databend-driver 0.34.1

Databend Driver Python Binding

databend-driver

Databend Python Client

Usage

Local Embedded Connection

The local embedded mode runs a full Databend engine in-process without any server. It is useful for local analytics, testing, and offline workflows.

Install the local extra to pull in the embedded engine:

pip install "databend-driver[local]"

The embedded dependency currently requires Python 3.12 or later.

from databend_driver import connect

# Persistent state stored under ./local-state
conn = connect("databend+local:///./local-state")
conn.exec("CREATE TABLE books(id INT, title STRING)")
conn.exec("INSERT INTO books VALUES (1, 'Databend')")

row = conn.query_row("SELECT title FROM books ORDER BY id LIMIT 1")
print(row.values())  # ('Databend',)

rows = [row.values() for row in conn.query_iter("SELECT * FROM books ORDER BY id")]

Supported local targets:

• connect(":memory:") — temporary in-memory instance (discarded on close)
• connect("databend+local:///:memory:") — explicit in-memory instance
• connect("databend+local:///./local-state") — persistent state under ./local-state
• connect("databend+local:///./local-state?tenant=default") — persistent state with an explicit tenant
• connect("databend+local:///./local-state?database=mydb") — open a specific database

You can also use connect_local() directly for more control:

from databend_driver import connect_local

conn = connect_local(database=":memory:")
conn = connect_local(data_path="./local-state", tenant="default")

If the optional databend package is not installed, connect() raises an ImportError with guidance about enabling the local extra and the Python version requirement.

For remote Databend, the same connect() entrypoint accepts standard DSNs:

from databend_driver import connect

conn = connect("databend://root:@localhost:8000/?sslmode=disable")
row = conn.query_row("SELECT 1")

Relation API

The local connection exposes an embedded-specific relation API for working with query results as DataFrames or Arrow tables:

relation = conn.sql("SELECT * FROM books")

df = relation.df()       # pandas DataFrame
pl = relation.pl()       # polars DataFrame
tbl = relation.arrow()   # pyarrow Table

rows = relation.fetchall()   # list[tuple]
row = relation.fetchone()    # tuple | None

Registering External Data

You can register files or in-memory data as virtual tables:

# Register a Parquet file
conn.register("sales", "./data/sales.parquet")
conn.sql("SELECT * FROM sales LIMIT 10").df()

# Register a CSV file
conn.register("events", "./data/events.csv")

# Register a pandas or polars DataFrame
import pandas as pd
df = pd.DataFrame({"id": [1, 2], "name": ["Alice", "Bob"]})
conn.register("users", df)

# Shorthand: register a DataFrame and return a relation immediately
relation = conn.from_df(df)

# Read helpers (register and return relation in one call)
relation = conn.read_parquet("./data/sales.parquet")
relation = conn.read_csv("./data/events.csv")
relation = conn.read_json("./data/logs.ndjson")
relation = conn.read_text("./data/raw.txt")

PEP 249 Cursor Object

from databend_driver import BlockingDatabendClient

client = BlockingDatabendClient('databend://root:root@localhost:8000/?sslmode=disable')
cursor = client.cursor()

cursor.execute(
    """
    CREATE TABLE test (
        i64 Int64,
        u64 UInt64,
        f64 Float64,
        s   String,
        s2  String,
        d   Date,
        t   DateTime
    )
    """
)
cursor.execute("INSERT INTO test VALUES (?, ?, ?, ?, ?, ?, ?)", (1, 1, 1.0, 'hello', 'world', '2021-01-01', '2021-01-01 00:00:00'))
cursor.execute("SELECT * FROM test")
rows = cursor.fetchall()
for row in rows:
    print(row.values())
cursor.close()

