diff --git ql/src/java/org/apache/hadoop/hive/ql/optimizer/calcite/HiveRexUtil.java ql/src/java/org/apache/hadoop/hive/ql/optimizer/calcite/HiveRexUtil.java index 3f6dd6a..f75303f 100644 --- ql/src/java/org/apache/hadoop/hive/ql/optimizer/calcite/HiveRexUtil.java +++ ql/src/java/org/apache/hadoop/hive/ql/optimizer/calcite/HiveRexUtil.java @@ -91,7 +91,7 @@ private static RexNode simplifyCase(RexBuilder rexBuilder, RexCall call) { assert newOperands.size() % 2 == 1; switch (newOperands.size()) { case 1: - return newOperands.get(0); + return rexBuilder.makeCast(call.getType(), newOperands.get(0)); } trueFalse: if (call.getType().getSqlTypeName() == SqlTypeName.BOOLEAN) { diff --git ql/src/java/org/apache/hadoop/hive/ql/optimizer/calcite/rules/HiveReduceExpressionsRule.java ql/src/java/org/apache/hadoop/hive/ql/optimizer/calcite/rules/HiveReduceExpressionsRule.java index 8f15ec7..1d2c4cc 100644 --- ql/src/java/org/apache/hadoop/hive/ql/optimizer/calcite/rules/HiveReduceExpressionsRule.java +++ ql/src/java/org/apache/hadoop/hive/ql/optimizer/calcite/rules/HiveReduceExpressionsRule.java @@ -582,7 +582,7 @@ protected static RexNode substitute(RexCall call, int ordinal, RexNode node) { } node = super.visitCall(call); if (node != call) { - node = RexUtil.simplify(rexBuilder, node); + node = HiveRexUtil.simplify(rexBuilder, node); } return node; } diff --git ql/src/test/queries/clientpositive/fold_case.q ql/src/test/queries/clientpositive/fold_case.q index 4732aa5..013a45f 100644 --- ql/src/test/queries/clientpositive/fold_case.q +++ ql/src/test/queries/clientpositive/fold_case.q @@ -18,3 +18,6 @@ explain select (case key when '238' then null else null end) from src where (case key when '238' then 2 = null else 3 = null end); explain select count(1) from src where (case key when '238' then null else 1=1 end); +explain +select (CASE WHEN (-2) >= 0 THEN SUBSTRING(key, 1,CAST((-2) AS INT)) ELSE NULL END) +from src; diff --git ql/src/test/results/clientpositive/fold_case.q.out ql/src/test/results/clientpositive/fold_case.q.out index 90ea0af..8e88dbf 100644 --- ql/src/test/results/clientpositive/fold_case.q.out +++ ql/src/test/results/clientpositive/fold_case.q.out @@ -475,3 +475,28 @@ STAGE PLANS: Processor Tree: ListSink +PREHOOK: query: explain +select (CASE WHEN (-2) >= 0 THEN SUBSTRING(key, 1,CAST((-2) AS INT)) ELSE NULL END) +from src +PREHOOK: type: QUERY +POSTHOOK: query: explain +select (CASE WHEN (-2) >= 0 THEN SUBSTRING(key, 1,CAST((-2) AS INT)) ELSE NULL END) +from src +POSTHOOK: type: QUERY +STAGE DEPENDENCIES: + Stage-0 is a root stage + +STAGE PLANS: + Stage: Stage-0 + Fetch Operator + limit: -1 + Processor Tree: + TableScan + alias: src + Statistics: Num rows: 500 Data size: 5312 Basic stats: COMPLETE Column stats: COMPLETE + Select Operator + expressions: null (type: void) + outputColumnNames: _col0 + Statistics: Num rows: 500 Data size: 0 Basic stats: PARTIAL Column stats: COMPLETE + ListSink + diff --git ql/src/test/results/clientpositive/spark/vectorization_short_regress.q.out ql/src/test/results/clientpositive/spark/vectorization_short_regress.q.out index f7eadaf..4cb21a1 100644 --- ql/src/test/results/clientpositive/spark/vectorization_short_regress.q.out +++ ql/src/test/results/clientpositive/spark/vectorization_short_regress.q.out @@ -2345,7 +2345,7 @@ STAGE PLANS: alias: alltypesorc Statistics: Num rows: 12288 Data size: 377237 Basic stats: COMPLETE Column stats: NONE Filter Operator - predicate: ((UDFToDouble(ctimestamp1) <> 0.0) and (((-257 <> UDFToInteger(ctinyint)) and cboolean2 is not null and cstring1 regexp '.