feat: don't pull invalid json owner data (#1535)

Co-authored-by: Derek Roberts <derek.roberts@gmail.com>

sync/config/SQL/SPAR/POSTGRES_SEEDLOT_OWNER_QUANTITY_EXTRACT.sql

@@ -61,6 +61,11 @@ SELECT drft.seedlot_number
  WHERE drft.seedlot_number = %(p_seedlot_number)s
    AND s.seedlot_status_code = 'PND'
    AND NOT(drft.all_step_data @? '$.ownershipStep[*].*.isInvalid ? (@ == true)')
+   AND TRIM(ownerdata->'ownerAgency'->>'value') != ''
+   AND TRIM(ownerdata->'ownerCode'->>'value') != ''
+   AND TRIM(ownerdata->'ownerPortion'->>'value') != ''
+   AND TRIM(ownerdata->'reservedPerc'->>'value') != ''
+   AND TRIM(ownerdata->'surplusPerc'->>'value') != ''
 ORDER BY 1
        , 2
        , 3

sync/src/main.py

@@ -155,6 +155,7 @@ def main() -> int:
         return job_return_code
 
     except Exception as err:
+        print(f"A fatal error has occurred ({type(err)}): {err}") 
         return 1 #failure
 
 # MAIN Execution

sync/src/module/data_sync_control.py

@@ -41,7 +41,7 @@ def get_scheduler(track_db_conn:object, database_schema:str) -> list:
     select_sync_id_stm = f"""
         select 
             COALESCE(MAX(to_timestamp) + INTERVAL '1 microsecond'
-                   , CURRENT_TIMESTAMP - INTERVAL '2 days') as current_start_time,
+                   , CURRENT_TIMESTAMP - INTERVAL '3 days') as current_start_time,
             CURRENT_TIMESTAMP as current_end_time,
             (select run_status
                from spar.etl_execution_log

sync/src/module/data_synchronization.py

@@ -51,6 +51,7 @@ def execute_instance(oracle_config, postgres_config, track_config):
 
         #if job is already running, stop
         if schedule_times['last_run_status'] == "RUNNING":
+            logger.info("Critical error: previous job still RUNNING or did not report SUCCESS or FAILURE")
             raise Exception("Critical error: previous job still RUNNING or did not report SUCCESS or FAILURE")
 
         logger.info('Insert execution log - signal RUNNING')