
.compress(CompressionKind.NONE)
.bufferSize(100)
.blockPadding(false)))
{
row.setFieldValue(1, union);
row.setFieldValue(0, Timestamp.Parse("2000-03-12 15:00:00"));
HiveDecimal value = HiveDecimal.Parse("12345678.6547456");
row.setFieldValue(2, value);
union.set((byte)0, 42);
writer.addRow(row);
row.setFieldValue(0, Timestamp.Parse("2000-03-20 12:00:00.123456789"));
union.set((byte)1, "hello");
value = HiveDecimal.Parse("-5643.234");
row.setFieldValue(2, value);
writer.addRow(row);
row.setFieldValue(0, null);
row.setFieldValue(1, null);
row.setFieldValue(2, null);
writer.addRow(row);
row.setFieldValue(1, union);
union.set((byte)0, null);
writer.addRow(row);
union.set((byte)1, null);
writer.addRow(row);
union.set((byte)0, 200000);
row.setFieldValue(0, Timestamp.Parse("1970-01-01 00:00:00"));
value = HiveDecimal.Parse("10000000000000000000");
row.setFieldValue(2, value);
writer.addRow(row);
rand = new Random(42);
for (int i = 1970; i < 2038; ++i)
{
row.setFieldValue(0, Timestamp.Parse(i + "-05-05 12:34:56." + i));
if ((i & 1) == 0)
{
union.set((byte)0, (i * i));
}
else
{
union.set((byte)1, (i * i).ToString());
}
value = HiveDecimal.create(rand.NextBigInteger(64), rand.Next(18));
row.setFieldValue(2, value);
if (maxValue.CompareTo(value) < 0)
{
maxValue = value;
}
writer.addRow(row);
}
// let's add a lot of constant rows to test the rle
row.setFieldValue(0, null);
union.set((byte)0, 1732050807);
row.setFieldValue(2, null);
for (int i = 0; i < 5000; ++i)
{
writer.addRow(row);
}
union.set((byte)0, 0);
writer.addRow(row);
union.set((byte)0, 10);
writer.addRow(row);
union.set((byte)0, 138);
writer.addRow(row);
writer.close();
TypeDescription schema = writer.getSchema();
Assert.Equal(5, schema.getMaximumId());
bool[] expected = new bool[] { false, false, false, false, false, false };
bool[] included = OrcUtils.includeColumns("", schema);
Assert.Equal(expected, included);
expected = new bool[] { false, true, false, false, false, true };
included = OrcUtils.includeColumns("time,decimal", schema);
Assert.Equal(expected, included);
expected = new bool[] { false, false, true, true, true, false };
included = OrcUtils.includeColumns("union", schema);
Assert.Equal(expected, included);
}
Reader reader = OrcFile.createReader(TestFilePath, OrcFile.readerOptions(conf));
Assert.Equal(0, reader.getMetadataKeys().Count);
Assert.Equal(5077, reader.getNumberOfRows());
DecimalColumnStatistics stats =
(DecimalColumnStatistics)reader.getStatistics()[5];
Assert.Equal(71, stats.getNumberOfValues());
Assert.Equal(HiveDecimal.Parse("-5643.234"), stats.getMinimum());
Assert.Equal(maxValue, stats.getMaximum());
// TODO: fix this
// Assert.Equal(null,stats.getSum());
int stripeCount = 0;
int rowCount = 0;
long currentOffset = -1;
foreach (StripeInformation stripe in reader.getStripes())
{
stripeCount += 1;
rowCount += (int)stripe.getNumberOfRows();
if (currentOffset < 0)
{
currentOffset = stripe.getOffset() + stripe.getLength();
}
else
{
Assert.Equal(currentOffset, stripe.getOffset());
currentOffset += stripe.getLength();
}
}
Assert.Equal(reader.getNumberOfRows(), rowCount);
Assert.Equal(2, stripeCount);
Assert.Equal(reader.getContentLength(), currentOffset);
using (RecordReader rows = reader.rows())
{
Assert.Equal(0, rows.getRowNumber());
Assert.Equal(0.0, rows.getProgress(), 6);
