soda_core_duckdb_programmatic_data_quality_scan.py

1from soda.scan import Scan
2
3def run_soda_scan():
4    scan = Scan()
5    scan.set_verbose(True)
6    
7    # Define the DuckDB data source connection directly in the scan
8    # You can point to a file (e.g., 'path/to/my_database.duckdb') or use ':memory:'
9    scan.add_configuration_yaml_str(
10        """
11        data_source my_duckdb_source:
12          type: duckdb
13          path: :memory:
14        """
15    )
16    
17    scan.set_data_source_name("my_duckdb_source")
18
19    # Define the data quality checks
20    # Note: Ensure the table (e.g., 'dim_customer') exists in your DuckDB instance
21    scan.add_sodacl_yaml_str(
22        """
23        checks for dim_customer:
24          - row_count > 0
25          - missing_count(last_name) = 0
26          - duplicate_count(phone_number) = 0
27        """
28    )
29
30    # Execute the scan
31    exit_code = scan.execute()
32    
33    # Inspect the results
34    print(f"Scan exit code: {exit_code}")
35    print(f"Scan results: {scan.get_scan_results()}")
36
37if __name__ == "__main__":
38    run_soda_scan()