private void setupSegmentList(int numberOfSegments) throws Exception {
final String filePath =
TestUtils.getFileFromResourceUrl(getClass().getClassLoader().getResource(SMALL_AVRO_DATA));
_indexSegmentList.clear();
if (INDEXES_DIR.exists()) {
FileUtils.deleteQuietly(INDEXES_DIR);
}
INDEXES_DIR.mkdir();
for (int i = 0; i < numberOfSegments; ++i) {
final File segmentDir = new File(INDEXES_DIR, "segment_" + i);
final SegmentGeneratorConfig config =
SegmentTestUtils.getSegmentGenSpecWithSchemAndProjectedColumns(
new File(filePath), segmentDir, "dim" + i, TimeUnit.DAYS, "midas");
config.setSegmentNamePostfix(String.valueOf(i));
final SegmentIndexCreationDriver driver = SegmentCreationDriverFactory.get(null);
driver.init(config);
driver.build();
File parent = new File(INDEXES_DIR, "segment_" + String.valueOf(i));
String segmentName = parent.list()[0];
_indexSegmentList.add(
ColumnarSegmentLoader.load(new File(parent, segmentName), ReadMode.mmap));
System.out.println("built at : " + segmentDir.getAbsolutePath());
}
}
private void setupSegmentList(int numberOfSegments) throws Exception {
final String filePath =
TestUtils.getFileFromResourceUrl(getClass().getClassLoader().getResource(AVRO_DATA));
_indexSegmentList.clear();
if (INDEXES_DIR.exists()) {
FileUtils.deleteQuietly(INDEXES_DIR);
}
INDEXES_DIR.mkdir();
for (int i = 0; i < numberOfSegments; ++i) {
final File segmentDir = new File(INDEXES_DIR, "segment_" + i);
final SegmentGeneratorConfig config =
SegmentTestUtils.getSegmentGenSpecWithSchemAndProjectedColumns(
new File(filePath), segmentDir, "time_day", TimeUnit.DAYS, "test");
config.setSegmentNamePostfix(String.valueOf(i));
final SegmentIndexCreationDriver driver = SegmentCreationDriverFactory.get(null);
driver.init(config);
driver.build();
LOGGER.debug("built at : {}", segmentDir.getAbsolutePath());
final File indexSegmentDir = new File(segmentDir, driver.getSegmentName());
_indexSegmentList.add(
new OfflineSegmentDataManager(
ColumnarSegmentLoader.load(indexSegmentDir, ReadMode.heap)));
}
}
@Override
public void indexRow(GenericRow row) {
// Find matching leaves in StarTree for row
currentMatchingNodes.clear();
StarTreeTableRow tableRow = extractValues(row);
findMatchingLeaves(starTreeBuilder.getTree(), tableRow.getDimensions(), currentMatchingNodes);
// Only write the raw value, maintaining sort order (we will write aggregates when sealing)
for (StarTreeIndexNode node : currentMatchingNodes) {
Map<Integer, Integer> pathValues = node.getPathValues();
if (!pathValues.containsValue(StarTreeIndexNode.all())) {
StarTreeTableRange range = starTreeBuilder.getDocumentIdRange(node.getNodeId());
StarTreeTable subTable =
starTreeBuilder.getTable().view(range.getStartDocumentId(), range.getDocumentCount());
Integer nextMatchingDocumentId =
starTreeBuilder.getNextDocumentId(tableRow.getDimensions());
if (nextMatchingDocumentId == null) {
throw new IllegalStateException("Could not assign document ID for row " + tableRow);
}
// Write using that document ID to all columns
for (final String column : dictionaryCreatorMap.keySet()) {
Object columnValueToIndex = row.getValue(column);
if (schema.getFieldSpecFor(column).isSingleValueField()) {
System.out.println(column + ": " + columnValueToIndex);
int dictionaryIndex = dictionaryCreatorMap.get(column).indexOfSV(columnValueToIndex);
((SingleValueForwardIndexCreator) forwardIndexCreatorMap.get(column))
.index(nextMatchingDocumentId, dictionaryIndex);
if (config.createInvertedIndexEnabled()) {
invertedIndexCreatorMap
.get(column)
.add(nextMatchingDocumentId, (Object) dictionaryIndex);
}
} else {
int[] dictionaryIndex = dictionaryCreatorMap.get(column).indexOfMV(columnValueToIndex);
((MultiValueForwardIndexCreator) forwardIndexCreatorMap.get(column))
