This crate intends to build a native python binding.

pip install databend

import databend
databend.init_service(local_dir = ".databend")
# or use config
# databend.init_service( config = "config.toml.sample" )

from databend import SessionContext
ctx = SessionContext()

df = ctx.sql("select number, number + 1, number::String as number_p_1 from numbers(8)")

df.show()
# convert to pyarrow
import pyarrow
df.to_py_arrow()

# convert to pandas
import pandas
df.to_pandas()

supported functions:

• register_parquet
• register_ndjson
• register_csv
• register_tsv

ctx.register_parquet("pa", "/home/sundy/dataset/hits_p/", pattern = ".*.parquet")
ctx.sql("select * from pa limit 10").collect()

Tenant has it's own catalog and tables

ctx = SessionContext(tenant = "your_tenant_name")

Setup virtualenv:

uv sync

Activate venv:

source .venv/bin/activate

Install maturin:

pip install "maturin[patchelf]"

Build bindings:

uvx maturin develop

Run tests:

uvx maturin develop -E test

Build API docs:

uvx maturin develop -E docs
uvx pdoc databend

Note:

databend.init_service must be initialized before SessionContext

databend.init_service must be called only once

• By default, you can init the service by a local directory, then data & catalogs will be stored inside the directory.

import databend

databend.init_service(local_dir = ".databend")

• You can also init by file

import databend
databend.init_service( config = "config.toml.sample" )

• And by config str

import databend

databend.init_service(config = """
[meta]
embedded_dir = "./.databend/"

# Storage config.
[storage]
# fs | s3 | azblob | obs | oss
type = "fs"
allow_insecure = true

[storage.fs]
data_path = "./.databend/"
""")

Read more about configs of databend in docs

Databend python api is inspired by arrow-datafusion-python, thanks for their great work.