use of com.linkedin.pinot.core.data.GenericRow in project pinot by linkedin.
the class DefaultGroupKeyGeneratorTest method setup.
@BeforeClass
private void setup() throws Exception {
GenericRow[] segmentData = new GenericRow[NUM_ROWS];
int value = _random.nextInt(MAX_STEP_LENGTH);
// Generate random values for the segment.
for (int i = 0; i < UNIQUE_ROWS; i++) {
Map<String, Object> map = new HashMap<>();
for (String singleValueColumn : SINGLE_VALUE_COLUMNS) {
map.put(singleValueColumn, value);
value += 1 + _random.nextInt(MAX_STEP_LENGTH);
}
for (String multiValueColumn : MULTI_VALUE_COLUMNS) {
int numMultiValues = 1 + _random.nextInt(MAX_NUM_MULTI_VALUES);
Integer[] values = new Integer[numMultiValues];
for (int k = 0; k < numMultiValues; k++) {
values[k] = value;
value += 1 + _random.nextInt(MAX_STEP_LENGTH);
}
map.put(multiValueColumn, values);
}
GenericRow genericRow = new GenericRow();
genericRow.init(map);
segmentData[i] = genericRow;
}
for (int i = UNIQUE_ROWS; i < NUM_ROWS; i += UNIQUE_ROWS) {
System.arraycopy(segmentData, 0, segmentData, i, UNIQUE_ROWS);
}
// Create an index segment with the random values.
Schema schema = new Schema();
for (String singleValueColumn : SINGLE_VALUE_COLUMNS) {
DimensionFieldSpec dimensionFieldSpec = new DimensionFieldSpec(singleValueColumn, FieldSpec.DataType.INT, true);
schema.addField(dimensionFieldSpec);
}
for (String multiValueColumn : MULTI_VALUE_COLUMNS) {
DimensionFieldSpec dimensionFieldSpec = new DimensionFieldSpec(multiValueColumn, FieldSpec.DataType.INT, false);
schema.addField(dimensionFieldSpec);
}
SegmentGeneratorConfig config = new SegmentGeneratorConfig(schema);
FileUtils.deleteQuietly(new File(INDEX_DIR_PATH));
config.setOutDir(INDEX_DIR_PATH);
config.setSegmentName(SEGMENT_NAME);
SegmentIndexCreationDriverImpl driver = new SegmentIndexCreationDriverImpl();
driver.init(config, new TestDataRecordReader(schema, segmentData));
driver.build();
IndexSegment indexSegment = Loaders.IndexSegment.load(new File(INDEX_DIR_PATH, SEGMENT_NAME), ReadMode.heap);
// Get a data fetcher for the index segment.
Map<String, BaseOperator> dataSourceMap = new HashMap<>();
Map<String, Block> blockMap = new HashMap<>();
for (String column : indexSegment.getColumnNames()) {
DataSource dataSource = indexSegment.getDataSource(column);
dataSourceMap.put(column, dataSource);
blockMap.put(column, dataSource.getNextBlock());
}
// Generate a random test doc id set.
int num1 = _random.nextInt(50);
int num2 = num1 + 1 + _random.nextInt(50);
for (int i = 0; i < 20; i += 2) {
_testDocIdSet[i] = num1 + 50 * i;
_testDocIdSet[i + 1] = num2 + 50 * i;
}
DataFetcher dataFetcher = new DataFetcher(dataSourceMap);
DocIdSetBlock docIdSetBlock = new DocIdSetBlock(_testDocIdSet, _testDocIdSet.length);
ProjectionBlock projectionBlock = new ProjectionBlock(blockMap, new DataBlockCache(dataFetcher), docIdSetBlock);
_transformBlock = new TransformBlock(projectionBlock, new HashMap<String, BlockValSet>());
}
use of com.linkedin.pinot.core.data.GenericRow in project pinot by linkedin.
the class PinotSegmentToAvroConverter method convert.
