diff --git a/push-pazans-to-mongo.py b/push-pazans-to-mongo.py index 202e411..b3ea851 100644 --- a/push-pazans-to-mongo.py +++ b/push-pazans-to-mongo.py @@ -23,36 +23,36 @@ ids = set() idsFile = sys.argv[3] if not os.path.isfile(idsFile): - idsRegex = re.compile('\\"_id\\": (.+?),') - with open(usersJsonFile, "r") as file_name: - for line in file_name: - groups = idsRegex.search(line) - uid = int(groups.group(1)) - ids.add(uid) + idsRegex = re.compile('\\"_id\\": (.+?),') + with open(usersJsonFile, "r") as file_name: + for line in file_name: + groups = idsRegex.search(line) + uid = int(groups.group(1)) + ids.add(uid) - with open(idsFile, "w") as file_name: - for uid in ids: - file_name.write(str(uid) + "\n") + with open(idsFile, "w") as file_name: + for uid in ids: + file_name.write(str(uid) + "\n") else: - with open(idsFile, "r") as f_ids: - for line in f_ids: - f_ids.add(int(line)) + with open(idsFile, "r") as f_ids: + for line in f_ids: + f_ids.add(int(line)) for file_name in os.listdir(dirWithIds): - print("parsing {}".format(file_name)) + print("parsing {}".format(file_name)) - with open(os.path.join(dirWithIds, file_name), "r") as file: - for line in file: - id = int(line) - if id in ids: - pazan = pazansCollection.find_one(id) - if pazan is None: - pazansCollection.insert_one({"_id": id, "groups": [file_name]}) - elif file_name not in pazan["groups"]: - pazan["groups"].append(file_name) - pazansCollection.update_one( - {"_id": pazan["_id"]}, - {"$set": {"groups": pazan["groups"]}} - ) + with open(os.path.join(dirWithIds, file_name), "r") as f_out: + for line in f_out: + uid = int(line) + if uid in ids: + pazan = pazansCollection.find_one(uid) + if pazan is None: + pazansCollection.insert_one({"_id": uid, "groups": [file_name]}) + elif file_name not in pazan["groups"]: + pazan["groups"].append(file_name) + pazansCollection.update_one( + {"_id": pazan["_id"]}, + {"$set": {"groups": pazan["groups"]}} + ) - print("- done") + print("- done")