dahyedahye commited on
Commit
438fb2c
Β·
1 Parent(s): 0be25e9
Files changed (6) hide show
  1. .env +2 -2
  2. .env.backup +2 -0
  3. app.py +35 -20
  4. app.py.backup +126 -0
  5. env_backup +2 -0
  6. requirements.txt +2 -1
.env CHANGED
@@ -1,2 +1,2 @@
1
- SMTP_USER=kyowoon.lee1924@gmail.com
2
- SMTP_PASSWORD=rhwy amyv hdwq pspa
 
1
+ FROM_EMAIL=leekwoon@unist.ac.kr
2
+ POSTMARK_API_KEY=db96eba8-a4b5-4b83-9377-2e7dcdc00d39
.env.backup ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ SMTP_USER=kyowoon.lee1924@gmail.com
2
+ SMTP_PASSWORD=rhwy amyv hdwq pspa
app.py CHANGED
@@ -10,6 +10,8 @@ from concurrent.futures import ThreadPoolExecutor
10
  from fastapi import FastAPI, HTTPException, BackgroundTasks
11
  from fastapi.responses import JSONResponse
12
  from dotenv import load_dotenv
 
 
13
 
14
  # ꡬ글 λ“œλΌμ΄λΈŒ 링크, μ–΄λ–€ μ–Έμ–΄, 받을 이메일 μ£Όμ†Œ
15
 
@@ -29,6 +31,8 @@ app = FastAPI(
29
  # Whisper λͺ¨λΈ λ‘œλ“œ
30
  model = whisper.load_model("tiny")
31
 
 
 
32
  executor = ThreadPoolExecutor(max_workers=3) # μ΅œλŒ€ 3개의 μŠ€λ ˆλ“œλ‘œ 비동기 μž‘μ—… 처리
33
 
34
  def extract_file_id(drive_url: str) -> str:
@@ -47,30 +51,41 @@ def extract_file_id(drive_url: str) -> str:
47
  raise ValueError("Invalid Google Drive URL")
48
 
49
  def send_email(to_email: str, srt_file_path: str, transcription_time: float):
50
- smtp_server = "smtp.gmail.com"
51
- smtp_port = 587
52
- smtp_user = os.getenv("SMTP_USER")
53
- smtp_password = os.getenv("SMTP_PASSWORD")
54
 
55
  subject = "[kyobody - μžλ§‰μƒμ„±] μž‘μ—…μ΄ μ™„λ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€."
56
  body = f"[kyobody - μžλ§‰μƒμ„±] μž‘μ—…μ΄ μ™„λ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€. 총 μ†Œμš” μ‹œκ°„: {transcription_time:.2f} 초. SRT νŒŒμΌμ„ μ²¨λΆ€ν•˜μ—¬ μ „λ‹¬λ“œλ¦½λ‹ˆλ‹€."
57
 
58
- msg = MIMEMultipart()
59
- msg["From"] = smtp_user
60
- msg["To"] = to_email
61
- msg["Subject"] = subject
62
-
63
- msg.attach(MIMEText(body, "plain"))
64
-
65
- with open(srt_file_path, "r") as file:
66
- attachment = MIMEText(file.read())
67
- attachment.add_header("Content-Disposition", "attachment", filename=os.path.basename(srt_file_path))
68
- msg.attach(attachment)
69
-
70
- with smtplib.SMTP(smtp_server, smtp_port) as server:
71
- server.starttls()
72
- server.login(smtp_user, smtp_password)
73
- server.sendmail(smtp_user, to_email, msg.as_string())
 
 
 
 
 
 
 
 
 
 
 
74
 
75
  def transcribe_and_send_email(temp_input_file: str, srt_file_path: str, email: str, language: str):
76
  try:
 
10
  from fastapi import FastAPI, HTTPException, BackgroundTasks
11
  from fastapi.responses import JSONResponse
12
  from dotenv import load_dotenv
13
+ from postmarker.core import PostmarkClient
14
+
15
 
16
  # ꡬ글 λ“œλΌμ΄λΈŒ 링크, μ–΄λ–€ μ–Έμ–΄, 받을 이메일 μ£Όμ†Œ
17
 
 
31
  # Whisper λͺ¨λΈ λ‘œλ“œ
32
  model = whisper.load_model("tiny")
33
 
34
+ postmark = PostmarkClient(server_token=os.getenv("POSTMARK_API_KEY"))
35
+
36
  executor = ThreadPoolExecutor(max_workers=3) # μ΅œλŒ€ 3개의 μŠ€λ ˆλ“œλ‘œ 비동기 μž‘μ—… 처리
37
 
38
  def extract_file_id(drive_url: str) -> str:
 
