Skip to content

HIVE-28911: Improve SEARCH expansion to exploit <> operator #6503

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
rubenada wants to merge 12 commits into
base: master
Choose a base branch
from
+152 −84
Open

HIVE-28911: Improve SEARCH expansion to exploit <> operator #6503

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
12 commits
Select commit Hold shift + click to select a range
827b879
HIVE-28911: Improve SEARCH expansion to exploit <> operator
rubenada May 21, 2026
2142444
minor
rubenada May 21, 2026
7a73f3b
Adjust tests
rubenada May 21, 2026
952a039
Adjust tests
rubenada May 23, 2026
d31ba37
In SearchTransformer, generate 'NOT (ref IN (value1, value2))' instea…
rubenada May 26, 2026
cbe2b37
Put back AND of NOT_EQUALS (instead of NOT IN), flatten ANDs / ORs in…
rubenada Jun 1, 2026
55bb8cc
Update tests files
rubenada Jun 1, 2026
da58715
Simplify
rubenada Jun 2, 2026
cdec829
Remove unused import
rubenada Jun 2, 2026
2635139
Apply fix (AND flattening) in ExprNodeDescUtils
rubenada Jun 2, 2026
8a9a283
empty
rubenada Jun 2, 2026
0578182
Apply OR flattening in ExprNodeConverter for consistency
rubenada Jun 2, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions ...ceberg-handler/src/test/results/positive/delete_iceberg_copy_on_write_unpartitioned.q.out
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -48,10 +48,10 @@ STAGE PLANS:
Map Operator Tree:
TableScan
alias: tbl_ice
filterExpr: (((b) IN ('four', 'one') or (a = 22)) is null or (((b < 'four') or ((b > 'four') and (b < 'one')) or (b > 'one')) and (a <> 22)) or (b) IN ('four', 'one') or (a = 22)) (type: boolean)
filterExpr: (((b) IN ('four', 'one') or (a = 22)) is null or ((b <> 'four') and (b <> 'one') and (a <> 22)) or (b) IN ('four', 'one') or (a = 22)) (type: boolean)
Statistics: Num rows: 7 Data size: 672 Basic stats: COMPLETE Column stats: COMPLETE
Filter Operator
predicate: ((((b) IN ('four', 'one') or (a = 22)) is null or (((b < 'four') or ((b > 'four') and (b < 'one')) or (b > 'one')) and (a <> 22))) and FILE__PATH is not null) (type: boolean)
predicate: ((((b) IN ('four', 'one') or (a = 22)) is null or ((b <> 'four') and (b <> 'one') and (a <> 22))) and FILE__PATH is not null) (type: boolean)
Statistics: Num rows: 7 Data size: 672 Basic stats: COMPLETE Column stats: COMPLETE
Select Operator
expressions: a (type: int), b (type: string), c (type: int), PARTITION__SPEC__ID (type: int), PARTITION__HASH (type: bigint), FILE__PATH (type: string), ROW__POSITION (type: bigint), PARTITION__PROJECTION (type: string)
Expand Down
80 changes: 40 additions & 40 deletions iceberg/iceberg-handler/src/test/results/positive/llap/iceberg_bucket_map_join_7.q.out
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -150,27 +150,27 @@ Stage-0
File Output Operator [FS_61]
Limit [LIM_60] (rows=20 width=447)
Number of rows:20
Select Operator [SEL_59] (rows=473 width=447)
Select Operator [SEL_59] (rows=791 width=447)
Output:["_col0","_col1","_col2","_col3","_col4"]
<-Map 1 [SIMPLE_EDGE] vectorized, llap
SHUFFLE [RS_58]
Top N Key Operator [TNK_57] (rows=473 width=447)
Top N Key Operator [TNK_57] (rows=791 width=447)
keys:_col0,top n:20
Map Join Operator [MAPJOIN_56] (rows=473 width=447)
Map Join Operator [MAPJOIN_56] (rows=791 width=447)
BucketMapJoin:true,Conds:SEL_55._col0, _col1=RS_53._col0, _col1(Inner),Output:["_col0","_col1","_col2","_col3","_col4"]
<-Map 3 [CUSTOM_EDGE] vectorized, llap
MULTICAST [RS_53]
PartitionCols:_col0, _col1
Select Operator [SEL_52] (rows=387 width=178)
Select Operator [SEL_52] (rows=500 width=178)
Output:["_col0","_col1"]
Filter Operator [FIL_51] (rows=387 width=178)
predicate:(((key < '0') or ((key > '0') and (key < '100')) or (key > '100')) and value is not null)
Filter Operator [FIL_51] (rows=500 width=178)
predicate:((key <> '0') and (key <> '100') and value is not null)
TableScan [TS_3] (rows=500 width=178)
default@src,b,Tbl:COMPLETE,Col:COMPLETE,Output:["key","value"]
<-Select Operator [SEL_55] (rows=387 width=269)
<-Select Operator [SEL_55] (rows=500 width=269)
Output:["_col0","_col1","_col2"]
Filter Operator [FIL_54] (rows=387 width=269)
predicate:(((key1 < '0') or ((key1 > '0') and (key1 < '100')) or (key1 > '100')) and key2 is not null)
Filter Operator [FIL_54] (rows=500 width=269)
predicate:((key1 <> '0') and (key1 <> '100') and key2 is not null)
TableScan [TS_0] (rows=500 width=269)
default@srcbucket_big,a,Tbl:COMPLETE,Col:COMPLETE,Grouping Num Buckets:8,Grouping Partition Columns:["key1","key2"],Output:["key1","key2","value"]

