schemadiff: normalize missing column collation

Signed-off-by: Shlomi Noach <[email protected]>

go/vt/schemadiff/column.go

@@ -104,7 +104,13 @@ func (c *ColumnDefinitionEntity) ColumnDiff(
 ) (*ModifyColumnDiff, error) {
 	if c.IsTextual() || other.IsTextual() {
 		// We will now denormalize the columns charset & collate as needed (if empty, populate from table.)
-
+		if c.columnDefinition.Type.Charset.Name != "" && c.columnDefinition.Type.Options.Collate == "" {
+			collation := env.CollationEnv().DefaultCollationForCharset(c.columnDefinition.Type.Charset.Name)
+			if collation == collations.Unknown {
+				return nil, &UnknownColumnCharsetCollationError{Column: c.columnDefinition.Name.String(), Charset: t1cc.charset}
+			}
+			c.columnDefinition.Type.Options.Collate = env.CollationEnv().LookupName(collation)
+		}
 		if c.columnDefinition.Type.Charset.Name == "" && c.columnDefinition.Type.Options.Collate != "" {
 			// Column has explicit collation but no charset. We can infer the charset from the collation.
 			collationID := env.CollationEnv().LookupByName(c.columnDefinition.Type.Options.Collate)
@@ -137,6 +143,13 @@ func (c *ColumnDefinitionEntity) ColumnDiff(
 				c.columnDefinition.Type.Options.Collate = env.CollationEnv().LookupName(collation)
 			}
 		}
+		if other.columnDefinition.Type.Charset.Name != "" && other.columnDefinition.Type.Options.Collate == "" {
+			collation := env.CollationEnv().DefaultCollationForCharset(other.columnDefinition.Type.Charset.Name)
+			if collation == collations.Unknown {
+				return nil, &UnknownColumnCharsetCollationError{Column: other.columnDefinition.Name.String(), Charset: t1cc.charset}
+			}
+			other.columnDefinition.Type.Options.Collate = env.CollationEnv().LookupName(collation)
+		}
 		if other.columnDefinition.Type.Charset.Name == "" && other.columnDefinition.Type.Options.Collate != "" {
 			// Column has explicit collation but no charset. We can infer the charset from the collation.
 			collationID := env.CollationEnv().LookupByName(other.columnDefinition.Type.Options.Collate)

go/vt/schemadiff/table_test.go

@@ -1896,6 +1896,26 @@ func TestCreateTableDiff(t *testing.T) {
 			from: "create table t (id int primary key, v varchar(64) character set utf8mb3 collate utf8mb3_bin)",
 			to:   "create table t (id int primary key, v varchar(64) collate utf8mb3_bin)",
 		},
+		{
+			name: "ignore identical implicit ascii charset",
+			from: "create table t (id int primary key, v varchar(64) character set ascii collate ascii_general_ci)",
+			to:   "create table t (id int primary key, v varchar(64) collate ascii_general_ci)",
+		},
+		{
+			name: "ignore identical implicit collation",
+			from: "create table t (id int primary key, v varchar(64) character set utf8mb3 collate utf8mb3_general_ci)",
+			to:   "create table t (id int primary key, v varchar(64) character set utf8mb3)",
+		},
+		{
+			name: "ignore identical implicit collation, reverse",
+			from: "create table t (id int primary key, v varchar(64) character set utf8mb3)",
+			to:   "create table t (id int primary key, v varchar(64) character set utf8mb3 collate utf8mb3_general_ci)",
+		},
+		{
+			name: "ignore identical implicit ascii collation",
+			from: "create table t (id int primary key, v varchar(64) character set ascii collate ascii_general_ci)",
+			to:   "create table t (id int primary key, v varchar(64) character set ascii)",
+		},
 		{
 			name:  "normalized unsigned attribute",
 			from:  "create table t1 (id int primary key)",
@@ -2925,6 +2945,11 @@ func TestNormalize(t *testing.T) {
 			from: "create table t (id int primary key, v varchar(255) charset utf8mb4 collate utf8mb4_german2_ci)",
 			to:   "CREATE TABLE `t` (\n\t`id` int,\n\t`v` varchar(255) COLLATE utf8mb4_german2_ci,\n\tPRIMARY KEY (`id`)\n)",
 		},
+		{
+			name: "ascii charset and collation",
+			from: "create table t (id int primary key, v varchar(255) charset ascii collate ascii_general_ci) charset utf8mb3 collate utf8_general_ci",
+			to:   "CREATE TABLE `t` (\n\t`id` int,\n\t`v` varchar(255) CHARACTER SET ascii COLLATE ascii_general_ci,\n\tPRIMARY KEY (`id`)\n) CHARSET utf8mb3,\n  COLLATE utf8mb3_general_ci",
+		},
 		{
 			name: "correct case table options for engine",
 			from: "create table t (id int signed primary key) engine innodb",