public override ObjectInspectorCategory getCategory()
{
return(wrapped.getCategory());
}
public void testMultiStripeWithNull()
{
ObjectInspector inspector = ObjectInspectorFactory.getReflectionObjectInspector(typeof(MyStruct));
using (Stream file = File.OpenWrite(TestFilePath))
using (Writer writer = OrcFile.createWriter(TestFilePath, file, OrcFile.writerOptions(conf)
.inspector(inspector)
.stripeSize(100000)
.compress(CompressionKind.NONE)
.bufferSize(10000)))
{
Random rand = new Random(100);
writer.addRow(new MyStruct(null, null, true, new List <InnerStruct> {
new InnerStruct(100)
}));
for (int i = 2; i < 20000; i++)
{
writer.addRow(new MyStruct(rand.Next(1), "a", true, new List <InnerStruct> {
new InnerStruct(100)
}));
}
writer.addRow(new MyStruct(null, null, true, new List <InnerStruct> {
new InnerStruct(100)
}));
}
Reader reader = OrcFile.createReader(TestFilePath,
OrcFile.readerOptions(conf));
// check the stats
ColumnStatistics[] stats = reader.getStatistics();
Assert.Equal(20000, reader.getNumberOfRows());
Assert.Equal(20000, stats[0].getNumberOfValues());
Assert.Equal(0, ((IntegerColumnStatistics)stats[1]).getMaximum());
Assert.Equal(0, ((IntegerColumnStatistics)stats[1]).getMinimum());
Assert.Equal(true, ((IntegerColumnStatistics)stats[1]).isSumDefined());
Assert.Equal(0, ((IntegerColumnStatistics)stats[1]).getSum());
Assert.Equal("count: 19998 hasNull: True min: 0 max: 0 sum: 0",
stats[1].ToString());
Assert.Equal("a", ((StringColumnStatistics)stats[2]).getMaximum());
Assert.Equal("a", ((StringColumnStatistics)stats[2]).getMinimum());
Assert.Equal(19998, stats[2].getNumberOfValues());
Assert.Equal("count: 19998 hasNull: True min: a max: a sum: 19998",
stats[2].ToString());
// check the inspectors
StructObjectInspector readerInspector =
(StructObjectInspector)reader.getObjectInspector();
Assert.Equal(ObjectInspectorCategory.STRUCT,
readerInspector.getCategory());
Assert.Equal("struct<a:int,b:string,c:boolean,list:array<struct<z:int>>>",
readerInspector.getTypeName());
using (RecordReader rows = reader.rows())
{
List <bool> expected = new List <bool>();
foreach (StripeInformation sinfo in reader.getStripes())
{
expected.Add(false);
}
// only the first and last stripe will have PRESENT stream
expected[0] = true;
expected[expected.Count - 1] = true;
List <bool> got = new List <bool>();
// check if the strip footer contains PRESENT stream
foreach (StripeInformation sinfo in reader.getStripes())
{
OrcProto.StripeFooter sf =
((RecordReaderImpl)rows).readStripeFooter(sinfo);
got.Add(sf.ToString().IndexOf(OrcProto.Stream.Types.Kind.PRESENT.ToString()) != -1);
}
Assert.Equal(expected, got);
// row 1
OrcStruct row = (OrcStruct)rows.next();
Assert.NotNull(row);
Assert.Null(row.getFieldValue(0));
Assert.Null(row.getFieldValue(1));
Assert.Equal(true, row.getFieldValue(2));
Assert.Equal(100, ((OrcStruct)((IList <object>)row.getFieldValue(3))[0]).
getFieldValue(0));
rows.seekToRow(19998);
// last-1 row
row = (OrcStruct)rows.next();
Assert.NotNull(row);
Assert.NotNull(row.getFieldValue(1));
Assert.Equal(0, row.getFieldValue(0));
Assert.Equal(true, row.getFieldValue(2));
Assert.Equal(100, ((OrcStruct)((IList <object>)row.getFieldValue(3))[0]).
getFieldValue(0));
// last row
row = (OrcStruct)rows.next();
Assert.NotNull(row);
Assert.Null(row.getFieldValue(0));
Assert.Null(row.getFieldValue(1));
Assert.Equal(true, row.getFieldValue(2));
Assert.Equal(100, ((OrcStruct)((IList <object>)row.getFieldValue(3))[0]).