Expand Down Expand Up @@ -346,27 +346,27 @@ Stage-0
File Output Operator [FS_41]
Limit [LIM_40] (rows=20 width=447)
Number of rows:20
Select Operator [SEL_39] (rows=473 width=447)
Select Operator [SEL_39] (rows=791 width=447)
Output:["_col0","_col1","_col2","_col3","_col4"]
<-Map 1 [SIMPLE_EDGE] vectorized, llap
SHUFFLE [RS_38]
Top N Key Operator [TNK_37] (rows=473 width=447)
Top N Key Operator [TNK_37] (rows=791 width=447)
keys:_col0,top n:20
Map Join Operator [MAPJOIN_36] (rows=473 width=447)
Map Join Operator [MAPJOIN_36] (rows=791 width=447)
BucketMapJoin:true,Conds:SEL_35._col0=RS_33._col0(Inner),Output:["_col0","_col1","_col2","_col3","_col4"]
<-Map 3 [CUSTOM_EDGE] vectorized, llap
MULTICAST [RS_33]
PartitionCols:_col0
Select Operator [SEL_32] (rows=387 width=178)
Select Operator [SEL_32] (rows=500 width=178)
Output:["_col0","_col1"]
Filter Operator [FIL_31] (rows=387 width=178)
predicate:((key < '0') or (key > '100') or ((key > '0') and (key < '100')))
Filter Operator [FIL_31] (rows=500 width=178)
predicate:((key <> '0') and (key <> '100'))
TableScan [TS_3] (rows=500 width=178)
default@src,b,Tbl:COMPLETE,Col:COMPLETE,Output:["key","value"]
<-Select Operator [SEL_35] (rows=387 width=269)
<-Select Operator [SEL_35] (rows=500 width=269)
Output:["_col0","_col1","_col2"]
Filter Operator [FIL_34] (rows=387 width=269)
predicate:((key1 < '0') or (key1 > '100') or ((key1 > '0') and (key1 < '100')))
Filter Operator [FIL_34] (rows=500 width=269)
predicate:((key1 <> '0') and (key1 <> '100'))
TableScan [TS_0] (rows=500 width=269)
default@srcbucket_big,a,Tbl:COMPLETE,Col:COMPLETE,Grouping Num Buckets:4,Grouping Partition Columns:["key1"],Output:["key1","key2","value"]

Expand Down Expand Up @@ -435,40 +435,40 @@ POSTHOOK: Input: default@srcbucket_big
Plan optimized by CBO.