.index(nextMatchingDocumentId, dictionaryIndex);
if (config.createInvertedIndexEnabled()) {
invertedIndexCreatorMap.get(column).add(nextMatchingDocumentId, dictionaryIndex);
}
}
}
}
}
}
@Override
public void seal() throws ConfigurationException, IOException {
for (final String column : forwardIndexCreatorMap.keySet()) {
forwardIndexCreatorMap.get(column).close();
if (config.createInvertedIndexEnabled()) {
invertedIndexCreatorMap.get(column).seal();
}
dictionaryCreatorMap.get(column).close();
}
writeMetadata();
}
@Override
public void indexRow(GenericRow row) {
for (final String column : dictionaryCreatorMap.keySet()) {
Object columnValueToIndex = row.getValue(column);
Object dictionaryIndex;
if (dictionaryCache.get(column).containsKey(columnValueToIndex)) {
dictionaryIndex = dictionaryCache.get(column).get(columnValueToIndex);
} else {
dictionaryIndex = dictionaryCreatorMap.get(column).indexOf(columnValueToIndex);
dictionaryCache.get(column).put(columnValueToIndex, dictionaryIndex);
}
forwardIndexCreatorMap.get(column).index(docIdCounter, dictionaryIndex);
if (config.createInvertedIndexEnabled()) {
invertedIndexCreatorMap.get(column).add(docIdCounter, dictionaryIndex);
}
}
docIdCounter++;
}
@Override
public void init(
SegmentGeneratorConfig config,
Map<String, ColumnIndexCreationInfo> columnInfo,
Schema schema,
int totalDocs,
File outDir)
throws Exception {
// Member variables
this.config = config;
this.columnInfo = columnInfo;
this.schema = schema;
this.outDir = outDir;
this.starTreeDimensionDictionary = new HashMap<String, Integer>();
this.starTreeMetricDictionary = new HashMap<String, Integer>();
// Dictionaries (will go in root segment)
initializeAndBuildDictionaries(schema, columnInfo, outDir);
// Compute dimension dictionary
for (int i = 0; i < schema.getDimensionNames().size(); i++) {
starTreeDimensionDictionary.put(schema.getDimensionNames().get(i), i);
}
LOG.info("StarTree dimension dictionary: {}", starTreeDimensionDictionary);
// Compute the metric dictionary
for (int i = 0; i < schema.getMetricNames().size(); i++) {
starTreeMetricDictionary.put(schema.getMetricNames().get(i), i);
}
LOG.info("StarTree metric dictionary: {}", starTreeDimensionDictionary);
// Compute StarTree split order
splitOrder = computeSplitOrder(columnInfo);
LOG.info("Computed split order {}", splitOrder);
List<Integer> splitOrderIndexes = new ArrayList<Integer>();
for (String dimensionName : splitOrder) {
Integer dimensionId = starTreeDimensionDictionary.get(dimensionName);
splitOrderIndexes.add(dimensionId);
}
Collections.reverse(splitOrderIndexes);
// StarTree builder / table
StarTreeTable table = new LinkedListStarTreeTable(); // TODO: ByteBuffer-based
StarTreeDocumentIdMap documentIdMap =
new HashMapStarTreeDocumentIdMap(); // TODO: ByteBuffer-based
starTreeBuilder.init(
splitOrderIndexes, starTreeIndexSpec.getMaxLeafRecords(), table, documentIdMap);
// Build the StarTree structure and table
LOG.info("Building StarTree table...");
int count = 0;
long startMillis = System.currentTimeMillis();
recordReader.rewind();
while (recordReader.hasNext()) {
GenericRow row = recordReader.next();
StarTreeTableRow starTreeTableRow = extractValues(row);
starTreeBuilder.append(starTreeTableRow);
count++;
}
long endMillis = System.currentTimeMillis();
LOG.info(
"Finished building StarTree table ({} documents, took {} ms)",
count,
endMillis - startMillis);
LOG.info("Building StarTree (computing aggregates)...");
startMillis = System.currentTimeMillis();
starTreeBuilder.build();
endMillis = System.currentTimeMillis();
LOG.info("Finished building StarTree, took {} ms", endMillis - startMillis);
// Re-compute the unique values for metrics including aggregates to allow for dictionary
