Merge #40528

40528: exec: omit some of the operators from EXPLAIN (VEC) output r=yuzefovich a=yuzefovich

This commits adds NonExplainable marker interface which operators
can choose to implement so that they are omitted from the output
of EXPLAIN (VEC). It is intended that non-core operators will be
marked. Additionally, now a VERBOSE explain option is supported
which overrides the omitting behavior.

Release note: None

Co-authored-by: Yahor Yuzefovich <yahor@cockroachlabs.com>

pkg/sql/distsqlrun/columnarizer.go

@@ -27,6 +27,7 @@ import (
 type columnarizer struct {
 	ProcessorBase
 	exec.ZeroInputNode
+	exec.NonExplainable
 
 	input RowSource
 	da    sqlbase.DatumAlloc

pkg/sql/distsqlrun/materializer.go

@@ -26,6 +26,7 @@ import (
 // materializer converts an exec.Operator input into a RowSource.
 type materializer struct {
 	ProcessorBase
+	exec.NonExplainable
 
 	input exec.Operator
 

pkg/sql/exec/bool_vec_to_sel.go

@@ -20,6 +20,7 @@ import (
 // an index to the selection for each true value in the boolean column.
 type boolVecToSelOp struct {
 	OneInputNode
+	NonExplainable
 
 	// outputCol is the boolean output column. It should be shared by other
 	// operators that write to it.
@@ -111,6 +112,7 @@ func NewBoolVecToSelOp(input Operator, colIdx int) Operator {
 // an operator that can see the inside of its input batch for NewBoolVecToSelOp.
 type selBoolOp struct {
 	OneInputNode
+	NonExplainable
 	boolVecToSelOp *boolVecToSelOp
 	colIdx         int
 }

pkg/sql/exec/buffer.go

@@ -21,6 +21,7 @@ import (
 // and makes it available to be read multiple times by downstream consumers.
 type bufferOp struct {
 	OneInputNode
+	NonExplainable
 
 	// read is true if someone has read the current batch already.
 	read     bool

pkg/sql/exec/cancel_checker.go

@@ -22,6 +22,7 @@ import (
 // occurred. The check happens on every batch.
 type CancelChecker struct {
 	OneInputNode
+	NonExplainable
 
 	// Number of times check() has been called since last context cancellation
 	// check.

pkg/sql/exec/coalescer.go

@@ -21,6 +21,7 @@ import (
 // to return full batches of coldata.BatchSize.
 type coalescerOp struct {
 	OneInputNode
+	NonExplainable
 
 	inputTypes []coltypes.T
 

pkg/sql/exec/deselector.go

@@ -23,6 +23,7 @@ import (
 // selection vector, it is a noop.
 type deselectorOp struct {
 	OneInputNode
+	NonExplainable
 	inputTypes []coltypes.T
 
 	output coldata.Batch

pkg/sql/exec/fn_op.go

@@ -20,6 +20,7 @@ import (
 // once per input batch, passing the input batch unmodified along.
 type fnOp struct {
 	OneInputNode
+	NonExplainable
 
 	fn func()
 }

pkg/sql/exec/mergejoiner.go

@@ -157,6 +157,7 @@ type mergeJoinInput struct {
 // the next batch.
 type feedOperator struct {
 	ZeroInputNode
+	NonExplainable
 	batch coldata.Batch
 }
 

pkg/sql/exec/operator.go

@@ -47,6 +47,14 @@ type OpNode interface {
 	Child(nth int) OpNode
 }
 
+// NonExplainable is a marker interface which identifies an Operator that
+// should be omitted from the output of EXPLAIN (VEC). Note that VERBOSE
+// explain option will override the omitting behavior.
+type NonExplainable interface {
+	// nonExplainableMarker is just a marker method. It should never be called.
+	nonExplainableMarker()
+}
+
 // NewOneInputNode returns an OpNode with a single Operator input.
 func NewOneInputNode(input Operator) OneInputNode {
 	return OneInputNode{input: input}
@@ -141,6 +149,7 @@ type resettableOperator interface {
 
 type noopOperator struct {
 	OneInputNode
+	NonExplainable
 }
 
 var _ Operator = &noopOperator{}
@@ -166,6 +175,7 @@ func (n *noopOperator) reset() {
 
 type zeroOperator struct {
 	OneInputNode
+	NonExplainable
 }
 
 var _ Operator = &zeroOperator{}
@@ -188,6 +198,7 @@ func (s *zeroOperator) Next(ctx context.Context) coldata.Batch {
 
 type singleTupleNoInputOperator struct {
 	ZeroInputNode
+	NonExplainable
 	batch  coldata.Batch
 	nexted bool
 }

pkg/sql/exec/simple_project.go

@@ -22,6 +22,7 @@ import (
 // columns that aren't needed by later operators.
 type simpleProjectOp struct {
 	OneInputNode
+	NonExplainable
 
 	batch *projectingBatch
 }

pkg/sql/exec/sort.go

@@ -81,6 +81,7 @@ type spooler interface {
 // by the general sorter over the whole input.
 type allSpooler struct {
 	OneInputNode
+	NonExplainable
 
