fn_registry/python/functions/bigquery/bq_create_routine.md

name, kind, lang, domain, version, purity, signature, description, tags, uses_functions, uses_types, returns, returns_optional, error_type, imports, params, output, tested, tests, test_file_path, file_path

name	kind	lang	domain	version	purity	signature	description	tags	uses_functions	uses_types	returns	returns_optional	error_type	imports	params	output	tested	tests	test_file_path	file_path
bq_create_routine	function	py	infra	1.0.0	impure	def bq_create_routine(client: BQClient, dataset_id: str, routine_id: str, body: str, routine_type: str = 'SCALAR_FUNCTION', language: str = 'SQL', arguments: list[dict] | None = None, return_type: str = '', description: str = '') -> dict	Crea una routine (UDF scalar, tabla o stored procedure) en BigQuery. Soporta SQL, JavaScript y Python.	bigquery gcp routine udf create google-cloud python				false	error_go_core	google-cloud-bigquery	name desc client instancia autenticada de BQClient name desc dataset_id nombre del dataset donde se crea la routine name desc routine_id nombre/identificador de la routine dentro del dataset name desc body cuerpo de la routine: expresion SQL, bloque JavaScript o codigo Python segun el lenguaje name desc routine_type tipo de routine: SCALAR_FUNCTION para UDFs escalares, TABLE_VALUED_FUNCTION para UDFs de tabla, PROCEDURE para stored procedures name desc language lenguaje de implementacion: SQL, JAVASCRIPT o PYTHON name desc arguments lista de argumentos, cada uno como dict con claves 'name' y 'data_type' (ej: INT64, STRING, FLOAT64) name desc return_type tipo de dato que retorna la funcion (ej: INT64, STRING); ignorado para PROCEDURE name desc description descripcion opcional de la routine	dict con routine_id, dataset_id, project, routine_type, language, body, description, created y modified (ISO 8601)	false			python/functions/bigquery/routines.py

Ejemplo

from bigquery.client import bq_auth
from bigquery.routines import bq_create_routine

client = bq_auth("my-project")

# UDF escalar SQL
fn = bq_create_routine(
    client,
    dataset_id="analytics",
    routine_id="double_value",
    body="x * 2",
    arguments=[{"name": "x", "data_type": "INT64"}],
    return_type="INT64",
    description="Duplica un entero",
)
print(fn["routine_id"], fn["routine_type"], fn["language"])
# double_value SCALAR_FUNCTION SQL

# Stored procedure SQL
bq_create_routine(
    client,
    dataset_id="analytics",
    routine_id="refresh_summary",
    body="INSERT INTO summary SELECT * FROM raw WHERE date = CURRENT_DATE();",
    routine_type="PROCEDURE",
)

Notas

Lanza google.api_core.exceptions.Conflict (409) si la routine ya existe. Para actualizar una routine existente, eliminarla primero con bq_delete_routine y recrearla, o usar client._client.update_routine() directamente.

Los data_type de los argumentos deben ser constantes de bigquery.StandardSqlTypeNames: INT64, FLOAT64, STRING, BOOL, BYTES, DATE, DATETIME, TIMESTAMP, TIME, NUMERIC, BIGNUMERIC, JSON, ARRAY, STRUCT.

Las routines JavaScript permiten librerias externas via imported_libraries (no expuesto en este wrapper).