// encoding
LOG.info("Re-computing unique metric values for dictionary encoding...");
startMillis = System.currentTimeMillis();
Map<String, Set<Object>> uniqueMetricValues = computeUniqueMetricValues();
resetMetricDictionaries(uniqueMetricValues);
endMillis = System.currentTimeMillis();
LOG.info("Finished re-computing unique metric values (took {} ms)", endMillis - startMillis);
// StarTree directory
starTreeDir = new File(outDir, V1Constants.STARTREE_DIR);
if (!starTreeDir.mkdir()) {
throw new RuntimeException(
"Could not create star tree directory " + starTreeDir.getAbsolutePath());
}
// For each column, build its dictionary and initialize a forwards and an inverted index for raw
// / agg segment
int totalAggDocs = starTreeBuilder.getTotalAggregateDocumentCount();
int totalRawDocs = starTreeBuilder.getTotalRawDocumentCount();
for (final String column : dictionaryCreatorMap.keySet()) {
ColumnIndexCreationInfo indexCreationInfo = columnInfo.get(column);
Object[] uniqueValues = indexCreationInfo.getSortedUniqueElementsArray();
if (schema.getMetricNames().contains(column)) {
// Use the unique values including the new aggregate values
uniqueValues = uniqueMetricValues.get(column).toArray();
}
if (schema.getFieldSpecFor(column).isSingleValueField()) {
if (indexCreationInfo.isSorted()) {
forwardIndexCreatorMap.put(
column,
new SingleValueSortedForwardIndexCreator(
outDir, uniqueValues.length, schema.getFieldSpecFor(column)));
aggregateForwardIndexCreatorMap.put(
column,
new SingleValueSortedForwardIndexCreator(
starTreeDir, uniqueValues.length, schema.getFieldSpecFor(column)));
} else {
forwardIndexCreatorMap.put(
column,
new SingleValueUnsortedForwardIndexCreator(
schema.getFieldSpecFor(column),
outDir,
uniqueValues.length,
totalRawDocs,
indexCreationInfo.getTotalNumberOfEntries(),
indexCreationInfo.hasNulls()));
aggregateForwardIndexCreatorMap.put(
column,
new SingleValueUnsortedForwardIndexCreator(
schema.getFieldSpecFor(column),
starTreeDir,
indexCreationInfo.getSortedUniqueElementsArray().length,
totalAggDocs,
indexCreationInfo.getTotalNumberOfEntries(),
indexCreationInfo.hasNulls()));
}
} else {
forwardIndexCreatorMap.put(
column,
new MultiValueUnsortedForwardIndexCreator(
schema.getFieldSpecFor(column),
outDir,
uniqueValues.length,
totalRawDocs,
indexCreationInfo.getTotalNumberOfEntries(),
indexCreationInfo.hasNulls()));
aggregateForwardIndexCreatorMap.put(
column,
new MultiValueUnsortedForwardIndexCreator(
schema.getFieldSpecFor(column),
starTreeDir,
uniqueValues.length,
totalAggDocs,
indexCreationInfo.getTotalNumberOfEntries(),
indexCreationInfo.hasNulls()));
}
if (config.createInvertedIndexEnabled()) {
invertedIndexCreatorMap.put(
column,
new BitmapInvertedIndexCreator(
outDir,
indexCreationInfo.getSortedUniqueElementsArray().length,
schema.getFieldSpecFor(column)));
aggregateInvertedIndexCreatorMap.put(
column,
new BitmapInvertedIndexCreator(
starTreeDir,
indexCreationInfo.getSortedUniqueElementsArray().length,
schema.getFieldSpecFor(column)));
}
}
}
/** Constructs the segment metadata file, and writes in outputDir */
private void writeMetadata(File outputDir, int totalDocs) throws ConfigurationException {
final PropertiesConfiguration properties =
new PropertiesConfiguration(
new File(outputDir, V1Constants.MetadataKeys.METADATA_FILE_NAME));
properties.setProperty(SEGMENT_NAME, segmentName);
properties.setProperty(TABLE_NAME, config.getTableName());
properties.setProperty(DIMENSIONS, config.getDimensions());
properties.setProperty(METRICS, config.getMetrics());
properties.setProperty(TIME_COLUMN_NAME, config.getTimeColumnName());
properties.setProperty(TIME_INTERVAL, "not_there");
properties.setProperty(SEGMENT_TOTAL_DOCS, String.valueOf(totalDocs));