Blocking Connection Object

from databend_driver import BlockingDatabendClient

client = BlockingDatabendClient('databend://root:root@localhost:8000/?sslmode=disable')
conn = client.get_conn()
conn.exec(
    """
    CREATE TABLE test (
        i64 Int64,
        u64 UInt64,
        f64 Float64,
        s   String,
        s2  String,
        d   Date,
        t   DateTime
    )
    """
)
rows = conn.query_iter("SELECT * FROM test")
for row in rows:
    print(row.values())
conn.close()

Asyncio Connection Object

import asyncio
from databend_driver import AsyncDatabendClient

async def main():
    client = AsyncDatabendClient('databend://root:root@localhost:8000/?sslmode=disable')
    conn = await client.get_conn()
    await conn.exec(
        """
        CREATE TABLE test (
            i64 Int64,
            u64 UInt64,
            f64 Float64,
            s   String,
            s2  String,
            d   Date,
            t   DateTime
        )
        """
    )
    rows = await conn.query_iter("SELECT * FROM test")
    async for row in rows:
        print(row.values())
    await conn.close()

asyncio.run(main())

Parameter bindings

# Positional parameters using ?
row = await context.conn.query_row("SELECT ?, ?, ?, ?", (3, False, 4, "55"))

# Named parameters using :name
row = await context.conn.query_row(
    "SELECT :a, :b, :c, :d", {"a": 3, "b": False, "c": 4, "d": "55"}
)

# Single value (no tuple needed)
row = await context.conn.query_row("SELECT ?", 3)

# Keyword argument form
row = await context.conn.query_row("SELECT ?, ?, ?, ?", params=(3, False, 4, "55"))

Named parameters use token-aware matching, so :a will not corrupt :ab. For local embedded connections, passing a mismatched number of ? placeholders and values raises a ValueError immediately.

Query ID tracking and query management

# Get the last executed query ID
query_id = conn.last_query_id()
print(f"Last query ID: {query_id}")

# Execute a query and get its ID
await conn.query_row("SELECT 1")
query_id = conn.last_query_id()
print(f"Query ID: {query_id}")

# Kill a running query (if needed)
try:
    await conn.kill_query("some-query-id")
    print("Query killed successfully")
except Exception as e:
    print(f"Failed to kill query: {e}")

Type Mapping

Databend Types

General Data Types

Databend	Python
BOOLEAN	bool
TINYINT	int
SMALLINT	int
INT	int
BIGINT	int
FLOAT	float
DOUBLE	float
DECIMAL	decimal.Decimal
DATE	datetime.date
TIMESTAMP	datetime.datetime
INTERVAL	datetime.timedelta
VARCHAR	str
BINARY	bytes

Semi-Structured Data Types

Databend	Python
ARRAY	list
TUPLE	tuple
MAP	dict
VARIANT	str
BITMAP	str
GEOMETRY	str / bytes
GEOGRAPHY	str / bytes

Note: VARIANT is a json encoded string. Example:

CREATE TABLE example (
    data VARIANT
);
INSERT INTO example VALUES ('{"a": 1, "b": "hello"}');

row = await conn.query_row("SELECT * FROM example limit 1;")
data = row.values()[0]
value = json.loads(data)
print(value)

GEOMETRY and GEOGRAPHY follow the current geometry_output_format setting. Text formats such as GeoJSON or WKT return str; binary formats such as WKB or EWKB return bytes.

For example:

row = await conn.query_row("settings(geometry_output_format='WKB') SELECT st_point(60, 37)")
assert isinstance(row.values()[0], bytes)

APIs

Exception Classes (PEP 249 Compliant)

The driver provides a complete set of exception classes that follow the PEP 249 standard for database interfaces:

# Base exceptions
class Warning(Exception): ...
class Error(Exception): ...

# Interface errors
class InterfaceError(Error): ...

# Database errors
class DatabaseError(Error): ...

# Specific database error types
class DataError(DatabaseError): ...
class OperationalError(DatabaseError): ...
class IntegrityError(DatabaseError): ...
class InternalError(DatabaseError): ...
class ProgrammingError(DatabaseError): ...
class NotSupportedError(DatabaseError): ...

