Add ObjectTypeArgument AST type for JSON/OBJECT type parameters (#99)

Author: kyleconroy

ast/ast.go

@@ -390,6 +390,17 @@ func (d *DataType) Pos() token.Position { return d.Position }
 func (d *DataType) End() token.Position { return d.Position }
 func (d *DataType) expressionNode()     {}
 
+// ObjectTypeArgument wraps an expression that is an argument to JSON/OBJECT types.
+// This matches ClickHouse's ASTObjectTypeArgument node structure.
+type ObjectTypeArgument struct {
+	Position token.Position `json:"-"`
+	Expr     Expression     `json:"expr"`
+}
+
+func (o *ObjectTypeArgument) Pos() token.Position { return o.Position }
+func (o *ObjectTypeArgument) End() token.Position { return o.Position }
+func (o *ObjectTypeArgument) expressionNode()     {}
+
 // NameTypePair represents a named type pair, used in Nested types.
 type NameTypePair struct {
 	Position token.Position `json:"-"`

internal/explain/explain.go

@@ -155,6 +155,8 @@ func Node(sb *strings.Builder, node interface{}, depth int) {
 	// Types
 	case *ast.DataType:
 		explainDataType(sb, n, indent, depth)
+	case *ast.ObjectTypeArgument:
+		explainObjectTypeArgument(sb, n, indent, depth)
 	case *ast.NameTypePair:
 		explainNameTypePair(sb, n, indent, depth)
 	case *ast.Parameter:

internal/explain/format.go

@@ -218,6 +218,10 @@ func FormatDataType(dt *ast.DataType) string {
 	}
 	var params []string
 	for _, p := range dt.Parameters {
+		// Unwrap ObjectTypeArgument if present (used for JSON/OBJECT types)
+		if ota, ok := p.(*ast.ObjectTypeArgument); ok {
+			p = ota.Expr
+		}
 		if lit, ok := p.(*ast.Literal); ok {
 			if lit.Type == ast.LiteralString {
 				// String parameters in type need extra escaping: 'val' -> \\\'val\\\'

internal/explain/statements.go

@@ -671,6 +671,11 @@ func explainDataType(sb *strings.Builder, n *ast.DataType, indent string, depth
 	}
 }
 
+func explainObjectTypeArgument(sb *strings.Builder, n *ast.ObjectTypeArgument, indent string, depth int) {
+	fmt.Fprintf(sb, "%sASTObjectTypeArgument (children %d)\n", indent, 1)
+	Node(sb, n.Expr, depth+1)
+}
+
 func explainNameTypePair(sb *strings.Builder, n *ast.NameTypePair, indent string, depth int) {
 	fmt.Fprintf(sb, "%sNameTypePair %s (children %d)\n", indent, n.Name, 1)
 	Node(sb, n.Type, depth+1)

parser/parser.go

@@ -2413,7 +2413,9 @@ func (p *Parser) parseDataType() *ast.DataType {
 
 		// Determine if this type uses named parameters (Nested, Tuple, JSON)
 		upperName := strings.ToUpper(dt.Name)
-		usesNamedParams := upperName == "NESTED" || upperName == "TUPLE" || upperName == "JSON"
+		usesNamedParams := upperName == "NESTED" || upperName == "TUPLE" || upperName == "JSON" || upperName == "OBJECT"
+		// JSON and OBJECT types wrap their parameters in ObjectTypeArgument
+		isObjectType := upperName == "JSON" || upperName == "OBJECT"
 
