.build(),
OrcProto.Type.newBuilder().setKind(OrcProto.Type.Kind.STRING)
.build()
);
ObjectInspector inspector;
synchronized (TestOrcFile.class) {
inspector = OrcLazyObjectInspectorUtils.createWritableObjectInspector(0, types);
}
ReaderWriterProfiler.setProfilerOptions(conf);
Writer writer = OrcFile.createWriter(fs, testFilePath, conf, inspector,
15 * 1024, CompressionKind.NONE, 100, 10000);
OrcStruct row = new OrcStruct(types.get(0).getFieldNamesList());
OrcUnion union = new OrcUnion();
row.setFieldValue(1, union);
row.setFieldValue(0, Timestamp.valueOf("2000-03-12 15:00:00"));
union.set((byte) 0, new IntWritable(42));
writer.addRow(row);
row.setFieldValue(0, Timestamp.valueOf("2000-03-20 12:00:00.123456789"));
union.set((byte)1, new Text("hello"));
writer.addRow(row);
row.setFieldValue(0, null);
row.setFieldValue(1, 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, new IntWritable(200000));
row.setFieldValue(0, Timestamp.valueOf("1900-01-01 00:00:00"));
writer.addRow(row);
for(int i=1900; i < 2200; ++i) {
row.setFieldValue(0, Timestamp.valueOf(i + "-05-05 12:34:56." + i));
if ((i & 1) == 0) {
union.set((byte) 0, new IntWritable(i*i));
} else {
union.set((byte) 1, new Text(new Integer(i*i).toString()));
}
writer.addRow(row);
}
// let's add a lot of constant rows to test the rle
row.setFieldValue(0, null);
union.set((byte) 0, new IntWritable(1732050807));
for(int i=0; i < 5000; ++i) {
writer.addRow(row);
}
union.set((byte) 0, new IntWritable(0));
writer.addRow(row);
union.set((byte) 0, new IntWritable(10));
writer.addRow(row);
union.set((byte) 0, new IntWritable(138));
writer.addRow(row);
writer.close();
ReaderWriterProfiler.setProfilerOptions(conf);
Reader reader = OrcFile.createReader(fs, testFilePath, conf);
assertEquals(false, reader.getMetadataKeys().iterator().hasNext());
assertEquals(5309, reader.getNumberOfRows());
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);
assertEquals(reader.getContentLength(), currentOffset);
RecordReader rows = reader.rows(null);
assertEquals(0, rows.getRowNumber());
assertEquals(0.0, rows.getProgress(), 0.000001);
assertEquals(true, rows.hasNext());
OrcLazyStruct lazyRow = (OrcLazyStruct) rows.next(null);
row = (OrcStruct) lazyRow.materialize();
inspector = reader.getObjectInspector();
assertEquals("struct<time:timestamp,union:uniontype<int,string>>",
inspector.getTypeName());
assertEquals(Timestamp.valueOf("2000-03-12 15:00:00"),
((TimestampWritable) ((OrcLazyTimestamp) row.getFieldValue(0)).materialize()).getTimestamp());
union = (OrcUnion) ((OrcLazyUnion) row.getFieldValue(1)).materialize();
assertEquals(0, union.getTag());
assertEquals(new IntWritable(42), union.getObject());