initial commit

verify_zero.py

@@ -0,0 +1,26 @@
+import duckdb
+import os
+import json
+
+DUCKDB_PATH = 'analytics.duckdb'
+
+def verify_zero_tenant():
+    conn = duckdb.connect(DUCKDB_PATH)
+    try:
+        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)")
+        
+        # Check values distribution for tenantname
+        res = conn.execute("SELECT tenantname, COUNT(*) FROM deliveries GROUP BY 1 ORDER BY 2 DESC LIMIT 10").df()
+        print(res)
+        
+    except Exception as e:
+        print("Error:", e)
+    finally:
+        conn.close()
+
+if __name__ == "__main__":
+    verify_zero_tenant()