lib/telemetry.py

import os
from uuid import uuid4
import json

from huggingface_hub import login, CommitScheduler

from lib.data_entry import DataEntry

login(os.environ["HF_TOKEN"])


class TelemetryManager:
    def __init__(self) -> None:
        self.__file_path = f"data/{uuid4()}.jsonl"
        self.__scheduler = CommitScheduler(
            repo_id="elsamueldev/confia-demo-data",
            repo_type="dataset",
            private=True,
            folder_path="data/",
            path_in_repo="/",
            every=5
        )
    
    def write_data(self, data: DataEntry) -> None:
        content = json.dumps(data.to_dict())
        
        with self.__scheduler.lock:
            with open(self.__file_path, "a") as file:
                file.write(f"{content}\n")