|
| 1 | +# small script to archive off old data from the JSON database |
| 2 | +import datetime |
| 3 | +import os |
| 4 | +import sys |
| 5 | +from dateutil.relativedelta import relativedelta |
| 6 | + |
| 7 | +from wmpl.Utils.TrajConversions import datetime2JD, jd2Date |
| 8 | +from wmpl.Trajectory.CorrelateRMS import DatabaseJSON |
| 9 | + |
| 10 | +# Name of json file with the list of processed directories |
| 11 | +JSON_DB_NAME = "processed_trajectories.json" |
| 12 | + |
| 13 | + |
| 14 | +def archiveOldRecords(db, db_dir, older_than=3): |
| 15 | + """ |
| 16 | + Archive off old records to keep the database size down |
| 17 | +
|
| 18 | + Keyword Arguments: |
| 19 | + older_than: [int] number of months to keep, default 3 |
| 20 | + """ |
| 21 | + class DummyMetObs(): |
| 22 | + def __init__(self, station, obs_id): |
| 23 | + self.station_code = station |
| 24 | + self.id = obs_id |
| 25 | + |
| 26 | + archdate = datetime.datetime.now(datetime.timezone.utc) - relativedelta(months=older_than) |
| 27 | + archdate_jd = datetime2JD(archdate) |
| 28 | + |
| 29 | + arch_db_path = os.path.join(db_dir, 'archive', f'{archdate.strftime("%Y%m")}_{JSON_DB_NAME}') |
| 30 | + os.makedirs(os.path.join(db_dir, 'archive'), exist_ok=True) |
| 31 | + archdb = DatabaseJSON(arch_db_path) |
| 32 | + |
| 33 | + for traj in [t for t in db.trajectories if t < archdate_jd]: |
| 34 | + if traj < archdate_jd: |
| 35 | + archdb.addTrajectory(None, db.trajectories[traj], False) |
| 36 | + db.removeTrajectory(db.trajectories[traj]) |
| 37 | + |
| 38 | + for traj in [t for t in db.failed_trajectories if t < archdate_jd]: |
| 39 | + if traj < archdate_jd: |
| 40 | + archdb.addTrajectory(None, db.failed_trajectories[traj], True) |
| 41 | + db.removeTrajectory(db.failed_trajectories[traj]) |
| 42 | + |
| 43 | + for station in db.processed_dirs: |
| 44 | + arch_processed = [dirname for dirname in db.processed_dirs[station] if |
| 45 | + datetime.datetime.strptime(dirname[14:22], '%Y%m%d').replace(tzinfo=datetime.timezone.utc) < archdate] |
| 46 | + for dirname in arch_processed: |
| 47 | + archdb.addProcessedDir(station, dirname) |
| 48 | + db.processed_dirs[station].remove(dirname) |
| 49 | + |
| 50 | + for station in db.paired_obs: |
| 51 | + arch_processed = [obs_id for obs_id in db.paired_obs[station] if |
| 52 | + datetime.datetime.strptime(obs_id[7:15], '%Y%m%d').replace(tzinfo=datetime.timezone.utc) < archdate] |
| 53 | + for obs_id in arch_processed: |
| 54 | + archdb.addPairedObservation(DummyMetObs(station, obs_id)) |
| 55 | + db.paired_obs[station].remove(obs_id) |
| 56 | + |
| 57 | + archdb.save() |
| 58 | + db.save() |
| 59 | + return |
| 60 | + |
| 61 | + |
| 62 | +if __name__ == '__main__': |
| 63 | + db_dir = sys.argv[1] |
| 64 | + database_path = os.path.join(db_dir, JSON_DB_NAME) |
| 65 | + db = DatabaseJSON(database_path) |
| 66 | + soonest = jd2Date(min(min(list(db.trajectories.keys())), min(list(db.failed_trajectories.keys()))), dt_obj=True) |
| 67 | + |
| 68 | + nowdt = datetime.datetime.now() |
| 69 | + mthsback = int((nowdt - soonest).days/30) |
| 70 | + for i in range(mthsback, 2, -1): |
| 71 | + print(f'archiving {i} months back') |
| 72 | + archiveOldRecords(db, db_dir, i) |
0 commit comments