Spark SQL: Add SparkSqlDialect

[andygrove]

Summary

• Adds SparkSqlDialect (src/dialect/spark.rs) with support for the key Spark SQL syntax features needed to parse real-world Spark SQL workloads
• Adds tests/sqlparser_spark.rs with tests based on tests from Comet project

New Dialect trait methods (all default false)

Method	Purpose
supports_create_table_using	CREATE TABLE t (...) USING parquet
supports_long_type_as_bigint	LONG as alias for BIGINT
supports_map_literal_with_angle_brackets	MAP<K, V> type syntax

Parser/AST changes

• New HiveIOFormat::Using { format: Ident } variant — renders as USING <format>, distinct from Hive's STORED AS <format>
• parse_hive_formats handles USING <format> when dialect opts in
• STRUCT type parsing now uses supports_struct_literal() trait method instead of dialect_is!(BigQueryDialect | DatabricksDialect | GenericDialect) — cleaner and extensible
• New MAP<K, V> angle-bracket parsing path in parse_data_type_helper

SparkSqlDialect capabilities

USING <format> in CREATE TABLE, lambda functions, DIV integer division, aggregate FILTER, GROUP BY expressions/modifiers, SELECT * EXCEPT, STRUCT<> and MAP<> types, LONG type alias, nested comments, ! as NOT, CTE without AS, multi-column aliases, IGNORE NULLS in window functions (already supported by the generic parser).

🤖 Generated with Claude Code