sql: implement EXPLAIN (FINGERPRINT) statement

pkg/sql/conn_executor.go

@@ -4458,7 +4458,7 @@ func (ex *connExecutor) serialize() serverpb.Session {
 				"serialization")
 			continue
 		}
-		sqlNoConstants := truncateSQL(formatStatementHideConstants(parsed.AST))
+		sqlNoConstants := truncateSQL(tree.FormatStatementHideConstants(parsed.AST))
 		nPlaceholders := 0
 		if query.placeholders != nil {
 			nPlaceholders = len(query.placeholders.Values)
@@ -4483,7 +4483,7 @@ func (ex *connExecutor) serialize() serverpb.Session {
 			ElapsedTime:    elapsedTime,
 			Sql:            sql,
 			SqlNoConstants: sqlNoConstants,
-			SqlSummary:     formatStatementSummary(parsed.AST),
+			SqlSummary:     tree.FormatStatementSummary(parsed.AST),
 			Placeholders:   placeholders,
 			IsDistributed:  query.isDistributed,
 			Phase:          (serverpb.ActiveQuery_Phase)(query.phase),
@@ -4498,7 +4498,7 @@ func (ex *connExecutor) serialize() serverpb.Session {
 	lastActiveQueryNoConstants := ""
 	if ex.mu.LastActiveQuery != nil {
 		lastActiveQuery = truncateSQL(ex.mu.LastActiveQuery.String())
-		lastActiveQueryNoConstants = truncateSQL(formatStatementHideConstants(ex.mu.LastActiveQuery))
+		lastActiveQueryNoConstants = truncateSQL(tree.FormatStatementHideConstants(ex.mu.LastActiveQuery))
 	}
 	status := serverpb.Session_IDLE
 	if len(activeQueries) > 0 {

pkg/sql/conn_executor_exec.go

@@ -4395,7 +4395,7 @@ func (ex *connExecutor) execWithProfiling(
 		if prepared != nil {
 			stmtNoConstants = prepared.StatementNoConstants
 		} else {
-			stmtNoConstants = formatStatementHideConstants(ast)
+			stmtNoConstants = tree.FormatStatementHideConstants(ast)
 		}
 		labels := pprof.Labels(
 			"appname", ex.sessionData().ApplicationName,

pkg/sql/exec_util.go

@@ -3481,35 +3481,6 @@ func HashForReporting(secret, appName string) string {
 	return hex.EncodeToString(hash.Sum(nil)[:4])
 }
 
-// formatStatementHideConstants formats the statement using
-// tree.FmtHideConstants. It does *not* anonymize the statement, since
-// the result will still contain names and identifiers.
-func formatStatementHideConstants(ast tree.Statement, optFlags ...tree.FmtFlags) string {
-	if ast == nil {
-		return ""
-	}
-	fmtFlags := tree.FmtHideConstants
-	for _, f := range optFlags {
-		fmtFlags |= f
-	}
-	return tree.AsStringWithFlags(ast, fmtFlags)
-}
-
-// formatStatementSummary formats the statement using tree.FmtSummary
-// and tree.FmtHideConstants. This returns a summarized version of the
-// query. It does *not* anonymize the statement, since the result will
-// still contain names and identifiers.
-func formatStatementSummary(ast tree.Statement, optFlags ...tree.FmtFlags) string {
-	if ast == nil {
-		return ""
-	}
-	fmtFlags := tree.FmtSummary | tree.FmtHideConstants
-	for _, f := range optFlags {
-		fmtFlags |= f
-	}
-	return tree.AsStringWithFlags(ast, fmtFlags)
-}
-
 // DescsTxn is a convenient method for running a transaction on descriptors
 // when you have an ExecutorConfig.
 //

pkg/sql/executor_statement_metrics_test.go

@@ -138,7 +138,7 @@ func makeTestQuery(i int) (*Statement, error) {
 	}
 
 	return &Statement{
-		StmtNoConstants: formatStatementHideConstants(stmt.AST, tree.FmtCollapseLists|tree.FmtConstantsAsUnderscores),
+		StmtNoConstants: tree.FormatStatementHideConstants(stmt.AST, tree.FmtCollapseLists|tree.FmtConstantsAsUnderscores),
 		Statement:       stmt,
 	}, nil
 }

pkg/sql/opt/exec/execbuilder/statement.go

@@ -168,6 +168,15 @@ func (b *Builder) buildExplain(
 		return b.buildExplainOpt(explainExpr)
 	}
 
