File size: 9,969 Bytes
78efe79
440418c
302cdd6
7edf3cc
dc80b35
6a13d8f
3cbcff7
7524905
b4bdc7a
302cdd6
7524905
f2ef6a6
4f7204d
848278a
 
f53f3e8
 
 
4f7204d
 
407a575
32c38ef
672b059
440418c
1831164
440418c
22dee1c
440418c
22dee1c
 
08baccf
dc80b35
b17754f
848278a
dc80b35
829fd20
 
 
1e3100b
302cdd6
 
f05454f
3cbcff7
 
 
302cdd6
 
 
 
3cbcff7
302cdd6
 
 
 
 
 
3cbcff7
 
 
 
302cdd6
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
fa5c0c1
302cdd6
fa5c0c1
 
 
 
 
848278a
 
 
 
 
 
 
 
 
302cdd6
fa5c0c1
302cdd6
 
 
 
 
 
 
 
 
 
3cbcff7
 
 
 
 
 
302cdd6
 
 
 
 
 
 
 
 
 
3cbcff7
302cdd6
 
fa5c0c1
 
 
 
 
6c4b884
 
 
fa5c0c1
 
 
 
7524905
6c4b884
fa5c0c1
 
 
 
6c4b884
fa5c0c1
 
 
 
302cdd6
6c4b884
fa5c0c1
6c4b884
 
 
fa5c0c1
 
6c4b884
 
302cdd6
fa5c0c1
6c4b884
fa5c0c1
 
 
 
7524905
fa5c0c1
7524905
f53f3e8
 
848278a
 
 
 
 
 
 
f53f3e8
 
 
 
 
 
 
 
 
 
 
 
 
 
848278a
 
 
302cdd6
f53f3e8
848278a
 
 
f53f3e8
6c4b884
302cdd6
 
 
 
 
 
 
 
 
 
 
829fd20
1e3100b
3cbcff7
302cdd6
 
 
 
 
 
7524905
302cdd6
 
e2de219
 
 
 
 
 
 
 
302cdd6
1e3100b
3cbcff7
302cdd6
 
3cbcff7
302cdd6
3cbcff7
 
302cdd6
 
 
 
 
7524905
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
import discord
import logging
import os
import re
import asyncio
import subprocess
import aiohttp
import time
from huggingface_hub import InferenceClient
from googleapiclient.discovery import build
from youtube_transcript_api import YouTubeTranscriptApi, TranscriptsDisabled, NoTranscriptFound
from youtube_transcript_api.formatters import TextFormatter
from dotenv import load_dotenv
from pytube import YouTube
import whisper
import torch
from transformers import AutoProcessor, AutoModelForSpeechSeq2Seq
import librosa
# ν™˜κ²½ λ³€μˆ˜ λ‘œλ“œ
load_dotenv()

# λ‘œκΉ… μ„€μ •
logging.basicConfig(level=logging.DEBUG, format='%(asctime)s:%(levelname)s:%(name)s:%(message)s', handlers=[logging.StreamHandler()])

# μΈν…νŠΈ μ„€μ •
intents = discord.Intents.default()
intents.message_content = True
intents.messages = True
intents.guilds = True
intents.guild_messages = True

# μΆ”λ‘  API ν΄λΌμ΄μ–ΈνŠΈ μ„€μ •
hf_client = InferenceClient("CohereForAI/c4ai-command-r-plus-08-2024", token=os.getenv("HF_TOKEN"))
whisper_client = InferenceClient("openai/whisper-large-v3", token=os.getenv("HF_TOKEN"))

# YouTube API μ„€μ •
API_KEY = os.getenv("YOUTUBE_API_KEY")
youtube_service = build('youtube', 'v3', developerKey=API_KEY)

# νŠΉμ • 채널 ID
SPECIFIC_CHANNEL_ID = int(os.getenv("DISCORD_CHANNEL_ID"))

# 전솑 μ‹€νŒ¨ μ‹œ μž¬μ‹œλ„ 횟수
MAX_RETRIES = 3

class MyClient(discord.Client):
    def __init__(self, *args, **kwargs):
        super().__init__(*args, **kwargs)
        self.is_processing = False
        self.session = None

    async def on_ready(self):
        logging.info(f'{self.user}둜 λ‘œκ·ΈμΈλ˜μ—ˆμŠ΅λ‹ˆλ‹€!')

        # web.py 파일 μ‹€ν–‰
        subprocess.Popen(["python", "web.py"])
        logging.info("Web.py μ„œλ²„κ°€ μ‹œμž‘λ˜μ—ˆμŠ΅λ‹ˆλ‹€.")

