Fix 10 parser tests with various EXPLAIN AST improvements

internal/explain/expressions.go

@@ -49,12 +49,27 @@ func explainLiteral(sb *strings.Builder, n *ast.Literal, indent string, depth in
 			}
 			hasComplexExpr := false
 			for _, e := range exprs {
-				lit, isLit := e.(*ast.Literal)
-				// Non-literals or tuple/array literals count as complex
-				if !isLit || (isLit && (lit.Type == ast.LiteralTuple || lit.Type == ast.LiteralArray)) {
-					hasComplexExpr = true
-					break
+				// Simple literals (numbers, strings, etc.) are OK
+				if lit, isLit := e.(*ast.Literal); isLit {
+					// Nested tuples/arrays are complex
+					if lit.Type == ast.LiteralTuple || lit.Type == ast.LiteralArray {
+						hasComplexExpr = true
+						break
+					}
+					// Other literals are simple
+					continue
+				}
+				// Unary negation of numeric literals is also simple
+				if unary, isUnary := e.(*ast.UnaryExpr); isUnary && unary.Op == "-" {
+					if lit, isLit := unary.Operand.(*ast.Literal); isLit {
+						if lit.Type == ast.LiteralInteger || lit.Type == ast.LiteralFloat {
+							continue
+						}
+					}
 				}
+				// Everything else is complex
+				hasComplexExpr = true
+				break
 			}
 			if hasComplexExpr {
 				// Render as Function tuple instead of Literal
@@ -329,6 +344,33 @@ func explainAliasedExpr(sb *strings.Builder, n *ast.AliasedExpr, depth int) {
 				}
 			}
 		}
+		// Check if this is an array containing specific expressions that need Function array format
+		if e.Type == ast.LiteralArray {
+			if exprs, ok := e.Value.([]ast.Expression); ok {
+				needsFunctionFormat := false
+				for _, expr := range exprs {
+					// Check for tuples - use Function array
+					if lit, ok := expr.(*ast.Literal); ok && lit.Type == ast.LiteralTuple {
+						needsFunctionFormat = true
+						break
+					}
+					// Check for identifiers - use Function array
+					if _, ok := expr.(*ast.Identifier); ok {
+						needsFunctionFormat = true
+						break
+					}
+				}
+				if needsFunctionFormat {
+					// Render as Function array with alias
+					fmt.Fprintf(sb, "%sFunction array (alias %s) (children %d)\n", indent, n.Alias, 1)
+					fmt.Fprintf(sb, "%s ExpressionList (children %d)\n", indent, len(exprs))
+					for _, expr := range exprs {
+						Node(sb, expr, depth+2)
+					}
+					return
+				}
+			}
+		}
 		fmt.Fprintf(sb, "%sLiteral %s (alias %s)\n", indent, FormatLiteral(e), n.Alias)
 	case *ast.BinaryExpr:
 		// Binary expressions become functions with alias
@@ -450,6 +492,8 @@ func explainWithElement(sb *strings.Builder, n *ast.WithElement, indent string,
 			fmt.Fprintf(sb, "%sSubquery (children %d)\n", indent, 1)
 		}
 		Node(sb, e.Query, depth+1)
+	case *ast.CastExpr:
+		explainCastExprWithAlias(sb, e, n.Name, indent, depth)
 	default:
 		// For other types, just output the expression (alias may be lost)
 		Node(sb, n.Query, depth)

internal/explain/format.go

@@ -21,7 +21,14 @@ func FormatFloat(val float64) string {
 	if math.IsNaN(val) {
 		return "nan"
 	}
-	// Use 'f' format to avoid scientific notation, -1 precision for smallest representation
+	// Use scientific notation for extremely small numbers (< 1e-10)
+	// This matches ClickHouse's behavior where numbers like 0.000001 stay decimal
+	// but extremely small numbers like 1e-38 use scientific notation
+	absVal := math.Abs(val)
+	if absVal > 0 && absVal < 1e-10 {
+		return strconv.FormatFloat(val, 'e', -1, 64)
+	}
+	// Use decimal notation for normal-sized numbers
 	return strconv.FormatFloat(val, 'f', -1, 64)
 }
 