// StarTree
Joiner csv = Joiner.on(",");
properties.setProperty(SPLIT_ORDER, csv.join(splitOrder));
properties.setProperty(SPLIT_EXCLUDES, csv.join(starTreeIndexSpec.getSplitExcludes()));
properties.setProperty(MAX_LEAF_RECORDS, starTreeIndexSpec.getMaxLeafRecords());
properties.setProperty(
EXCLUDED_DIMENSIONS, csv.join(starTreeIndexSpec.getExcludedDimensions()));
String timeColumn = config.getTimeColumnName();
if (columnInfo.get(timeColumn) != null) {
properties.setProperty(SEGMENT_START_TIME, columnInfo.get(timeColumn).getMin());
properties.setProperty(SEGMENT_END_TIME, columnInfo.get(timeColumn).getMax());
properties.setProperty(TIME_UNIT, config.getTimeUnitForSegment());
}
if (config.containsKey(SEGMENT_START_TIME)) {
properties.setProperty(SEGMENT_START_TIME, config.getStartTime());
}
if (config.containsKey(SEGMENT_END_TIME)) {
properties.setProperty(SEGMENT_END_TIME, config.getStartTime());
}
if (config.containsKey(TIME_UNIT)) {
properties.setProperty(TIME_UNIT, config.getTimeUnitForSegment());
}
for (final String key : config.getAllCustomKeyValuePair().keySet()) {
properties.setProperty(key, config.getAllCustomKeyValuePair().get(key));
}
for (final String column : columnInfo.keySet()) {
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, CARDINALITY),
String.valueOf(columnInfo.get(column).getSortedUniqueElementsArray().length));
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, TOTAL_DOCS), String.valueOf(totalDocs));
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, DATA_TYPE),
schema.getFieldSpecFor(column).getDataType().toString());
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, BITS_PER_ELEMENT),
String.valueOf(
SingleValueUnsortedForwardIndexCreator.getNumOfBits(
columnInfo.get(column).getSortedUniqueElementsArray().length)));
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, DICTIONARY_ELEMENT_SIZE),
String.valueOf(dictionaryCreatorMap.get(column).getStringColumnMaxLength()));
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, COLUMN_TYPE),
String.valueOf(schema.getFieldSpecFor(column).getFieldType().toString()));
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, IS_SORTED),
String.valueOf(columnInfo.get(column).isSorted()));
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, HAS_NULL_VALUE),
String.valueOf(columnInfo.get(column).hasNulls()));
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(
column, V1Constants.MetadataKeys.Column.HAS_DICTIONARY),
String.valueOf(columnInfo.get(column).isCreateDictionary()));
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, HAS_INVERTED_INDEX),
String.valueOf(true));
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, IS_SINGLE_VALUED),
String.valueOf(schema.getFieldSpecFor(column).isSingleValueField()));
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, MAX_MULTI_VALUE_ELEMTS),
String.valueOf(columnInfo.get(column).getMaxNumberOfMutiValueElements()));
properties.setProperty(
V1Constants.MetadataKeys.Column.getKeyFor(column, TOTAL_NUMBER_OF_ENTRIES),
String.valueOf(columnInfo.get(column).getTotalNumberOfEntries()));
}
properties.save();
}
@Override
public void seal() throws ConfigurationException, IOException {
// Write all the aggregate rows to the aggregate segment
LOG.info("Writing aggregate segment...");
long startMillis = System.currentTimeMillis();
int currentAggregateDocumentId = 0;
Iterator<StarTreeTableRow> itr = starTreeBuilder.getTable().getAllCombinations();
while (itr.hasNext()) {
StarTreeTableRow next = itr.next();
if (next.getDimensions().contains(StarTreeIndexNode.all())) {
// Write using that document ID to all columns
for (final String column : dictionaryCreatorMap.keySet()) {
Object dictionaryIndex = null; // TODO: Is this okay?
