Add PostgreSQL PARTITION OF syntax support

[fmguerreiro]

Summary

This PR adds support for PostgreSQL's CREATE TABLE ... PARTITION OF syntax for creating child partition tables.

Closes#2042

Changes

• Added MODULUS and REMAINDER keywords to src/keywords.rs
• Added ForValues enum and PartitionBoundValue enum to src/ast/ddl.rs with Display implementations
• Added partition_of and for_values fields to CreateTable struct
• Updated CreateTableBuilder in src/ast/helpers/stmt_create_table.rs
• Added parser logic in src/parser/mod.rs for PARTITION OF, FOR VALUES IN/FROM/TO/WITH, and DEFAULT
• Updated src/ast/spans.rs to include new fields
• Exported new types from src/ast/mod.rs
• Added 7 comprehensive tests in tests/sqlparser_postgres.rs

Supported Syntax

-- RANGE partitionCREATETABLEt PARTITION OF parent FOR VALUESFROM ('2023-01-01') TO ('2024-01-01');
-- RANGE with MINVALUE/MAXVALUECREATETABLEt PARTITION OF parent FOR VALUESFROM (MINVALUE) TO ('2020-01-01');
-- LIST partitionCREATETABLEt PARTITION OF parent FOR VALUESIN ('US', 'CA', 'MX');
-- HASH partitionCREATETABLEt PARTITION OF parent FOR VALUES WITH (MODULUS 4, REMAINDER 0);
-- DEFAULT partitionCREATETABLEt PARTITION OF parent DEFAULT;
-- Multi-column rangeCREATETABLEt PARTITION OF parent FOR VALUESFROM ('2023-01-01', 1) TO ('2023-04-01', 1);
-- With table constraintsCREATETABLEt PARTITION OF parent (
CONSTRAINT check_date CHECK (order_date >='2023-01-01')
) FOR VALUESFROM ('2023-01-01') TO ('2024-01-01');

Testing

• All 7 new partition tests pass
• All 1312 existing tests pass (no regressions)
• cargo fmt passes
• Works with both PostgreSqlDialect and GenericDialect as required

[iffyio]

Suggested change

	// PostgreSQL PARTITION OF for child partition tables
	let partition_of = ifdialect_of!(self is PostgreSqlDialect | GenericDialect)
	&&self.parse_keywords(&[Keyword::PARTITION,Keyword::OF])
	let partition_of = ifself.parse_keywords(&[Keyword::PARTITION,Keyword::OF])

If possible we can skiip the dialect check and let the parser accept the PARTITION OF clause whenever it shows up in an input sql

[fmguerreiro]

0bbb684

[iffyio]

Suggested change

	/// Parse a single partition bound value (MINVALUE, MAXVALUE, or expression).
	/// Parse a single [PartitionBoundValue].

[fmguerreiro]

0c6857f

[iffyio]

Suggested change

	/// PostgreSQL partition bound specification for PARTITION OF.
	/// PostgreSQL partition bound specification for `PARTITION OF`.

[fmguerreiro]

0c6857f

[iffyio]

Suggested change

	/// Parse PostgreSQL partition bound specification for PARTITION OF.
	///
	/// Parses: `FOR VALUES partition_bound_spec | DEFAULT`
	/// Parse [ForValues] of a `PARTITION OF` clause.

[fmguerreiro]

0c6857f

[iffyio]

just in case @fmguerreiro please feel free to re-request review when ready

[iffyio]

If the intent is to test all dialects I think the test could live in sqlparser_common.rs and shouldn't explicitly list the dialects rather it can use all_dialects() - if the syntax conflicts across dialects (thinking probably snowflake for example) then I would suggest we just skip this test and rely on the pg_and_generic coverage after all

[fmguerreiro]

true, removed f74fd1a

[iffyio]

LGTM! Thanks @fmguerreiro!