@@ -218,35 +225,65 @@ func FormatDataType(dt *ast.DataType) string {
 		} else if ntp, ok := p.(*ast.NameTypePair); ok {
 			// Named tuple field: "name Type"
 			params = append(params, ntp.Name+" "+FormatDataType(ntp.Type))
+		} else if binExpr, ok := p.(*ast.BinaryExpr); ok {
+			// Binary expression (e.g., 'hello' = 1 for Enum types)
+			params = append(params, formatBinaryExprForType(binExpr))
 		} else {
 			params = append(params, fmt.Sprintf("%v", p))
 		}
 	}
 	return fmt.Sprintf("%s(%s)", dt.Name, strings.Join(params, ", "))
 }
 
+// formatBinaryExprForType formats a binary expression for use in type parameters
+func formatBinaryExprForType(expr *ast.BinaryExpr) string {
+	var left, right string
+
+	// Format left side
+	if lit, ok := expr.Left.(*ast.Literal); ok {
+		if lit.Type == ast.LiteralString {
+			left = fmt.Sprintf("\\\\\\'%s\\\\\\'", lit.Value)
+		} else {
+			left = fmt.Sprintf("%v", lit.Value)
+		}
+	} else if ident, ok := expr.Left.(*ast.Identifier); ok {
+		left = ident.Name()
+	} else {
+		left = fmt.Sprintf("%v", expr.Left)
+	}
+
+	// Format right side
+	if lit, ok := expr.Right.(*ast.Literal); ok {
+		right = fmt.Sprintf("%v", lit.Value)
+	} else if ident, ok := expr.Right.(*ast.Identifier); ok {
+		right = ident.Name()
+	} else {
+		right = fmt.Sprintf("%v", expr.Right)
+	}
+
+	return left + " " + expr.Op + " " + right
+}
+
 // NormalizeFunctionName normalizes function names to match ClickHouse's EXPLAIN AST output
 func NormalizeFunctionName(name string) string {
 	// ClickHouse normalizes certain function names in EXPLAIN AST
 	normalized := map[string]string{
-		"ltrim":       "trimLeft",
-		"rtrim":       "trimRight",
-		"lcase":       "lower",
-		"ucase":       "upper",
-		"mid":         "substring",
-		"ceiling":     "ceil",
-		"ln":          "log",
-		"log10":       "log10",
-		"log2":        "log2",
-		"rand":        "rand",
-		"ifnull":      "ifNull",
-		"nullif":      "nullIf",
-		"coalesce":    "coalesce",
-		"greatest":    "greatest",
-		"least":       "least",
-		"concat_ws":   "concat",
-		"length":      "length",
-		"char_length": "length",
+		"ltrim":      "trimLeft",
+		"rtrim":      "trimRight",
+		"lcase":      "lower",
+		"ucase":      "upper",
+		"mid":        "substring",
+		"ceiling":    "ceil",
+		"ln":         "log",
+		"log10":      "log10",
+		"log2":       "log2",
+		"rand":       "rand",
+		"ifnull":     "ifNull",
+		"nullif":     "nullIf",
+		"coalesce":   "coalesce",
+		"greatest":   "greatest",
+		"least":      "least",
+		"concat_ws":  "concat",
 	}
 	if n, ok := normalized[strings.ToLower(name)]; ok {
 		return n

