kysely-converter（凯瑟利转换器）

Kysely Converter Capabilities

This agent is capable of transforming SQL queries into idiomatic Kysely TypeScript code. It understands SQL syntax and maps it to the corresponding Kysely query builder methods.

Core Capabilities

• SQL Parsing: Understands raw SQL structure including CTEs, subqueries, and complex clauses.
• Kysely API Mapping: Maps SQL keywords and clauses to specific Kysely methods (e.g., SELECT -> .select(), WHERE -> .where()).
• Type-Safe Construction: Generates code that utilizes Kysely's type inference capabilities.
• Dialect Handling: Adapts conversion strategies for specific SQL dialects (PostgreSQL, MySQL).

Supported Query Types

SELECT Queries

• Column Selection: Capable of selecting specific columns, all columns (*), and handling table prefixes.
• Aliasing: Handles column and table aliases using as syntax.
• Distinct: Supports DISTINCT and PostgreSQL-specific DISTINCT ON.
• Joins: Converts INNER JOIN, LEFT JOIN, RIGHT JOIN, and other join types with complex ON conditions.
• Filtering:

- Handles basic comparison operators (=, >, <, etc.). - Supports complex boolean logic (AND, OR) using expression builders. - Manages NULL checks (IS NULL, IS NOT NULL). - Handles IN clauses and pattern matching (LIKE).

• Aggregation: Converts GROUP BY and HAVING clauses with aggregate functions (COUNT, SUM, AVG, MAX, MIN).
• Ordering & Pagination: Maps ORDER BY (asc/desc), LIMIT, and OFFSET.

INSERT Operations

• Single & Batch Insert: Can convert single-row and multi-row value insertions.
• Return Values: Handles RETURNING clauses for PostgreSQL to return inserted data.

UPDATE Operations

• Set Clauses: Converts SET assignments for simple values and expressions (e.g., incrementing a counter).
• Complex Updates: Supports updates involving subqueries or complex WHERE conditions.

DELETE Operations

• Basic Deletion: Maps standard DELETE FROM statements with conditions.
• Return Values: Handles RETURNING clauses for deleted rows (PostgreSQL).

Advanced Features

Window Functions

• Ranking Functions: ROW_NUMBER, RANK, DENSE_RANK, NTILE, PERCENT_RANK, CUME_DIST.
• Value Functions: LAG, LEAD, FIRST_VALUE, LAST_VALUE, NTH_VALUE.
• Aggregate Windows: Windowed versions of AVG, COUNT, MAX, MIN, SUM.
• Window Clauses: Correctly constructs OVER clauses with PARTITION BY, ORDER BY, and frame specifications using the Kysely expression builder (eb.fn.agg).

Expressions & Logic

• CASE Statements: Converts CASE WHEN... THEN... ELSE logic into Kysely's eb.case() chain.
• Raw SQL: Identifies when sql template tags are needed for unsupported or complex raw fragments.

Reference Patterns

The agent applies specific patterns found in the references/ directory:

• select.md: Query structure and clause mapping.
• insert.md: Insertion patterns and return value handling.
• update.md: Update logic and assignment expressions.
• delete.md: Deletion mapping.
• window_function.md: Complex window function construction using the expression builder.