Vertex dependency in root stage
Map 2 <- Map 1 (BROADCAST_EDGE)
Reducer 3 <- Map 2 (SIMPLE_EDGE)
Map 1 <- Map 3 (CUSTOM_EDGE)
Reducer 2 <- Map 1 (SIMPLE_EDGE)

Stage-0
Fetch Operator
limit:20
Stage-1
Reducer 3 vectorized, llap
Reducer 2 vectorized, llap
File Output Operator [FS_41]
Limit [LIM_40] (rows=20 width=447)
Number of rows:20
Select Operator [SEL_39] (rows=612 width=447)
Select Operator [SEL_39] (rows=791 width=447)
Output:["_col0","_col1","_col2","_col3","_col4"]
<-Map 2 [SIMPLE_EDGE] vectorized, llap
<-Map 1 [SIMPLE_EDGE] vectorized, llap
SHUFFLE [RS_38]
Top N Key Operator [TNK_37] (rows=612 width=447)
Top N Key Operator [TNK_37] (rows=791 width=447)
keys:_col0,top n:20
Map Join Operator [MAPJOIN_36] (rows=612 width=447)
Conds:RS_33._col0=SEL_35._col0(Inner),Output:["_col0","_col1","_col2","_col3","_col4"]
<-Map 1 [BROADCAST_EDGE] vectorized, llap
BROADCAST [RS_33]
Map Join Operator [MAPJOIN_36] (rows=791 width=447)
BucketMapJoin:true,Conds:SEL_35._col0=RS_33._col0(Inner),Output:["_col0","_col1","_col2","_col3","_col4"]
<-Map 3 [CUSTOM_EDGE] vectorized, llap
MULTICAST [RS_33]
PartitionCols:_col0
Select Operator [SEL_32] (rows=387 width=269)
Output:["_col0","_col1","_col2"]
Filter Operator [FIL_31] (rows=387 width=269)
predicate:(((key2 < 'val_0') or ((key2 > 'val_0') and (key2 < 'val_100')) or (key2 > 'val_100')) and key1 is not null)
TableScan [TS_0] (rows=500 width=269)
default@srcbucket_big,a,Tbl:COMPLETE,Col:COMPLETE,Output:["key1","key2","value"]
<-Select Operator [SEL_35] (rows=500 width=178)
Output:["_col0","_col1"]
Filter Operator [FIL_34] (rows=500 width=178)
predicate:key is not null
TableScan [TS_3] (rows=500 width=178)
default@src,b,Tbl:COMPLETE,Col:COMPLETE,Output:["key","value"]
Select Operator [SEL_32] (rows=500 width=178)
Output:["_col0","_col1"]
Filter Operator [FIL_31] (rows=500 width=178)
predicate:key is not null
TableScan [TS_3] (rows=500 width=178)
default@src,b,Tbl:COMPLETE,Col:COMPLETE,Output:["key","value"]
<-Select Operator [SEL_35] (rows=500 width=269)
Output:["_col0","_col1","_col2"]
Filter Operator [FIL_34] (rows=500 width=269)
predicate:((key2 <> 'val_0') and (key2 <> 'val_100') and key1 is not null)
TableScan [TS_0] (rows=500 width=269)
default@srcbucket_big,a,Tbl:COMPLETE,Col:COMPLETE,Grouping Num Buckets:4,Grouping Partition Columns:["key1"],Output:["key1","key2","value"]

