/** Converts a raw row into its (possibly partial) dimension and complete metric values */
private StarTreeTableRow extractValues(GenericRow row) {
List<Integer> dimensions = new ArrayList<Integer>();
for (String dimensionName : schema.getDimensionNames()) {
Integer valueId;
if (schema.getFieldSpecFor(dimensionName).isSingleValueField()
&& !starTreeIndexSpec.getExcludedDimensions().contains(dimensionName)) {
Object value = row.getValue(dimensionName);
valueId = dictionaryCreatorMap.get(dimensionName).indexOfSV(value);
} else {
// Multi-value fields are not supported - always ALL
valueId = V1Constants.STARTREE_ALL_NUMBER.intValue();
}
dimensions.add(valueId);
}
List<Number> metrics = new ArrayList<Number>(schema.getMetricNames().size());
for (MetricFieldSpec metricFieldSpec : schema.getMetricFieldSpecs()) {
Object value = row.getValue(metricFieldSpec.getName());
switch (metricFieldSpec.getDataType()) {
case INT:
metrics.add((Integer) value);
break;
case LONG:
metrics.add((Long) value);
break;
case DOUBLE:
metrics.add((Double) value);
break;
case FLOAT:
metrics.add((Float) value);
break;
default:
throw new IllegalStateException("Unsupported data type " + metricFieldSpec.getDataType());
}
}
return new StarTreeTableRow(dimensions, metrics);
}
/** 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();
}