logical: Add the EXCLUDE COLUMNS option to SHOW EXPERIMENTAL_FINGERPRINT

Currently tables used for logical data replication can't be fingerprinted
as they contain a column crdb_replication_origin_timestamp which will vary
based on the internal MVCC timestamp. Adding an option to SHOW
EXPERIMENTAL FINGERPRINT will allow LDR tables to be fingerprinted by excluding
the replication column. A follow up will use this to verify tables in roachtests

Release note: none
Fixes: cockroachdb#129497

docs/generated/sql/bnf/fingerprint_options.bnf

@@ -1,2 +1,3 @@
 fingerprint_options ::=
 	'START' 'TIMESTAMP' '=' d_expr
+	| 'EXCLUDE' 'COLUMNS' '=' string_or_placeholder_opt_list

pkg/sql/logictest/testdata/logic_test/show_fingerprints

@@ -18,6 +18,28 @@ SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE t
 t_pkey   -7903300865687235210
 t_b_idx  -5073888452016928166
 
+# Test excluded columns
+query TT rowsort
+SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE t WITH EXCLUDE COLUMNS = ('c')
+----
+t_pkey   -2938394162542358272
+t_b_idx  -5073888452016928166
+
+# Test multiple excluded columns
+query TT rowsort
+SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE t WITH EXCLUDE COLUMNS = ('a', 'b')
+----
+t_pkey   -3539648437866042702
+t_b_idx  590700560494856532
+
+# START TIMESTAMP is only for VIRTUAL CLUSTERS
+query error pgcode 22023 cannot use the START TIMESTAMP option when fingerprinting a table.
+SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE t WITH START TIMESTAMP = '132412341234.000000'
+
+# EXCLUDE COLUMNS is only for tables
+query error pgcode 22023 cannot use the EXCLUDE COLUMNS option when fingerprinting a tenant.
+SHOW EXPERIMENTAL_FINGERPRINTS FROM VIRTUAL CLUSTER t WITH EXCLUDE COLUMNS = ('a', 'b')
+
 # Test a partial index. We expect this index to have the same fingerprint
 # as t_b_idx since the predicate covers all values.
 statement ok

pkg/sql/parser/sql.y

@@ -9634,10 +9634,10 @@ show_locality_stmt:
   }
 
 show_fingerprints_stmt:
-  SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE table_name
+  SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE table_name opt_with_show_fingerprints_options
   {
     /* SKIP DOC */
-    $$.val = &tree.ShowFingerprints{Table: $5.unresolvedObjectName()}
+    $$.val = &tree.ShowFingerprints{Table: $5.unresolvedObjectName(), Options: *$6.showFingerprintOptions()}
   }
 | SHOW EXPERIMENTAL_FINGERPRINTS FROM virtual_cluster virtual_cluster_spec opt_with_show_fingerprints_options
   {
@@ -9678,6 +9678,11 @@ fingerprint_options:
   {
     $$.val = &tree.ShowFingerprintOptions{StartTimestamp: $4.expr()}
   }
+| EXCLUDE COLUMNS '=' string_or_placeholder_opt_list
+  {
+    $$.val = &tree.ShowFingerprintOptions{ExcludedUserColumns: $4.stringOrPlaceholderOptList()}
+  }
+
 
 
 show_full_scans_stmt:

pkg/sql/parser/testdata/show

@@ -1323,6 +1323,14 @@ SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE d.t -- fully parenthesized
 SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE d.t -- literals removed
 SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE _._ -- identifiers removed
 
