/**
* Returns the user-defined split order, or dimensions in order of descending cardinality (removes
* excludes too)
*/
private List<String> computeSplitOrder(final Map<String, ColumnIndexCreationInfo> columnInfo) {
List<String> splitOrder;
if (starTreeIndexSpec.getSplitOrder() == null) {
splitOrder = new ArrayList<String>(schema.getDimensionNames());
Collections.sort(
splitOrder,
new Comparator<String>() {
@Override
public int compare(String o1, String o2) {
Object[] o1UniqueElements = columnInfo.get(o1).getSortedUniqueElementsArray();
Object[] o2UniqueElements = columnInfo.get(o2).getSortedUniqueElementsArray();
return o2UniqueElements.length - o1UniqueElements.length; // descending
}
});
} else {
splitOrder = new ArrayList<String>(starTreeIndexSpec.getSplitOrder());
}
if (starTreeIndexSpec.getSplitExcludes() != null) {
splitOrder.removeAll(starTreeIndexSpec.getSplitExcludes());
}
return splitOrder;
}
/** 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();
}