from airflow.sdk import dag
from airflow.sdk.bases.operator import chain
from iqvia_nlp_provider.operators.i2e import (
I2EDeleteResourcesOperator,
I2EMakeIndexOperator,
I2EPrintTaskLogOperator,
I2ERunQueryOperator,
)
from iqvia_nlp_provider.transfers.local import I2EToLocalFilesystemOperator, LocalFilesystemToI2EOperator
@dag(dag_id="example_dags.i2e_query", schedule=None, catchup=False)
def i2e_query():
"""
DAG showing how to use I2E operators to upload source data from a local file to I2E,
index the source data, run a query against the index, write the query results
to a local file, and delete all transient resources from I2E.
Requirements to use this DAG:
- An I2E connection named `i2e_default`.
- A source file at '/work_dir/source.txt' on the Airflow worker's filesystem.
- A directory named '/work_dir/results/' on the Airflow worker's filesystem.
- An index template named `data-factory-workflow-example` installed on the I2E server.
- A query template named `data-factory-workflow-example` installed on the I2E server.
"""
local_filesystem_to_i2e_task = i2e_source_data_uri = LocalFilesystemToI2EOperator(
i2e_conn_id="i2e_default",
local_file_path="/work_dir/source.txt",
)
i2e_make_index_task = i2e_index_uri = I2EMakeIndexOperator(
i2e_conn_id="i2e_default",
i2e_source_data_uri=i2e_source_data_uri.output,
i2e_index_template="data-factory-workflow-example",
i2e_index_settings_overrides={
"useDaemon": True,
"checkDaemon": True,
"ontologyIndexSuppression": "All",
},
)
i2e_print_task_log_task = I2EPrintTaskLogOperator(
i2e_conn_id="i2e_default",
)
i2e_run_query_task = i2e_query_results_uri = I2ERunQueryOperator(
i2e_conn_id="i2e_default",
i2e_index_uri=i2e_index_uri.output,
i2e_query_template="data-factory-workflow-example",
i2e_query_settings_overrides={
"checkClasses": False,
},
)
i2e_to_local_filesystem_task = I2EToLocalFilesystemOperator(
i2e_conn_id="i2e_default",
i2e_query_results_uri=i2e_query_results_uri.output,
local_folder="/work_dir/results/",
)
i2e_delete_resources_task = I2EDeleteResourcesOperator(
i2e_conn_id="i2e_default",
i2e_resource_uris=[i2e_source_data_uri.output, i2e_index_uri.output, i2e_query_results_uri.output],
)
# Print task log runs only if make index or query fails
i2e_make_index_task >> i2e_print_task_log_task
i2e_run_query_task >> i2e_print_task_log_task
chain(
local_filesystem_to_i2e_task,
i2e_make_index_task,
i2e_run_query_task,
i2e_to_local_filesystem_task,
i2e_delete_resources_task,
)
i2e_query()