sql: enable storage for tsvector/tsquery

This commit adds the ability to store tsvector and tsquery data in
ordinary, unindexed columns.

This functionality is gated behind the 23.1 version.

Release note (sql change): permit non-indexed storage of tsvector and
tsquery datatypes

pkg/sql/alter_column_type.go

@@ -104,7 +104,7 @@ func AlterColumnType(
 		}
 	}
 
-	err = colinfo.ValidateColumnDefType(typ)
+	err = colinfo.ValidateColumnDefType(ctx, params.EvalContext().Settings.Version, typ)
 	if err != nil {
 		return err
 	}

pkg/sql/catalog/colinfo/BUILD.bazel

@@ -18,6 +18,7 @@ go_library(
     importpath = "github.com/cockroachdb/cockroach/pkg/sql/catalog/colinfo",
     visibility = ["//visibility:public"],
     deps = [
+        "//pkg/clusterversion",
         "//pkg/sql/catalog",
         "//pkg/sql/catalog/catpb",
         "//pkg/sql/catalog/descpb",

pkg/sql/catalog/colinfo/col_type_info.go

@@ -11,8 +11,10 @@
 package colinfo
 
 import (
+	"context"
 	"fmt"
 
+	"github.com/cockroachdb/cockroach/pkg/clusterversion"
 	"github.com/cockroachdb/cockroach/pkg/sql/catalog"
 	"github.com/cockroachdb/cockroach/pkg/sql/pgwire/pgcode"
 	"github.com/cockroachdb/cockroach/pkg/sql/pgwire/pgerror"
@@ -68,7 +70,7 @@ func (ti ColTypeInfo) Type(idx int) *types.T {
 
 // ValidateColumnDefType returns an error if the type of a column definition is
 // not valid. It is checked when a column is created or altered.
-func ValidateColumnDefType(t *types.T) error {
+func ValidateColumnDefType(ctx context.Context, version clusterversion.Handle, t *types.T) error {
 	switch t.Family() {
 	case types.StringFamily, types.CollatedStringFamily:
 		if t.Family() == types.CollatedStringFamily {
@@ -100,7 +102,7 @@ func ValidateColumnDefType(t *types.T) error {
 		if err := types.CheckArrayElementType(t.ArrayContents()); err != nil {
 			return err
 		}
-		return ValidateColumnDefType(t.ArrayContents())
+		return ValidateColumnDefType(ctx, version, t.ArrayContents())
 
 	case types.BitFamily, types.IntFamily, types.FloatFamily, types.BoolFamily, types.BytesFamily, types.DateFamily,
 		types.INetFamily, types.IntervalFamily, types.JsonFamily, types.OidFamily, types.TimeFamily,
@@ -116,6 +118,12 @@ func ValidateColumnDefType(t *types.T) error {
 			return unimplemented.NewWithIssue(70099, "cannot use table record type as table column")
 		}
 
+	case types.TSQueryFamily, types.TSVectorFamily:
+		if !version.IsActive(ctx, clusterversion.V23_1) {
+			return pgerror.Newf(pgcode.FeatureNotSupported,
+				"TSVector/TSQuery not supported until version 23.1")
+		}
+
 	default:
 		return pgerror.Newf(pgcode.InvalidTableDefinition,
 			"value type %s cannot be used for table columns", t.String())

pkg/sql/catalog/tabledesc/table.go

@@ -151,7 +151,7 @@ func MakeColumnDefDescs(
 	if err != nil {
 		return nil, err
 	}
-	if err = colinfo.ValidateColumnDefType(resType); err != nil {
+	if err = colinfo.ValidateColumnDefType(ctx, evalCtx.Settings.Version, resType); err != nil {
 		return nil, err
 	}
 	col.Type = resType

pkg/sql/logictest/testdata/logic_test/tsvector

@@ -3,27 +3,103 @@ SELECT 'foo:1,2 bar:3'::tsvector @@ 'foo <-> bar'::tsquery, 'foo <-> bar'::tsque
 ----
 true  true
 
