Delta Arrow Reader for Python

Read Delta Lake tables as PyArrow record batches with the Rust Delta Arrow Reader. Use filters directly or query registered tables with SQL. The same package includes the native delta-arrow-reader command.

Install in an activated Python virtual environment:

python -m pip install --only-binary=:all: delta-arrow-reader
delta-arrow-reader --help

For a separate CLI environment, choose a persistent installation or a single run:

uv tool install --no-build delta-arrow-reader
delta-arrow-reader --version

uvx --no-build delta-arrow-reader --help

If uv reports that its executable directory is missing from PATH, run uv tool update-shell and restart your shell. Tool environments do not install the Python API into your project's environment.

Wheels require CPython 3.10 or newer and target Linux x86_64 with glibc 2.28 or newer, Windows x86_64, and macOS arm64 with macOS 12.0 or newer. See the installation guide for tested Python versions and platform limits. Installing a wheel needs no Rust toolchain. Every installation includes the Python extension, native CLI, and PyArrow 18 or newer, even if you only use the command. This uses more disk space than the standalone Linux dar archive.

from delta_arrow_reader import DeltaTable

table = DeltaTable("/path/to/delta-table")
with table.to_reader() as reader:
    for batch in reader:
        print(batch)

See the Python quickstart for column selection, filters, and snapshot refresh. For SQL, use the included Rust DataFusion engine:

from delta_arrow_reader import Session

session = Session()
session.register_table("events", table)
with session.sql("SELECT COUNT(*) AS rows FROM events") as reader:
    print(reader.read_all().to_pydict())

SQL supports read-only queries over registered snapshots, including joins and aggregations. It does not require the separate datafusion Python package. See the SQL guide for execution and memory behavior. The contributor guide covers local builds and tests.