find_workspace_objs_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():
    _, host, db, user = sys.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 printobj(nonemeta, o, kv):
    print(f'{nonemeta}, {o["ws"]}, {o["id"]}, {o["ver"]}, {o["savedate"]}, {o["type"]}, {kv["k"]}, {kv["v"]}')


def main():
    col = get_ws_obj_collection()
    count = 0
    for o in col.find():
        meta = o.get('meta')
        if meta:
            for kv in meta:
                # type field still exists at this time
                if kv['k'] is None or kv['v'] is None:
                    printobj(1, o, kv)
                elif len(kv['k']) + len(kv['v']) > 900:
                    printobj(0, o, kv)
        count += 1
        if count % 1000 == 0:
            print(count, file=sys.stderr)


if __name__ == '__main__':
    main()