PRQL — Pipelined Relational Query Language

Introduction

PRQL (pronounced "prequel") is a modern query language that compiles to SQL. It was created to address the readability and composability limitations of SQL by introducing a pipelined, functional syntax that makes data transformations more intuitive and maintainable.

What PRQL Does

• Compiles a readable pipeline syntax into standard SQL for any database
• Supports variables, functions, and modular query composition
• Provides a type system that catches errors before queries reach the database
• Integrates with Python, JavaScript, Rust, and other languages via bindings
• Offers editor extensions with syntax highlighting and live SQL preview

Architecture Overview

PRQL is built as a compiler in Rust. Source code passes through a lexer, parser, semantic analyzer, and finally a SQL backend that emits dialect-specific output. The compiler is packaged as a CLI tool (prqlc), a Rust library, and WASM bindings that power integrations for Python (prqlc-python), JavaScript, and database tools like DBeaver and Jupyter.

Self-Hosting & Configuration

• Install the CLI via pip, cargo, brew, or npm depending on your ecosystem
• Use prqlc compile to convert .prql files to .sql as part of your CI pipeline
• Integrate with dbt via the dbt-prql plugin for transformation workflows
• Configure target SQL dialect (PostgreSQL, MySQL, SQLite, BigQuery, etc.) with the --target flag
• Embed in Jupyter notebooks using the %%prql magic command from pyprql

Key Features

• Pipeline syntax reads top-to-bottom like data actually flows
• S-strings allow raw SQL escape hatches when needed
• Automatic column tracking eliminates repetitive GROUP BY clauses
• Dialect-aware compilation targets 15+ SQL databases
• Composable functions let you build reusable query libraries

Comparison with Similar Tools

• SQL — PRQL compiles to SQL; it is not a replacement but a more ergonomic authoring layer
• Malloy — Malloy is a semantic modeling language; PRQL focuses on ad-hoc query composition
• dbt — dbt orchestrates SQL transformations; PRQL can be used inside dbt via a plugin
• LINQ — LINQ is embedded in .NET languages; PRQL is language-agnostic with multi-platform bindings
• Kusto (KQL) — KQL is Azure-specific; PRQL targets any SQL database

FAQ

Q: Can I use PRQL with my existing database? A: Yes. PRQL compiles to standard SQL, so it works with PostgreSQL, MySQL, SQLite, BigQuery, ClickHouse, DuckDB, and many others.

Q: Does PRQL support joins? A: Yes. PRQL supports join operations with a cleaner syntax than SQL, including automatic column disambiguation.

Q: Is PRQL production-ready? A: PRQL is at version 0.x but is actively used in production data pipelines. The compiler is well-tested with thousands of test cases.

Q: Can I mix PRQL and raw SQL? A: Yes. S-strings let you embed raw SQL expressions anywhere in a PRQL query for cases the language does not yet cover.

Sources

• https://github.com/PRQL/prql
• https://prql-lang.org