feat(parser): ClickHouse SETTINGS/TTL/INSERT FORMAT tail clauses (#482)

pkg/sql/parser/clickhouse_settings_ttl_format_test.go

@@ -0,0 +1,56 @@
+// Copyright 2026 GoSQLX Authors
+//
+// Licensed under the Apache License, Version 2.0 (the "License");
+
+package parser_test
+
+import (
+	"testing"
+
+	"github.com/ajitpratap0/GoSQLX/pkg/gosqlx"
+	"github.com/ajitpratap0/GoSQLX/pkg/sql/keywords"
+)
+
+// TestClickHouseSettingsTTLFormat verifies ClickHouse tail clauses parse for
+// both CREATE TABLE and SELECT/INSERT. Regression for #482.
+func TestClickHouseSettingsTTLFormat(t *testing.T) {
+	queries := map[string]string{
+		"create_table_with_settings": `CREATE TABLE events (
+			id UInt64,
+			ts DateTime
+		) ENGINE = MergeTree()
+		ORDER BY ts
+		SETTINGS index_granularity = 8192, storage_policy = 'default'`,
+
+		"create_table_with_ttl": `CREATE TABLE logs (
+			event_date Date,
+			message String
+		) ENGINE = MergeTree()
+		ORDER BY event_date
+		TTL event_date + INTERVAL 90 DAY`,
+
+		"create_table_ttl_then_settings": `CREATE TABLE logs (
+			event_date Date,
+			message String
+		) ENGINE = MergeTree()
+		ORDER BY event_date
+		TTL event_date + INTERVAL 90 DAY
+		SETTINGS index_granularity = 8192`,
+
+		"select_with_settings": `SELECT * FROM events
+		WHERE id > 100
+		SETTINGS max_threads = 4, distributed_aggregation_memory_efficient = 1`,
+
+		"insert_format_values": `INSERT INTO events (id, name) VALUES (1, 'a') FORMAT Values`,
+
+		"insert_format_json_each_row": `INSERT INTO events FORMAT JSONEachRow`,
+	}
+	for name, q := range queries {
+		q := q
+		t.Run(name, func(t *testing.T) {
+			if _, err := gosqlx.ParseWithDialect(q, keywords.DialectClickHouse); err != nil {
+				t.Fatalf("parse failed: %v", err)
+			}
+		})
+	}
+}

pkg/sql/parser/ddl.go

@@ -297,6 +297,31 @@ func (p *Parser) parseCreateTable(temporary bool) (*ast.CreateTableStatement, er
 			}
 			continue
 		}
+		if p.isTokenMatch("TTL") {
+			p.advance()
+			if err := p.skipClickHouseClauseExpr(); err != nil {
+				return nil, err
+			}
+			continue
+		}
+		if p.isTokenMatch("SETTINGS") {
+			p.advance()
+			// SETTINGS is a comma-separated list of name=value assignments.
+			// Consume each k=v pair until the next clause, EOF, or ';'.
+			for {
+				t := p.currentToken.Token.Type
+				val := strings.ToUpper(p.currentToken.Token.Value)
+				if t == models.TokenTypeEOF || t == models.TokenTypeSemicolon {
+					break
+				}
+				if val == "ORDER" || val == "PARTITION" || val == "PRIMARY" ||
+					val == "SAMPLE" || val == "TTL" {
+					break
+				}
+				p.advance()
+			}
+			continue
+		}
 		break
 	}
 

pkg/sql/parser/dml_insert.go

@@ -83,6 +83,21 @@ func (p *Parser) parseInsertStatement() (ast.Statement, error) {
 	var values [][]ast.Expression
 	var query ast.QueryExpression
 
+	// ClickHouse-only: INSERT INTO t [(cols)] FORMAT <name> with the data
+	// payload being external (network / file). Short-circuit here — there is
+	// no VALUES or SELECT to follow.
+	if p.dialect == string(keywords.DialectClickHouse) &&
+		strings.EqualFold(p.currentToken.Token.Value, "FORMAT") {
+		p.advance() // Consume FORMAT
+		if p.isIdentifier() || p.isType(models.TokenTypeKeyword) {
+			p.advance() // Consume format name
+		}
+		return &ast.InsertStatement{
+			TableName: tableName,
+			Columns:   columns,
+		}, nil
+	}
+
 	switch {
 	case p.isType(models.TokenTypeSelect):
 		// INSERT ... SELECT syntax
@@ -190,6 +205,17 @@ func (p *Parser) parseInsertStatement() (ast.Statement, error) {
 		}
 	}
 
+	// ClickHouse INSERT ... VALUES (...) FORMAT <name> trailing format hint.
+	// Parse-only; the trailing payload is not consumed.
+	if p.dialect == string(keywords.DialectClickHouse) &&
+		strings.EqualFold(p.currentToken.Token.Value, "FORMAT") {
+		p.advance() // Consume FORMAT
+		if p.isIdentifier() || p.isType(models.TokenTypeKeyword) ||
+			p.isType(models.TokenTypeValues) {
+			p.advance() // Consume format name (may tokenize as VALUES keyword)
+		}
+	}
+
 	// Create INSERT statement
 	return &ast.InsertStatement{
 		TableName:      tableName,

pkg/sql/parser/parser.go

@@ -591,6 +591,19 @@ func (p *Parser) parseStatement() (ast.Statement, error) {
 				ss.Pos = stmtPos
 			}
 		}
+		// ClickHouse trailing SETTINGS k=v [, k=v]... on SELECT. Parse-only;
+		// the settings are consumed but not modeled on the AST.
+		if p.dialect == string(keywords.DialectClickHouse) && p.isTokenMatch("SETTINGS") {
+			p.advance() // SETTINGS
+			for {
+				t := p.currentToken.Token.Type
+				if t == models.TokenTypeEOF || t == models.TokenTypeSemicolon ||
+					t == models.TokenTypeRParen {
+					break
+				}
+				p.advance()
+			}
+		}
 		return stmt, nil
 	case models.TokenTypeInsert:
 		stmtPos := p.currentLocation()