-statement error cannot be used for table columns
-CREATE TABLE a (v tsvector)
-
-statement error cannot be used for table columns
-CREATE TABLE a (q tsquery)
-
-# Uncomment once tsvector and tsquery are usable as table columns.
-#
-# statement ok
-# INSERT INTO a VALUES('foo:1,2 bar:4B'::tsvector, 'foo <2> bar'::tsquery)
-#
-# query TT
-# SELECT * FROM a
-# ----
-# 'bar':4B 'foo':1,2  'foo' <2> 'bar'
+statement ok
+CREATE TABLE a (v tsvector, q tsquery)
+
+statement ok
+INSERT INTO a VALUES('foo:1,2 bar:4B'::tsvector, 'foo <2> bar'::tsquery)
+
+query TT
+SELECT * FROM a
+----
+'bar':4B 'foo':1,2  'foo' <2> 'bar'
 
 query BB
 SELECT 'foo:1,2 bar:4B'::tsvector @@ 'foo <2> bar'::tsquery, 'foo:1,2 bar:4B' @@ 'foo <-> bar'::tsquery
 ----
 true  false
 
+query BB
+SELECT v @@ 'foo <2> bar'::tsquery, v @@ 'foo <-> bar'::tsquery FROM a
+----
+true  false
+
+query B
+SELECT v @@ q FROM a
+----
+true
+
+# Test column modifiers.
+
+statement ok
+CREATE TABLE b (a INT PRIMARY KEY DEFAULT 1, v tsvector DEFAULT 'foo:1' ON UPDATE 'bar:2', v2 tsvector AS (v) STORED, v3 tsvector AS (v) VIRTUAL)
+
+statement ok
+CREATE TABLE c (a INT PRIMARY KEY DEFAULT 1, q tsquery DEFAULT 'foo' ON UPDATE 'bar', q2 tsquery AS (q) STORED, q3 tsquery AS (q) VIRTUAL)
+
+statement ok
+INSERT INTO b DEFAULT VALUES
+
+statement ok
+INSERT INTO c DEFAULT VALUES
+
+query ITTT
+SELECT * FROM b
+----
+1  'foo':1  'foo':1  'foo':1
+
+query ITTT
+SELECT * FROM c
+----
+1  'foo'  'foo'  'foo'
+
+statement ok
+UPDATE b SET a = 2 WHERE a = 1
+
+statement ok
+UPDATE c SET a = 2 WHERE a = 1
+
+query ITTT
+SELECT * FROM b
+----
+2  'bar':2  'bar':2  'bar':2
+
+query ITTT
+SELECT * FROM c
+----
+2  'bar'  'bar'  'bar'
+
+statement ok
+INSERT INTO b VALUES (3, 'foo:1,5 zoop:3')
+
+statement error can't order by column type TSVECTOR
+SELECT * FROM b ORDER BY v
+
+statement error can't order by column type TSQUERY
+SELECT * FROM c ORDER BY q
+
+statement error arrays of tsvector not allowed
+CREATE TABLE tsarray(a tsvector[])
+
+statement error arrays of tsquery not allowed
+CREATE TABLE tsarray(a tsquery[])
+
+statement error unsupported comparison operator
+SELECT a, v FROM b WHERE v > 'bar:2'::tsvector
+
+statement error unsupported comparison operator
+SELECT a, q FROM c WHERE q > 'abc'::tsquery
+
+query IT
+SELECT a, v FROM b WHERE v = 'bar:2'::tsvector
+----
+2  'bar':2
+
+query IT
+SELECT a, q FROM c WHERE q = 'bar'::tsquery
+----
+2  'bar'
+
 # Ensure truncation of long position lists.
 query T
 SELECT ('foo:' || string_agg(g::TEXT,','))::tsvector from generate_series(1,280) g(g);

pkg/sql/opt/optbuilder/orderby.go

@@ -285,8 +285,13 @@ func (b *Builder) analyzeExtraArgument(
 
 func ensureColumnOrderable(e tree.TypedExpr) {
 	typ := e.ResolvedType()
-	if typ.Family() == types.JsonFamily ||
-		(typ.Family() == types.ArrayFamily && typ.ArrayContents().Family() == types.JsonFamily) {
+	if typ.Family() == types.ArrayFamily {
+		typ = typ.ArrayContents()
+	}
+	switch typ.Family() {
+	case types.JsonFamily:
 		panic(unimplementedWithIssueDetailf(35706, "", "can't order by column type jsonb"))
+	case types.TSQueryFamily, types.TSVectorFamily:
+		panic(unimplementedWithIssueDetailf(92165, "", "can't order by column type %s", typ.SQLString()))
 	}
 }