+	if explainExpr.Options.Mode == tree.ExplainFingerprint {
+		var ep execPlan
+		ep.root, err = b.factory.ConstructExplainOpt(explainExpr.Fingerprint, exec.ExplainEnvData{})
+		if err != nil {
+			return execPlan{}, colOrdMap{}, err
+		}
+		return ep, b.outputColsFromList(explainExpr.ColList), nil
+	}
+
 	var ep execPlan
 	ep.root, err = b.factory.ConstructExplain(
 		&explainExpr.Options,

pkg/sql/opt/exec/execbuilder/testdata/explain

@@ -2347,7 +2347,7 @@ quality of service: regular
 query T
 EXPLAIN (OPT, MEMO) SELECT * FROM tc JOIN t ON k=a
 ----
-memo (optimized, ~19KB, required=[presentation: info:14] [distribution: test])
+memo (optimized, ~20KB, required=[presentation: info:14] [distribution: test])
  ├── G1: (explain G2 [presentation: a:1,b:2,k:8,v:9] [distribution: test])
  │    └── [presentation: info:14] [distribution: test]
  │         ├── best: (explain G2="[presentation: a:1,b:2,k:8,v:9] [distribution: test]" [presentation: a:1,b:2,k:8,v:9] [distribution: test])
@@ -2428,7 +2428,7 @@ TABLE t
  ├── crdb_internal_origin_timestamp decimal [hidden] [system]
  └── PRIMARY INDEX t_pkey
       └── k int not null
-memo (optimized, ~19KB, required=[presentation: info:14] [distribution: test])
+memo (optimized, ~20KB, required=[presentation: info:14] [distribution: test])
  ├── G1: (explain G2 [presentation: a:1,b:2,k:8,v:9] [distribution: test])
  │    └── [presentation: info:14] [distribution: test]
  │         ├── best: (explain G2="[presentation: a:1,b:2,k:8,v:9] [distribution: test]" [presentation: a:1,b:2,k:8,v:9] [distribution: test])