+parse
+SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE d.t WITH START TIMESTAMP = '132412341234.000000', EXCLUDE COLUMNS = ('crdb_original_replication_timestamp', 'other_column')
+----
+SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE d.t WITH OPTIONS (START TIMESTAMP = '132412341234.000000', EXCLUDE COLUMNS = ('crdb_original_replication_timestamp', 'other_column')) -- normalized!
+SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE d.t WITH OPTIONS (START TIMESTAMP = ('132412341234.000000'), EXCLUDE COLUMNS = (('crdb_original_replication_timestamp'), ('other_column'))) -- fully parenthesized
+SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE d.t WITH OPTIONS (START TIMESTAMP = '_', EXCLUDE COLUMNS = ('_', '_')) -- literals removed
+SHOW EXPERIMENTAL_FINGERPRINTS FROM TABLE _._ WITH OPTIONS (START TIMESTAMP = '132412341234.000000', EXCLUDE COLUMNS = ('crdb_original_replication_timestamp', 'other_column')) -- identifiers removed
+
 parse
 SHOW EXPERIMENTAL_FINGERPRINTS FROM VIRTUAL CLUSTER t
 ----

pkg/sql/sem/tree/show.go

@@ -1160,11 +1160,21 @@ func (node *ShowFingerprints) Format(ctx *FmtCtx) {
 // ShowFingerprintOptions describes options for the SHOW EXPERIMENTAL_FINGERPINT
 // execution.
 type ShowFingerprintOptions struct {
-	StartTimestamp Expr
+	StartTimestamp      Expr
+	ExcludedUserColumns StringOrPlaceholderOptList
 }
 
 func (s *ShowFingerprintOptions) Format(ctx *FmtCtx) {
+	var addSep bool
+	maybeAddSep := func() {
+		if addSep {
+			ctx.WriteString(", ")
+		}
+		addSep = true
+	}
+
 	if s.StartTimestamp != nil {
+		maybeAddSep()
 		ctx.WriteString("START TIMESTAMP = ")
 		_, canOmitParentheses := s.StartTimestamp.(alreadyDelimitedAsSyntacticDExpr)
 		if !canOmitParentheses {
@@ -1175,6 +1185,11 @@ func (s *ShowFingerprintOptions) Format(ctx *FmtCtx) {
 			ctx.WriteByte(')')
 		}
 	}
+	if s.ExcludedUserColumns != nil {
+		maybeAddSep()
+		ctx.WriteString("EXCLUDE COLUMNS = ")
+		s.ExcludedUserColumns.Format(ctx)
+	}
 }
 
 // CombineWith merges other TenantReplicationOptions into this struct.
@@ -1188,13 +1203,18 @@ func (s *ShowFingerprintOptions) CombineWith(other *ShowFingerprintOptions) erro
 		s.StartTimestamp = other.StartTimestamp
 	}
 
+	var err error
+	s.ExcludedUserColumns, err = combineStringOrPlaceholderOptList(s.ExcludedUserColumns, other.ExcludedUserColumns, "excluded_user_columns")
+	if err != nil {
+		return err
+	}
 	return nil
 }
 
 // IsDefault returns true if this backup options struct has default value.
 func (s ShowFingerprintOptions) IsDefault() bool {
 	options := ShowFingerprintOptions{}
-	return s.StartTimestamp == options.StartTimestamp
+	return s.StartTimestamp == options.StartTimestamp && cmp.Equal(s.ExcludedUserColumns, options.ExcludedUserColumns)
 }
 
 var _ NodeFormatter = &ShowFingerprintOptions{}

pkg/sql/show_fingerprints.go

@@ -23,6 +23,7 @@ import (
 	"github.com/cockroachdb/cockroach/pkg/sql/catalog"
 	"github.com/cockroachdb/cockroach/pkg/sql/catalog/colinfo"
 	"github.com/cockroachdb/cockroach/pkg/sql/clusterunique"
+	"github.com/cockroachdb/cockroach/pkg/sql/exprutil"
 	"github.com/cockroachdb/cockroach/pkg/sql/isql"
 	"github.com/cockroachdb/cockroach/pkg/sql/pgwire/pgcode"
 	"github.com/cockroachdb/cockroach/pkg/sql/pgwire/pgerror"
@@ -71,8 +72,27 @@ type showFingerprintsNode struct {
 func (p *planner) ShowFingerprints(
 	ctx context.Context, n *tree.ShowFingerprints,
 ) (planNode, error) {
+
+	op := "SHOW EXPERIMENTAL_FINGERPRINTS"
+	evalOptions, err := evalShowFingerprintOptions(ctx, n.Options, p.EvalContext(), p.SemaCtx(),
+		op, p.ExprEvaluator(op))
+	if err != nil {
+		return nil, err
+	}
+
 	if n.TenantSpec != nil {
-		return p.planShowTenantFingerprint(ctx, n.TenantSpec, n.Options)
+		// Tenant fingerprints use the KV fingerprint method and can't exclude columns this way
+		if evalOptions.excludedUserColumns != nil {
+			err = pgerror.New(pgcode.InvalidParameterValue, "cannot use the EXCLUDE COLUMNS option when fingerprinting a tenant.")
+			return nil, err
+		}
+		return p.planShowTenantFingerprint(ctx, n.TenantSpec, evalOptions)
+	}
+
+	// Only allow this for virtual clusters as it uses the KV fingerprint method instead of SQL
+	if !evalOptions.startTimestamp.IsEmpty() {
+		err = pgerror.New(pgcode.InvalidParameterValue, "cannot use the START TIMESTAMP option when fingerprinting a table.")
+		return nil, err
 	}
 
