use of com.linkedin.pinot.core.indexsegment.generator.SegmentGeneratorConfig 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.indexsegment.generator.SegmentGeneratorConfig in project pinot by linkedin.
the class RealtimeSegmentConverter method build.
public void build(SegmentVersion segmentVersion) throws Exception {
// lets create a record reader
RecordReader reader;
if (sortedColumn == null) {
reader = new RealtimeSegmentRecordReader(realtimeSegmentImpl, dataSchema);
} else {
reader = new RealtimeSegmentRecordReader(realtimeSegmentImpl, dataSchema, sortedColumn);
}
SegmentGeneratorConfig genConfig = new SegmentGeneratorConfig(dataSchema);
if (invertedIndexColumns != null && !invertedIndexColumns.isEmpty()) {
for (String column : invertedIndexColumns) {
genConfig.createInvertedIndexForColumn(column);
}
}
if (noDictionaryColumns != null) {
genConfig.setRawIndexCreationColumns(noDictionaryColumns);
}
genConfig.setTimeColumnName(dataSchema.getTimeFieldSpec().getOutgoingTimeColumnName());
genConfig.setSegmentTimeUnit(dataSchema.getTimeFieldSpec().getOutgoingGranularitySpec().getTimeType());
genConfig.setSegmentVersion(segmentVersion);
genConfig.setTableName(tableName);
genConfig.setOutDir(outputPath);
genConfig.setSegmentName(segmentName);
final SegmentIndexCreationDriverImpl driver = new SegmentIndexCreationDriverImpl();
driver.init(genConfig, reader);
driver.build();
}
use of com.linkedin.pinot.core.indexsegment.generator.SegmentGeneratorConfig in project pinot by linkedin.
the class SegmentMetadataImplTest method setUp.
@BeforeMethod
public void setUp() throws Exception {
INDEX_DIR = Files.createTempDirectory(SegmentMetadataImplTest.class.getName() + "_segmentDir").toFile();
final String filePath = TestUtils.getFileFromResourceUrl(SegmentMetadataImplTest.class.getClassLoader().getResource(AVRO_DATA));
// intentionally changed this to TimeUnit.Hours to make it non-default for testing
final SegmentGeneratorConfig config = SegmentTestUtils.getSegmentGenSpecWithSchemAndProjectedColumns(new File(filePath), INDEX_DIR, "daysSinceEpoch", TimeUnit.HOURS, "testTable");
config.setSegmentNamePostfix("1");
config.setTimeColumnName("daysSinceEpoch");
final SegmentIndexCreationDriver driver = SegmentCreationDriverFactory.get(null);
driver.init(config);
driver.build();
segmentDirectory = new File(INDEX_DIR, driver.getSegmentName());
}
use of com.linkedin.pinot.core.indexsegment.generator.SegmentGeneratorConfig 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;
}
use of com.linkedin.pinot.core.indexsegment.generator.SegmentGeneratorConfig in project pinot by linkedin.
the class TestStarTreeMetadata method setupSegment.
private void setupSegment(File segmentDir) throws Exception {
final String filePath = TestUtils.getFileFromResourceUrl(getClass().getClassLoader().getResource(AVRO_DATA));
if (segmentDir.exists()) {
FileUtils.deleteQuietly(segmentDir);
}
final SegmentGeneratorConfig config = SegmentTestUtils.getSegmentGenSpecWithSchemAndProjectedColumns(new File(filePath), segmentDir, "time_day", TimeUnit.DAYS, TABLE_NAME);
config.setTableName(TABLE_NAME);
config.setSegmentName(SEGMENT_NAME);
StarTreeIndexSpec starTreeIndexSpec = new StarTreeIndexSpec();
starTreeIndexSpec.setDimensionsSplitOrder(DIMENSIONS_SPLIT_ORDER);
starTreeIndexSpec.setMaxLeafRecords(MAX_LEAF_RECORDS);
starTreeIndexSpec.setSkipMaterializationCardinalityThreshold(SKIP_CARDINALITY_THRESHOLD);
starTreeIndexSpec.setSkipStarNodeCreationForDimensions(SKIP_STAR_NODE_CREATION_DIMENSTIONS);
starTreeIndexSpec.setSkipMaterializationForDimensions(SKIP_MATERIALIZATION_DIMENSIONS);
config.setEnableStarTreeIndex(true);
config.setStarTreeIndexSpec(starTreeIndexSpec);
final SegmentIndexCreationDriver driver = SegmentCreationDriverFactory.get(null);
driver.init(config);
driver.build();
}
Aggregations