tree-sitter-sql-polyglot

Appearance

tree-sitter-sql-polyglotA multi-dialect SQL grammar for tree-sitter

An ANSI SQL base plus 22 independently compiled dialect grammars, layered with tree-sitter's grammar(parent, overrides) composition.

Dialect coverage
View on GitHub
Downloads

22 dialects on a shared base

postgres, mysql, mariadb, oracle, db2, tsql, bigquery, snowflake, redshift, sqlite, duckdb, trino, athena, clickhouse, flink, hive, spark, databricks, cockroachdb, spanner, teradata, and hana. Each extends the ANSI base or a real parent dialect.

Inheritance follows real dialects

Each dialect extends its real-world parent (mariadb -> mysql, databricks -> spark -> hive, cockroachdb -> postgres, spanner -> bigquery) rather than sitting in a flat list, so shared syntax is written once.

Coverage checked against other parsers

Every feature probe is parsed by this grammar and by SQLGlot, ANTLR grammars-v4, pglast and sqlfluff. A feature counts as covered only when an outside parser agrees.

What this is ​

The grammar restructures the upstream DerekStride/tree-sitter-sql "permissive" grammar into a strict ANSI SQL base plus dialect extensions. Each dialect compiles to its own <dialect>/src/parser.c and can be used independently. The README has the full dialect table and install instructions, and AGENTS.md covers the grammar architecture.

  • Usage, install/import snippets for Rust, Node.js, Python, Go, Swift, and CMake/C, plus how per-dialect lazy loading works in each.
  • Dialect coverage, generated on every CI run, with per-dialect scores, the inheritance tree and the full feature-by-dialect matrix.
  • Changelog, the release history.
  • Downloads, the parser build artifacts (grammar sources, bindings, queries) mirrored from the latest main.