Parse compression codecs with no parameters (#197)

Compression codecs such as `DoubleDelta` do not require any additional positional parameters and infer defaults, such as `CODEC(DoubleDelta)`. This commit doesn't continue parsing a codec if there's no comma indicating additional parameters to the codec.

Author: swgillespie

parser/ast.go

@@ -4823,11 +4823,13 @@ func (c *CompressionCodec) String() string {
 		builder.WriteByte(',')
 		builder.WriteByte(' ')
 	}
-	builder.WriteString(c.Name.String())
-	if c.Level != nil {
-		builder.WriteByte('(')
-		builder.WriteString(c.Level.String())
-		builder.WriteByte(')')
+	if c.Name != nil {
+		builder.WriteString(c.Name.String())
+		if c.Level != nil {
+			builder.WriteByte('(')
+			builder.WriteString(c.Level.String())
+			builder.WriteByte(')')
+		}
 	}
 	builder.WriteByte(')')
 	return builder.String()
@@ -4836,16 +4838,20 @@ func (c *CompressionCodec) String() string {
 func (c *CompressionCodec) Accept(visitor ASTVisitor) error {
 	visitor.Enter(c)
 	defer visitor.Leave(c)
-	if err := c.Type.Accept(visitor); err != nil {
-		return err
+	if c.Type != nil {
+		if err := c.Type.Accept(visitor); err != nil {
+			return err
+		}
 	}
 	if c.TypeLevel != nil {
 		if err := c.TypeLevel.Accept(visitor); err != nil {
 			return err
 		}
 	}
-	if err := c.Name.Accept(visitor); err != nil {
-		return err
+	if c.Name != nil {
+		if err := c.Name.Accept(visitor); err != nil {
+			return err
+		}
 	}
 	if c.Level != nil {
 		if err := c.Level.Accept(visitor); err != nil {

parser/parser_column.go

@@ -1166,39 +1166,48 @@ func (p *Parser) tryParseCompressionCodecs(pos Pos) (*CompressionCodec, error) {
 	}
 
 	// parse codec name
-	name, err := p.parseIdent()
+	codecType, err := p.parseIdent()
 	if err != nil {
 		return nil, err
 	}
 	// parse DELTA if  CODEC(Delta, ZSTD(1))
 	// or CODEC(Delta(9), ZSTD(1)) or CODEC(T64, ZSTD(1))
-	var codecType *Ident
+	var name *Ident
 	var typeLevel *NumberLiteral
-	switch strings.ToUpper(name.Name) {
+	switch strings.ToUpper(codecType.Name) {
 	case "DELTA", "DOUBLEDELTA", "T64", "GORILLA":
-		codecType = name
 		// try parse delta level
 		typeLevel, err = p.tryParseCompressionLevel(p.Pos())
 		if err != nil {
 			return nil, err
 		}
-		// consume comma
-		if err := p.expectTokenKind(TokenKindComma); err != nil {
-			return nil, err
+
+		if p.matchTokenKind(TokenKindComma) {
+			if err := p.expectTokenKind(TokenKindComma); err != nil {
+				return nil, err
+			}
+			name, err = p.parseIdent()
+			if err != nil {
+				return nil, err
+			}
 		}
-		name, err = p.parseIdent()
+	case "ZSTD", "LZ4HC", "LH4":
+		// For compression codecs, try to parse level
+		typeLevel, err = p.tryParseCompressionLevel(p.Pos())
 		if err != nil {
 			return nil, err
 		}
 	}
 
 	var level *NumberLiteral
 	// TODO: check if the codec name is valid
-	switch strings.ToUpper(name.Name) {
-	case "ZSTD", "LZ4HC", "LH4":
-		level, err = p.tryParseCompressionLevel(p.Pos())
-		if err != nil {
-			return nil, err
+	if name != nil {
+		switch strings.ToUpper(name.Name) {
+		case "ZSTD", "LZ4HC", "LH4":
+			level, err = p.tryParseCompressionLevel(p.Pos())
+			if err != nil {
+				return nil, err
+			}
 		}
 	}
 
@@ -1207,6 +1216,21 @@ func (p *Parser) tryParseCompressionCodecs(pos Pos) (*CompressionCodec, error) {
 		return nil, err
 	}
 
+	// If there's only one codec (no second name), store it in Name field instead of Type
+	// This handles cases like CODEC(DoubleDelta) or CODEC(ZSTD(1))
+	if name == nil {
+		return &CompressionCodec{
+			CodecPos:      pos,
+			RightParenPos: rightParenPos,
+			Type:          nil,
+			TypeLevel:     nil,
+			Name:          codecType,
+			Level:         typeLevel,
+		}, nil
+	}
+
+	// Two codecs: Type is the preprocessing codec, Name is the compression codec
+	// e.g., CODEC(DoubleDelta, ZSTD(1))
 	return &CompressionCodec{
 		CodecPos:      pos,
 		RightParenPos: rightParenPos,

parser/testdata/ddl/create_table_codec_no_args.sql

@@ -0,0 +1,3 @@
+CREATE TABLE shark_attacks (
+    timestamp DateTime CODEC(DoubleDelta),
+);

parser/testdata/ddl/format/create_table_codec_no_args.sql

@@ -0,0 +1,7 @@
+-- Origin SQL:
+CREATE TABLE shark_attacks (
+    timestamp DateTime CODEC(DoubleDelta),
+);
+
+-- Format SQL:
+CREATE TABLE shark_attacks (timestamp DateTime CODEC(DoubleDelta));

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

@@ -0,0 +1,73 @@
+[
+  {
+    "CreatePos": 0,
+    "StatementEnd": 0,
+    "OrReplace": false,
+    "Name": {
+      "Database": null,
+      "Table": {
+        "Name": "shark_attacks",
+        "QuoteType": 1,
+        "NamePos": 13,
+        "NameEnd": 26
+      }
+    },
+    "IfNotExists": false,
+    "UUID": null,
+    "OnCluster": null,
+    "TableSchema": {
+      "SchemaPos": 27,
+      "SchemaEnd": 72,
+      "Columns": [
+        {
+          "NamePos": 33,
+          "ColumnEnd": 70,
+          "Name": {
+            "Ident": {
+              "Name": "timestamp",
+              "QuoteType": 1,
+              "NamePos": 33,
+              "NameEnd": 42
+            },
+            "DotIdent": null
+          },
+          "Type": {
+            "Name": {
+              "Name": "DateTime",
+              "QuoteType": 1,
+              "NamePos": 43,
+              "NameEnd": 51
+            }
+          },
+          "NotNull": null,
+          "Nullable": null,
+          "DefaultExpr": null,
+          "MaterializedExpr": null,
+          "AliasExpr": null,
+          "Codec": {
+            "CodecPos": 52,
+            "RightParenPos": 70,
+            "Type": null,
+            "TypeLevel": null,
+            "Name": {
+              "Name": "DoubleDelta",
+              "QuoteType": 1,
+              "NamePos": 58,
+              "NameEnd": 69
+            },
+            "Level": null
+          },
+          "TTL": null,
+          "Comment": null,
+          "CompressionCodec": null
+        }
+      ],
+      "AliasTable": null,
+      "TableFunction": null
+    },
+    "Engine": null,
+    "SubQuery": null,
+    "HasTemporary": false,
+    "Comment": null
+  }
+]