51
  raise ValueError("Invalid Google Drive URL")
52
 
53
  def send_email(to_email: str, srt_file_path: str, transcription_time: float):
54
+ # smtp_server = "smtp.gmail.com"
55
+ # smtp_port = 587
56
+ # smtp_user = os.getenv("SMTP_USER")
57
+ # smtp_password = os.getenv("SMTP_PASSWORD")
58
 
59
  subject = "[kyobody - μžλ§‰μƒμ„±] μž‘μ—…μ΄ μ™„λ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€."
60
  body = f"[kyobody - μžλ§‰μƒμ„±] μž‘μ—…μ΄ μ™„λ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€. 총 μ†Œμš” μ‹œκ°„: {transcription_time:.2f} 초. SRT νŒŒμΌμ„ μ²¨λΆ€ν•˜μ—¬ μ „λ‹¬λ“œλ¦½λ‹ˆλ‹€."
61
 
62
+ # msg = MIMEMultipart()
63
+ # msg["From"] = smtp_user
64
+ # msg["To"] = to_email
65
+ # msg["Subject"] = subject
66
+
67
+ # msg.attach(MIMEText(body, "plain"))
68
+
69
+ # with open(srt_file_path, "r") as file:
70
+ # attachment = MIMEText(file.read())
71
+ # attachment.add_header("Content-Disposition", "attachment", filename=os.path.basename(srt_file_path))
72
+ # msg.attach(attachment)
73
+
74
+ # with smtplib.SMTP(smtp_server, smtp_port) as server:
75
+ # server.starttls()
76
+ # server.login(smtp_user, smtp_password)
77
+ # server.sendmail(smtp_user, to_email, msg.as_string())
78
+
79
+ email = postmark.emails.Email(
80
+ From=os.getenv("FROM_EMAIL"),
81
+ To=to_email,
82
+ Subject=subject,
83
+ # HtmlBody='<html><body><strong>Hello</strong> dear Postmark user.</body></html>'
84
+ HtmlBody=body
85
+ )
86
+ email['X-Accept-Language'] = 'ko'
87
+ email.attach(srt_file_path)
88
+ email.send()
89
 
