Disallow creating VECTOR indexes on nullable columns

enginetest/queries/vector_index_queries.go

@@ -23,7 +23,7 @@ var VectorIndexQueries = []ScriptTest{
 	{
 		Name: "basic JSON vector index",
 		SetUpScript: []string{
-			"create table vectors (id int primary key, v json);",
+			"create table vectors (id int primary key, v json not null);",
 			`insert into vectors values (1, '[4.0,3.0]'), (2, '[0.0,0.0]'), (3, '[-1.0,1.0]'), (4, '[0.0,-2.0]');`,
 			`create vector index v_idx on vectors(v);`,
 			`set @query_vec = '[0.0,0.0]';`,
@@ -32,7 +32,7 @@ var VectorIndexQueries = []ScriptTest{
 			{
 				Query: "show create table vectors",
 				Expected: []sql.Row{
-					{"vectors", "CREATE TABLE `vectors` (\n  `id` int NOT NULL,\n  `v` json,\n  PRIMARY KEY (`id`),\n  VECTOR KEY `v_idx` (`v`)\n) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_bin"},
+					{"vectors", "CREATE TABLE `vectors` (\n  `id` int NOT NULL,\n  `v` json NOT NULL,\n  PRIMARY KEY (`id`),\n  VECTOR KEY `v_idx` (`v`)\n) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_bin"},
 				},
 			},
 			{
@@ -119,7 +119,7 @@ var VectorIndexQueries = []ScriptTest{
 	{
 		Name: "basic VECTOR vector index",
 		SetUpScript: []string{
-			"create table vectors (id int primary key, v vector(2));",
+			"create table vectors (id int primary key, v vector(2) not null);",
 			`insert into vectors values (1, STRING_TO_VECTOR('[4.0,3.0]')), (2, STRING_TO_VECTOR('[0.0,0.0]')), (3, STRING_TO_VECTOR('[-1.0,1.0]')), (4, STRING_TO_VECTOR('[0.0,-2.0]'));`,
 			`create vector index v_idx on vectors(v);`,
 			`set @query_vec = '[0.0,0.0]';`,
@@ -128,7 +128,7 @@ var VectorIndexQueries = []ScriptTest{
 			{
 				Query: "show create table vectors",
 				Expected: []sql.Row{
-					{"vectors", "CREATE TABLE `vectors` (\n  `id` int NOT NULL,\n  `v` VECTOR(2),\n  PRIMARY KEY (`id`),\n  VECTOR KEY `v_idx` (`v`)\n) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_bin"},
+					{"vectors", "CREATE TABLE `vectors` (\n  `id` int NOT NULL,\n  `v` VECTOR(2) NOT NULL,\n  PRIMARY KEY (`id`),\n  VECTOR KEY `v_idx` (`v`)\n) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4 COLLATE=utf8mb4_0900_bin"},
 				},
 			},
 			{
@@ -233,4 +233,20 @@ var VectorIndexQueries = []ScriptTest{
 			},
 		},
 	},
+	{
+		Name: "vector index on nullable column errors",
+		SetUpScript: []string{
+			"create table vectors_nullable (id int primary key, v vector(2) null);",
+		},
+		Assertions: []ScriptTestAssertion{
+			{
+				Query:       `create vector index v_idx on vectors_nullable(v);`,
+				ExpectedErr: sql.ErrNullableVectorIdx,
+			},
+			{
+				Query:       `create table bad_vector_idx (id int primary key, v vector(2), vector index (v));`,
+				ExpectedErr: sql.ErrNullableVectorIdx,
+			},
+		},
+	},
 }

enginetest/scriptgen/setup/scripts/comp_index_tables

@@ -98,11 +98,11 @@ create table pref_index_t4 (i int primary key, v1 varchar(10), v2 varchar(10), u
 ----
 
 exec
-CREATE TABLE comp_vector_index_t0 (pk BIGINT PRIMARY KEY, v1 BIGINT, v2 JSON);
+CREATE TABLE comp_vector_index_t0 (pk BIGINT PRIMARY KEY, v1 BIGINT, json_column JSON NOT NULL, vector_column VECTOR(2) NOT NULL GENERATED ALWAYS AS (STRING_TO_VECTOR(json_column)) STORED);
 ----
 
