Skip to content

BigQuery

Local/Built-in Scheduler

Engine Adapter Type: bigquery

Installation

pip install "sqlmesh[bigquery]"

Connection options

Option Description Type Required
type Engine type name - must be bigquery string Y
method Connection methods - see allowed values below. Default: ouath. string N
project The name of the GCP project string N
location The location of for the datasets (can be regional or multi-regional) string N
keyfile Path to the keyfile to be used with service-account method string N
keyfile_json Keyfile information provided inline (not recommended) dict N
token OAuth 2.0 access token string N
refresh_token OAuth 2.0 refresh token string N
client_id OAuth 2.0 client ID string N
client_secret OAuth 2.0 client secret string N
token_uri OAuth 2.0 authorization server's toke endpoint URI string N
scopes The scopes used to obtain authorization list N
job_creation_timeout_seconds The maximum amount of time, in seconds, to wait for the underlying job to be created. int N
job_execution_timeout_seconds The maximum amount of time, in seconds, to wait for the underlying job to complete. int N
job_retries The number of times to retry the underlying job if it fails. (Default: 1) int N
priority The priority of the underlying job. (Default: INTERACTIVE) string N
maximum_bytes_billed The maximum number of bytes to be billed for the underlying job. int N

Airflow Scheduler

Engine Name: bigquery

In order to share a common implementation across local and Airflow, SQLMesh BigQuery implements its own hook and operator.

Installation

To enable support for this operator, the Airflow BigQuery provider package should be installed on the target Airflow cluster along with SQLMesh with the BigQuery extra: 

pip install "apache-airflow-providers-google"
pip install "sqlmesh[bigquery]"

Connection info

The operator requires an Airflow connection to determine the target BigQuery account. Please see GoogleBaseHook and GCP connectionfor more details. The goal is to create a working GCP connection on Airflow. The only difference is that you should use the sqlmesh_google_cloud_bigquery_default (by default) connection ID instead of the google_cloud_default one in the Airflow guide.

By default, the connection ID is set to sqlmesh_google_cloud_bigquery_default, but it can be overridden using the engine_operator_args parameter to the SQLMeshAirflow instance as in the example below: 

1
2
3
4
5
6
sqlmesh_airflow = SQLMeshAirflow(
    "bigquery",
    engine_operator_args={
        "bigquery_conn_id": "<Connection ID>"
    },
)

Connection Methods

  • oauth (default)
    • Related Credential Configuration:
      • scopes (Optional)
  • oauth-secrets
    • Related Credential Configuration:
      • token (Optional): Can be None if refresh information is provided.
      • refresh_token (Optional): If specified, credentials can be refreshed.
      • client_id (Optional): Must be specified for refresh, can be left as None if the token can not be refreshed.
      • client_secret (Optional): Must be specified for refresh, can be left as None if the token can not be refreshed.
      • token_uri (Optional): Must be specified for refresh, can be left as None if the token can not be refreshed.
      • scopes (Optional): OAuth 2.0 credentials can not request additional scopes after authorization. The scopes must be derivable from the refresh token if refresh information is provided (e.g. The refresh token scopes are a superset of this or contain a wild card scope like 'https://www.googleapis.com/auth/any-api')
  • service-account
    • Related Credential Configuration:
      • keyfile (Required)
      • scopes (Optional)
  • service-account-json
    • Related Credential Configuration:
      • keyfile_json (Required)
      • scopes (Optional)