use of org.apache.hadoop.hive.ql.exec.vector.BytesColumnVector in project hive by apache.
the class OctetLength method evaluate.
// Calculate the length of the UTF-8 strings in input vector and place results in output vector.
@Override
public void evaluate(VectorizedRowBatch batch) {
if (childExpressions != null) {
super.evaluateChildren(batch);
}
BytesColumnVector inputColVector = (BytesColumnVector) batch.cols[colNum];
LongColumnVector outV = (LongColumnVector) batch.cols[outputColumn];
int[] sel = batch.selected;
int n = batch.size;
int[] length = inputColVector.length;
long[] resultLen = outV.vector;
if (n == 0) {
//Nothing to do
return;
}
if (inputColVector.noNulls) {
outV.noNulls = true;
if (inputColVector.isRepeating) {
outV.isRepeating = true;
resultLen[0] = length[0];
} else if (batch.selectedInUse) {
for (int j = 0; j != n; j++) {
int i = sel[j];
resultLen[i] = length[i];
}
outV.isRepeating = false;
} else {
for (int i = 0; i != n; i++) {
resultLen[i] = length[i];
}
outV.isRepeating = false;
}
} else {
/*
* Handle case with nulls. Don't do function if the value is null, to save time,
* because calling the function can be expensive.
*/
outV.noNulls = false;
if (inputColVector.isRepeating) {
outV.isRepeating = true;
outV.isNull[0] = inputColVector.isNull[0];
if (!inputColVector.isNull[0]) {
resultLen[0] = length[0];
}
} else if (batch.selectedInUse) {
for (int j = 0; j != n; j++) {
int i = sel[j];
if (!inputColVector.isNull[i]) {
resultLen[i] = length[i];
}
outV.isNull[i] = inputColVector.isNull[i];
}
outV.isRepeating = false;
} else {
for (int i = 0; i != n; i++) {
if (!inputColVector.isNull[i]) {
resultLen[i] = length[i];
}
outV.isNull[i] = inputColVector.isNull[i];
}
outV.isRepeating = false;
}
}
}
use of org.apache.hadoop.hive.ql.exec.vector.BytesColumnVector in project hive by apache.
the class CastStringToDecimal method evaluate.
@Override
public void evaluate(VectorizedRowBatch batch) {
if (childExpressions != null) {
super.evaluateChildren(batch);
}
BytesColumnVector inV = (BytesColumnVector) batch.cols[inputColumn];
int[] sel = batch.selected;
int n = batch.size;
DecimalColumnVector outV = (DecimalColumnVector) batch.cols[outputColumn];
if (n == 0) {
// Nothing to do
return;
}
if (inV.noNulls) {
outV.noNulls = true;
if (inV.isRepeating) {
outV.isRepeating = true;
func(outV, inV, 0);
} else if (batch.selectedInUse) {
for (int j = 0; j != n; j++) {
int i = sel[j];
func(outV, inV, i);
}
outV.isRepeating = false;
} else {
for (int i = 0; i != n; i++) {
func(outV, inV, i);
}
outV.isRepeating = false;
}
} else {
// Handle case with nulls. Don't do function if the value is null,
// because the data may be undefined for a null value.
outV.noNulls = false;
if (inV.isRepeating) {
outV.isRepeating = true;
outV.isNull[0] = inV.isNull[0];
if (!inV.isNull[0]) {
func(outV, inV, 0);
}
} else if (batch.selectedInUse) {
for (int j = 0; j != n; j++) {
int i = sel[j];
outV.isNull[i] = inV.isNull[i];
if (!inV.isNull[i]) {
func(outV, inV, i);
}
}
outV.isRepeating = false;
} else {
System.arraycopy(inV.isNull, 0, outV.isNull, 0, n);
for (int i = 0; i != n; i++) {
if (!inV.isNull[i]) {
func(outV, inV, i);
}
}
outV.isRepeating = false;
}
}
}
use of org.apache.hadoop.hive.ql.exec.vector.BytesColumnVector in project hive by apache.
the class CastStringToIntervalDayTime method evaluate.
