parallel insert

mongo.py

import json
import pandas as pd
from pymongo import MongoClient


def mongo_to_pandas(client, database, collection):
    """ Reads mongo collection to pandas data frame."""
    cur = client[database][collection].find()
    df = pd.DataFrame(list(cur))
    df["_id"] = df["_id"].apply(lambda x: str(x)) #convert ids to strings
    return df


def pandas_to_mongo(df, database, collection):
    """ Parallel inserts a data frame into a mongo collection. """
    docs = json.loads(df.to_json(orient="records"))
    cli = MongoClient()
    cli[database][collection].insert_many(docs, ordered=False)
    return