Fix should allow column list before ENGINE in CREATE MATERIALIZED VIEW

parser/ast.go

@@ -1356,6 +1356,7 @@ type CreateMaterializedView struct {
 	Settings     *SettingsClause
 	HasAppend    bool
 	Engine       *EngineExpr
+	TableSchema  *TableSchemaClause // column list for ENGINE-based MVs (no TO clause)
 	HasEmpty     bool
 	Destination  *DestinationClause
 	SubQuery     *SubQuery
@@ -1410,6 +1411,11 @@ func (c *CreateMaterializedView) Accept(visitor ASTVisitor) error {
 			return err
 		}
 	}
+	if c.TableSchema != nil {
+		if err := c.TableSchema.Accept(visitor); err != nil {
+			return err
+		}
+	}
 	if c.Engine != nil {
 		if err := c.Engine.Accept(visitor); err != nil {
 			return err

parser/format.go

@@ -907,6 +907,10 @@ func (c *CreateMaterializedView) FormatSQL(formatter *Formatter) {
 		formatter.Break()
 		formatter.WriteString("APPEND")
 	}
+	if c.TableSchema != nil {
+		formatter.Break()
+		formatter.WriteExpr(c.TableSchema)
+	}
 	if c.Engine != nil {
 		formatter.Break()
 		formatter.WriteExpr(c.Engine)

parser/parser_view.go

@@ -66,6 +66,7 @@ func (p *Parser) parseCreateMaterializedView(pos Pos) (*CreateMaterializedView,
 		}
 		dependsOnTables = append(dependsOnTables, table)
 		for p.matchTokenKind(TokenKindComma) {
+			_ = p.lexer.consumeToken()
 			table, err := p.parseTableIdentifier(p.Pos())
 			if err != nil {
 				return nil, err
@@ -96,6 +97,22 @@ func (p *Parser) parseCreateMaterializedView(pos Pos) (*CreateMaterializedView,
 			}
 			createMaterializedView.Destination.TableSchema = tableSchema
 		}
+	case p.matchTokenKind(TokenKindLParen):
+		// Column list before ENGINE (e.g. SHOW CREATE TABLE output for RMVs with ENGINE = Memory)
+		tableSchema, err := p.parseTableSchemaClause(p.Pos())
+		if err != nil {
+			return nil, err
+		}
+		createMaterializedView.TableSchema = tableSchema
+		if !p.matchKeyword(KeywordEngine) {
+			return nil, fmt.Errorf("unexpected token: %q, expected ENGINE after column list", p.lastTokenKind())
+		}
+		engineExpr, err := p.parseEngineExpr(p.Pos())
+		if err != nil {
+			return nil, err
+		}
+		createMaterializedView.Engine = engineExpr
+		createMaterializedView.StatementEnd = engineExpr.End()
 	case p.matchKeyword(KeywordEngine):
 		engineExpr, err := p.parseEngineExpr(p.Pos())
 		if err != nil {

parser/testdata/ddl/create_materialized_view_rmv_depends_on_multi.sql

@@ -0,0 +1,10 @@
+CREATE MATERIALIZED VIEW db1.attrs_rollup_v0
+REFRESH EVERY 5 MINUTE
+DEPENDS ON db1.metadata_mv_v0, db1.metadata_mv_v1
+ENGINE = Memory
+AS SELECT DISTINCT
+    service_name,
+    attr_key,
+    'profiles' AS dataset,
+    'string' AS attr_type
+FROM db1.metadata_v0

