zabetak commented on code in PR #4442:
URL: https://github.com/apache/hive/pull/4442#discussion_r1335607615
##########
ql/src/java/org/apache/hadoop/hive/ql/ppd/OpProcFactory.java:
##########
@@ -385,6 +385,23 @@ public Object process(Node nd, Stack<Node> stack,
NodeProcessorCtx procCtx,
}
+ public static class LateralViewJoinerPPD extends JoinerPPD implements
SemanticNodeProcessor {
+ @Override
+ public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx procCtx,
+ Object... nodeOutputs) throws SemanticException {
+ Object o = super.process(nd, stack, procCtx, nodeOutputs);
+ Operator<?> operator = (Operator<?>) nd;
+ OpWalkerInfo owi = (OpWalkerInfo) procCtx;
+ if (HiveConf.getBoolVar(owi.getParseContext().getConf(),
+ HiveConf.ConfVars.HIVEPPDREMOVEDUPLICATEFILTERS)) {
+ // remove all the candidate filter operators
+ // when we get to the TS
Review Comment:
I don't think we are in the TS op here? I think we are in the
`LateralViewJoinOperator` (LVJ).
##########
ql/src/java/org/apache/hadoop/hive/ql/ppd/OpProcFactory.java:
##########
@@ -385,6 +385,23 @@ public Object process(Node nd, Stack<Node> stack,
NodeProcessorCtx procCtx,
}
+ public static class LateralViewJoinerPPD extends JoinerPPD implements
SemanticNodeProcessor {
+ @Override
+ public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx procCtx,
+ Object... nodeOutputs) throws SemanticException {
+ Object o = super.process(nd, stack, procCtx, nodeOutputs);
+ Operator<?> operator = (Operator<?>) nd;
Review Comment:
unused variable/dead code?
##########
ql/src/test/results/clientpositive/llap/lateral_view_cbo.q.out:
##########
@@ -1,34 +1,113 @@
-PREHOOK: query: EXPLAIN CBO SELECT myTable.myCol FROM src
+PREHOOK: query: CREATE TABLE simple_table (col string, array_col array<string>)
+PREHOOK: type: CREATETABLE
+PREHOOK: Output: database:default
+PREHOOK: Output: default@simple_table
+POSTHOOK: query: CREATE TABLE simple_table (col string, array_col
array<string>)
+POSTHOOK: type: CREATETABLE
+POSTHOOK: Output: database:default
+POSTHOOK: Output: default@simple_table
+PREHOOK: query: EXPLAIN CBO SELECT myTable.myCol FROM simple_table
LATERAL VIEW explode(array(1,2,3)) myTable AS myCol
PREHOOK: type: QUERY
-PREHOOK: Input: default@src
+PREHOOK: Input: default@simple_table
#### A masked pattern was here ####
-POSTHOOK: query: EXPLAIN CBO SELECT myTable.myCol FROM src
+POSTHOOK: query: EXPLAIN CBO SELECT myTable.myCol FROM simple_table
LATERAL VIEW explode(array(1,2,3)) myTable AS myCol
POSTHOOK: type: QUERY
-POSTHOOK: Input: default@src
+POSTHOOK: Input: default@simple_table
#### A masked pattern was here ####
CBO PLAN:
HiveProject(mycol=[$6])
- HiveTableFunctionScan(invocation=[LATERAL(explode(ARRAY(1, 2, 3)), $0, $1,
$2, $3, $4, $5)], rowType=[RecordType(VARCHAR(2147483647) key,
VARCHAR(2147483647) value, BIGINT BLOCK__OFFSET__INSIDE__FILE,
VARCHAR(2147483647) INPUT__FILE__NAME, RecordType(BIGINT writeid, INTEGER
bucketid, BIGINT rowid) ROW__ID, BOOLEAN ROW__IS__DELETED, INTEGER
mytable.mycol)])
- HiveTableScan(table=[[default, src]], table:alias=[src])
+ HiveTableFunctionScan(invocation=[LATERAL(explode(ARRAY(1, 2, 3)), $0, $1,
$2, $3, $4, $5)], rowType=[RecordType(VARCHAR(2147483647) col,
