SQL Dialect Converter to YQL

The SQL Dialect Converter is a service that transforms SQL queries written in other dialects (PostgreSQL, MySQL, ClickHouse, and others) into YQL syntax. The service is designed to simplify the migration of existing queries to YDB and lower the entry barrier for developers already familiar with other DBMS.

The converter is used in the DBeaver plugin for YDB on the Convert Dialect tab. The user pastes a query in the source dialect, selects the dialect from the list, and receives a version of the same query in YQL.

Principle of Operation

The transformation is performed in three stages:

1. Parsing the Source SQL. The source query is parsed by a dialect-specific parser into an abstract syntax tree (AST) — a structure that describes the meaning of the query independently of the syntax.
2. AST Transformation. Constructions specific to the source dialect are replaced with their equivalents in YQL. For example, type casting functions, work with dates and strings, LIMIT/OFFSET operators, and WITH constructions are adapted to YQL rules.
3. YQL Generation. A textual representation of the query in YQL is generated from the modified AST.

The stages are performed using the SQLGlot library, which includes a separate dialect module for YDB. SQLGlot is an open-source SQL parser and transpiler that supports over twenty dialects. It acts as both the parser for the source query and the generator for YQL.

Supported Dialects

The list of source dialects is determined by the set supported by SQLGlot. Among the main ones:

• PostgreSQL;
• MySQL;
• ClickHouse;
• Microsoft SQL Server (T-SQL);
• Oracle;
• Snowflake;
• BigQuery;
• Presto / Trino;
• Spark SQL / Databricks;
• SQLite.

The up-to-date full list of dialects can be obtained via the converter's API or viewed in the SQLGlot sources. The target dialect is always YQL.

Supported Constructions

The converter covers typical constructions of analytical and OLTP queries:

• Selection operators: SELECT, JOIN (all types), WHERE, GROUP BY, HAVING, ORDER BY, LIMIT, OFFSET.
• Window functions (OVER, PARTITION BY, ROWS BETWEEN).
• Subqueries and CTE (WITH).
• Type casting (CAST, ::), arithmetic and logical operations.
• Functions for strings, numbers, dates, and times — mapped to the closest equivalent in YQL.
• Aggregate functions (COUNT, SUM, AVG, MIN, MAX, and others).
• DML operations: INSERT, UPDATE, DELETE (taking into account YDB transaction features).

Limitations

Full automatic conversion is not always possible, as YQL and other SQL dialects differ in data models and operation semantics. In particular:

• Dialect-specific functions. Functions that do not have a direct equivalent in YQL (for example, PostgreSQL arrays, MySQL JSON_EXTRACT with special path syntax) are translated into approximate equivalents or left for manual refinement.
• Stored procedures and triggers. Not supported, as YDB uses a different code execution model in the database.
• Specific DDL. CREATE TABLE for complex types may require manual adjustments taking into account columnar and row tables in YDB.
• Optimizer hints. Ignored: the YDB optimizer has its own control mechanisms.

The conversion result should be considered a draft: complex queries require manual verification and adaptation before execution.

See Also

• DBeaver Plugin for YDB
• YQL Reference
• SQLGlot