getFieldValue(0));
}
}
public void testStringAndBinaryStatistics()
{
ObjectInspector inspector = ObjectInspectorFactory.getReflectionObjectInspector(typeof(SimpleStruct));
using (Stream file = File.OpenWrite(TestFilePath))
using (Writer writer = OrcFile.createWriter(TestFilePath, file, OrcFile.writerOptions(conf)
.inspector(inspector)
.stripeSize(100000)
.bufferSize(10000)))
{
writer.addRow(new SimpleStruct(bytes(0, 1, 2, 3, 4), "foo"));
writer.addRow(new SimpleStruct(bytes(0, 1, 2, 3), "bar"));
writer.addRow(new SimpleStruct(bytes(0, 1, 2, 3, 4, 5), null));
writer.addRow(new SimpleStruct(null, "hi"));
writer.close();
Assert.Equal(4, writer.getNumberOfRows());
Assert.Equal(273, writer.getRawDataSize());
}
Reader reader = OrcFile.createReader(TestFilePath,
OrcFile.readerOptions(conf));
Assert.Equal(4, reader.getNumberOfRows());
Assert.Equal(273, reader.getRawDataSize());
Assert.Equal(15, reader.getRawDataSizeOfColumns(Lists.newArrayList("bytes1")));
Assert.Equal(258, reader.getRawDataSizeOfColumns(Lists.newArrayList("string1")));
Assert.Equal(273, reader.getRawDataSizeOfColumns(Lists.newArrayList("bytes1", "string1")));
// check the stats
ColumnStatistics[] stats = reader.getStatistics();
Assert.Equal(4, stats[0].getNumberOfValues());
Assert.Equal("count: 4 hasNull: False", stats[0].ToString());
Assert.Equal(3, stats[1].getNumberOfValues());
Assert.Equal(15, ((BinaryColumnStatistics)stats[1]).getSum());
Assert.Equal("count: 3 hasNull: True sum: 15", stats[1].ToString());
Assert.Equal(3, stats[2].getNumberOfValues());
Assert.Equal("bar", ((StringColumnStatistics)stats[2]).getMinimum());
Assert.Equal("hi", ((StringColumnStatistics)stats[2]).getMaximum());
Assert.Equal(8, ((StringColumnStatistics)stats[2]).getSum());
Assert.Equal("count: 3 hasNull: True min: bar max: hi sum: 8",
stats[2].ToString());
// check the inspectors
StructObjectInspector readerInspector =
(StructObjectInspector)reader.getObjectInspector();
Assert.Equal(ObjectInspectorCategory.STRUCT, readerInspector.getCategory());
Assert.Equal("struct<bytes1:binary,string1:string>", readerInspector.getTypeName());
IList <StructField> fields = readerInspector.getAllStructFieldRefs();
BinaryObjectInspector bi = (BinaryObjectInspector)readerInspector.
getStructFieldRef("bytes1").getFieldObjectInspector();
StringObjectInspector st = (StringObjectInspector)readerInspector.
getStructFieldRef("string1").getFieldObjectInspector();
using (RecordReader rows = reader.rows())
{
object row = rows.next();
Assert.NotNull(row);
// check the contents of the first row
Assert.Equal(bytes(0, 1, 2, 3, 4), bi.get(
readerInspector.getStructFieldData(row, fields[0])));
Assert.Equal("foo", st.getPrimitiveJavaObject(readerInspector.
getStructFieldData(row, fields[1])));
// check the contents of second row
Assert.Equal(true, rows.hasNext());
row = rows.next();
Assert.Equal(bytes(0, 1, 2, 3), bi.get(
readerInspector.getStructFieldData(row, fields[0])));
Assert.Equal("bar", st.getPrimitiveJavaObject(readerInspector.
getStructFieldData(row, fields[1])));
// check the contents of second row
Assert.Equal(true, rows.hasNext());
row = rows.next();
Assert.Equal(bytes(0, 1, 2, 3, 4, 5), bi.get(
readerInspector.getStructFieldData(row, fields[0])));
Assert.Null(st.getPrimitiveJavaObject(readerInspector.
getStructFieldData(row, fields[1])));
// check the contents of second row
Assert.Equal(true, rows.hasNext());
row = rows.next();
Assert.Null(bi.get(
readerInspector.getStructFieldData(row, fields[0])));
Assert.Equal("hi", st.getPrimitiveJavaObject(readerInspector.