@Override
public void evaluate(VectorizedRowBatch batch) {
if (childExpressions != null) {
super.evaluateChildren(batch);
}
BytesColumnVector inV = (BytesColumnVector) batch.cols[inputColumn];
int[] sel = batch.selected;
int n = batch.size;
IntervalDayTimeColumnVector outV = (IntervalDayTimeColumnVector) batch.cols[outputColumn];
if (n == 0) {
// Nothing to do
return;
}
if (inV.noNulls) {
outV.noNulls = true;
if (inV.isRepeating) {
outV.isRepeating = true;
evaluate(outV, inV, 0);
} else if (batch.selectedInUse) {
for (int j = 0; j != n; j++) {
int i = sel[j];
evaluate(outV, inV, i);
}
outV.isRepeating = false;
} else {
for (int i = 0; i != n; i++) {
evaluate(outV, inV, i);
}
outV.isRepeating = false;
}
} else {
// Handle case with nulls. Don't do function if the value is null,
// because the data may be undefined for a null value.
outV.noNulls = false;
if (inV.isRepeating) {
outV.isRepeating = true;
outV.isNull[0] = inV.isNull[0];
if (!inV.isNull[0]) {
evaluate(outV, inV, 0);
}
} else if (batch.selectedInUse) {
for (int j = 0; j != n; j++) {
int i = sel[j];
outV.isNull[i] = inV.isNull[i];
if (!inV.isNull[i]) {
evaluate(outV, inV, i);
}
}
outV.isRepeating = false;
} else {
System.arraycopy(inV.isNull, 0, outV.isNull, 0, n);
for (int i = 0; i != n; i++) {
if (!inV.isNull[i]) {
evaluate(outV, inV, i);
}
}
outV.isRepeating = false;
}
}
}
use of org.apache.hadoop.hive.ql.exec.vector.BytesColumnVector in project hive by apache.
the class TestInputOutputFormat method testSchemaEvolution.
/**
* Test schema evolution when using the reader directly.
*/
@Test
public void testSchemaEvolution() throws Exception {
TypeDescription fileSchema = TypeDescription.fromString("struct<a:int,b:struct<c:int>,d:string>");
Writer writer = OrcFile.createWriter(testFilePath, OrcFile.writerOptions(conf).fileSystem(fs).setSchema(fileSchema).compress(org.apache.orc.CompressionKind.NONE));
VectorizedRowBatch batch = fileSchema.createRowBatch(1000);
batch.size = 1000;
LongColumnVector lcv = ((LongColumnVector) ((StructColumnVector) batch.cols[1]).fields[0]);
for (int r = 0; r < 1000; r++) {
((LongColumnVector) batch.cols[0]).vector[r] = r * 42;
lcv.vector[r] = r * 10001;
((BytesColumnVector) batch.cols[2]).setVal(r, Integer.toHexString(r).getBytes(StandardCharsets.UTF_8));
}
writer.addRowBatch(batch);
writer.close();
TypeDescription readerSchema = TypeDescription.fromString("struct<a:int,b:struct<c:int,future1:int>,d:string,future2:int>");
Reader reader = OrcFile.createReader(testFilePath, OrcFile.readerOptions(conf).filesystem(fs));
RecordReader rows = reader.rowsOptions(new Reader.Options().schema(readerSchema));
batch = readerSchema.createRowBatch();
lcv = ((LongColumnVector) ((StructColumnVector) batch.cols[1]).fields[0]);
LongColumnVector future1 = ((LongColumnVector) ((StructColumnVector) batch.cols[1]).fields[1]);
assertEquals(true, rows.nextBatch(batch));
assertEquals(1000, batch.size);
assertEquals(true, future1.isRepeating);
assertEquals(true, future1.isNull[0]);
assertEquals(true, batch.cols[3].isRepeating);
assertEquals(true, batch.cols[3].isNull[0]);
for (int r = 0; r < batch.size; ++r) {
assertEquals("row " + r, r * 42, ((LongColumnVector) batch.cols[0]).vector[r]);
assertEquals("row " + r, r * 10001, lcv.vector[r]);
assertEquals("row " + r, r * 10001, lcv.vector[r]);
assertEquals("row " + r, Integer.toHexString(r), ((BytesColumnVector) batch.cols[2]).toString(r));
}
assertEquals(false, rows.nextBatch(batch));
rows.close();
// try it again with an include vector
rows = reader.rowsOptions(new Reader.Options().schema(readerSchema).include(new boolean[] { false, true, true, true, false, false, true }));
batch = readerSchema.createRowBatch();
lcv = ((LongColumnVector) ((StructColumnVector) batch.cols[1]).fields[0]);
future1 = ((LongColumnVector) ((StructColumnVector) batch.cols[1]).fields[1]);
assertEquals(true, rows.nextBatch(batch));
assertEquals(1000, batch.size);
assertEquals(true, future1.isRepeating);
assertEquals(true, future1.isNull[0]);
assertEquals(true, batch.cols[3].isRepeating);
assertEquals(true, batch.cols[3].isNull[0]);
assertEquals(true, batch.cols[2].isRepeating);
assertEquals(true, batch.cols[2].isNull[0]);
for (int r = 0; r < batch.size; ++r) {
assertEquals("row " + r, r * 42, ((LongColumnVector) batch.cols[0]).vector[r]);
assertEquals("row " + r, r * 10001, lcv.vector[r]);
}
assertEquals(false, rows.nextBatch(batch));
rows.close();
}
use of org.apache.hadoop.hive.ql.exec.vector.BytesColumnVector in project hive by apache.
the class TestInputOutputFormat method testVectorizationWithAcid.