if (starTreeDimensionDictionary.containsKey(column)) {
// Index the dimension value
Integer dimensionId = starTreeDimensionDictionary.get(column);
Integer dimensionValue = next.getDimensions().get(dimensionId);
if (dimensionValue == StarTreeIndexNode.all()) {
// Use all value
Object allValue = StarTreeIndexNode.getAllValue(schema.getFieldSpecFor(column));
if (schema.getFieldSpecFor(column).isSingleValueField()) {
dictionaryIndex = dictionaryCreatorMap.get(column).indexOfSV(allValue);
} else {
dictionaryIndex = dictionaryCreatorMap.get(column).indexOfMV(allValue);
}
} else {
dictionaryIndex = dimensionValue;
}
} else if (starTreeMetricDictionary.containsKey(column)) {
// Index the aggregate metric
Integer metricId = starTreeMetricDictionary.get(column);
Object columnValueToIndex = next.getMetrics().get(metricId);
if (schema.getFieldSpecFor(column).isSingleValueField()) {
dictionaryIndex = dictionaryCreatorMap.get(column).indexOfSV(columnValueToIndex);
} else {
dictionaryIndex = dictionaryCreatorMap.get(column).indexOfMV(columnValueToIndex);
}
} else {
// Just index the raw value
Object columnValueToIndex =
StarTreeIndexNode.getAllValue(schema.getFieldSpecFor(column));
if (schema.getFieldSpecFor(column).isSingleValueField()) {
dictionaryIndex = dictionaryCreatorMap.get(column).indexOfSV(columnValueToIndex);
} else {
dictionaryIndex = dictionaryCreatorMap.get(column).indexOfMV(columnValueToIndex);
}
}
if (schema.getFieldSpecFor(column).isSingleValueField()) {
((SingleValueForwardIndexCreator) aggregateForwardIndexCreatorMap.get(column))
.index(currentAggregateDocumentId, (Integer) dictionaryIndex);
} else {
((MultiValueForwardIndexCreator) aggregateForwardIndexCreatorMap.get(column))
.index(currentAggregateDocumentId, (int[]) dictionaryIndex);
}
if (config.createInvertedIndexEnabled()) {
aggregateInvertedIndexCreatorMap
.get(column)
.add(currentAggregateDocumentId, dictionaryIndex);
}
}
currentAggregateDocumentId++;
}
}
long endMillis = System.currentTimeMillis();
LOG.info("Done writing aggregate segment (took {} ms)", endMillis - startMillis);
for (final String column : forwardIndexCreatorMap.keySet()) {
forwardIndexCreatorMap.get(column).close();
if (config.createInvertedIndexEnabled()) {
invertedIndexCreatorMap.get(column).seal();
}
dictionaryCreatorMap.get(column).close();
}
for (final String column : aggregateForwardIndexCreatorMap.keySet()) {
aggregateForwardIndexCreatorMap.get(column).close();
if (config.createInvertedIndexEnabled()) {
aggregateInvertedIndexCreatorMap.get(column).seal();
}
// n.b. The dictionary from raw data is used
}
writeMetadata(outDir, starTreeBuilder.getTotalRawDocumentCount());
// Write star tree
LOG.info("Writing " + V1Constants.STARTREE_FILE);
startMillis = System.currentTimeMillis();
File starTreeFile = new File(starTreeDir, V1Constants.STARTREE_FILE);
OutputStream starTreeOutputStream = new FileOutputStream(starTreeFile);
starTreeBuilder.getTree().writeTree(starTreeOutputStream);
starTreeOutputStream.close();
endMillis = System.currentTimeMillis();
LOG.info("Wrote StarTree file (took {} ms)", endMillis - startMillis);
// Copy the dictionary files into startree directory