        # aiohttp ν΄λΌμ΄μ–ΈνŠΈ μ„Έμ…˜ 생성
        self.session = aiohttp.ClientSession()

        # 봇이 μ‹œμž‘λ  λ•Œ μ•ˆλ‚΄ λ©”μ‹œμ§€λ₯Ό 전솑
        channel = self.get_channel(SPECIFIC_CHANNEL_ID)
        if channel:
            await channel.send("유튜브 λΉ„λ””μ˜€ URL을 μž…λ ₯ν•˜λ©΄, μžλ§‰κ³Ό λŒ“κΈ€μ„ 기반으둜 닡글을 μž‘μ„±ν•©λ‹ˆλ‹€.")

    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:
            return
        self.is_processing = True
        try:
            video_id = extract_video_id(message.content)
            if video_id:
                transcript, language = await get_best_available_transcript(video_id)
                comments = await get_video_comments(video_id)
                if comments:
                    if transcript:
                        replies = await generate_replies(comments, transcript)
                        await create_thread_and_send_replies(message, video_id, comments, replies, self.session)
                    else:
                        await message.channel.send("μžλ§‰μ„ κ°€μ Έμ˜¬ 수 μ—†μŠ΅λ‹ˆλ‹€. Whisper λͺ¨λΈμ„ μ‚¬μš©ν•˜μ—¬ μžλ§‰μ„ μƒμ„±ν•©λ‹ˆλ‹€.")
                        transcript = await generate_whisper_transcript(video_id)
                        if transcript:
                            replies = await generate_replies(comments, transcript)
                            await create_thread_and_send_replies(message, video_id, comments, replies, self.session)
                        else:
                            await message.channel.send("Whisper λͺ¨λΈλ‘œλ„ μžλ§‰μ„ 생성할 수 μ—†μŠ΅λ‹ˆλ‹€. λŒ“κΈ€λ§Œμ„ 기반으둜 닡변을 μƒμ„±ν•©λ‹ˆλ‹€.")
                            replies = await generate_replies(comments, "")
                            await create_thread_and_send_replies(message, video_id, comments, replies, self.session)
                else:
                    await message.channel.send("λŒ“κΈ€μ„ κ°€μ Έμ˜¬ 수 μ—†μŠ΅λ‹ˆλ‹€.")
            else:
                await message.channel.send("μœ νš¨ν•œ 유튜브 λΉ„λ””μ˜€ URL을 μ œκ³΅ν•΄ μ£Όμ„Έμš”.")
        finally:
            self.is_processing = False

    def is_message_in_specific_channel(self, message):
        return message.channel.id == SPECIFIC_CHANNEL_ID or (
            isinstance(message.channel, discord.Thread) and message.channel.parent_id == SPECIFIC_CHANNEL_ID
        )

    async def close(self):
        # aiohttp ν΄λΌμ΄μ–ΈνŠΈ μ„Έμ…˜ μ’…λ£Œ
        if self.session:
            await self.session.close()
        await super().close()

def extract_video_id(url):
    video_id = None
    youtube_regex = (
        r'(https?://)?(www\.)?'
        '(youtube|youtu|youtube-nocookie)\.(com|be)/'
        '(watch\?v=|embed/|v/|.+\?v=)?([^&=%\?]{11})')

    match = re.match(youtube_regex, url)
    if match:
        video_id = match.group(6)
    logging.debug(f'μΆ”μΆœλœ λΉ„λ””μ˜€ ID: {video_id}')
    return video_id

async def get_best_available_transcript(video_id, max_retries=5, delay=10):
    async def fetch_transcript(language):
        try:
            transcript = await asyncio.to_thread(YouTubeTranscriptApi.get_transcript, video_id, languages=[language])
            return transcript, language
        except (NoTranscriptFound, TranscriptsDisabled):
            logging.warning(f'{language} μžλ§‰μ΄ μ œκ³΅λ˜μ§€ μ•ŠμŒ.')
            return None, None
        except Exception as e:
            logging.warning(f'{language} μžλ§‰ κ°€μ Έμ˜€κΈ° 였λ₯˜: {e}')
            return None, None

    for attempt in range(max_retries):
        # μš°μ„  ν•œκ΅­μ–΄ μžλ§‰μ„ μ‹œλ„
        ko_transcript, ko_lang = await fetch_transcript('ko')
        if ko_transcript:
            return ko_transcript, ko_lang

        # μ˜μ–΄ μžλ§‰ μ‹œλ„
        en_transcript, en_lang = await fetch_transcript('en')
        if en_transcript:
            return en_transcript, en_lang

        try:
            # λΉ„λ””μ˜€μ— μžλ§‰ λͺ©λ‘μ΄ μžˆλŠ”μ§€ 확인
            transcripts = await asyncio.to_thread(YouTubeTranscriptApi.list_transcripts, video_id)