 	// We avoid the cache so that we can observe the fingerprints without
@@ -91,19 +111,22 @@ func (p *planner) ShowFingerprints(
 		columns:   colinfo.ShowFingerprintsColumns,
 		tableDesc: tableDesc,
 		indexes:   tableDesc.ActiveIndexes(),
+		options:   evalOptions,
 	}, nil
 }
 
 type resolvedShowTenantFingerprintOptions struct {
-	startTimestamp hlc.Timestamp
+	startTimestamp      hlc.Timestamp
+	excludedUserColumns []string
 }
 
-func evalShowTenantFingerprintOptions(
+func evalShowFingerprintOptions(
 	ctx context.Context,
 	options tree.ShowFingerprintOptions,
 	evalCtx *eval.Context,
 	semaCtx *tree.SemaContext,
 	op string,
+	eval exprutil.Evaluator,
 ) (*resolvedShowTenantFingerprintOptions, error) {
 	r := &resolvedShowTenantFingerprintOptions{}
 	if options.StartTimestamp != nil {
@@ -114,11 +137,21 @@ func evalShowTenantFingerprintOptions(
 		r.startTimestamp = ts
 	}
 
+	if options.ExcludedUserColumns != nil {
+		cols, err := eval.StringArray(
+			ctx, tree.Exprs(options.ExcludedUserColumns))
+
+		if err != nil {
+			return nil, err
+		}
+		r.excludedUserColumns = cols
+	}
+
 	return r, nil
 }
 
 func (p *planner) planShowTenantFingerprint(
-	ctx context.Context, ts *tree.TenantSpec, options tree.ShowFingerprintOptions,
+	ctx context.Context, ts *tree.TenantSpec, evalOptions *resolvedShowTenantFingerprintOptions,
 ) (planNode, error) {
 	if err := CanManageTenant(ctx, p); err != nil {
 		return nil, err
@@ -133,12 +166,6 @@ func (p *planner) planShowTenantFingerprint(
 		return nil, err
 	}
 
-	evalOptions, err := evalShowTenantFingerprintOptions(ctx, options, p.EvalContext(), p.SemaCtx(),
-		"SHOW EXPERIMENTAL_FINGERPRINTS FROM VIRTUAL CLUSTER")
-	if err != nil {
-		return nil, err
-	}
-
 	return &showFingerprintsNode{
 		columns:    colinfo.ShowTenantFingerprintsColumns,
 		tenantSpec: tspec,
@@ -290,7 +317,11 @@ func (n *showFingerprintsNode) Next(params runParams) (bool, error) {
 		return false, nil
 	}
 	index := n.indexes[n.run.rowIdx]
-	sql, err := BuildFingerprintQueryForIndex(n.tableDesc, index, []string{})
+	excludedColumns := []string{}
+	if n.options != nil && len(n.options.excludedUserColumns) > 0 {
+		excludedColumns = append(excludedColumns, n.options.excludedUserColumns...)
+	}
+	sql, err := BuildFingerprintQueryForIndex(n.tableDesc, index, excludedColumns)
 	if err != nil {
 		return false, err
 	}