fix_workspace_objects_w_bad_meta.py

import os
import sys

from pymongo.mongo_client import MongoClient

WS_MONGO_PWD = 'WS_MONGO_PWD'

MONGO_COL_WSOBJS = 'workspaceObjVersions'

def get_ws_obj_collection(argv):
    _, host, db, user = argv
    pwd = os.environ[WS_MONGO_PWD] 
    client = MongoClient(host, authSource=db, username=user, password=pwd)
    client.admin.command('ismaster')  # check connection
    return client[db][MONGO_COL_WSOBJS]


def main():
    dry_run = len(sys.argv) < 5
    if dry_run:
        print("Dry run - not making any modifications. Add a 4th argument to the command " +
              "line to turn off dry run mode")
    col = get_ws_obj_collection(sys.argv[:4])
    # bad idea usually, but we know this is < 10K objects for this case
    objs = []
    count = 0
    for o in col.find():
        meta = o.get('meta')
        if meta:
            for kv in meta:
                if (
                        kv['k'] is not None
                        and kv['v'] is not None
                        and len(kv['k']) + len(kv['v']) > 900
                ):
                    objs.append((o, kv))
        count += 1
        if count % 1000 == 0:
            print(count, file=sys.stderr)
    print(f"Updating {len(objs)} meta keys")
    count = 0
    for o, kv in objs:
        if not dry_run:
            col.update_one(
                {'ws': o["ws"], 'id': o['id'], 'ver': o['ver']},
                {'$pull': {'meta': {'k': kv['k'], 'v': kv['v']}}})
        print(f'Removed key {kv["k"]} from {o["ws"]}/{o["id"]}/{o["ver"]}')
        count += 1
        if count % 100 == 0:
            print(count, file=sys.stderr)


if __name__ == '__main__':
    main()