PREHOOK: query: SELECT *
FROM srcbucket_big a
Expand Down
4 changes: 2 additions & 2 deletions .../iceberg-handler/src/test/results/positive/update_iceberg_copy_on_write_partitioned.q.out
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,10 +71,10 @@ STAGE PLANS:
Map Operator Tree:
TableScan
alias: tbl_ice
filterExpr: (((b) IN ('four', 'one') or (a = 22)) is null or (((b < 'four') or ((b > 'four') and (b < 'one')) or (b > 'one')) and (a <> 22))) (type: boolean)
filterExpr: (((b) IN ('four', 'one') or (a = 22)) is null or ((b <> 'four') and (b <> 'one') and (a <> 22))) (type: boolean)
Statistics: Num rows: 1 Data size: 84 Basic stats: COMPLETE Column stats: PARTIAL
Filter Operator
predicate: ((((b) IN ('four', 'one') or (a = 22)) is null or (((b < 'four') or ((b > 'four') and (b < 'one')) or (b > 'one')) and (a <> 22))) and FILE__PATH is not null) (type: boolean)
predicate: ((((b) IN ('four', 'one') or (a = 22)) is null or ((b <> 'four') and (b <> 'one') and (a <> 22))) and FILE__PATH is not null) (type: boolean)
Statistics: Num rows: 1 Data size: 84 Basic stats: COMPLETE Column stats: PARTIAL
Select Operator
expressions: a (type: int), b (type: string), c (type: int), PARTITION__SPEC__ID (type: int), PARTITION__HASH (type: bigint), FILE__PATH (type: string), ROW__POSITION (type: bigint), PARTITION__PROJECTION (type: string)
Expand Down
4 changes: 2 additions & 2 deletions ...ceberg-handler/src/test/results/positive/update_iceberg_copy_on_write_unpartitioned.q.out
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -71,7 +71,7 @@ STAGE PLANS:
Map Operator Tree:
TableScan
alias: tbl_ice
filterExpr: ((a = 22) or (b) IN ('four', 'one') or ((b) IN ('four', 'one') or (a = 22)) is null or (((b < 'four') or ((b > 'four') and (b < 'one')) or (b > 'one')) and (a <> 22))) (type: boolean)
filterExpr: ((a = 22) or (b) IN ('four', 'one') or ((b) IN ('four', 'one') or (a = 22)) is null or ((b <> 'four') and (b <> 'one') and (a <> 22))) (type: boolean)
Statistics: Num rows: 7 Data size: 672 Basic stats: COMPLETE Column stats: COMPLETE
Filter Operator
predicate: ((a = 22) or (b) IN ('four', 'one')) (type: boolean)
Expand All @@ -93,7 +93,7 @@ STAGE PLANS:
Map-reduce partition columns: FILE__PATH (type: string)
Statistics: Num rows: 4 Data size: 368 Basic stats: COMPLETE Column stats: COMPLETE
Filter Operator
predicate: ((((b) IN ('four', 'one') or (a = 22)) is null or (((b < 'four') or ((b > 'four') and (b < 'one')) or (b > 'one')) and (a <> 22))) and FILE__PATH is not null) (type: boolean)
predicate: ((((b) IN ('four', 'one') or (a = 22)) is null or ((b <> 'four') and (b <> 'one') and (a <> 22))) and FILE__PATH is not null) (type: boolean)
Statistics: Num rows: 7 Data size: 672 Basic stats: COMPLETE Column stats: COMPLETE
Select Operator
expressions: a (type: int), b (type: string), c (type: int), PARTITION__SPEC__ID (type: int), PARTITION__HASH (type: bigint), FILE__PATH (type: string), ROW__POSITION (type: bigint), PARTITION__PROJECTION (type: string)
Expand Down
44 changes: 29 additions & 15 deletions ql/src/java/org/apache/hadoop/hive/ql/optimizer/calcite/SearchTransformer.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -72,30 +72,44 @@ public SearchTransformer(RexBuilder rexBuilder, RexCall search, final RexUnknown
this.unknownContext = unknownContext;
}

/**
* Transforms the SEARCH expression into an equivalent RexNode expression.
* Warning: when called from a shuttle, callers of this method should consider flattening AND/OR expressions
* afterward, to get the same result as applying {@link SearchTransformer.Shuttle}.
*/
public RexNode transform() {
PerfLogger perfLogger = SessionState.getPerfLogger();
perfLogger.perfLogBegin(this.getClass().getName(), PerfLogger.SEARCH_TRANSFORMER);

RangeConverter<C> consumer = new RangeConverter<>(rexBuilder, operandType, ref);
RangeSets.forEach(sarg.rangeSet, consumer);

List<RexNode> orList = new ArrayList<>();
if (sarg.nullAs == RexUnknownAs.TRUE && unknownContext != RexUnknownAs.TRUE) {
orList.add(rexBuilder.makeCall(SqlStdOperatorTable.IS_NULL, ref));
}
switch (consumer.inLiterals.size()) {
case 0:
break;
case 1:
orList.add(rexBuilder.makeCall(SqlStdOperatorTable.EQUALS, ref, consumer.inLiterals.get(0)));
break;
default:
List<RexNode> operands = new ArrayList<>(consumer.inLiterals.size() + 1);
operands.add(ref);
operands.addAll(consumer.inLiterals);
orList.add(rexBuilder.makeCall(HiveIn.INSTANCE, operands));

if (sarg.isComplementedPoints()) {
Copy link
Copy Markdown
Member

@zabetak zabetak Jun 4, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

When the sarg is not strictly a complement then I guess we cannot handle it. Examples:

... WHERE NOT ( x = 10 OR x = 20 OR (x > 30 AND x < 50))
... WHERE (x <> 10 AND x <> 20 AND (x <= 30 OR x >= 50))

One way to cover those would be to apply the existing RangeConverter on sarg.rangeSet.complement() and invert the EQUALS and HiveIn operators in the switch. The challenge here is when to use the sarg.rangeSet and when its complement. A naive choice could be to base the decision on sarg.rangeSet.asRanges().size() or something along these lines.

Anyways, the change here is fine as it is so we can log a another ticket and follow-up there if its worth to do it or not.

Copy link
Copy Markdown
Member

@zabetak zabetak Jun 5, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

In fact we could negate the entire sarg (not only the rangeset) if that leads to a simpler more efficient expansion. This would evolve negating the operators (IS NULL, =, IN, etc.), and changing the orList into an andList.

Copy link
Copy Markdown
Contributor Author

@rubenada rubenada Jun 5, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

What that would look like? Try systematically the "normal" sarg expansion and the negated one, and pick the "simpler" one?
I guess we could try... but could we consider that a follow-up, separate task, out of the scope of the current PR?

// Generate 'ref <> value1 AND ... AND ref <> valueN'
List<RexNode> list = sarg.rangeSet.complement().asRanges().stream().map(
range -> rexBuilder.makeCall(SqlStdOperatorTable.NOT_EQUALS, ref,
rexBuilder.makeLiteral(range.lowerEndpoint(), operandType, true, true))).toList();
orList.add(RexUtil.composeConjunction(rexBuilder, list));
} else {
RangeConverter<C> consumer = new RangeConverter<>(rexBuilder, operandType, ref);
RangeSets.forEach(sarg.rangeSet, consumer);

switch (consumer.inLiterals.size()) {
case 0:
break;
case 1:
orList.add(rexBuilder.makeCall(SqlStdOperatorTable.EQUALS, ref, consumer.inLiterals.get(0)));
break;
default:
List<RexNode> operands = new ArrayList<>(consumer.inLiterals.size() + 1);
operands.add(ref);
operands.addAll(consumer.inLiterals);
orList.add(rexBuilder.makeCall(HiveIn.INSTANCE, operands));
}
orList.addAll(consumer.nodes);
}
orList.addAll(consumer.nodes);
RexNode x = RexUtil.composeDisjunction(rexBuilder, orList);

if (sarg.nullAs == RexUnknownAs.FALSE && unknownContext != RexUnknownAs.FALSE) {
Expand Down
26 changes: 19 additions & 7 deletions ...rc/java/org/apache/hadoop/hive/ql/optimizer/calcite/stats/FilterSelectivityEstimator.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -628,14 +628,23 @@ private RexNode makeLiteral(C value) {
private double compute() {
final List<RexNode> inLiterals = new ArrayList<>();
final List<Double> rangeSelectivities = new ArrayList<>();
for (Range<C> range : sarg.rangeSet.asRanges()) {
if (!range.hasLowerBound() && !range.hasUpperBound()) {
return 1.0; // "all" range
final List<Double> searchSelectivities = new ArrayList<>();

if (sarg.isComplementedPoints()) {
// Generate 'ref <> value1 AND ... AND ref <> valueN'
List<RexNode> notEq = sarg.rangeSet.complement().asRanges().stream()
.map(range -> rexBuilder.makeCall(SqlStdOperatorTable.NOT_EQUALS, ref, makeLiteral(range.lowerEndpoint())))
.toList();
searchSelectivities.add(RexUtil.composeConjunction(rexBuilder, notEq).accept(FilterSelectivityEstimator.this));
} else {
Comment on lines +633 to +639
Copy link
Copy Markdown
Member

@zabetak zabetak Jun 4, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Does this form lead to better selectivity estimates than what we had before? If not then we don't necessarily need to change it. The purpose of this code is to compute a good selectivity for the SEARCH predicate not to transform it to something else.