@Override
public void convert() throws Exception {
PinotSegmentRecordReader recordReader = new PinotSegmentRecordReader(new File(_segmentDir));
try {
recordReader.init();
Schema avroSchema = buildAvroSchemaFromPinotSchema(recordReader.getSchema());
try (DataFileWriter<Record> recordWriter = new DataFileWriter<>(new GenericDatumWriter<Record>(avroSchema))) {
recordWriter.create(avroSchema, new File(_outputFile));
while (recordReader.hasNext()) {
GenericRow row = recordReader.next();
Record record = new Record(avroSchema);
for (String field : row.getFieldNames()) {
Object value = row.getValue(field);
if (value instanceof Object[]) {
record.put(field, Arrays.asList((Object[]) value));
} else {
record.put(field, value);
}
}
recordWriter.append(record);
}
}
} finally {
recordReader.close();
}
}
use of com.linkedin.pinot.core.data.GenericRow in project pinot by linkedin.
the class RawIndexCreatorTest method testFixedLengthRawIndexCreator.
/**
* Helper method to perform actual tests for a given column.
*
* @param column Column for which to perform the test
* @param dataType Data type of the column
* @throws Exception
*/
private void testFixedLengthRawIndexCreator(String column, FieldSpec.DataType dataType) throws Exception {
PinotDataBuffer indexBuffer = getIndexBufferForColumn(column);
FixedByteChunkSingleValueReader rawIndexReader = new FixedByteChunkSingleValueReader(indexBuffer, ChunkCompressorFactory.getDecompressor("snappy"));
_recordReader.rewind();
for (int row = 0; row < NUM_ROWS; row++) {
GenericRow expectedRow = _recordReader.next();
Object expected = expectedRow.getValue(column);
Object actual;
actual = readValueFromIndex(rawIndexReader, dataType, row);
Assert.assertEquals(actual, expected);
}
}
use of com.linkedin.pinot.core.data.GenericRow in project pinot by linkedin.
the class RawIndexCreatorTest method testStringRawIndexCreator.
/**
* Test for string raw index creator.
* Compares values read from the raw index against expected value.
* @throws Exception
*/
@Test
public void testStringRawIndexCreator() throws Exception {
PinotDataBuffer indexBuffer = getIndexBufferForColumn(STRING_COLUMN);
ChunkDecompressor uncompressor = ChunkCompressorFactory.getDecompressor("snappy");
VarByteChunkSingleValueReader rawIndexReader = new VarByteChunkSingleValueReader(indexBuffer, uncompressor);
_recordReader.rewind();
ChunkReaderContext context = rawIndexReader.createContext();
for (int row = 0; row < NUM_ROWS; row++) {
GenericRow expectedRow = _recordReader.next();
Object expected = expectedRow.getValue(STRING_COLUMN);
Object actual = rawIndexReader.getString(row, context);
Assert.assertEquals(actual, expected);
}
}
use of com.linkedin.pinot.core.data.GenericRow in project pinot by linkedin.
the class RawIndexCreatorTest method buildIndex.
/**
* Helper method to build a segment containing a single valued string column with RAW (no-dictionary) index.
*
* @return Array of string values for the rows in the generated index.
* @throws Exception
*/
private RecordReader buildIndex(Schema schema) throws Exception {
SegmentGeneratorConfig config = new SegmentGeneratorConfig(schema);
config.setRawIndexCreationColumns(schema.getDimensionNames());
config.setOutDir(SEGMENT_DIR_NAME);
config.setSegmentName(SEGMENT_NAME);
final List<GenericRow> rows = new ArrayList<>();
for (int row = 0; row < NUM_ROWS; row++) {
HashMap<String, Object> map = new HashMap<>();
for (FieldSpec fieldSpec : schema.getAllFieldSpecs()) {
Object value;
FieldSpec.DataType dataType = fieldSpec.getDataType();
value = getRandomValue(dataType);
map.put(fieldSpec.getName(), value);
}
GenericRow genericRow = new GenericRow();
genericRow.init(map);
rows.add(genericRow);
}
SegmentIndexCreationDriverImpl driver = new SegmentIndexCreationDriverImpl();
RecordReader reader = new TestRecordReader(rows, schema);
driver.init(config, reader);
driver.build();
_segmentDirectory = SegmentDirectory.createFromLocalFS(driver.getOutputDirectory(), ReadMode.mmap);
_segmentReader = _segmentDirectory.createReader();
reader.rewind();
return reader;
}
Aggregations