pkg/sql/opt/exec/execbuilder/testdata/explain_fingerprint

@@ -0,0 +1,249 @@
+# LogicTest: local
+
+statement ok
+CREATE TABLE t (a INT PRIMARY KEY, b STRING, c FLOAT)
+
+statement ok
+CREATE TABLE u (x INT, y STRING, INDEX idx_y (y))
+
+# Basic EXPLAIN (FINGERPRINT) test
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM t
+----
+SELECT * FROM t
+
+# Test with constants - they should be replaced with placeholders
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM t WHERE a = 123
+----
+SELECT * FROM t WHERE a = _
+
+# Test with string constants
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM t WHERE b = 'hello'
+----
+SELECT * FROM t WHERE b = _
+
+# Test with multiple constants
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM t WHERE a = 1 AND b = 'test' AND c = 3.14
+----
+SELECT * FROM t WHERE ((a = _) AND (b = _)) AND (c = _)
+
+# Test with JOIN
+query T
+EXPLAIN (FINGERPRINT) SELECT t.a, u.y FROM t JOIN u ON t.a = u.x
+----
+SELECT t.a, u.y FROM t JOIN u ON t.a = u.x
+
+# Test with GROUP BY and aggregation
+query T
+EXPLAIN (FINGERPRINT) SELECT b, count(*) FROM t GROUP BY b
+----
+SELECT b, count(*) FROM t GROUP BY b
+
+# Test with ORDER BY
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM t ORDER BY a DESC
+----
+SELECT * FROM t ORDER BY a DESC
+
+# Test with LIMIT
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM t LIMIT 10
+----
+SELECT * FROM t LIMIT _
+
+# Test with subquery
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM t WHERE a IN (SELECT x FROM u WHERE y = 'test')
+----
+SELECT * FROM t WHERE a IN (SELECT x FROM u WHERE y = _)
+
+# Test with INSERT
+query T
+EXPLAIN (FINGERPRINT) INSERT INTO t VALUES (1, 'foo', 2.5)
+----
+INSERT INTO t VALUES (_, __more__)
+
+# Test with UPDATE
+query T
+EXPLAIN (FINGERPRINT) UPDATE t SET b = 'updated' WHERE a = 100
+----
+UPDATE t SET b = _ WHERE a = _
+
+# Test with DELETE
+query T
+EXPLAIN (FINGERPRINT) DELETE FROM t WHERE a > 50
+----
+DELETE FROM t WHERE a > _
+
+# Test with index hint
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM u@idx_y WHERE y = 'search'
+----
+SELECT * FROM u@idx_y WHERE y = _
+
+# Test with UNION
+query T
+EXPLAIN (FINGERPRINT) SELECT a FROM t UNION SELECT x FROM u
+----
+SELECT a FROM t UNION SELECT x FROM u
+
+# Test with CTE
+query T
+EXPLAIN (FINGERPRINT) WITH cte AS (SELECT * FROM t WHERE a < 10) SELECT * FROM cte
+----
+WITH cte AS (SELECT * FROM t WHERE a < _) SELECT * FROM cte
+
+# Test prepared statement functionality
+statement ok
+PREPARE stmt1 AS SELECT * FROM t WHERE a = $1
+
+query T
+EXPLAIN (FINGERPRINT) EXECUTE stmt1
+----
+EXECUTE stmt1
+
+statement ok
+PREPARE stmt2 AS INSERT INTO t VALUES ($1, $2, $3)
+
+query T
+EXPLAIN (FINGERPRINT) EXECUTE stmt2
+----
+EXECUTE stmt2
+
+statement ok
+PREPARE stmt3 AS SELECT t.a, u.y FROM t JOIN u ON t.a = u.x WHERE t.b = $1
+
+query T
+EXPLAIN (FINGERPRINT) EXECUTE stmt3
+----
+EXECUTE stmt3
+
+# Test prepare of EXPLAIN (FINGERPRINT)
+statement ok
+PREPARE stmt4 AS EXPLAIN (FINGERPRINT) SELECT 1
+
+query T
+EXECUTE stmt4
+----
+SELECT _
+
+# Test that invalid combinations are rejected
+statement error EXPLAIN \(FINGERPRINT\) cannot be used with VERBOSE
+EXPLAIN (FINGERPRINT, VERBOSE) SELECT * FROM t
+
+statement error EXPLAIN \(FINGERPRINT\) cannot be used with TYPES
+EXPLAIN (FINGERPRINT, TYPES) SELECT * FROM t
+
+statement error pq: at or near "EOF": syntax error: the JSON flag can only be used with DISTSQL
+EXPLAIN (FINGERPRINT, JSON) SELECT * FROM t
+
+# Test that regular EXPLAIN EXECUTE is still rejected while FINGERPRINT works
+statement error EXPLAIN EXECUTE is not supported; use EXPLAIN ANALYZE
+EXPLAIN EXECUTE stmt1
+
+# But EXPLAIN (FINGERPRINT) EXECUTE should work (already tested above)
