Skip to main content

Python SDK Reference

pygnok is the standalone Python client for external scripts. Gnok support provides the package when your client connectivity is set up (email support@gnok.io); see installation and connection setup.

Working inside a Gnok Studio notebook?

Use the pre-installed gnok helper instead — it auto-connects as the signed-in user (no connect(), no token) and adds a gnok.ml namespace to register trained models for SQL inference. gnok wraps pygnok.

pygnok.connect()​

Create a connection to Gnok. The signature below shows selected SDK defaults, not hosted connection values: always pass the host, port, and token provided for your organization, and tls=True.

pygnok.connect(
    host: str = "localhost",
    port: int = 50070,
    token: str | None = None,
    tls: bool = False,
    tls_root_certs: bytes | None = None,
    connect_timeout: float = 10.0,
    query_timeout: float = 300.0,
) -> Connection
ParameterTypeDefaultDescription
hoststr"localhost"Flight SQL host provided for your organization. Always set it.
portint50070Port provided for your organization. Always set it.
tokenstrNoneJWT bearer token
tlsboolFalseSet True for the hosted TLS endpoint
tls_root_certsbytesNoneOptional CA certificates supplied for your endpoint
connect_timeoutfloat10.0Connection timeout in seconds
query_timeoutfloat300.0Query timeout in seconds

Connection​

connection.cursor()​

Create a new cursor for executing queries.

cursor = conn.cursor()

connection.close()​

Close the connection and release resources.

Cursor​

cursor.execute(sql, parameters=None)​

Execute a SQL statement.

cursor.execute("SELECT * FROM orders WHERE region = ?", ["US"])

cursor.executemany(sql, seq_of_parameters)​

Execute a SQL statement with multiple parameter sets.

cursor.executemany(
    "INSERT INTO orders (id, amount) VALUES (?, ?)",
    [(1, 99.99), (2, 149.50)]
)

cursor.fetchone()​

Fetch the next row as a tuple, or None if no more rows.

cursor.fetchmany(size=cursor.arraysize)​

Fetch the next size rows as a list of tuples.

cursor.fetchall()​

Fetch all remaining rows as a list of tuples.

cursor.fetch_arrow_table()​

Fetch all results as a PyArrow Table for zero-copy access.

table = cursor.fetch_arrow_table()
df = table.to_pandas()

cursor.description​

Column metadata after executing a query:

# [(name, type_code, display_size, internal_size, precision, scale, null_ok), ...]
cursor.description

cursor.rowcount​

Number of rows affected by the last DML statement, or -1 for queries.

cursor.close()​

Close the cursor and release resources.

Last updated on