// n.b. this is done so the segment is as stand-alone as possible, though could be removed as an
// optimization
File[] dictionaryFiles =
outDir.listFiles(
new FilenameFilter() {
@Override
public boolean accept(File dir, String name) {
return name.endsWith(V1Constants.Dict.FILE_EXTENTION);
}
});
for (File dictionaryFile : dictionaryFiles) {
FileUtils.copyFile(dictionaryFile, new File(starTreeDir, dictionaryFile.getName()));
}
// Write star tree metadata
writeMetadata(starTreeDir, starTreeBuilder.getTotalAggregateDocumentCount());
}
@Override
public void init(
SegmentGeneratorConfig segmentCreationSpec,
Map<String, ColumnIndexCreationInfo> indexCreationInfoMap,
Schema schema,
int totalDocs,
File outDir)
throws Exception {
docIdCounter = 0;
config = segmentCreationSpec;
this.indexCreationInfoMap = indexCreationInfoMap;
dictionaryCreatorMap = new HashMap<String, SegmentDictionaryCreator>();
forwardIndexCreatorMap = new HashMap<String, ForwardIndexCreator>();
this.indexCreationInfoMap = indexCreationInfoMap;
invertedIndexCreatorMap = new HashMap<String, InvertedIndexCreator>();
file = outDir;
// Check that the output directory does not exist
if (file.exists()) {
throw new RuntimeException(
"Segment output directory " + file.getAbsolutePath() + " already exists.");
}
file.mkdir();
this.schema = schema;
this.totalDocs = totalDocs;
// Initialize and build dictionaries
for (final FieldSpec spec : schema.getAllFieldSpecs()) {
final ColumnIndexCreationInfo info = indexCreationInfoMap.get(spec.getName());
if (info.isCreateDictionary()) {
dictionaryCreatorMap.put(
spec.getName(),
new SegmentDictionaryCreator(
info.hasNulls(), info.getSortedUniqueElementsArray(), spec, file));
} else {
throw new RuntimeException("Creation of indices without dictionaries is not implemented!");
}
}
// For each column, build its dictionary and initialize a forwards and an inverted index
for (final String column : dictionaryCreatorMap.keySet()) {
dictionaryCreatorMap.get(column).build();
dictionaryCache.put(column, new HashMap<Object, Object>());
ColumnIndexCreationInfo indexCreationInfo = indexCreationInfoMap.get(column);
if (schema.getFieldSpecFor(column).isSingleValueField()) {
if (indexCreationInfo.isSorted()) {
forwardIndexCreatorMap.put(
column,
new SingleValueSortedForwardIndexCreator(
file,
indexCreationInfo.getSortedUniqueElementsArray().length,
schema.getFieldSpecFor(column)));
} else {
forwardIndexCreatorMap.put(
column,
new SingleValueUnsortedForwardIndexCreator(
schema.getFieldSpecFor(column),
file,
indexCreationInfo.getSortedUniqueElementsArray().length,
totalDocs,
indexCreationInfo.getTotalNumberOfEntries(),
indexCreationInfo.hasNulls()));
}
} else {
forwardIndexCreatorMap.put(
column,
new MultiValueUnsortedForwardIndexCreator(
schema.getFieldSpecFor(column),
file,
indexCreationInfo.getSortedUniqueElementsArray().length,
totalDocs,
indexCreationInfo.getTotalNumberOfEntries(),
indexCreationInfo.hasNulls()));
}
if (config.createInvertedIndexEnabled()) {
invertedIndexCreatorMap.put(
column,
new BitmapInvertedIndexCreator(
file,
indexCreationInfo.getSortedUniqueElementsArray().length,
schema.getFieldSpecFor(column)));
}
}
}
