Applying implicit prefix lengths for TEXT columns in secondary indexes

memory/index.go

@@ -137,6 +137,10 @@ func (idx *Index) PrefixLengths() []uint16 {
 	return idx.PrefixLens
 }
 
+func (idx *Index) SetPrefixLengths(prefixLengths []uint16) {
+	idx.PrefixLens = prefixLengths
+}
+
 func (idx *Index) IndexType() string {
 	if len(idx.DriverName) > 0 {
 		return idx.DriverName

sql/analyzer/analyzer.go

@@ -274,7 +274,7 @@ func (ab *Builder) Build() *Analyzer {
 		Parallelism:  ab.parallelism,
 		Coster:       memo.NewDefaultCoster(),
 		ExecBuilder:  rowexec.DefaultBuilder,
-		DatabaseType: sql.EngineType_MySql,
+		EngineType:   sql.EngineType_MySql,
 	}
 }
 
@@ -299,9 +299,9 @@ type Analyzer struct {
 	// EventScheduler is used to communiate with the event scheduler
 	// for any EVENT related statements. It can be nil if EventScheduler is not defined.
 	EventScheduler sql.EventScheduler
-	// DatabaseType indicates whether the analyzer's behavior is compatible with a MySQL
+	// EngineType indicates whether the analyzer's behavior is compatible with a MySQL
 	// database or a PostgreSQL database.
-	DatabaseType sql.EngineType
+	EngineType sql.EngineType
 }
 
 // NewDefault creates a default Analyzer instance with all default Rules and configuration.

sql/analyzer/index_analyzer_test.go

@@ -152,8 +152,9 @@ func (i *dummyIdx) Comment() string                       { return "" }
 func (i *dummyIdx) IsGenerated() bool                     { return false }
 func (i *dummyIdx) CanSupportOrderBy(sql.Expression) bool { return false }
 
-func (i *dummyIdx) IndexType() string       { return "BTREE" }
-func (i *dummyIdx) PrefixLengths() []uint16 { return nil }
+func (i *dummyIdx) IndexType() string                { return "BTREE" }
+func (i *dummyIdx) PrefixLengths() []uint16          { return nil }
+func (i dummyIdx) SetPrefixLengths(uint16s []uint16) {}
 
 func (i *dummyIdx) NewLookup(*sql.Context, ...sql.Range) (sql.IndexLookup, error) {
 	panic("not implemented")

sql/analyzer/validate_create_table.go

@@ -26,6 +26,10 @@ import (
 
 const MaxBytePrefix = 3072
 
+// defaultPrefixLength is used for PostgreSQL compatibility. We don't support secondary indexes including address
+// encoded types such as TEXT/BLOB, so we must apply an implicit prefix length.
+const defaultPrefixLength = 200
+
 // validateCreateTable validates various constraints about CREATE TABLE statements.
 func validateCreateTable(ctx *sql.Context, a *Analyzer, n sql.Node, scope *plan.Scope, sel RuleSelector, qFlags *sql.QueryFlags) (sql.Node, transform.TreeIdentity, error) {
 	ct, ok := n.(*plan.CreateTable)
@@ -44,7 +48,10 @@ func validateCreateTable(ctx *sql.Context, a *Analyzer, n sql.Node, scope *plan.
 	if err != nil {
 		return nil, transform.SameTree, err
 	}
-	validatePrefixLengths := a.DatabaseType == sql.EngineType_MySql
+
+	// For MySQL compatibility, users must specify a prefix length for any TEXT or BLOB columns used in an index,
+	// otherwise we will apply an implicit prefix length.
+	validatePrefixLengths := a.EngineType == sql.EngineType_MySql
 	err = validateIndexes(ctx, sch, idxs, strictMySQLCompat, validatePrefixLengths)
 	if err != nil {
 		return nil, transform.SameTree, err
@@ -225,9 +232,11 @@ func resolveAlterColumn(ctx *sql.Context, a *Analyzer, n sql.Node, scope *plan.S
 
 	sch = sch.Copy() // Make a copy of the original schema to deal with any references to the original table.
 	initialSch := sch
-
 	addedColumn := false
-	validatePrefixLengths := a.DatabaseType == sql.EngineType_MySql
+
+	// For MySQL compatibility, users must specify a prefix length for any TEXT or BLOB columns used in an index,
+	// otherwise we will apply an implicit prefix length.
+	validatePrefixLengths := a.EngineType == sql.EngineType_MySql
 