internal/explain/functions.go

@@ -21,6 +21,10 @@ func explainFunctionCallWithAlias(sb *strings.Builder, n *ast.FunctionCall, alia
 	}
 	// Normalize function name
 	fnName := NormalizeFunctionName(n.Name)
+	// Append "Distinct" if the function has DISTINCT modifier
+	if n.Distinct {
+		fnName = fnName + "Distinct"
+	}
 	if alias != "" {
 		fmt.Fprintf(sb, "%sFunction %s (alias %s) (children %d)\n", indent, fnName, alias, children)
 	} else {
@@ -774,7 +778,6 @@ func explainExtractExpr(sb *strings.Builder, n *ast.ExtractExpr, indent string,
 func explainWindowSpec(sb *strings.Builder, n *ast.WindowSpec, indent string, depth int) {
 	// Window spec is represented as WindowDefinition
 	// For simple cases like OVER (), just output WindowDefinition without children
-	// Note: ClickHouse's EXPLAIN AST does not output frame info (ROWS BETWEEN etc)
 	children := 0
 	if n.Name != "" {
 		children++
@@ -785,6 +788,10 @@ func explainWindowSpec(sb *strings.Builder, n *ast.WindowSpec, indent string, de
 	if len(n.OrderBy) > 0 {
 		children++
 	}
+	// Count frame offset as child if present
+	if n.Frame != nil && n.Frame.StartBound != nil && n.Frame.StartBound.Offset != nil {
+		children++
+	}
 	if children > 0 {
 		fmt.Fprintf(sb, "%sWindowDefinition (children %d)\n", indent, children)
 		if n.Name != "" {
@@ -802,7 +809,10 @@ func explainWindowSpec(sb *strings.Builder, n *ast.WindowSpec, indent string, de
 				explainOrderByElement(sb, o, strings.Repeat(" ", depth+2), depth+2)
 			}
 		}
-		// Frame handling would go here if needed
+		// Frame start offset
+		if n.Frame != nil && n.Frame.StartBound != nil && n.Frame.StartBound.Offset != nil {
+			Node(sb, n.Frame.StartBound.Offset, depth+1)
+		}
 	} else {
 		fmt.Fprintf(sb, "%sWindowDefinition\n", indent)
 	}

lexer/lexer.go

@@ -429,7 +429,19 @@ func (l *Lexer) readQuotedIdentifier() Item {
 	var sb strings.Builder
 	l.readChar() // skip opening quote
 
-	for !l.eof && l.ch != '"' {
+	for !l.eof {
+		if l.ch == '"' {
+			// Check for SQL-style doubled quote escape ""
+			l.readChar()
+			if l.ch == '"' {
+				// Doubled quote - add single quote and continue
+				sb.WriteRune('"')
+				l.readChar()
+				continue
+			}
+			// Single quote - end of identifier
+			break
+		}
 		if l.ch == '\\' {
 			l.readChar()
 			if !l.eof {
@@ -441,9 +453,6 @@ func (l *Lexer) readQuotedIdentifier() Item {
 		sb.WriteRune(l.ch)
 		l.readChar()
 	}
-	if l.ch == '"' {
-		l.readChar() // skip closing quote
-	}
 	return Item{Token: token.IDENT, Value: sb.String(), Pos: pos}
 }
 

parser/testdata/00160_decode_xml_component/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00267_tuple_array_access_operators_priority/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00317_in_tuples_and_out_of_range_values/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00324_hashing_enums/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00471_sql_style_quoting/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00502_string_concat_with_array/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00511_get_size_of_enum/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00526_array_join_with_arrays_of_nullable/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00548_slice_of_nested/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00566_enum_min_max/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00642_cast/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00674_has_array_enum/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/00726_length_aliases/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/01259_combinator_distinct/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/01353_topk_enum/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/01651_group_uniq_array_enum/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/01700_point_in_polygon_ubsan/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/01717_global_with_subquery_fix/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/01914_ubsan_quantile_timing/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/02158_interval_length_sum/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/02306_window_move_row_number_fix/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/02467_cross_join_three_table_functions/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{"todo": true}

parser/testdata/02476_fix_cast_parser_bug/metadata.json

@@ -1 +1 @@
-{"todo": true, "parse_error": true}
+{"todo": true, "parse_error": true}

parser/testdata/02958_transform_enum/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/03003_enum_and_string_compatible/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}

parser/testdata/03747_float_parsing_subnormal/metadata.json

@@ -1 +1 @@
-{"todo": true}
+{}