 	// inputTypes contains the types of all of the columns from the input.
 	inputTypes []coltypes.T

pkg/sql/exec/sort_chunks.go

@@ -151,6 +151,7 @@ const (
 // buffer when appropriate.
 type chunker struct {
 	OneInputNode
+	NonExplainable
 	// inputTypes contains the types of all of the columns from input.
 	inputTypes []coltypes.T
 	// inputDone indicates whether input has been fully consumed.

pkg/sql/exec/stats.go

@@ -26,6 +26,7 @@ import (
 // StopWatch.
 type VectorizedStatsCollector struct {
 	Operator
+	NonExplainable
 	execpb.VectorizedStats
 
 	// inputWatch is a single stop watch that is shared with all the input

pkg/sql/explain_vec.go

@@ -30,7 +30,8 @@ import (
 type explainVecNode struct {
 	optColumnsSlot
 
-	plan planNode
+	options *tree.ExplainOptions
+	plan    planNode
 
 	stmtType tree.StatementType
 
@@ -120,27 +121,41 @@ func (n *explainVecNode) startExec(params runParams) error {
 	sort.Slice(sortedFlows, func(i, j int) bool { return sortedFlows[i].nodeID < sortedFlows[j].nodeID })
 	tp := treeprinter.New()
 	root := tp.Child("")
+	verbose := n.options.Flags.Contains(tree.ExplainFlagVerbose)
 	for _, flow := range sortedFlows {
 		node := root.Childf("Node %d", flow.nodeID)
 		opChains, err := distsqlrun.SupportsVectorized(params.ctx, flowCtx, flow.flow.Processors)
 		if err != nil {
 			return err
 		}
 		for _, op := range opChains {
-			formatOpChain(op, node)
+			formatOpChain(op, node, verbose)
 		}
 	}
 	n.run.lines = tp.FormattedRows()
 	return nil
 }
 
-func formatOpChain(operator exec.OpNode, node treeprinter.Node) {
-	doFormatOpChain(operator, node.Child(reflect.TypeOf(operator).String()))
+func shouldOutput(operator exec.OpNode, verbose bool) bool {
+	_, nonExplainable := operator.(exec.NonExplainable)
+	return !nonExplainable || verbose
 }
-func doFormatOpChain(operator exec.OpNode, node treeprinter.Node) {
+
+func formatOpChain(operator exec.OpNode, node treeprinter.Node, verbose bool) {
+	if shouldOutput(operator, verbose) {
+		doFormatOpChain(operator, node.Child(reflect.TypeOf(operator).String()), verbose)
+	} else {
+		doFormatOpChain(operator, node, verbose)
+	}
+}
+func doFormatOpChain(operator exec.OpNode, node treeprinter.Node, verbose bool) {
 	for i := 0; i < operator.ChildCount(); i++ {
 		child := operator.Child(i)
-		doFormatOpChain(child, node.Child(reflect.TypeOf(child).String()))
+		if shouldOutput(child, verbose) {
+			doFormatOpChain(child, node.Child(reflect.TypeOf(child).String()), verbose)
+		} else {
+			doFormatOpChain(child, node, verbose)
+		}
 	}
 }
 

pkg/sql/logictest/testdata/logic_test/dist_vectorize

@@ -87,7 +87,7 @@ statement ok
 RESET optimizer
 
 query T
-EXPLAIN (VEC) SELECT count(*) FROM kv
+EXPLAIN (VEC, VERBOSE) SELECT count(*) FROM kv
 ----
  │
  ├── Node 1
@@ -134,7 +134,7 @@ EXPLAIN (VEC) SELECT count(*) FROM kv
                                └── *distsqlrun.colBatchScan
 
 query T
-EXPLAIN (VEC) SELECT count(*) FROM kv NATURAL INNER HASH JOIN kv kv2
+EXPLAIN (VEC, VERBOSE) SELECT count(*) FROM kv NATURAL INNER HASH JOIN kv kv2
 ----
  │
  ├── Node 1