 		// Parse type parameters, but stop on keywords that can't be part of type params
 		for !p.currentIs(token.RPAREN) && !p.currentIs(token.EOF) && !p.currentIs(token.COLLATE) {
@@ -2422,10 +2424,11 @@ func (p *Parser) parseDataType() *ast.DataType {
 			isNamedParam := false
 			if usesNamedParams && (p.currentIs(token.IDENT) || p.current.Token.IsKeyword()) {
 				// Check if current is NOT a type name and peek IS a type name or LPAREN follows for complex types
-				if !p.isDataTypeName(p.current.Value) {
+				// But NOT if peek is '=' which indicates an expression like max_dynamic_paths=8
+				if !p.isDataTypeName(p.current.Value) && !p.peekIs(token.EQ) {
 					// Current is a name (not a type), next should be a type
 					isNamedParam = true
-				} else if p.peekIs(token.IDENT) || p.peekIs(token.LPAREN) {
+				} else if !p.peekIs(token.EQ) && (p.peekIs(token.IDENT) || p.peekIs(token.LPAREN)) {
 					// Current looks like a type name but is followed by another identifier
 					// This happens with things like "a Tuple(...)" where "a" looks like it could be a type
 					// Check if peek is a known type name
@@ -2441,6 +2444,7 @@ func (p *Parser) parseDataType() *ast.DataType {
 				}
 			}
 
+			var param ast.Expression
 			if isNamedParam {
 				// Parse as name + type pair
 				pos := p.current.Pos
@@ -2449,19 +2453,29 @@ func (p *Parser) parseDataType() *ast.DataType {
 				// Parse the type for this parameter
 				paramType := p.parseDataType()
 				if paramType != nil {
-					ntp := &ast.NameTypePair{
+					param = &ast.NameTypePair{
 						Position: pos,
 						Name:     paramName,
 						Type:     paramType,
 					}
-					dt.Parameters = append(dt.Parameters, ntp)
 				}
 			} else if (p.currentIs(token.IDENT) || p.current.Token.IsKeyword()) && p.isDataTypeName(p.current.Value) {
 				// It's a type name, parse as data type
-				dt.Parameters = append(dt.Parameters, p.parseDataType())
+				param = p.parseDataType()
 			} else {
 				// Parse as expression (for things like Decimal(10, 2))
-				dt.Parameters = append(dt.Parameters, p.parseExpression(LOWEST))
+				param = p.parseExpression(LOWEST)
+			}
+
+			// Wrap in ObjectTypeArgument for JSON/OBJECT types
+			if param != nil {
+				if isObjectType {
+					param = &ast.ObjectTypeArgument{
+						Position: param.Pos(),
+						Expr:     param,
+					}
+				}
+				dt.Parameters = append(dt.Parameters, param)
 			}
 
 			if p.currentIs(token.COMMA) {

parser/testdata/03100_lwu_36_json_skip_indexes/metadata.json

@@ -4,7 +4,6 @@
     "stmt12": true,
     "stmt14": true,
     "stmt16": true,
-    "stmt3": true,
     "stmt8": true
   }
 }

parser/testdata/03205_json_cast_from_string/metadata.json

@@ -3,12 +3,6 @@
     "stmt10": true,
     "stmt11": true,
     "stmt12": true,
-    "stmt14": true,
-    "stmt15": true,
-    "stmt16": true,
-    "stmt17": true,
-    "stmt18": true,
-    "stmt19": true,
     "stmt5": true,
     "stmt6": true,
     "stmt7": true,

parser/testdata/03205_json_syntax/metadata.json

@@ -1,7 +1,5 @@
 {
   "explain_todo": {
-    "stmt11": true,
-    "stmt13": true,
     "stmt15": true,
     "stmt17": true,
     "stmt19": true,
@@ -13,9 +11,6 @@
     "stmt31": true,
     "stmt33": true,
     "stmt35": true,
-    "stmt37": true,
-    "stmt5": true,
-    "stmt7": true,
-    "stmt9": true
+    "stmt37": true
   }
 }

parser/testdata/03214_json_typed_dynamic_path/metadata.json

@@ -1,6 +1 @@
-{
-  "explain_todo": {
-    "stmt10": true,
-    "stmt4": true
-  }
-}
+{}

parser/testdata/03222_json_squashing/metadata.json

@@ -1,19 +1,11 @@
 {
   "explain_todo": {
-    "stmt21": true,
-    "stmt22": true,
-    "stmt30": true,
-    "stmt38": true,
-    "stmt46": true,
-    "stmt47": true,
     "stmt49": true,
     "stmt51": true,
     "stmt53": true,
-    "stmt55": true,
     "stmt57": true,
     "stmt59": true,
     "stmt61": true,
-    "stmt63": true,
     "stmt65": true,
     "stmt67": true,
     "stmt69": true