These exceptions are automatically mapped from Databend error codes to appropriate PEP 249 exception types based on the nature of the error.

Note: stream_load and load_file support an optional method parameter, it accepts two string values:

• stage: Data is first uploaded to a temporary stage and then loaded. This is the default behavior.
• streaming: Data is directly streamed to the Databend server.

AsyncDatabendClient

class AsyncDatabendClient:
    def __init__(self, dsn: str): ...
    async def get_conn(self) -> AsyncDatabendConnection: ...

AsyncDatabendConnection

class AsyncDatabendConnection:
    async def info(self) -> ConnectionInfo: ...
    async def version(self) -> str: ...
    async def close(self) -> None: ...
    def last_query_id(self) -> str | None: ...
    async def kill_query(self, query_id: str) -> None: ...
    async def exec(self, sql: str, params: list[string] | tuple[string] | any = None) -> int: ...
    async def query_row(self, sql: str, params: list[string] | tuple[string] | any = None) -> Row: ...
    async def query_iter(self, sql: str, params: list[string] | tuple[string] | any = None) -> RowIterator: ...
    async def stream_load(self, sql: str, data: list[list[str]], method: str = None) -> ServerStats: ...
    async def load_file(self, sql: str, file: str, method: str = None) -> ServerStats: ...

BlockingDatabendClient

class BlockingDatabendClient:
    def __init__(self, dsn: str): ...
    def get_conn(self) -> BlockingDatabendConnection: ...
    def cursor(self) -> BlockingDatabendCursor: ...

BlockingDatabendConnection

class BlockingDatabendConnection:
    def info(self) -> ConnectionInfo: ...
    def version(self) -> str: ...
    def close(self) -> None: ...
    def last_query_id(self) -> str | None: ...
    def kill_query(self, query_id: str) -> None: ...
    def exec(self, sql: str, params: list[string] | tuple[string] | any = None) -> int: ...
    def query_row(self, sql: str, params: list[string] | tuple[string] | any = None) -> Row: ...
    def query_iter(self, sql: str, params: list[string] | tuple[string] | any = None) -> RowIterator: ...
    def stream_load(self, sql: str, data: list[list[str]], method: str = None) -> ServerStats: ...
    def load_file(self, sql: str, file: str, method: str = None, format_option: dict = None, copy_options: dict = None) -> ServerStats: ...

BlockingDatabendCursor

class BlockingDatabendCursor:
    @property
    def description(self) -> list[tuple[str, str, int | None, int | None, int | None, int | None, bool | None]] | None: ...
    @property
    def rowcount(self) -> int: ...
    def close(self) -> None: ...
    def execute(self, operation: str, params: list[string] | tuple[string] = None) -> None | int: ...
    def executemany(self, operation: str, params: list[string] | tuple[string] = None, values: list[list[string] | tuple[string]]) -> None | int: ...
    def fetchone(self) -> Row | None: ...
    def fetchmany(self, size: int = 1) -> list[Row]: ...
    def fetchall(self) -> list[Row]: ...

    # Optional DB API Extensions
    def next(self) -> Row: ...
    def __next__(self) -> Row: ...
    def __iter__(self) -> BlockingDatabendCursor: ...

Row

class Row:
    def values(self) -> tuple: ...
    def __len__(self) -> int: ...
    def __iter__(self) -> Row: ...
    def __next__(self) -> any: ...
    def __dict__(self) -> dict: ...
    def __getitem__(self, key: int | str) -> any: ...

RowIterator

class RowIterator:
    def schema(self) -> Schema: ...

    def __iter__(self) -> RowIterator: ...
    def __next__(self) -> Row: ...

    def __aiter__(self) -> RowIterator: ...
    async def __anext__(self) -> Row: ...

Field

class Field:
    @property
    def name(self) -> str: ...
    @property
    def data_type(self) -> str: ...

