openlineage_sql_parse_extract_input_output_datasets.py

1import openlineage.sql as sql
2
3# The SQL query to be parsed
4sql_query = "INSERT INTO table_out SELECT * FROM table_in WHERE id > 10"
5
6# Parse the SQL statement
7# The parse method returns a SqlMeta object containing inputs and outputs
8metadata = sql.parse([sql_query], default_schema="public")
9
10# Extracting input and output datasets
11inputs = [f"{dataset.namespace}/{dataset.name}" for dataset in metadata.in_tables]
12outputs = [f"{dataset.namespace}/{dataset.name}" for dataset in metadata.out_tables]
13
14print(f"Inputs: {inputs}")
15print(f"Outputs: {outputs}")