import discord import logging import os from huggingface_hub import InferenceClient import asyncio import subprocess from datasets import load_dataset # 현재 작업 디렉토리 출력 print("Current Working Directory:", os.getcwd()) # 데이터셋 파일 이름 data_files = ['train_0.csv', 'train_1.csv', 'train_2.csv', 'train_3.csv', 'train_4.csv', 'train_5.csv'] # 현재 작업 디렉토리에 모든 파일이 있는지 확인 missing_files = [file for file in data_files if not os.path.exists(file)] if missing_files: print(f"Missing files: {missing_files}") # 필요한 경우 작업 디렉토리 변경 os.chdir('/home/user/app') print("Changed directory to:", os.getcwd()) else: print("All files are present in the current directory.") # 데이터셋 로드 law_dataset = load_dataset('csv', data_files=data_files) print("Dataset loaded successfully.") # 로깅 설정 logging.basicConfig(level=logging.DEBUG, format='%(asctime)s:%(levelname)s:%(name)s: %(message)s', handlers=[logging.StreamHandler()]) # 인텐트 설정 intents = discord.Intents.default() intents.message_content = True intents.messages = True intents.guilds = True intents.guild_messages = True # 추론 API 클라이언트 설정 hf_client = InferenceClient("CohereForAI/c4ai-command-r-plus", token=os.getenv("HF_TOKEN")) # 특정 채널 ID SPECIFIC_CHANNEL_ID = int(os.getenv("DISCORD_CHANNEL_ID")) # 대화 히스토리를 저장할 전역 변수 conversation_history = [] class MyClient(discord.Client): def __init__(self, *args, **kwargs): super().__init__(*args, **kwargs) self.is_processing = False async def on_ready(self): logging.info(f'{self.user}로 로그인되었습니다!') subprocess.Popen(["python", "web.py"]) logging.info("Web.py server has been started.") async def on_message(self, message): if message.author == self.user: return if not self.is_message_in_specific_channel(message): return if self.is_processing: logging.debug("Currently processing another message, skipping this one.") return self.is_processing = True try: response = await generate_response(message) if response.strip() == "": response = "죄송합니다, 제공할 수 있는 정보가 없습니다." await message.channel.send(response) finally: self.is_processing = False logging.debug("Message processing completed, ready for the next one.") def is_message_in_specific_channel(self, message): channel_condition = message.channel.id == SPECIFIC_CHANNEL_ID thread_condition = isinstance(message.channel, discord.Thread) and message.channel.parent_id == SPECIFIC_CHANNEL_ID return channel_condition or thread_condition async def generate_response(message): global conversation_history case_number = message.content.strip() user_mention = message.author.mention system_message = f"{user_mention}, 다음은 요청하신 사건의 전문입니다:" # 데이터 검색 및 응답 준비 full_text = search_in_dataset_by_case_number(case_number, law_dataset) full_response_text = f"{system_message}\n\n{full_text}" max_length = 2000 if len(full_response_text) > max_length: for i in range(0, len(full_response_text), max_length): part_response = full_response_text[i:i+max_length] await message.channel.send(part_response) logging.debug(f'Full model response sent: {full_response_text}') conversation_history.append({"role": "assistant", "content": full_response_text}) else: await message.channel.send(full_response_text) logging.debug(f'Full model response sent: {full_response_text}') conversation_history.append({"role": "assistant", "content": full_response_text}) def search_in_dataset(query, dataset): for record in dataset['train']: # '사건명'이 None이 아닌지 확인하고, query가 해당 값에 포함되어 있는지 검사 if record['사건명'] and query in record['사건명']: return record['사건번호'] return "관련 법률 정보를 찾을 수 없습니다." def search_in_dataset_by_case_number(case_number, dataset): for record in dataset['train']: if record['사건번호'] == case_number: return record.get('전문', "해당 사건에 대한 전문 정보가 없습니다.") return "관련 법률 정보를 찾을 수 없습니다." if __name__ == "__main__": discord_client = MyClient(intents=intents) discord_client.run(os.getenv('DISCORD_TOKEN'))