Schema

class Schema:
    def fields(self) -> list[Field]: ...

ServerStats

class ServerStats:
    @property
    def total_rows(self) -> int: ...
    @property
    def total_bytes(self) -> int: ...
    @property
    def read_rows(self) -> int: ...
    @property
    def read_bytes(self) -> int: ...
    @property
    def write_rows(self) -> int: ...
    @property
    def write_bytes(self) -> int: ...
    @property
    def running_time_ms(self) -> float: ...

ConnectionInfo

class ConnectionInfo:
    @property
    def handler(self) -> str: ...
    @property
    def host(self) -> str: ...
    @property
    def port(self) -> int: ...
    @property
    def user(self) -> str: ...
    @property
    def database(self) -> str | None: ...
    @property
    def warehouse(self) -> str | None: ...

connect_local

def connect_local(
    database: str = ":memory:",
    *,
    data_path: str | None = None,
    tenant: str | None = None,
) -> LocalConnection: ...

LocalConnection

class LocalConnection:
    def sql(self, query: str) -> LocalRelation: ...
    def table(self, name: str) -> LocalRelation: ...
    def format_sql(self, sql: str, params: Any = None) -> str: ...
    def execute(self, query: str, params: Any = None) -> None: ...
    def exec(self, sql: str, params: Any = None) -> None: ...
    def query_row(self, sql: str, params: Any = None) -> LocalRow | None: ...
    def query_all(self, sql: str, params: Any = None) -> list[LocalRow]: ...
    def query_iter(self, sql: str, params: Any = None) -> LocalRowIterator: ...
    def close(self) -> None: ...
    def last_query_id(self) -> None: ...  # always None for local mode
    def kill_query(self, query_id: str) -> None: ...  # raises NotImplementedError
    def register(
        self,
        name: str,
        source: Any,           # path str/Path, pandas/polars DataFrame, or pyarrow Table
        *,
        format: str | None = None,
        pattern: str | None = None,
        connection: str | None = None,
    ) -> LocalConnection: ...
    def from_df(self, source: Any, *, name: str | None = None) -> LocalRelation: ...
    def read_parquet(
        self, path: str | Path, *, pattern: str | None = None,
        connection: str | None = None, name: str | None = None,
    ) -> LocalRelation: ...
    def read_csv(
        self, path: str | Path, *, pattern: str | None = None,
        connection: str | None = None, name: str | None = None,
    ) -> LocalRelation: ...
    def read_json(
        self, path: str | Path, *, pattern: str | None = None,
        connection: str | None = None, name: str | None = None,
    ) -> LocalRelation: ...
    def read_text(
        self, path: str | Path, *, pattern: str | None = None,
        connection: str | None = None, name: str | None = None,
    ) -> LocalRelation: ...

LocalRelation

class LocalRelation:
    def df(self) -> Any: ...          # pandas DataFrame
    def pl(self) -> Any: ...          # polars DataFrame
    def arrow(self) -> Any: ...       # pyarrow Table
    def fetchall(self) -> list[tuple]: ...
    def fetchone(self) -> tuple | None: ...

LocalRow

class LocalRow:
    def values(self) -> tuple[Any, ...]: ...
    def __len__(self) -> int: ...
    def __iter__(self) -> LocalRow: ...
    def __next__(self) -> Any: ...
    def __getitem__(self, key: int) -> Any: ...

LocalRowIterator

class LocalRowIterator:
    def schema(self) -> Any: ...  # not yet implemented for local mode
    def close(self) -> None: ...
    def __iter__(self) -> LocalRowIterator: ...
    def __next__(self) -> LocalRow: ...

Development

cd tests
make up

cd bindings/python
uv python install 3.12
uv venv --python 3.12
uv sync --extra local
source .venv/bin/activate
maturin develop

behave tests/asyncio
behave tests/blocking
behave tests/cursor
behave tests/local