public static IndexPredicateAnalyzer createAnalyzer(boolean equalOnly) {
IndexPredicateAnalyzer analyzer = new IndexPredicateAnalyzer();
analyzer.addComparisonOp("org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPEqual");
if (equalOnly) {
return analyzer;
}
analyzer.addComparisonOp(
"org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPEqualOrGreaterThan");
analyzer.addComparisonOp("org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPEqualOrLessThan");
analyzer.addComparisonOp("org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan");
analyzer.addComparisonOp("org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPGreaterThan");
return analyzer;
}
/**
* Instantiates a new predicate analyzer suitable for determining how to push a filter down into
* the HBase scan, based on the rules for what kinds of pushdown we currently support.
*
* @param keyColumnName name of the Hive column mapped to the HBase row key
* @return preconfigured predicate analyzer
*/
static IndexPredicateAnalyzer newIndexPredicateAnalyzer(
String keyColumnName, String keyColType, boolean isKeyBinary) {
IndexPredicateAnalyzer analyzer = new IndexPredicateAnalyzer();
// We can always do equality predicate. Just need to make sure we get appropriate
// BA representation of constant of filter condition.
analyzer.addComparisonOp("org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPEqual");
// We can do other comparisons only if storage format in hbase is either binary
// or we are dealing with string types since there lexographic ordering will suffice.
if (isKeyBinary || (keyColType.equalsIgnoreCase("string"))) {
analyzer.addComparisonOp(
"org.apache.hadoop.hive.ql.udf.generic." + "GenericUDFOPEqualOrGreaterThan");
analyzer.addComparisonOp("org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPEqualOrLessThan");
analyzer.addComparisonOp("org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPLessThan");
analyzer.addComparisonOp("org.apache.hadoop.hive.ql.udf.generic.GenericUDFOPGreaterThan");
}
// and only on the key column
analyzer.clearAllowedColumnNames();
analyzer.allowColumnName(keyColumnName);
return analyzer;
}