Assert.Equal(true, rows.hasNext());
row = (OrcStruct)rows.next();
Assert.Equal(1, rows.getRowNumber());
inspector = reader.getObjectInspector();
Assert.Equal("struct<time:timestamp,union:uniontype<int,string>,decimal:decimal(38,18)>",
inspector.getTypeName());
Assert.Equal(Timestamp.Parse("2000-03-12 15:00:00"), row.getFieldValue(0));
union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(0, union.getTag());
Assert.Equal(42, union.getObject());
Assert.Equal(HiveDecimal.Parse("12345678.6547456"), row.getFieldValue(2));
row = (OrcStruct)rows.next();
Assert.Equal(2, rows.getRowNumber());
Assert.Equal(Timestamp.Parse("2000-03-20 12:00:00.123456789"), row.getFieldValue(0));
Assert.Equal(1, union.getTag());
Assert.Equal("hello", union.getObject());
Assert.Equal(HiveDecimal.Parse("-5643.234"), row.getFieldValue(2));
row = (OrcStruct)rows.next();
Assert.Equal(null, row.getFieldValue(0));
Assert.Equal(null, row.getFieldValue(1));
Assert.Equal(null, row.getFieldValue(2));
row = (OrcStruct)rows.next();
Assert.Equal(null, row.getFieldValue(0));
union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(0, union.getTag());
Assert.Equal(null, union.getObject());
Assert.Equal(null, row.getFieldValue(2));
row = (OrcStruct)rows.next();
Assert.Equal(null, row.getFieldValue(0));
Assert.Equal(1, union.getTag());
Assert.Equal(null, union.getObject());
Assert.Equal(null, row.getFieldValue(2));
row = (OrcStruct)rows.next();
Assert.Equal(Timestamp.Parse("1970-01-01 00:00:00"), row.getFieldValue(0));
Assert.Equal(200000, union.getObject());
Assert.Equal(HiveDecimal.Parse("10000000000000000000"), row.getFieldValue(2));
rand = new Random(42);
for (int i = 1970; i < 2038; ++i)
{
row = (OrcStruct)rows.next();
Assert.Equal(Timestamp.Parse(i + "-05-05 12:34:56." + i), row.getFieldValue(0));
if ((i & 1) == 0)
{
Assert.Equal(0, union.getTag());
Assert.Equal(i * i, union.getObject());
}
else
{
Assert.Equal(1, union.getTag());
Assert.Equal((i * i).ToString(), union.getObject());
}
Assert.Equal(HiveDecimal.create(rand.NextBigInteger(64), rand.Next(18)), row.getFieldValue(2));
}
for (int i = 0; i < 5000; ++i)
{
row = (OrcStruct)rows.next();
Assert.Equal(1732050807, union.getObject());
}
row = (OrcStruct)rows.next();
Assert.Equal(0, union.getObject());
row = (OrcStruct)rows.next();
Assert.Equal(10, union.getObject());
row = (OrcStruct)rows.next();
Assert.Equal(138, union.getObject());
Assert.Equal(false, rows.hasNext());
Assert.Equal(1.0, rows.getProgress(), 5);
Assert.Equal(reader.getNumberOfRows(), rows.getRowNumber());
rows.seekToRow(1);
row = (OrcStruct)rows.next();
Assert.Equal(Timestamp.Parse("2000-03-20 12:00:00.123456789"), row.getFieldValue(0));
Assert.Equal(1, union.getTag());
Assert.Equal("hello", union.getObject());
Assert.Equal(HiveDecimal.Parse("-5643.234"), row.getFieldValue(2));
}
}
public void testUnionAndTimestamp()
{
TypeDescription schema = TypeDescription.createStruct()
.addField("time", TypeDescription.createTimestamp())
.addField("union", TypeDescription.createUnion()
.addUnionChild(TypeDescription.createInt())
.addUnionChild(TypeDescription.createString()))
.addField("decimal", TypeDescription.createDecimal()
.withPrecision(38)
.withScale(18));
HiveDecimal maxValue = HiveDecimal.Parse("10000000000000000000");
Random rand = new Random(42);
