# SQLGlot — SQL Parser, Transpiler & Optimizer in Pure Python

> SQLGlot is a no-dependency Python library that parses, transpiles, and optimizes SQL across 20+ dialects. Convert queries between Snowflake, BigQuery, DuckDB, Spark, Postgres, and more without touching the database.

## Install

Save as a script file and run:

# SQLGlot — SQL Parser, Transpiler & Optimizer in Pure Python

## Quick Use
```bash
pip install sqlglot
python -c "
import sqlglot
# Transpile Spark SQL to BigQuery
print(sqlglot.transpile('SELECT EPOCH(ts) FROM t', read='spark', write='bigquery')[0])
# Parse into AST
ast = sqlglot.parse_one('SELECT a, b FROM t WHERE x > 1')
print(ast)
"
```

## Introduction
SQLGlot solves the universal pain of SQL dialect differences. When migrating between data platforms or building tools that must understand SQL, hand-translating queries is error-prone and slow. SQLGlot parses SQL into an AST, rewrites dialect-specific syntax, and outputs clean SQL for the target engine — all in pure Python with zero native dependencies.

## What SQLGlot Does
- Parses SQL from 20+ dialects into a typed abstract syntax tree
- Transpiles queries between dialects (Snowflake to BigQuery, Spark to DuckDB, etc.)
- Optimizes queries by simplifying expressions, pushing down predicates, and eliminating subqueries
- Formats SQL with consistent style and indentation
- Provides a Python API for programmatic SQL generation and transformation

## Architecture Overview
SQLGlot's core is a recursive-descent parser that produces a tree of expression nodes. Each SQL dialect is defined as a subclass that overrides tokenization and generation rules. The optimizer applies relational algebra transformations (predicate pushdown, column pruning, join reordering) on the AST. The generator walks the tree and emits dialect-specific SQL. Everything is pure Python with no C extensions or external dependencies.

## Self-Hosting & Configuration
- Install with pip install sqlglot — no system dependencies required
- Use sqlglot.transpile(sql, read=source, write=target) for quick conversions
- Parse SQL into an AST with sqlglot.parse_one() for programmatic manipulation
- Apply the optimizer with sqlglot.optimizer.optimize() for query simplification
- Extend with custom dialects by subclassing the Dialect class

## Key Features
- Supports 20+ SQL dialects including Snowflake, BigQuery, Spark, Presto, DuckDB, MySQL, and Postgres
- Zero dependencies — pure Python, easy to embed anywhere
- Full AST with typed expression nodes for safe programmatic SQL manipulation
- Query optimizer with predicate pushdown, constant folding, and scope analysis
- Semantic diff tool for comparing SQL queries structurally

## Comparison with Similar Tools
- **sqlparse** — tokenizer and formatter only; SQLGlot builds a full typed AST and transpiles
- **sqlfluff** — SQL linter focused on style; SQLGlot is a parser, transpiler, and optimizer
- **Apache Calcite** — JVM-based SQL framework; SQLGlot is pure Python and lighter for scripting
- **jOOQ** — Java SQL builder for runtime query generation; SQLGlot is a static analysis and transpilation tool
- **dbt** — SQL templating for warehouses; SQLGlot operates on raw SQL syntax without execution

## FAQ
**Q: How accurate is the transpilation?**
A: SQLGlot handles the vast majority of common syntax differences. Edge cases in proprietary functions may need custom mappings, but the dialect system makes extending straightforward.

**Q: Can SQLGlot execute SQL?**
A: SQLGlot includes an in-memory executor for testing, but it is primarily a parsing and transpilation library, not a database engine.

**Q: Is it fast enough for production use?**
A: Yes. SQLGlot parses thousands of queries per second in pure Python. Tools like Tobiko's SQLMesh and multiple data catalogs use it in production.

**Q: Does it support stored procedures or DDL?**
A: SQLGlot supports DDL (CREATE, ALTER, DROP) and many procedural extensions. Coverage varies by dialect.

## Sources
- https://github.com/tobymao/sqlglot
- https://sqlglot.com/sqlglot.html

---
Source: https://tokrepo.com/en/workflows/16b7b083-39ec-11f1-9bc6-00163e2b0d79
Author: Script Depot