*ss' and (-3.0 < UDFToDouble(ctimestamp1))) or (UDFToDouble(ctimestamp2) = -5.0) or ((UDFToDouble(ctimestamp1) < 0.0) and (cstring2 like '%b%')) or (cdouble = UDFToDouble(cint)) or (cboolean1 is null and (cfloat < UDFToFloat(cint))))) (type: boolean) + predicate: ((UDFToDouble(ctimestamp1) <> 0.0) and (((-257 <> UDFToInteger(ctinyint)) and cstring1 regexp '.*ss' and (-3.0 < UDFToDouble(ctimestamp1)) and cboolean2 is not null) or (UDFToDouble(ctimestamp2) = -5.0) or ((UDFToDouble(ctimestamp1) < 0.0) and (cstring2 like '%b%')) or (cdouble = UDFToDouble(cint)) or (cboolean1 is null and (cfloat < UDFToFloat(cint))))) (type: boolean) Statistics: Num rows: 12288 Data size: 377237 Basic stats: COMPLETE Column stats: NONE Select Operator expressions: cstring1 (type: string), ctimestamp1 (type: timestamp), cint (type: int), csmallint (type: smallint), ctinyint (type: tinyint), cfloat (type: float), cdouble (type: double) @@ -2678,7 +2678,7 @@ STAGE PLANS: alias: alltypesorc Statistics: Num rows: 12288 Data size: 377237 Basic stats: COMPLETE Column stats: NONE Filter Operator - predicate: ((((cdouble < UDFToDouble(csmallint)) and (cboolean2 = cboolean1) and (UDFToDouble(cbigint) <= -863.257)) or ((cint >= -257) and cstring1 is not null and (cboolean1 >= 1)) or cstring2 regexp 'b' or ((csmallint >= UDFToShort(ctinyint)) and ctimestamp2 is null)) and cboolean1 is not null) (type: boolean) + predicate: ((((cdouble < UDFToDouble(csmallint)) and (cboolean2 = cboolean1) and (UDFToDouble(cbigint) <= -863.257)) or ((cint >= -257) and (cboolean1 >= 1) and cstring1 is not null) or cstring2 regexp 'b' or ((csmallint >= UDFToShort(ctinyint)) and ctimestamp2 is null)) and cboolean1 is not null) (type: boolean) Statistics: Num rows: 10239 Data size: 314333 Basic stats: COMPLETE Column stats: NONE Select Operator expressions: cboolean1 (type: boolean), cfloat (type: float), cbigint (type: bigint), cint (type: int), cdouble (type: double), ctinyint (type: tinyint), csmallint (type: smallint) diff --git ql/src/test/results/clientpositive/tez/vectorization_short_regress.q.out ql/src/test/results/clientpositive/tez/vectorization_short_regress.q.out index 483b7bc..3b69947 100644 --- ql/src/test/results/clientpositive/tez/vectorization_short_regress.q.out +++ ql/src/test/results/clientpositive/tez/vectorization_short_regress.q.out @@ -2355,7 +2355,7 @@ STAGE PLANS: alias: alltypesorc Statistics: Num rows: 12288 Data size: 2641964 Basic stats: COMPLETE Column stats: NONE Filter Operator - predicate: ((UDFToDouble(ctimestamp1) <> 0.0) and (((-257 <> UDFToInteger(ctinyint)) and cboolean2 is not null and cstring1 regexp '.*ss' and (-3.0 < UDFToDouble(ctimestamp1))) or (UDFToDouble(ctimestamp2) = -5.0) or ((UDFToDouble(ctimestamp1) < 0.0) and (cstring2 like '%b%')) or (cdouble = UDFToDouble(cint)) or (cboolean1 is null and (cfloat < UDFToFloat(cint))))) (type: boolean) + predicate: ((UDFToDouble(ctimestamp1) <> 0.0) and (((-257 <> UDFToInteger(ctinyint)) and cstring1 regexp '.*ss' and (-3.0 < UDFToDouble(ctimestamp1)) and cboolean2 is not null) or (UDFToDouble(ctimestamp2) = -5.0) or ((UDFToDouble(ctimestamp1) < 0.0) and (cstring2 like '%b%')) or (cdouble = UDFToDouble(cint)) or (cboolean1 is null and (cfloat < UDFToFloat(cint))))) (type: boolean) Statistics: Num rows: 12288 Data size: 2641964 Basic stats: COMPLETE Column stats: NONE Select Operator expressions: cstring1 (type: string), ctimestamp1 (type: timestamp), cint (type: int), csmallint (type: smallint), ctinyint (type: tinyint), cfloat (type: float), cdouble (type: double) @@ -2689,7 +2689,7 @@ STAGE PLANS: alias: alltypesorc Statistics: Num rows: 12288 