---
id: datafusion-query-builder
version: "0.2.0"
license: MIT
license_treatment: permissive
maintenance: active
---

# datafusion-query-builder — Programmatic, injection-safe builder for DataFusion SQL.

License: permissive · Maintenance: active · Downloads: 104.4K/mo

## What it is and what it does

datafusion-query-builder is a typed SQL query builder for DataFusion, written in Rust with a Python API exposed via PyO3. It replaces hand-rolled SQL strings with a composable, type-safe interface that automatically escapes values by construction—untrusted input like column values or user-supplied filters are safe to embed without manual quoting. The builder emits actual SQL text (not an AST or opaque object), so queries remain visible in logs, greppable, and cache-keyable.

The library provides a façade layer over sqlparser, handling expressions, filters, selections, grouping, and ordering through method chaining. Python scalars auto-promote to literals; specialized operators like JSONB key-exists checks (`has_key`, `has_any_key`, `has_all_keys`) are first-class. An explicit `raw()` escape hatch and `param()` placeholders allow stepping outside the standard grammar when needed. The crate is correctness-critical and tested against DataFusion itself as an oracle to catch silent encoding errors.

Use it for:

- Build dynamic SQL filters from user input without manual escaping or string concatenation.
- Generate analytics queries programmatically while keeping SQL visible in application logs.
- Construct complex DataFusion queries with type safety and composable method chaining.
- Embed untrusted column names or filter values safely without SQL injection risk.
- Test query generation logic with deterministic, re-parseable SQL output.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Builds DataFusion SQL queries programmatically with injection-safe value handling, emitting readable SQL text rather than requiring string concatenation or templates.

Yes, if you build DataFusion queries from user input or dynamic parameters. The injection-safe-by-construction design and readable SQL output make it worth the medium install friction. Alpha status and 26-day-old release suggest active development; verify dialect coverage for your specific query patterns before production use.

## Install

pip install datafusion-query-builder
uv add datafusion-query-builder
poetry add datafusion-query-builder

## Installing datafusion-query-builder

Before you install: Medium install friction due to compiled Rust wheels; prebuilt abi3 wheels available for CPython 3.9+ on macOS (both architectures) and Linux (x86_64 and aarch64). Active maintenance with a release 26 days ago.

License in practice: MIT license permits commercial and private use with minimal restrictions; suitable for most projects.

Quickstart:

from datafusion_query_builder import col, table, functions as f

q = table("records").filter(col("kind") == "span").select(
    f.coalesce(col("service_name"), "(unknown)").alias("service")
).limit(200)
print(q.to_sql())

Requires Python 3.9 or later; prebuilt wheels available for common platforms (macOS arm64/x86_64, Linux x86_64/aarch64).

Verify before relying:

- Whether the library's DataFusion SQL dialect coverage matches your specific query patterns.
- Performance characteristics when building very large or deeply nested queries.
- Stability guarantees given the Alpha development status.

## Package facts

- License: MIT (permissive)
- Python support: supports_current
- Install friction: medium
- Maintenance: active
- Downloads: 104.4K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

sql query builder, datafusion query construction, injection-safe sql generation, typed sql builder, programmatic query building, sql injection prevention, datafusion python api, sql-injection-safe, query-builder, rust-python-binding

[View on SkillFed](https://skillfed.io/packages/datafusion-query-builder) · [View on PyPI](https://pypi.org/project/datafusion-query-builder/)