// test acid with vectorization, no combine
@Test
public void testVectorizationWithAcid() throws Exception {
StructObjectInspector inspector = new BigRowInspector();
JobConf conf = createMockExecutionEnvironment(workDir, new Path("mock:///"), "vectorizationAcid", inspector, true, 1);
// write the orc file to the mock file system
Path partDir = new Path(conf.get("mapred.input.dir"));
OrcRecordUpdater writer = new OrcRecordUpdater(partDir, new AcidOutputFormat.Options(conf).maximumTransactionId(10).writingBase(true).bucket(0).inspector(inspector).finalDestination(partDir));
for (int i = 0; i < 100; ++i) {
BigRow row = new BigRow(i);
writer.insert(10, row);
}
writer.close(false);
Path path = new Path("mock:/vectorizationAcid/p=0/base_0000010/bucket_00000");
setBlocks(path, conf, new MockBlock("host0", "host1"));
// call getsplits
HiveInputFormat<?, ?> inputFormat = new HiveInputFormat<WritableComparable, Writable>();
InputSplit[] splits = inputFormat.getSplits(conf, 10);
assertEquals(1, splits.length);
conf.set(IOConstants.SCHEMA_EVOLUTION_COLUMNS, BigRow.getColumnNamesProperty());
conf.set(IOConstants.SCHEMA_EVOLUTION_COLUMNS_TYPES, BigRow.getColumnTypesProperty());
HiveConf.setBoolVar(conf, HiveConf.ConfVars.HIVE_TRANSACTIONAL_TABLE_SCAN, true);
org.apache.hadoop.mapred.RecordReader<NullWritable, VectorizedRowBatch> reader = inputFormat.getRecordReader(splits[0], conf, Reporter.NULL);
NullWritable key = reader.createKey();
VectorizedRowBatch value = reader.createValue();
assertEquals(true, reader.next(key, value));
assertEquals(100, value.count());
LongColumnVector booleanColumn = (LongColumnVector) value.cols[0];
LongColumnVector byteColumn = (LongColumnVector) value.cols[1];
LongColumnVector shortColumn = (LongColumnVector) value.cols[2];
LongColumnVector intColumn = (LongColumnVector) value.cols[3];
LongColumnVector longColumn = (LongColumnVector) value.cols[4];
DoubleColumnVector floatColumn = (DoubleColumnVector) value.cols[5];
DoubleColumnVector doubleCoulmn = (DoubleColumnVector) value.cols[6];
BytesColumnVector stringColumn = (BytesColumnVector) value.cols[7];
DecimalColumnVector decimalColumn = (DecimalColumnVector) value.cols[8];
LongColumnVector dateColumn = (LongColumnVector) value.cols[9];
TimestampColumnVector timestampColumn = (TimestampColumnVector) value.cols[10];
for (int i = 0; i < 100; i++) {
assertEquals("checking boolean " + i, i % 2 == 0 ? 1 : 0, booleanColumn.vector[i]);
assertEquals("checking byte " + i, (byte) i, byteColumn.vector[i]);
assertEquals("checking short " + i, (short) i, shortColumn.vector[i]);
assertEquals("checking int " + i, i, intColumn.vector[i]);
assertEquals("checking long " + i, i, longColumn.vector[i]);
assertEquals("checking float " + i, i, floatColumn.vector[i], 0.0001);
assertEquals("checking double " + i, i, doubleCoulmn.vector[i], 0.0001);
Text strValue = new Text();
strValue.set(stringColumn.vector[i], stringColumn.start[i], stringColumn.length[i]);
assertEquals("checking string " + i, new Text(Long.toHexString(i)), strValue);
assertEquals("checking decimal " + i, HiveDecimal.create(i), decimalColumn.vector[i].getHiveDecimal());
assertEquals("checking date " + i, i, dateColumn.vector[i]);
long millis = (long) i * MILLIS_IN_DAY;
millis -= LOCAL_TIMEZONE.getOffset(millis);
assertEquals("checking timestamp " + i, millis, timestampColumn.getTime(i));
}
assertEquals(false, reader.next(key, value));
}
Aggregations