using (Stream file = File.OpenWrite(TestFilePath))
using (Writer writer = OrcFile.createWriter(TestFilePath, file, OrcFile.writerOptions(conf)
.setSchema(schema)
.stripeSize(1000)
.compress(CompressionKind.NONE)
.bufferSize(100)
.blockPadding(false)))
{
VectorizedRowBatch batch = schema.createRowBatch();
batch.size = 6;
setUnion(batch, 0, Timestamp.Parse("2000-03-12 15:00:00"), 0, 42, null,
HiveDecimal.Parse("12345678.6547456"));
setUnion(batch, 1, Timestamp.Parse("2000-03-20 12:00:00.123456789"),
1, null, "hello", HiveDecimal.Parse("-5643.234"));
setUnion(batch, 2, null, null, null, null, null);
setUnion(batch, 3, null, 0, null, null, null);
setUnion(batch, 4, null, 1, null, null, null);
setUnion(batch, 5, Timestamp.Parse("1970-01-01 00:00:00"), 0, 200000,
null, HiveDecimal.Parse("10000000000000000000"));
writer.addRowBatch(batch);
batch.reset();
for (int i = 1970; i < 2038; ++i)
{
Timestamp ts = Timestamp.Parse(i + "-05-05 12:34:56." + i);
HiveDecimal dec = HiveDecimal.create(rand.NextBigInteger(64), rand.Next(18));
if ((i & 1) == 0)
{
setUnion(batch, batch.size++, ts, 0, i * i, null, dec);
}
else
{
setUnion(batch, batch.size++, ts, 1, null, (i * i).ToString(), dec);
}
if (maxValue.CompareTo(dec) < 0)
{
maxValue = dec;
}
}
writer.addRowBatch(batch);
batch.reset();
// let's add a lot of constant rows to test the rle
batch.size = 1000;
for (int c = 0; c < batch.cols.Length; ++c)
{
batch.cols[c].setRepeating(true);
}
setUnion(batch, 0, null, 0, 1732050807, null, null);
for (int i = 0; i < 5; ++i)
{
writer.addRowBatch(batch);
}
batch.reset();
batch.size = 3;
setUnion(batch, 0, null, 0, 0, null, null);
setUnion(batch, 1, null, 0, 10, null, null);
setUnion(batch, 2, null, 0, 138, null, null);
writer.addRowBatch(batch);
schema = writer.getSchema();
}
Reader reader = OrcFile.createReader(TestFilePath, OrcFile.readerOptions(conf));
Assert.Equal(5, schema.getMaximumId());
bool[] expected = new bool[] { false, false, false, false, false, false };
bool[] included = OrcUtils.includeColumns("", schema);
Assert.Equal(expected, included);
expected = new bool[] { false, true, false, false, false, true };
included = OrcUtils.includeColumns("time,decimal", schema);
Assert.Equal(expected, included);
expected = new bool[] { false, false, true, true, true, false };
included = OrcUtils.includeColumns("union", schema);
Assert.Equal(expected, included);
Assert.Equal(0, reader.getMetadataKeys().Count);
Assert.Equal(5077, reader.getNumberOfRows());
DecimalColumnStatistics stats =
(DecimalColumnStatistics)reader.getStatistics()[5];
Assert.Equal(71, stats.getNumberOfValues());
Assert.Equal(HiveDecimal.Parse("-5643.234"), stats.getMinimum());
Assert.Equal(maxValue, stats.getMaximum());
// TODO: fix this
// Assert.Equal(null,stats.getSum());
int stripeCount = 0;
int rowCount = 0;
long currentOffset = -1;
foreach (StripeInformation stripe in reader.getStripes())
{
stripeCount += 1;
rowCount += (int)stripe.getNumberOfRows();
if (currentOffset < 0)
{
currentOffset = stripe.getOffset() + stripe.getLength();
}
else
{
Assert.Equal(currentOffset, stripe.getOffset());
currentOffset += stripe.getLength();
}
}
Assert.Equal(reader.getNumberOfRows(), rowCount);
Assert.Equal(2, stripeCount);
Assert.Equal(reader.getContentLength(), currentOffset);