Copy link
Copy Markdown
Contributor Author

@rubenada rubenada Jun 5, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Uhmmm... you're right, the original version (which uses histograms) would have a more accurate estimation than the proposed one with NOT_EQUALS (which is estimated simply with ndv-1/ndv , which can be quite off). However, it might be possible that histograms are not available in general (so the original version would default to a hadcoded selectivity), whereas the sub-optimal optimization with NOT_EQUALS uses a more generally available ndv estimated value (and this estimation, although not perfect, would be better than the hardcoded value of the original version).

Having said that, I guess we should try to aim for the better solution, and trust that statistics would be available, so I lean towards reverting the change in this file.

for (Range<C> range : sarg.rangeSet.asRanges()) {
if (!range.hasLowerBound() && !range.hasUpperBound()) {
return 1.0; // "all" range
}
processRangeSelectivity(range, rangeSelectivities, inLiterals);
}
processRangeSelectivity(range, rangeSelectivities, inLiterals);
}

final List<Double> searchSelectivities = new ArrayList<>();
if (!rangeSelectivities.isEmpty() && rangeSelectivities.stream().noneMatch(Objects::isNull)) {
// Aggregate all ranges selectivity, respecting the max value of 1
double total = Math.min(1.0, rangeSelectivities.stream().mapToDouble(Double::doubleValue).sum());
Expand All @@ -655,7 +664,8 @@ private double compute() {
List<RexNode> operands = new ArrayList<>(inLiterals.size() + 1);
operands.add(ref);
operands.addAll(inLiterals);
searchSelectivities.add(rexBuilder.makeCall(HiveIn.INSTANCE, operands).accept(FilterSelectivityEstimator.this));
searchSelectivities.add(
rexBuilder.makeCall(HiveIn.INSTANCE, operands).accept(FilterSelectivityEstimator.this));
}
}

Expand All @@ -664,7 +674,9 @@ private double compute() {
rexBuilder.makeCall(SqlStdOperatorTable.IS_NULL, ref).accept(FilterSelectivityEstimator.this));
}

return searchSelectivities.size() == 1 ? searchSelectivities.get(0) : computeDisjunctionSelectivity(searchSelectivities);
return searchSelectivities.size() == 1
? searchSelectivities.get(0)
: computeDisjunctionSelectivity(searchSelectivities);
}

private void processRangeSelectivity(Range<C> range, List<Double> rangeSelectivities, List<RexNode> inLiterals) {
Expand Down
7 changes: 7 additions & 0 deletions ql/src/java/org/apache/hadoop/hive/ql/optimizer/calcite/translator/ExprNodeConverter.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -82,6 +82,7 @@
import org.apache.hadoop.hive.ql.plan.ExprNodeColumnDesc;
import org.apache.hadoop.hive.ql.plan.ExprNodeConstantDesc;
import org.apache.hadoop.hive.ql.plan.ExprNodeDesc;
import org.apache.hadoop.hive.ql.plan.ExprNodeDescUtils;
import org.apache.hadoop.hive.ql.plan.ExprNodeFieldDesc;
import org.apache.hadoop.hive.ql.plan.ExprNodeGenericFuncDesc;
import org.apache.hadoop.hive.ql.udf.generic.GenericUDF;
Expand Down Expand Up @@ -214,6 +215,12 @@ public ExprNodeDesc visitCall(RexCall call) {
&& SqlTypeUtil.equalSansNullability(dTFactory, call.getType(),
call.operands.get(0).getType())) {
return args.get(0);
} else if (call.isA(SqlKind.AND)) {
// Make sure AND is flattened (we may have nested ANDs due to SearchTransformer conversion above)
return ExprNodeDescUtils.and(args);
} else if (call.isA(SqlKind.OR)) {
// Make sure OR is flattened (we may have nested ORs due to SearchTransformer conversion above)
return ExprNodeDescUtils.or(args);
} else {
GenericUDF hiveUdf = SqlFunctionConverter.getHiveUDF(call.getOperator(), call.getType(),
args.size());
Expand Down
Loading
Loading