getStructFieldData(row, fields[1])));
Assert.Equal(false, rows.hasNext());
}
}
public void testColumnsWithNullAndCompression()
{
ObjectInspector inspector = ObjectInspectorFactory.getReflectionObjectInspector(typeof(MyStruct));
using (Stream file = File.OpenWrite(TestFilePath))
using (Writer writer = OrcFile.createWriter(TestFilePath, file, OrcFile.writerOptions(conf)
.inspector(inspector)
.stripeSize(100000)
.bufferSize(10000)))
{
writer.addRow(new MyStruct(3, "a", true,
Lists.newArrayList(new InnerStruct(100))));
writer.addRow(new MyStruct(null, "b", true,
Lists.newArrayList(new InnerStruct(100))));
writer.addRow(new MyStruct(3, null, false,
Lists.newArrayList(new InnerStruct(100))));
writer.addRow(new MyStruct(3, "d", true,
Lists.newArrayList(new InnerStruct(100))));
writer.addRow(new MyStruct(2, "e", true,
Lists.newArrayList(new InnerStruct(100))));
writer.addRow(new MyStruct(2, "f", true,
Lists.newArrayList(new InnerStruct(100))));
writer.addRow(new MyStruct(2, "g", true,
Lists.newArrayList(new InnerStruct(100))));
writer.addRow(new MyStruct(2, "h", true,
Lists.newArrayList(new InnerStruct(100))));
}
Reader reader = OrcFile.createReader(TestFilePath,
OrcFile.readerOptions(conf));
// check the stats
ColumnStatistics[] stats = reader.getStatistics();
Assert.Equal(8, reader.getNumberOfRows());
Assert.Equal(8, stats[0].getNumberOfValues());
Assert.Equal(3, ((IntegerColumnStatistics)stats[1]).getMaximum());
Assert.Equal(2, ((IntegerColumnStatistics)stats[1]).getMinimum());
Assert.Equal(true, ((IntegerColumnStatistics)stats[1]).isSumDefined());
Assert.Equal(17, ((IntegerColumnStatistics)stats[1]).getSum());
Assert.Equal("count: 7 hasNull: True min: 2 max: 3 sum: 17",
stats[1].ToString());
Assert.Equal("h", ((StringColumnStatistics)stats[2]).getMaximum());
Assert.Equal("a", ((StringColumnStatistics)stats[2]).getMinimum());
Assert.Equal(7, stats[2].getNumberOfValues());
Assert.Equal("count: 7 hasNull: True min: a max: h sum: 7",
stats[2].ToString());
// check the inspectors
StructObjectInspector readerInspector = (StructObjectInspector)reader.getObjectInspector();
Assert.Equal(ObjectInspectorCategory.STRUCT,
readerInspector.getCategory());
Assert.Equal("struct<a:int,b:string,c:boolean,list:array<struct<z:int>>>",
readerInspector.getTypeName());
using (RecordReader rows = reader.rows())
{
// only the last strip will have PRESENT stream
List <bool> expected = new List <bool>();
foreach (StripeInformation sinfo in reader.getStripes())
{
expected.Add(false);
}
expected[expected.Count - 1] = true;
List <bool> got = new List <bool>();
// check if the strip footer contains PRESENT stream
foreach (StripeInformation sinfo in reader.getStripes())
{
OrcProto.StripeFooter sf = ((RecordReaderImpl)rows).readStripeFooter(sinfo);
got.Add(sf.ToString().IndexOf(OrcProto.Stream.Types.Kind.PRESENT.ToString()) != -1);
}
Assert.Equal(expected, got);
// row 1
OrcStruct row = (OrcStruct)rows.next();
Assert.NotNull(row);
Assert.Equal(3, row.getFieldValue(0));
Assert.Equal("a", row.getFieldValue(1).ToString());
Assert.Equal(true, row.getFieldValue(2));
Assert.Equal(100, ((OrcStruct)((IList <object>)row.getFieldValue(3))[0]).
getFieldValue(0));
// row 2
row = (OrcStruct)rows.next();
Assert.NotNull(row);
Assert.Null(row.getFieldValue(0));
Assert.Equal("b", row.getFieldValue(1).ToString());
Assert.Equal(true, row.getFieldValue(2));
Assert.Equal(100, ((OrcStruct)((IList <object>)row.getFieldValue(3))[0]).
getFieldValue(0));
// row 3
row = (OrcStruct)rows.next();
Assert.NotNull(row);
Assert.Null(row.getFieldValue(1));
Assert.Equal(3, row.getFieldValue(0));
Assert.Equal(false, row.getFieldValue(2));
Assert.Equal(100, ((OrcStruct)((IList <object>)row.getFieldValue(3))[0]).
getFieldValue(0));
}
}