VARCHAR(2147483647) ARRAY array_col, BIGINT BLOCK__OFFSET__INSIDE__FILE,
VARCHAR(2147483647) INPUT__FILE__NAME, RecordType(BIGINT writeid, INTEGER
bucketid, BIGINT rowid) ROW__ID, BOOLEAN ROW__IS__DELETED, INTEGER
mytable.mycol)])
+ HiveTableScan(table=[[default, simple_table]], table:alias=[simple_table])
-PREHOOK: query: EXPLAIN CBO SELECT myTable.myCol, myTable2.myCol2 FROM src
+PREHOOK: query: EXPLAIN CBO SELECT myTable.myCol, myTable2.myCol2 FROM
simple_table
LATERAL VIEW explode(array(1,2,3)) myTable AS myCol
LATERAL VIEW explode(array(1,2,3)) myTable2 AS myCol2
PREHOOK: type: QUERY
-PREHOOK: Input: default@src
+PREHOOK: Input: default@simple_table
#### A masked pattern was here ####
-POSTHOOK: query: EXPLAIN CBO SELECT myTable.myCol, myTable2.myCol2 FROM src
+POSTHOOK: query: EXPLAIN CBO SELECT myTable.myCol, myTable2.myCol2 FROM
simple_table
LATERAL VIEW explode(array(1,2,3)) myTable AS myCol
LATERAL VIEW explode(array(1,2,3)) myTable2 AS myCol2
POSTHOOK: type: QUERY
-POSTHOOK: Input: default@src
+POSTHOOK: Input: default@simple_table
#### A masked pattern was here ####
CBO PLAN:
HiveProject(mycol=[$6], mycol2=[$7])
- HiveTableFunctionScan(invocation=[LATERAL(explode(ARRAY(1, 2, 3)), $0, $1,
$2, $3, $4, $5, $6)], rowType=[RecordType(VARCHAR(2147483647) key,
VARCHAR(2147483647) value, BIGINT BLOCK__OFFSET__INSIDE__FILE,
VARCHAR(2147483647) INPUT__FILE__NAME, RecordType(BIGINT writeid, INTEGER
bucketid, BIGINT rowid) ROW__ID, BOOLEAN ROW__IS__DELETED, INTEGER
mytable.mycol, INTEGER mytable2.mycol2)])
- HiveProject(key=[$0], value=[$1], BLOCK__OFFSET__INSIDE__FILE=[$2],
INPUT__FILE__NAME=[$3], ROW__ID=[$4], ROW__IS__DELETED=[$5], mytable.mycol=[$6])
- HiveTableFunctionScan(invocation=[LATERAL(explode(ARRAY(1, 2, 3)), $0,
$1, $2, $3, $4, $5)], rowType=[RecordType(VARCHAR(2147483647) key,
VARCHAR(2147483647) value, BIGINT BLOCK__OFFSET__INSIDE__FILE,
VARCHAR(2147483647) INPUT__FILE__NAME, RecordType(BIGINT writeid, INTEGER
bucketid, BIGINT rowid) ROW__ID, BOOLEAN ROW__IS__DELETED, INTEGER
mytable.mycol)])
- HiveTableScan(table=[[default, src]], table:alias=[src])
+ HiveTableFunctionScan(invocation=[LATERAL(explode(ARRAY(1, 2, 3)), $0, $1,
$2, $3, $4, $5, $6)], rowType=[RecordType(VARCHAR(2147483647) col,
VARCHAR(2147483647) ARRAY array_col, BIGINT BLOCK__OFFSET__INSIDE__FILE,
VARCHAR(2147483647) INPUT__FILE__NAME, RecordType(BIGINT writeid, INTEGER
bucketid, BIGINT rowid) ROW__ID, BOOLEAN ROW__IS__DELETED, INTEGER
mytable.mycol, INTEGER mytable2.mycol2)])
+ HiveProject(col=[$0], array_col=[$1], BLOCK__OFFSET__INSIDE__FILE=[$2],
INPUT__FILE__NAME=[$3], ROW__ID=[$4], ROW__IS__DELETED=[$5], mytable.mycol=[$6])
+ HiveTableFunctionScan(invocation=[LATERAL(explode(ARRAY(1, 2, 3)), $0,
$1, $2, $3, $4, $5)], rowType=[RecordType(VARCHAR(2147483647) col,
VARCHAR(2147483647) ARRAY array_col, BIGINT BLOCK__OFFSET__INSIDE__FILE,
VARCHAR(2147483647) INPUT__FILE__NAME, RecordType(BIGINT writeid, INTEGER
bucketid, BIGINT rowid) ROW__ID, BOOLEAN ROW__IS__DELETED, INTEGER
mytable.mycol)])
+ HiveTableScan(table=[[default, simple_table]],
table:alias=[simple_table])
+
+PREHOOK: query: EXPLAIN CBO SELECT tf.col1, tf.col2, tf.col3
+FROM simple_table
+ LATERAL TABLE(VALUES('A', 10, simple_table.col),('B', 20, simple_table.col))