            # μˆ˜λ™μœΌλ‘œ μƒμ„±λœ μžλ§‰μ΄ μžˆλŠ”μ§€ 확인
            manual_transcript = transcripts.find_manually_created_transcript(['ko', 'en'])
            transcript = await asyncio.to_thread(manual_transcript.fetch)
            return transcript, manual_transcript.language_code
        except (NoTranscriptFound, TranscriptsDisabled) as e:
            logging.warning(f'μˆ˜λ™ μžλ§‰μ„ 찾을 수 μ—†μŒ: {e}')
        except Exception as e:
            if attempt < max_retries - 1:
                logging.error(f'μžλ§‰ κ°€μ Έμ˜€κΈ° μ‹€νŒ¨ (μ‹œλ„ {attempt + 1}/{max_retries}): {e}')
                await asyncio.sleep(delay)
            else:
                logging.error(f'μ΅œμ’… μžλ§‰ κ°€μ Έμ˜€κΈ° μ‹€νŒ¨: {e}')
                return None, None

    return None, None



async def generate_whisper_transcript(video_id):
    try:
        # YouTube λΉ„λ””μ˜€ λ‹€μš΄λ‘œλ“œ
        yt = YouTube(f'https://www.youtube.com/watch?v={video_id}')
        audio_stream = yt.streams.filter(only_audio=True).first()
        audio_file = audio_stream.download(output_path='temp', filename=f'{video_id}.mp3')

        # μ˜€λ””μ˜€ 파일 λ‘œλ“œ
        audio, sr = librosa.load(audio_file, sr=16000)

        # Whisper λͺ¨λΈ 및 ν”„λ‘œμ„Έμ„œ λ‘œλ“œ
        device = "cuda" if torch.cuda.is_available() else "cpu"
        processor = AutoProcessor.from_pretrained("openai/whisper-large-v3")
        model = AutoModelForSpeechSeq2Seq.from_pretrained("openai/whisper-large-v3").to(device)

        # μ˜€λ””μ˜€ 처리
        input_features = processor(audio, sampling_rate=sr, return_tensors="pt").input_features.to(device)

        # 생성
        predicted_ids = model.generate(input_features)
        transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)

        # μž„μ‹œ 파일 μ‚­μ œ
        os.remove(audio_file)

        return transcription[0]
    except Exception as e:
        logging.error(f'Whisper μžλ§‰ 생성 μ‹€νŒ¨: {e}')
        return None
        

async def get_video_comments(video_id):
    comments = []
    response = youtube_service.commentThreads().list(
        part='snippet',
        videoId=video_id,
        maxResults=100  # μ΅œλŒ€ 100개의 λŒ“κΈ€ κ°€μ Έμ˜€κΈ°
    ).execute()

    for item in response.get('items', []):
        comment = item['snippet']['topLevelComment']['snippet']['textOriginal']
        comment_id = item['snippet']['topLevelComment']['id']
        comments.append((comment, comment_id))
    
    logging.debug(f'κ°€μ Έμ˜¨ λŒ“κΈ€: {comments}')
    return comments

async def generate_replies(comments, transcript):
    replies = []
    for comment, _ in comments:
        messages = [
            {"role": "system", "content": f"""λ„ˆμ˜ 이름은 OpenFreeAI이닀. λ‹΅κΈ€ 생성후 κ°€μž₯ λ§ˆμ§€λ§‰μ— λ„ˆμ˜ 이름을 밝히고 κ³΅μ†ν•˜κ²Œ μΈμ‚¬ν•˜λΌ. λΉ„λ””μ˜€ μžλ§‰: {transcript}"""},
            {"role": "user", "content": comment}
        ]
        loop = asyncio.get_event_loop()
        response = await loop.run_in_executor(None, lambda: hf_client.chat_completion(
            messages, max_tokens=250, temperature=0.7, top_p=0.85))

        if response.choices and response.choices[0].message:
            reply = response.choices[0].message['content'].strip()
        else:
            reply = "닡글을 생성할 수 μ—†μŠ΅λ‹ˆλ‹€."
        replies.append(reply)
    
    logging.debug(f'μƒμ„±λœ λ‹΅κΈ€: {replies}')
    return replies

async def create_thread_and_send_replies(message, video_id, comments, replies, session):
    thread = await message.channel.create_thread(name=f"{message.author.name}의 λŒ“κΈ€ λ‹΅κΈ€", message=message)

    for (comment, comment_id), reply in zip(comments, replies):
        embed = discord.Embed(description=f"**λŒ“κΈ€**: {comment}\n**λ‹΅κΈ€**: {reply}")
        await thread.send(embed=embed)

if __name__ == "__main__":
    discord_client = MyClient(intents=intents)
    discord_client.run(os.getenv('DISCORD_TOKEN'))