initial commit

verify_ridertime.py

@@ -0,0 +1,28 @@
+import duckdb
+import os
+import json
+
+DUCKDB_PATH = 'analytics.duckdb'
+
+def verify_ridertime():
+    conn = duckdb.connect(DUCKDB_PATH)
+    try:
+        # Same setup as service
+        conn.execute("SET s3_region='sgp1';")
+        conn.execute("SET s3_endpoint='sgp1.digitaloceanspaces.com';")
+        conn.execute("SET s3_url_style='path';")
+        
+        s3_path = 's3://nearle/parquet/deliveries/*.parquet'
+        conn.execute(f"CREATE OR REPLACE VIEW deliveries AS SELECT * FROM read_parquet('{s3_path}', union_by_name = true)")
+        
+        ridertime_gt_0 = int(conn.execute("SELECT COUNT(*) FROM deliveries WHERE ridertime > 0").fetchone()[0])
+        total = int(conn.execute("SELECT COUNT(*) FROM deliveries").fetchone()[0])
+        print(f"Total: {total}, ridertime > 0: {ridertime_gt_0}")
+        
+    except Exception as e:
+        print("Error:", e)
+    finally:
+        conn.close()
+
+if __name__ == "__main__":
+    verify_ridertime()