parser/testdata/ddl/create_materialized_view_rmv_engine_with_columns.sql

@@ -0,0 +1,36 @@
+CREATE MATERIALIZED VIEW db1.config_memory_v0
+REFRESH EVERY 1 SECOND
+(
+    `schedule_id` String,
+    `sample_rate` Int8,
+    `start_at` DateTime64(9),
+    `end_at` DateTime64(9),
+    `created_at` DateTime64(9),
+    `created_by` String,
+    `properties` Map(String, String),
+    `cluster_percentage` Float64,
+    `schedule_filters` String
+)
+ENGINE = Memory
+SETTINGS min_rows_to_keep = 250000, max_rows_to_keep = 500000
+DEFINER = default SQL SECURITY DEFINER
+COMMENT 'test comment'
+AS SELECT
+    schedule_id,
+    sample_rate,
+    start_at,
+    end_at,
+    created_at,
+    created_by,
+    properties,
+    cluster_percentage,
+    schedule_filters
+FROM
+(
+    SELECT
+        *,
+        row_number() OVER (PARTITION BY schedule_filters ORDER BY created_at DESC) AS rn
+    FROM db1.config_v0
+    WHERE schedule_filters != '{}'
+)
+WHERE rn = 1

parser/testdata/ddl/format/beautify/create_materialized_view_rmv_depends_on_multi.sql

@@ -0,0 +1,26 @@
+-- Origin SQL:
+CREATE MATERIALIZED VIEW db1.attrs_rollup_v0
+REFRESH EVERY 5 MINUTE
+DEPENDS ON db1.metadata_mv_v0, db1.metadata_mv_v1
+ENGINE = Memory
+AS SELECT DISTINCT
+    service_name,
+    attr_key,
+    'profiles' AS dataset,
+    'string' AS attr_type
+FROM db1.metadata_v0
+
+
+-- Beautify SQL:
+CREATE MATERIALIZED VIEW db1.attrs_rollup_v0
+REFRESH EVERY 5 MINUTE
+DEPENDS ON db1.metadata_mv_v0, db1.metadata_mv_v1
+ENGINE = Memory
+AS
+  SELECT DISTINCT
+    service_name,
+    attr_key,
+    'profiles' AS dataset,
+    'string' AS attr_type
+  FROM
+    db1.metadata_v0;

parser/testdata/ddl/format/beautify/create_materialized_view_rmv_engine_with_columns.sql

@@ -0,0 +1,82 @@
+-- Origin SQL:
+CREATE MATERIALIZED VIEW db1.config_memory_v0
+REFRESH EVERY 1 SECOND
+(
+    `schedule_id` String,
+    `sample_rate` Int8,
+    `start_at` DateTime64(9),
+    `end_at` DateTime64(9),
+    `created_at` DateTime64(9),
+    `created_by` String,
+    `properties` Map(String, String),
+    `cluster_percentage` Float64,
+    `schedule_filters` String
+)
+ENGINE = Memory
+SETTINGS min_rows_to_keep = 250000, max_rows_to_keep = 500000
+DEFINER = default SQL SECURITY DEFINER
+COMMENT 'test comment'
+AS SELECT
+    schedule_id,
+    sample_rate,
+    start_at,
+    end_at,
+    created_at,
+    created_by,
+    properties,
+    cluster_percentage,
+    schedule_filters
+FROM
+(
+    SELECT
+        *,
+        row_number() OVER (PARTITION BY schedule_filters ORDER BY created_at DESC) AS rn
+    FROM db1.config_v0
+    WHERE schedule_filters != '{}'
+)
+WHERE rn = 1
+
+
+-- Beautify SQL:
+CREATE MATERIALIZED VIEW db1.config_memory_v0
+REFRESH EVERY 1 SECOND
+(
+  `schedule_id` String,
+  `sample_rate` Int8,
+  `start_at` DateTime64(9),
+  `end_at` DateTime64(9),
+  `created_at` DateTime64(9),
+  `created_by` String,
+  `properties` Map(String, String),
+  `cluster_percentage` Float64,
+  `schedule_filters` String
+)
+ENGINE = Memory
+SETTINGS
+  min_rows_to_keep=250000,
+  max_rows_to_keep=500000
+DEFINER = default
+SQL SECURITY DEFINER
+AS
+  SELECT
+    schedule_id,
+    sample_rate,
+    start_at,
+    end_at,
+    created_at,
+    created_by,
+    properties,
+    cluster_percentage,
+    schedule_filters
+  FROM
+    (SELECT
+      *,
+      row_number() OVER (PARTITION BY schedule_filters ORDER BY
+        created_at DESC) AS rn
+    FROM
+      db1.config_v0
+    WHERE
+      schedule_filters != '{}')
+  WHERE
+    rn = 1
+COMMENT 'test comment';