using (RecordReader rows = reader.rows())
{
Assert.Equal(0, rows.getRowNumber());
Assert.Equal(0.0, rows.getProgress(), 6);
Assert.Equal(true, rows.hasNext());
OrcStruct row = (OrcStruct)rows.next();
Assert.Equal(1, rows.getRowNumber());
ObjectInspector inspector = reader.getObjectInspector();
Assert.Equal("struct<time:timestamp,union:uniontype<int,string>,decimal:decimal(38,18)>",
inspector.getTypeName());
Assert.Equal(Timestamp.Parse("2000-03-12 15:00:00"), row.getFieldValue(0));
OrcUnion union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(0, union.getTag());
Assert.Equal(42, union.getObject());
Assert.Equal(HiveDecimal.Parse("12345678.6547456"), row.getFieldValue(2));
row = (OrcStruct)rows.next();
Assert.Equal(2, rows.getRowNumber());
Assert.Equal(Timestamp.Parse("2000-03-20 12:00:00.123456789"), row.getFieldValue(0));
union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(1, union.getTag());
Assert.Equal("hello", union.getObject());
Assert.Equal(HiveDecimal.Parse("-5643.234"), row.getFieldValue(2));
row = (OrcStruct)rows.next();
Assert.Null(row.getFieldValue(0));
Assert.Null(row.getFieldValue(1));
Assert.Null(row.getFieldValue(2));
row = (OrcStruct)rows.next();
Assert.Null(row.getFieldValue(0));
union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(0, union.getTag());
Assert.Null(union.getObject());
Assert.Null(row.getFieldValue(2));
row = (OrcStruct)rows.next();
Assert.Null(row.getFieldValue(0));
union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(1, union.getTag());
Assert.Null(union.getObject());
Assert.Null(row.getFieldValue(2));
row = (OrcStruct)rows.next();
Assert.Equal(Timestamp.Parse("1970-01-01 00:00:00"), row.getFieldValue(0));
union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(200000, union.getObject());
Assert.Equal(HiveDecimal.Parse("10000000000000000000"), row.getFieldValue(2));
rand = new Random(42);
for (int i = 1970; i < 2038; ++i)
{
row = (OrcStruct)rows.next();
Assert.Equal(Timestamp.Parse(i + "-05-05 12:34:56." + i),
row.getFieldValue(0));
union = (OrcUnion)row.getFieldValue(1);
if ((i & 1) == 0)
{
Assert.Equal(0, union.getTag());
Assert.Equal(i * i, union.getObject());
}
else
{
Assert.Equal(1, union.getTag());
Assert.Equal((i * i).ToString(), union.getObject());
}
Assert.Equal(
HiveDecimal.create(rand.NextBigInteger(64), rand.Next(18)),
row.getFieldValue(2));
}
for (int i = 0; i < 5000; ++i)
{
row = (OrcStruct)rows.next();
union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(1732050807, union.getObject());
}
row = (OrcStruct)rows.next();
union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(0, union.getObject());
row = (OrcStruct)rows.next();
union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(10, union.getObject());
row = (OrcStruct)rows.next();
union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(138, union.getObject());
Assert.Equal(false, rows.hasNext());
Assert.Equal(1.0, rows.getProgress(), 5);
Assert.Equal(reader.getNumberOfRows(), rows.getRowNumber());
rows.seekToRow(1);
row = (OrcStruct)rows.next();
Assert.Equal(Timestamp.Parse("2000-03-20 12:00:00.123456789"), row.getFieldValue(0));
union = (OrcUnion)row.getFieldValue(1);
Assert.Equal(1, union.getTag());
Assert.Equal("hello", union.getObject());
Assert.Equal(HiveDecimal.Parse("-5643.234"), row.getFieldValue(2));
}
}