 exec
-INSERT INTO comp_vector_index_t0 VALUES (0,0,"[3,16]"),(1,2,"[65,9]"),(2,3,"[38,37]"),(3,3,"[99,99]"),(4,5,"[17,42]"),(5,6,"[6,76]"),(6,6,"[81,33]"),
+INSERT INTO comp_vector_index_t0 (pk, v1, json_column) VALUES (0,0,"[3,16]"),(1,2,"[65,9]"),(2,3,"[38,37]"),(3,3,"[99,99]"),(4,5,"[17,42]"),(5,6,"[6,76]"),(6,6,"[81,33]"),
 (7,7,"[33,51]"),(8,7,"[37,42]"),(9,8,"[9,21]"),(10,8,"[37,90]"),(11,9,"[39,20]"),(12,9,"[71,82]"),(13,10,"[16,21]"),(14,10,"[32,46]"),(15,10,"[47,36]"),
 (16,12,"[44,84]"),(17,12,"[66,40]"),(18,13,"[47,30]"),(19,13,"[56,41]"),(20,14,"[38,24]"),(21,14,"[91,1]"),(22,15,"[2,69]"),(23,16,"[40,36]"),
 (24,20,"[29,93]"),(25,21,"[9,89]"),(26,21,"[42,76]"),(27,23,"[13,53]"),(28,23,"[28,68]"),(29,23,"[28,90]"),(30,23,"[30,44]"),(31,24,"[20,8]"),
@@ -118,7 +118,11 @@ INSERT INTO comp_vector_index_t0 VALUES (0,0,"[3,16]"),(1,2,"[65,9]"),(2,3,"[38,
 ----
 
 exec
-create VECTOR INDEX v_idx on comp_vector_index_t0 (v2)
+create VECTOR INDEX v_idx_json on comp_vector_index_t0 (json_column)
+----
+
+exec
+create VECTOR INDEX v_idx_vec on comp_vector_index_t0 (vector_column)
 ----
 
 exec

sql/analyzer/validate_create_table.go

@@ -15,6 +15,7 @@
 package analyzer
 
 import (
+	"fmt"
 	"strings"
 
 	"github.com/dolthub/go-mysql-server/sql"
@@ -888,6 +889,16 @@ func validateIndex(ctx *sql.Context, colMap map[string]*sql.Column, idxDef *sql.
 		}
 	}
 
+	if idxDef.IsVector() {
+		if len(idxDef.Columns) != 1 {
+			return fmt.Errorf("a vector index must have exactly one column")
+		}
+		schCol, _ := colMap[strings.ToLower(idxDef.Columns[0].Name)]
+		if schCol.Nullable {
+			return sql.ErrNullableVectorIdx.New()
+		}
+	}
+
 	if idxDef.IsSpatial() {
 		if len(idxDef.Columns) != 1 {
 			return sql.ErrTooManyKeyParts.New(1)

sql/errors.go

@@ -651,6 +651,9 @@ var (
 	// ErrNullableSpatialIdx is thrown when creating a SPATIAL index with a nullable column
 	ErrNullableSpatialIdx = errors.NewKind("All parts of a SPATIAL index must be NOT NULL")
 
+	// ErrNullableVectorIdx is thrown when creating a VECTOR index with a nullable column
+	ErrNullableVectorIdx = errors.NewKind("All parts of a VECTOR index must be NOT NULL")
+
 	// ErrBadSpatialIdxCol is thrown when attempting to define a SPATIAL index over a non-geometry column
 	ErrBadSpatialIdxCol = errors.NewKind("a SPATIAL index may only contain a geometrical type column")
 

sql/rowexec/ddl_iters.go

@@ -2163,6 +2163,9 @@ func (b *BaseBuilder) executeAlterIndex(ctx *sql.Context, n *plan.AlterIndex) er
 					if !types.IsVectorConvertable(tblCol.Type) {
 						return sql.ErrVectorInvalidColumnType.New()
 					}
+					if tblCol.Nullable {
+						return sql.ErrNullableVectorIdx.New()
+					}
 					break
 				}
 			}