parser/testdata/ddl/format/create_materialized_view_rmv_depends_on_multi.sql

@@ -0,0 +1,15 @@
+-- Origin SQL:
+CREATE MATERIALIZED VIEW db1.attrs_rollup_v0
+REFRESH EVERY 5 MINUTE
+DEPENDS ON db1.metadata_mv_v0, db1.metadata_mv_v1
+ENGINE = Memory
+AS SELECT DISTINCT
+    service_name,
+    attr_key,
+    'profiles' AS dataset,
+    'string' AS attr_type
+FROM db1.metadata_v0
+
+
+-- Format SQL:
+CREATE MATERIALIZED VIEW db1.attrs_rollup_v0 REFRESH EVERY 5 MINUTE DEPENDS ON db1.metadata_mv_v0, db1.metadata_mv_v1 ENGINE = Memory AS SELECT DISTINCT service_name, attr_key, 'profiles' AS dataset, 'string' AS attr_type FROM db1.metadata_v0;

parser/testdata/ddl/format/create_materialized_view_rmv_engine_with_columns.sql

@@ -0,0 +1,41 @@
+-- Origin SQL:
+CREATE MATERIALIZED VIEW db1.config_memory_v0
+REFRESH EVERY 1 SECOND
+(
+    `schedule_id` String,
+    `sample_rate` Int8,
+    `start_at` DateTime64(9),
+    `end_at` DateTime64(9),
+    `created_at` DateTime64(9),
+    `created_by` String,
+    `properties` Map(String, String),
+    `cluster_percentage` Float64,
+    `schedule_filters` String
+)
+ENGINE = Memory
+SETTINGS min_rows_to_keep = 250000, max_rows_to_keep = 500000
+DEFINER = default SQL SECURITY DEFINER
+COMMENT 'test comment'
+AS SELECT
+    schedule_id,
+    sample_rate,
+    start_at,
+    end_at,
+    created_at,
+    created_by,
+    properties,
+    cluster_percentage,
+    schedule_filters
+FROM
+(
+    SELECT
+        *,
+        row_number() OVER (PARTITION BY schedule_filters ORDER BY created_at DESC) AS rn
+    FROM db1.config_v0
+    WHERE schedule_filters != '{}'
+)
+WHERE rn = 1
+
+
+-- Format SQL:
+CREATE MATERIALIZED VIEW db1.config_memory_v0 REFRESH EVERY 1 SECOND (`schedule_id` String, `sample_rate` Int8, `start_at` DateTime64(9), `end_at` DateTime64(9), `created_at` DateTime64(9), `created_by` String, `properties` Map(String, String), `cluster_percentage` Float64, `schedule_filters` String) ENGINE = Memory SETTINGS min_rows_to_keep=250000, max_rows_to_keep=500000 DEFINER = default SQL SECURITY DEFINER AS SELECT schedule_id, sample_rate, start_at, end_at, created_at, created_by, properties, cluster_percentage, schedule_filters FROM (SELECT *, row_number() OVER (PARTITION BY schedule_filters ORDER BY created_at DESC) AS rn FROM db1.config_v0 WHERE schedule_filters != '{}') WHERE rn = 1 COMMENT 'test comment';

parser/testdata/ddl/output/bug_001.sql.golden.json

@@ -31,6 +31,7 @@
     "Settings": null,
     "HasAppend": false,
     "Engine": null,
+    "TableSchema": null,
     "HasEmpty": false,
     "Destination": {
       "ToPos": 89,

parser/testdata/ddl/output/create_materialized_view_basic.sql.golden.json

@@ -31,6 +31,7 @@
     "Settings": null,
     "HasAppend": false,
     "Engine": null,
+    "TableSchema": null,
     "HasEmpty": false,
     "Destination": {
       "ToPos": 78,