duckgresql 1.4.4.0

Python SDK for DuckGresQL — DuckDB-compatible API for remote databases

DuckGresQL Python SDK

Python client for DuckGresQL — a DuckDB-compatible API for remote databases. Uses Arrow Flight SQL for fast query execution and a REST API for async job management.

Installation

Requires Python 3.11+

pip install duckgresql

With optional extras:

pip install duckgresql[pandas]   # adds fetchdf() support
pip install duckgresql[all]      # pandas + numpy

Quick Start

Synchronous

import duckgresql

conn = duckgresql.connect(
    token="dkgql_your_api_token",
    database="my_database",
)

# Execute a query
result = conn.execute("SELECT * FROM users LIMIT 10")
print(result.fetchall())

# Get a pandas DataFrame
df = conn.execute("SELECT * FROM sales").fetchdf()

# DML
result = conn.execute("INSERT INTO users (name) VALUES ('Alice')")
print(result.rowcount)  # affected rows

conn.close()

Async

import asyncio
import duckgresql

async def main():
    conn = await duckgresql.connect_async(
        token="dkgql_your_api_token",
        database="my_database",
    )

    result = await conn.execute("SELECT * FROM users")
    print(result.fetchall())

    await conn.close()

asyncio.run(main())

Context Manager

with duckgresql.connect(token="...", database="...") as conn:
    result = conn.execute("SELECT 1")
    print(result.fetchone())

Async Jobs (Long-Running Queries)

conn = duckgresql.connect(token="...", database="...")

job = conn.execute_async("SELECT * FROM very_large_table")
print(f"Job submitted: {job.job_id}")

# Poll until done (with exponential backoff)
result = job.result(timeout=600)
print(result.fetchall())

Connection Parameters

Parameter	Default	Description
token	required	API token (dkgql_…)
database	required	Database name or UUID

Result Methods

Method	Returns	Description
fetchone()	tuple | None	Next row
fetchmany(size)	list[tuple]	Up to size rows
fetchall()	list[tuple]	All remaining rows
fetchdf()	DataFrame	Pandas DataFrame (requires pandas)
fetchnumpy()	dict[str, ndarray]	NumPy arrays (requires numpy)
fetch_arrow_table()	pyarrow.Table	Zero-copy Arrow table

Result Properties

Property	Type	Description
description	list | None	DB-API 2.0 column metadata
rowcount	int	Row count or affected rows
columns	list[str]	Column names

Architecture

• Arrow Flight SQL (gRPC) — Primary transport for execute(). Fast, columnar, zero-copy.
• REST API — Used only for execute_async() (job submission/polling), since Flight SQL has no native job queue pattern.

Both transports authenticate via the same API token. Flight SQL uses BasicAuth handshake; REST uses the /connect endpoint to exchange the token for a session token.

---

Developing the SDK? See CONTRIBUTING.md for setup, testing, building, and publishing.