 	// Need a TransformUp here because multiple of these statement types can be nested under a Block node.
 	// It doesn't look it, but this is actually an iterative loop over all the independent clauses in an ALTER statement
@@ -520,6 +529,24 @@ func validateModifyColumn(ctx *sql.Context, initialSch sql.Schema, schema sql.Sc
 				}
 			}
 		}
+
+		// If we aren't validating user-specified prefix lengths, then apply an implicit prefix length for TEXT columns
+		if !validatePrefixLengths {
+			prefixLengths := make([]uint16, len(index.Expressions()))
+			for i, expr := range index.Expressions() {
+				col := plan.GetColumnFromIndexExpr(expr, tbl)
+				if !strings.EqualFold(col.Name, oldColName) {
+					continue
+				}
+				if types.IsJSON(newCol.Type) {
+					return nil, sql.ErrJSONIndex.New(col.Name)
+				}
+				if types.IsText(newCol.Type) {
+					prefixLengths[i] = defaultPrefixLength
+				}
+			}
+			index.SetPrefixLengths(prefixLengths)
+		}
 	}
 
 	return newSch, nil
@@ -906,6 +933,25 @@ func validateIndex(ctx *sql.Context, colMap map[string]*sql.Column, idxDef *sql.
 		}
 	}
 
+	// If we aren't validating user-specified prefix lengths, then we need to apply implicit prefix lengths for
+	// any TEXT columns.
+	if !validatePrefixLengths {
+		prefixLengths := make([]uint16, len(idxDef.Columns))
+		for i, idxCol := range idxDef.Columns {
+			schCol, exists := colMap[strings.ToLower(idxCol.Name)]
+			if !exists {
+				return sql.ErrKeyColumnDoesNotExist.New(idxCol.Name)
+			}
+			if types.IsJSON(schCol.Type) {
+				return sql.ErrJSONIndex.New(schCol.Name)
+			}
+			if types.IsText(schCol.Type) {
+				prefixLengths[i] = defaultPrefixLength
+			}
+			idxDef.Columns[i].Length = int64(prefixLengths[i])
+		}
+	}
+
 	if idxDef.IsSpatial() {
 		if len(idxDef.Columns) != 1 {
 			return sql.ErrTooManyKeyParts.New(1)

sql/index.go

@@ -126,8 +126,10 @@ type Index interface {
 	// Verifying that the expression's children match the index columns are done separately.
 	CanSupportOrderBy(expr Expression) bool
 
-	// PrefixLengths returns the prefix lengths for each column in this index
+	// PrefixLengths returns the prefix lengths for each column in this index.
 	PrefixLengths() []uint16
+	// SetPrefixLengths sets the prefix lengths for each column in this index.
+	SetPrefixLengths([]uint16)
 }
 
 // ExtendedIndex is an extension of Index, that allows access to appended primary keys. MySQL internally represents an

sql/index_builder_test.go

@@ -225,4 +225,6 @@ func (testIndex) PrefixLengths() []uint16 {
 	return nil
 }
 
+func (i testIndex) SetPrefixLengths(uint16s []uint16) {}
+
 var _ sql.Index = testIndex{}

sql/index_registry_test.go

@@ -443,17 +443,18 @@ func (i dummyIdx) Expressions() []string {
 	}
 	return exprs
 }
-func (i dummyIdx) ID() string              { return i.id }
-func (i dummyIdx) Database() string        { return i.database }
-func (i dummyIdx) Table() string           { return i.table }
-func (i dummyIdx) Driver() string          { return "dummy" }
-func (i dummyIdx) IsUnique() bool          { return false }
-func (i dummyIdx) IsSpatial() bool         { return false }
-func (i dummyIdx) IsFullText() bool        { return false }
-func (i dummyIdx) Comment() string         { return "" }
-func (i dummyIdx) IsGenerated() bool       { return false }
-func (i dummyIdx) IndexType() string       { return "BTREE" }
-func (i dummyIdx) PrefixLengths() []uint16 { return nil }
+func (i dummyIdx) ID() string                        { return i.id }
+func (i dummyIdx) Database() string                  { return i.database }
+func (i dummyIdx) Table() string                     { return i.table }
+func (i dummyIdx) Driver() string                    { return "dummy" }
+func (i dummyIdx) IsUnique() bool                    { return false }
+func (i dummyIdx) IsSpatial() bool                   { return false }
+func (i dummyIdx) IsFullText() bool                  { return false }
+func (i dummyIdx) Comment() string                   { return "" }
+func (i dummyIdx) IsGenerated() bool                 { return false }
+func (i dummyIdx) IndexType() string                 { return "BTREE" }
+func (i dummyIdx) PrefixLengths() []uint16           { return nil }
+func (i dummyIdx) SetPrefixLengths(uint16s []uint16) {}
 
 func (i dummyIdx) NewLookup(ctx *Context, ranges ...Range) (IndexLookup, error) {
 	panic("not implemented")

sql/memo/rel_props_test.go

@@ -257,5 +257,6 @@ func (i dummyIndex) ColumnExpressionTypes() []sql.ColumnExpressionType {
 func (dummyIndex) PrefixLengths() []uint16 {
 	return nil
 }
+func (i dummyIndex) SetPrefixLengths(uint16s []uint16) {}
 
 var _ sql.Index = dummyIndex{}