90
  def transcribe_and_send_email(temp_input_file: str, srt_file_path: str, email: str, language: str):
91
  try:
app.py.backup ADDED
@@ -0,0 +1,126 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import os
2
+ import re
3
+ import uuid
4
+ import gdown
5
+ import whisper
6
+ import smtplib
7
+ from email.mime.text import MIMEText
8
+ from email.mime.multipart import MIMEMultipart
9
+ from concurrent.futures import ThreadPoolExecutor
10
+ from fastapi import FastAPI, HTTPException, BackgroundTasks
11
+ from fastapi.responses import JSONResponse
12
+ from dotenv import load_dotenv
13
+
14
+ # ꡬ글 λ“œλΌμ΄λΈŒ 링크, μ–΄λ–€ μ–Έμ–΄, 받을 이메일 μ£Όμ†Œ
15
+
16
+ # .env νŒŒμΌμ—μ„œ μ€‘μš”ν•œ ν™˜κ²½ λ³€μˆ˜ λ‘œλ“œ
17
+ load_dotenv()
18
+
19
+ app = FastAPI(
20
+ version="0.0.1",
21
+ servers=[
22
+ {
23
+ "url": "https://leekwoon-whisper-api.hf.space",
24
+ "description": "video/audio transcription API",
25
+ }
26
+ ],
27
+ )
28
+
29
+ # Whisper λͺ¨λΈ λ‘œλ“œ
30
+ model = whisper.load_model("tiny")
31
+
32
+ executor = ThreadPoolExecutor(max_workers=3) # μ΅œλŒ€ 3개의 μŠ€λ ˆλ“œλ‘œ 비동기 μž‘μ—… 처리
33
+
34
+ def extract_file_id(drive_url: str) -> str:
35
+ """
36
+ Google Drive URLμ—μ„œ 파일 IDλ₯Ό μΆ”μΆœν•©λ‹ˆλ‹€.
37
+ """
38
+ match = re.search(r'/d/([a-zA-Z0-9_-]+)', drive_url)
39
+ if match:
40
+ return match.group(1)
41
+ match = re.search(r'file/d/([a-zA-Z0-9_-]+)', drive_url)
42
+ if match:
43
+ return match.group(1)
44
+ match = re.search(r'([a-zA-Z0-9_-]{33,})', drive_url)
45
+ if match:
46
+ return match.group(1)
47
+ raise ValueError("Invalid Google Drive URL")
48
+
49
+ def send_email(to_email: str, srt_file_path: str, transcription_time: float):
50
+ smtp_server = "smtp.gmail.com"
51
+ smtp_port = 587
52
+ smtp_user = os.getenv("SMTP_USER")
53
+ smtp_password = os.getenv("SMTP_PASSWORD")
54
+
55
+ subject = "[kyobody - μžλ§‰μƒμ„±] μž‘μ—…μ΄ μ™„λ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€."
56
+ body = f"[kyobody - μžλ§‰μƒμ„±] μž‘μ—…μ΄ μ™„λ£Œλ˜μ—ˆμŠ΅λ‹ˆλ‹€. 총 μ†Œμš” μ‹œκ°„: {transcription_time:.2f} 초. SRT νŒŒμΌμ„ μ²¨λΆ€ν•˜μ—¬ μ „λ‹¬λ“œλ¦½λ‹ˆλ‹€."
57
+
58
+ msg = MIMEMultipart()
59
+ msg["From"] = smtp_user
60
+ msg["To"] = to_email
61
+ msg["Subject"] = subject
62
+
63
+ msg.attach(MIMEText(body, "plain"))
64
+
65
+ with open(srt_file_path, "r") as file:
66
+ attachment = MIMEText(file.read())
67
+ attachment.add_header("Content-Disposition", "attachment", filename=os.path.basename(srt_file_path))
68
+ msg.attach(attachment)
69
+
70
+ with smtplib.SMTP(smtp_server, smtp_port) as server:
71
+ server.starttls()
72
+ server.login(smtp_user, smtp_password)
73
+ server.sendmail(smtp_user, to_email, msg.as_string())
74
+
75
+ def transcribe_and_send_email(temp_input_file: str, srt_file_path: str, email: str, language: str):
76
+ try:
77
+ # Transcribe the video/audio file
78
+ import time
79
+ start_time = time.time()
80
+ result = model.transcribe(temp_input_file, language=language)
81
+ transcription_time = time.time() - start_time
82
+
83
+ # Save the transcription to an SRT file
84
+ with open(srt_file_path, "w") as srt_file:
85
+ for i, segment in enumerate(result["segments"]):
86
+ start = segment['start']
87
+ end = segment['end']
88
+ text = segment['text'][1:]
89
+
90
+ start_time = f"{int(start // 3600):02}:{int((start % 3600) // 60):02}:{int(start % 60):02},{int((start * 1000) % 1000):03}"
91
+ end_time = f"{int(end // 3600):02}:{int((end % 3600) // 60):02}:{int(end % 60):02},{int((end * 1000) % 1000):03}"
92
+
93
+ srt_file.write(f"{i + 1}\n")
94
+ srt_file.write(f"{start_time} --> {end_time}\n")
95
+ srt_file.write(f"{text}\n\n")
96
+
97
+ # Send the result via email
98
+ send_email(email, srt_file_path, transcription_time)
99
+
100
+ # Clean up the temporary files
101
+ os.remove(temp_input_file)
102
+ os.remove(srt_file_path)
103
+
104
+ except Exception as e:
105
+ raise e
106
+
107
+ @app.post("/transcribe/")
108
+ def transcribe_video(url: str, email: str, background_tasks: BackgroundTasks, language: str = "en"):
109
+ try:
110
+ # Extract file ID and download the file
111
+ file_id = extract_file_id(url)
112
+ download_url = f"https://drive.google.com/uc?id={file_id}"
113
+ temp_input_file = f'/tmp/{uuid.uuid4()}.mp4'
114
+ gdown.download(download_url, temp_input_file, quiet=False)
115
+
116
+ # Define SRT file path
117
+ srt_file_path = f'/tmp/{uuid.uuid4()}.srt'
118
+
119
+ # Schedule the transcription and email sending in the background
120
+ background_tasks.add_task(executor.submit, transcribe_and_send_email, temp_input_file, srt_file_path, email, language)
121
+
122
+ # Respond to the client immediately
123
+ return JSONResponse(status_code=202, content={"message": "Transcription started, you will receive an email when it's done."})
124
+
125
+ except Exception as e:
126
+ raise HTTPException(status_code=500, detail=str(e))
env_backup ADDED
@@ -0,0 +1,2 @@
 
 
 
1
+ SMTP_USER=kyowoon.lee1924@gmail.com
2
+ SMTP_PASSWORD=rhwy amyv hdwq pspa
requirements.txt CHANGED
@@ -4,4 +4,5 @@ openai-whisper # Whisper 라이브러리λ₯Ό μ„€μΉ˜ν•©λ‹ˆλ‹€.
4
  fastapi # FastAPI ν”„λ ˆμž„μ›Œν¬
5
  uvicorn # FastAPI μ„œλ²„ 싀행을 μœ„ν•œ Uvicorn ASGI μ„œλ²„
6
  python-dotenv
7
- gdown
 
 
4
  fastapi # FastAPI ν”„λ ˆμž„μ›Œν¬
5
  uvicorn # FastAPI μ„œλ²„ 싀행을 μœ„ν•œ Uvicorn ASGI μ„œλ²„
6
  python-dotenv
7
+ gdown
8
+ requests