import sqlite3 def clean_sales(): conn = sqlite3.connect('data/hiffi_catalog.db') cur = conn.cursor() # Delete everything except the strictly verified pure sales records cur.execute(""" DELETE FROM commercial_sales WHERE provenance_flag NOT IN ('VERIFIED_CSPC', 'VERIFIED_RIAA') """) deleted_count = cur.rowcount conn.commit() print(f"✅ Purged {deleted_count} unverified and streaming-based records.") # Verify the final clean state print("\n--- Final Pure Sales Cohort ---") cur.execute("SELECT provenance_flag, COUNT(*) FROM commercial_sales GROUP BY provenance_flag") for flag, count in cur.fetchall(): print(f" - {flag}: {count} records") cur.execute("SELECT COUNT(*) FROM commercial_sales") total = cur.fetchone()[0] print(f"\nTotal Verified Pure Sales Records: {total} / 530 expected") conn.close() if __name__ == "__main__": clean_sales()