/**
* Creates an instance of StatsValues which supports values from the specified {@link StatsField}
*
* @param statsField {@link StatsField} whose statistics will be created by the resulting {@link
* StatsValues}
* @return Instance of {@link StatsValues} that will create statistics from values from the
* specified {@link StatsField}
*/
public static StatsValues createStatsValues(StatsField statsField) {
final SchemaField sf = statsField.getSchemaField();
if (null == sf) {
// function stats
return new NumericStatsValues(statsField);
}
final FieldType fieldType = sf.getType(); // TODO: allow FieldType to provide impl.
if (TrieDateField.class.isInstance(fieldType)) {
return new DateStatsValues(statsField);
} else if (TrieField.class.isInstance(fieldType)) {
return new NumericStatsValues(statsField);
} else if (StrField.class.isInstance(fieldType)) {
return new StringStatsValues(statsField);
} else if (sf.getType().getClass().equals(EnumField.class)) {
return new EnumStatsValues(statsField);
} else {
throw new SolrException(
SolrException.ErrorCode.BAD_REQUEST,
"Field type " + fieldType + " is not currently supported");
}
}
protected AbstractStatsValues(StatsField statsField) {
this.statsField = statsField;
this.computeCount = statsField.calculateStats(Stat.count);
this.computeMissing = statsField.calculateStats(Stat.missing);
this.computeCalcDistinct =
statsField.calculateStats(Stat.countDistinct)
|| statsField.calculateStats(Stat.distinctValues);
this.computeMin = statsField.calculateStats(Stat.min);
this.computeMax = statsField.calculateStats(Stat.max);
this.computeMinOrMax = computeMin || computeMax;
this.distinctValues = computeCalcDistinct ? new TreeSet<T>() : null;
this.computeCardinality = statsField.calculateStats(Stat.cardinality);
if (computeCardinality) {
hasher = statsField.getHllOptions().getHasher();
hll = statsField.getHllOptions().newHLL();
assert null != hll : "Cardinality requires an HLL";
} else {
hll = null;
hasher = null;
}
// alternatively, we could refactor a common base class that doesn't know/care
// about either SchemaField or ValueSource - but then there would be a lot of
// duplicate code between "NumericSchemaFieldStatsValues" and
// "NumericValueSourceStatsValues" which would have diff parent classes
//
// part of the complexity here being that the StatsValues API serves two
// masters: collecting concrete Values from things like DocValuesStats and
// the distributed aggregation logic, but also collecting docIds which it
// then
// uses to go out and pull concreate values from the ValueSource
// (from a func, or single valued field)
if (null != statsField.getSchemaField()) {
assert null == statsField.getValueSource();
this.sf = statsField.getSchemaField();
this.ft = sf.getType();
} else {
assert null != statsField.getValueSource();
assert null == statsField.getSchemaField();
this.sf = null;
this.ft = null;
}
}
