|
| 1 | +from mapswipe_workers import auth |
| 2 | +from mapswipe_workers.definitions import logger, sentry |
| 3 | + |
| 4 | + |
| 5 | +def get_project_ids_from_postgres(): |
| 6 | + """Get project ids.""" |
| 7 | + |
| 8 | + p_con = auth.postgresDB() |
| 9 | + |
| 10 | + query = """ |
| 11 | + SELECT project_id FROM projects |
| 12 | + LIMIT 10 |
| 13 | + """ |
| 14 | + data = p_con.retr_query(query) |
| 15 | + project_ids = [] |
| 16 | + for item in data: |
| 17 | + project_ids.append(item[0]) |
| 18 | + |
| 19 | + logger.info("Got projects from postgres.") |
| 20 | + logger.info(project_ids) |
| 21 | + return project_ids |
| 22 | + |
| 23 | + |
| 24 | +def update_tasks_table(project_id: str): |
| 25 | + """Remove duplicates in 'project_types_specifics' attribute in tasks table.""" |
| 26 | + |
| 27 | + logger.info(f"Start process for project: '{project_id}'") |
| 28 | + p_con = auth.postgresDB() |
| 29 | + |
| 30 | + query = """ |
| 31 | + UPDATE tasks |
| 32 | + SET project_type_specifics = project_type_specifics::jsonb |
| 33 | + #- '{projectId}' |
| 34 | + #- '{groupId}' |
| 35 | + #- '{taskId}' |
| 36 | + #- '{geometry}' |
| 37 | + #- '{wkt}' |
| 38 | + WHERE project_id = %(project_id)s |
| 39 | + """ |
| 40 | + try: |
| 41 | + p_con.query(query, {"project_id": project_id}) |
| 42 | + logger.info(f"Updated tasks table for project '{project_id}'.") |
| 43 | + except Exception as e: |
| 44 | + sentry.capture_exception(e) |
| 45 | + sentry.capture_message( |
| 46 | + f"Could NOT update tasks table for project '{project_id}'." |
| 47 | + ) |
| 48 | + logger.exception(e) |
| 49 | + logger.warning(f"Could NOT update tasks table for project '{project_id}'.") |
| 50 | + |
| 51 | + |
| 52 | +def update_groups_table(project_id: str): |
| 53 | + """Remove duplicates in 'project_types_specifics' attribute in groups table.""" |
| 54 | + |
| 55 | + logger.info(f"Start process for project: '{project_id}'") |
| 56 | + p_con = auth.postgresDB() |
| 57 | + |
| 58 | + # removing only the duplicated keys would be better |
| 59 | + """ |
| 60 | + select |
| 61 | + project_type_specifics::jsonb #- '{projectId}' #- '{groupId}' |
| 62 | + from tasks |
| 63 | + limit 10 |
| 64 | + """ |
| 65 | + |
| 66 | + query = """ |
| 67 | + UPDATE groups |
| 68 | + SET project_type_specifics = project_type_specifics::jsonb |
| 69 | + #- '{projectId}' |
| 70 | + #- '{id}' |
| 71 | + #- '{requiredCount}' |
| 72 | + #- '{finishedCount}' |
| 73 | + #- '{neededCount}' |
| 74 | + #- '{reportCount}' |
| 75 | + #- '{distributedCount}' |
| 76 | + WHERE project_id = %(project_id)s |
| 77 | + """ |
| 78 | + try: |
| 79 | + p_con.query(query, {"project_id": project_id}) |
| 80 | + logger.info(f"Updated tasks table for project '{project_id}'.") |
| 81 | + except Exception as e: |
| 82 | + sentry.capture_exception(e) |
| 83 | + sentry.capture_message( |
| 84 | + f"Could NOT update tasks table for project '{project_id}'." |
| 85 | + ) |
| 86 | + logger.exception(e) |
| 87 | + logger.warning(f"Could NOT update tasks table for project '{project_id}'.") |
| 88 | + |
| 89 | + |
| 90 | +def run_vacuum_tasks_table(): |
| 91 | + """Run vacuum to reclaim storage.""" |
| 92 | + logger.info("Start vacuum on tasks table.") |
| 93 | + p_con = auth.postgresDB() |
| 94 | + # isolation_level 0 will move you out of a transaction block |
| 95 | + old_isolation_level = p_con._db_connection.isolation_level |
| 96 | + p_con._db_connection.set_isolation_level(0) |
| 97 | + query = """ |
| 98 | + VACUUM FULL tasks |
| 99 | + """ |
| 100 | + p_con.query(query) |
| 101 | + # set isolation_level back to initial value |
| 102 | + p_con._db_connection.set_isolation_level(old_isolation_level) |
| 103 | + logger.info("Finish vacuum on tasks table.") |
| 104 | + |
| 105 | + |
| 106 | +def run_vacuum_groups_table(): |
| 107 | + """Run vacuum to reclaim storage.""" |
| 108 | + logger.info("Start vacuum on groups table.") |
| 109 | + p_con = auth.postgresDB() |
| 110 | + # isolation_level 0 will move you out of a transaction block |
| 111 | + old_isolation_level = p_con._db_connection.isolation_level |
| 112 | + p_con._db_connection.set_isolation_level(0) |
| 113 | + query = """ |
| 114 | + VACUUM FULL groups |
| 115 | + """ |
| 116 | + p_con.query(query) |
| 117 | + # set isolation_level back to initial value |
| 118 | + p_con._db_connection.set_isolation_level(old_isolation_level) |
| 119 | + logger.info("Finish vacuum on groups table.") |
| 120 | + |
| 121 | + |
| 122 | +if __name__ == "__main__": |
| 123 | + project_ids_list = get_project_ids_from_postgres() |
| 124 | + for i, project_id in enumerate(project_ids_list): |
| 125 | + update_tasks_table(project_id) |
| 126 | + update_groups_table(project_id) |
| 127 | + # vacuum every 50th project |
| 128 | + if i % 50 == 0 or i == len(project_ids_list) - 1: |
| 129 | + run_vacuum_tasks_table() |
| 130 | + run_vacuum_groups_table() |
0 commit comments