+
+# Test with more complex queries
+query T
+EXPLAIN (FINGERPRINT)
+SELECT t.a, t.b, count(u.x) as cnt
+FROM t
+LEFT JOIN u ON t.a = u.x
+WHERE t.c > 1.0 AND (t.b LIKE 'prefix%' OR t.b IS NULL)
+GROUP BY t.a, t.b
+HAVING count(u.x) > 5
+ORDER BY cnt DESC, t.a ASC
+LIMIT 20 OFFSET 10
+----
+SELECT t.a, t.b, count(u.x) AS cnt FROM t LEFT JOIN u ON t.a = u.x WHERE (t.c > _) AND ((t.b LIKE _) OR (t.b IS NULL)) GROUP BY t.a, t.b HAVING count(u.x) > _ ORDER BY cnt DESC, t.a ASC LIMIT _ OFFSET _
+
+# Test with window functions
+query T
+EXPLAIN (FINGERPRINT)
+SELECT a, b, row_number() OVER (PARTITION BY b ORDER BY a) as rn
+FROM t
+----
+SELECT a, b, row_number() OVER (PARTITION BY b ORDER BY a) AS rn FROM t
+
+query T
+EXPLAIN (FINGERPRINT)
+SELECT
+  department_id,
+  employee_id,
+  salary,
+  avg(salary) OVER (
+    PARTITION BY department_id
+    ORDER BY employee_id
+    ROWS BETWEEN 2 PRECEDING AND CURRENT ROW
+  ) AS moving_avg_salary,
+  round(avg(salary) OVER (
+    PARTITION BY department_id
+  ), 2) AS dept_avg_salary,
+  1.05 * salary AS adjusted_salary,
+  salary + 1000 AS bonus_salary
+FROM employees
+----
+SELECT department_id, employee_id, salary, avg(salary) OVER (PARTITION BY department_id ORDER BY employee_id ROWS BETWEEN _ PRECEDING AND CURRENT ROW) AS moving_avg_salary, round(avg(salary) OVER (PARTITION BY department_id), _) AS dept_avg_salary, _ * salary AS adjusted_salary, salary + _ AS bonus_salary FROM employees
+
+
+# Test with CASE expression
+query T
+EXPLAIN (FINGERPRINT)
+SELECT a,
+       CASE
+         WHEN a > 100 THEN 'high'
+         WHEN a > 50 THEN 'medium'
+         ELSE 'low'
+       END as category
+FROM t
+----
+SELECT a, CASE WHEN a > _ THEN _ WHEN a > _ THEN _ ELSE _ END AS category FROM t
+
+# Test that different constant values produce the same fingerprint
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM t WHERE a = 999
+----
+SELECT * FROM t WHERE a = _
+
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM t WHERE a = 111
+----
+SELECT * FROM t WHERE a = _
+
+# Test with cluster setting sql.stats.statement_fingerprint.format_mask set to 0
+statement ok
+SET CLUSTER SETTING sql.stats.statement_fingerprint.format_mask = 0
+
+# With format_mask=0, long lists should not be collapsed to __more__
+query T
+EXPLAIN (FINGERPRINT) INSERT INTO t VALUES (1, 'a', 1.1), (2, 'b', 2.2), (3, 'c', 3.3), (4, 'd', 4.4), (5, 'e', 5.5)
+----
+INSERT INTO t VALUES (_, '_', __more1_10__), (__more1_10__)
+
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM t WHERE a IN (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
+----
+SELECT * FROM t WHERE a IN (_, _, __more1_10__)
+
+query T
+EXPLAIN (FINGERPRINT) UPDATE t SET b = 'updated' WHERE a IN (10, 20, 30, 40, 50, 60, 70, 80, 90, 100)
+----
+UPDATE t SET b = '_' WHERE a IN (_, _, __more1_10__)
+
+# Reset cluster setting to default
+statement ok
+RESET CLUSTER SETTING sql.stats.statement_fingerprint.format_mask
+
+# Verify default behavior is restored (long lists should be collapsed)
+query T
+EXPLAIN (FINGERPRINT) INSERT INTO t VALUES (1, 'a', 1.1), (2, 'b', 2.2), (3, 'c', 3.3), (4, 'd', 4.4), (5, 'e', 5.5)
+----
+INSERT INTO t VALUES (_, __more__), (__more__)
+
+query T
+EXPLAIN (FINGERPRINT) SELECT * FROM t WHERE a IN (1, 2, 3, 4, 5, 6, 7, 8, 9, 10)
+----
+SELECT * FROM t WHERE a IN (_, __more__)

pkg/sql/opt/ops/statement.opt

@@ -129,6 +129,9 @@ define ExplainPrivate {
 
     # StmtType stores the type of the statement return we are explaining.
     StmtType StatementReturnType
+
+    # Fingerprint stores the statement fingerprint for EXPLAIN (FINGERPRINT).
+    Fingerprint string
 }
 
 # ShowTraceForSession returns the current session traces.