Data size: 2641964 Basic stats: COMPLETE Column stats: NONE Filter Operator - predicate: ((((cdouble < UDFToDouble(csmallint)) and (cboolean2 = cboolean1) and (UDFToDouble(cbigint) <= -863.257)) or ((cint >= -257) and cstring1 is not null and (cboolean1 >= 1)) or cstring2 regexp 'b' or ((csmallint >= UDFToShort(ctinyint)) and ctimestamp2 is null)) and cboolean1 is not null) (type: boolean) + predicate: ((((cdouble < UDFToDouble(csmallint)) and (cboolean2 = cboolean1) and (UDFToDouble(cbigint) <= -863.257)) or ((cint >= -257) and (cboolean1 >= 1) and cstring1 is not null) or cstring2 regexp 'b' or ((csmallint >= UDFToShort(ctinyint)) and ctimestamp2 is null)) and cboolean1 is not null) (type: boolean) Statistics: Num rows: 10239 Data size: 2201421 Basic stats: COMPLETE Column stats: NONE Select Operator expressions: cboolean1 (type: boolean), cfloat (type: float), cbigint (type: bigint), cint (type: int), cdouble (type: double), ctinyint (type: tinyint), csmallint (type: smallint) diff --git ql/src/test/results/clientpositive/vectorization_short_regress.q.out ql/src/test/results/clientpositive/vectorization_short_regress.q.out index 155e4d3..4b8ff69 100644 --- ql/src/test/results/clientpositive/vectorization_short_regress.q.out +++ ql/src/test/results/clientpositive/vectorization_short_regress.q.out @@ -2294,7 +2294,7 @@ STAGE PLANS: alias: alltypesorc Statistics: Num rows: 12288 Data size: 2641964 Basic stats: COMPLETE Column stats: NONE Filter Operator - predicate: ((UDFToDouble(ctimestamp1) <> 0.0) and (((-257 <> UDFToInteger(ctinyint)) and cboolean2 is not null and cstring1 regexp '.*ss' and (-3.0 < UDFToDouble(ctimestamp1))) or (UDFToDouble(ctimestamp2) = -5.0) or ((UDFToDouble(ctimestamp1) < 0.0) and (cstring2 like '%b%')) or (cdouble = UDFToDouble(cint)) or (cboolean1 is null and (cfloat < UDFToFloat(cint))))) (type: boolean) + predicate: ((UDFToDouble(ctimestamp1) <> 0.0) and (((-257 <> UDFToInteger(ctinyint)) and cstring1 regexp '.*ss' and (-3.0 < UDFToDouble(ctimestamp1)) and cboolean2 is not null) or (UDFToDouble(ctimestamp2) = -5.0) or ((UDFToDouble(ctimestamp1) < 0.0) and (cstring2 like '%b%')) or (cdouble = UDFToDouble(cint)) or (cboolean1 is null and (cfloat < UDFToFloat(cint))))) (type: boolean) Statistics: Num rows: 12288 Data size: 2641964 Basic stats: COMPLETE Column stats: NONE Select Operator expressions: cstring1 (type: string), ctimestamp1 (type: timestamp), cint (type: int), csmallint (type: smallint), ctinyint (type: tinyint), cfloat (type: float), cdouble (type: double) @@ -2630,7 +2630,7 @@ STAGE PLANS: alias: alltypesorc Statistics: Num rows: 12288 Data size: 2641964 Basic stats: COMPLETE Column stats: NONE Filter Operator - predicate: ((((cdouble < UDFToDouble(csmallint)) and (cboolean2 = cboolean1) and (UDFToDouble(cbigint) <= -863.257)) or ((cint >= -257) and cstring1 is not null and (cboolean1 >= 1)) or cstring2 regexp 'b' or ((csmallint >= UDFToShort(ctinyint)) and ctimestamp2 is null)) and cboolean1 is not null) (type: boolean) + predicate: ((((cdouble < UDFToDouble(csmallint)) and (cboolean2 = cboolean1) and (UDFToDouble(cbigint) <= -863.257)) or ((cint >= -257) and (cboolean1 >= 1) and cstring1 is not null) or cstring2 regexp 'b' or ((csmallint >= UDFToShort(ctinyint)) and ctimestamp2 is null)) and cboolean1 is not null) (type: boolean) Statistics: Num rows: 10239 Data size: 2201421 Basic stats: COMPLETE Column stats: NONE Select Operator expressions: cboolean1 (type: boolean), cfloat (type: float), cbigint (type: bigint), cint (type: int), cdouble (type: double), ctinyint (type: tinyint), csmallint (type: smallint)