Trigger a Dataflow job when a file is uploaded to Cloud Storage using a Cloud Function. This works by launching pre-configured Dataflow templates. This can be used to trigger Dataprep flows from Cloud Storage (or other) events.

For triggering Dataprep flows using the Dataprep API rather than Dataflow, see gcp-dataprep-gcf-trigger.

Run the following command to deploy this Cloud Function:

$ gcloud functions deploy <function-name> \
    --entry-point trigger \
    --trigger-bucket gs://<my-bucket> \
    --set-env-vars DATAFLOW_TEMPLATE=<dataflow-template-gcs-path>,BIGQUERY_DATASET=<bigquery-output-dataset>,BIGQUERY_TABLE=<bigquery-output-table>,TEMP_LOCATION=<temp-output-gcs-path> \
    --runtime python37

This function assumes a Dataflow template with a single input and output location, i.e. a Dataprep template.

As shown above, this Cloud Function requires four environment variables:

• DATAFLOW_TEMPLATE: GCS path for Dataflow template
• BIGQUERY_DATASET: BigQuery dataset for output
• BIGQUERY_TABLE: BigQuery table for output
• TEMP_LOCATION: GCS path for temporary output