AS tf(col1, col2, col3)
+PREHOOK: type: QUERY
+PREHOOK: Input: default@simple_table
+#### A masked pattern was here ####
+POSTHOOK: query: EXPLAIN CBO SELECT tf.col1, tf.col2, tf.col3
+FROM simple_table
+ LATERAL TABLE(VALUES('A', 10, simple_table.col),('B', 20, simple_table.col))
AS tf(col1, col2, col3)
+POSTHOOK: type: QUERY
+POSTHOOK: Input: default@simple_table
+#### A masked pattern was here ####
+CBO PLAN:
+HiveProject(col1=[$6], col2=[$7], col3=[$8])
+ HiveTableFunctionScan(invocation=[inline(ARRAY(ROW($0, $1, $2, $3, $4, $5,
_UTF-16LE'A':VARCHAR(2147483647) CHARACTER SET "UTF-16LE", 10, $0), ROW($0, $1,
$2, $3, $4, $5, _UTF-16LE'B':VARCHAR(2147483647) CHARACTER SET "UTF-16LE", 20,
$0)))], rowType=[RecordType(VARCHAR(2147483647) col, VARCHAR(2147483647) ARRAY
array_col, BIGINT BLOCK__OFFSET__INSIDE__FILE, VARCHAR(2147483647)
INPUT__FILE__NAME, RecordType(BIGINT writeid, INTEGER bucketid, BIGINT rowid)
ROW__ID, BOOLEAN ROW__IS__DELETED, VARCHAR(2147483647) tf.col1, INTEGER
tf.col2, VARCHAR(2147483647) tf.col3)])
Review Comment:
Do we need to keep `$0, $1, $2, $3, $4, $5,` in this case. They seem
redundant should we pass them as an input to the table function?
Does this affect runtime plan/performance? Are we going to copy all these
columns and pass them as input to the function?
##########
ql/src/test/results/clientpositive/llap/lateral_view_ppd.q.out:
##########
@@ -316,29 +332,36 @@ STAGE PLANS:
filterExpr: (key = '0') (type: boolean)
Filter Operator
predicate: (key = '0') (type: boolean)
- Lateral View Forward
- Select Operator
- expressions: value (type: string)
- outputColumnNames: value
- Lateral View Join Operator
- outputColumnNames: _col1, _col6
- Select Operator
- expressions: _col1 (type: string), _col6 (type: int)
- outputColumnNames: _col0, _col1
- ListSink
- Select Operator
- expressions: array(1,2,3) (type: array<int>)
- outputColumnNames: _col0
- UDTF Operator
- function name: explode
- Filter Operator
- predicate: (col > 1) (type: boolean)
Review Comment:
Can't find the other comment with the extra details. Can you please share
the link?
##########
ql/src/java/org/apache/hadoop/hive/ql/ppd/OpProcFactory.java:
##########
@@ -385,6 +385,23 @@ public Object process(Node nd, Stack<Node> stack,
NodeProcessorCtx procCtx,
}
+ public static class LateralViewJoinerPPD extends JoinerPPD implements
SemanticNodeProcessor {
+ @Override
+ public Object process(Node nd, Stack<Node> stack, NodeProcessorCtx procCtx,
+ Object... nodeOutputs) throws SemanticException {
+ Object o = super.process(nd, stack, procCtx, nodeOutputs);
+ Operator<?> operator = (Operator<?>) nd;
+ OpWalkerInfo owi = (OpWalkerInfo) procCtx;
+ if (HiveConf.getBoolVar(owi.getParseContext().getConf(),
+ HiveConf.ConfVars.HIVEPPDREMOVEDUPLICATEFILTERS)) {
+ // remove all the candidate filter operators
+ // when we get to the TS
Review Comment:
Why do we need to do this here and not as part of `TableScanPPD` as it is
done in every other case? Why is it specific to LVJ?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: gitbox-unsubscr...@hive.apache.org
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: gitbox-unsubscr...@hive.apache.org
For additional commands, e-mail: gitbox-h...@hive.apache.org