pkg/sql/randgen/BUILD.bazel

@@ -22,6 +22,7 @@ go_library(
         "//pkg/geo/geopb",
         "//pkg/keys",
         "//pkg/roachpb",
+        "//pkg/settings/cluster",
         "//pkg/sql/catalog",
         "//pkg/sql/catalog/colinfo",
         "//pkg/sql/catalog/descpb",

pkg/sql/randgen/doc.go

@@ -11,5 +11,7 @@
 /*
 Package randgen provides utility functions for generating random syntax trees,
 datums, encoded datums, types, and more. It useful in randomized tests.
+
+It should not be used outside of a testing context.
 */
 package randgen

pkg/sql/randgen/type.go

@@ -11,9 +11,11 @@
 package randgen
 
 import (
+	"context"
 	"math/rand"
 	"sort"
 
+	clustersettings "github.com/cockroachdb/cockroach/pkg/settings/cluster"
 	"github.com/cockroachdb/cockroach/pkg/sql/catalog/colinfo"
 	"github.com/cockroachdb/cockroach/pkg/sql/rowenc/valueside"
 	"github.com/cockroachdb/cockroach/pkg/sql/types"
@@ -152,15 +154,19 @@ func IsLegalColumnType(typ *types.T) bool {
 		// user-created tables.
 		return false
 	}
-	return colinfo.ValidateColumnDefType(typ) == nil
+	ctx := context.Background()
+	version := clustersettings.MakeTestingClusterSettings().Version
+	return colinfo.ValidateColumnDefType(ctx, version, typ) == nil
 }
 
 // RandArrayType generates a random array type.
 func RandArrayType(rng *rand.Rand) *types.T {
+	ctx := context.Background()
+	version := clustersettings.MakeTestingClusterSettings().Version
 	for {
 		typ := RandColumnType(rng)
 		resTyp := types.MakeArray(typ)
-		if err := colinfo.ValidateColumnDefType(resTyp); err == nil {
+		if err := colinfo.ValidateColumnDefType(ctx, version, resTyp); err == nil {
 			return resTyp
 		}
 	}
@@ -247,11 +253,13 @@ func RandEncodableType(rng *rand.Rand) *types.T {
 // TODO(andyk): Remove this workaround once #36736 is resolved. Replace calls to
 // it with calls to RandColumnTypes.
 func RandEncodableColumnTypes(rng *rand.Rand, numCols int) []*types.T {
+	ctx := context.Background()
+	version := clustersettings.MakeTestingClusterSettings().Version
 	types := make([]*types.T, numCols)
 	for i := range types {
 		for {
 			types[i] = RandEncodableType(rng)
-			if err := colinfo.ValidateColumnDefType(types[i]); err == nil {
+			if err := colinfo.ValidateColumnDefType(ctx, version, types[i]); err == nil {
 				break
 			}
 		}

pkg/sql/virtual_schema.go

@@ -196,10 +196,14 @@ func (t virtualSchemaTable) initVirtualTableDesc(
 			privilege.List{},
 			username.NodeUserName(),
 		),
-		nil,                        /* affected */
-		&semaCtx,                   /* semaCtx */
-		nil,                        /* evalCtx */
-		&sessiondata.SessionData{}, /* sessionData */
+		nil,      /* affected */
+		&semaCtx, /* semaCtx */
+		// We explicitly pass in a half-baked EvalContext because we don't need to
+		// evaluate any expressions to initialize virtual tables. We do need to
+		// pass in the cluster settings to make sure that functions can properly
+		// evaluate version gates, though.
+		&eval.Context{Settings: st}, /* evalCtx */
+		&sessiondata.SessionData{},  /* sessionData */
 		tree.PersistencePermanent,
 	)
 	if err != nil {
@@ -323,7 +327,11 @@ func (v virtualSchemaView) initVirtualTableDesc(
 			username.NodeUserName(),
 		),
 		nil, // semaCtx
-		nil, // evalCtx
+		// We explicitly pass in a half-baked EvalContext because we don't need to
+		// evaluate any expressions to initialize virtual tables. We do need to
+		// pass in the cluster settings to make sure that functions can properly
+		// evaluate version gates, though.
+		&eval.Context{Settings: st}, /* evalCtx */
 		st,
 		tree.PersistencePermanent,
 		false, // isMultiRegion