isync/parse-mail.py

   1 #!/usr/bin/env python3
   2
   3 from datetime import timedelta
   4 from email import message_from_file
   5 import re
   6 import sys
   7 import youtube_dl
   8
   9
  10 FROM    = "{uploader} <noreply@youtube.com>"
  11 SUBJECT = "{title}"
  12 CONTENT = """\
  13 {uploader} just uploaded a video
  14 {title} ({duration})
  15
  16 {webpage_url}
  17
  18 Description
  19 ───────────────────────────────────────────────────────────────────────
  20
  21 {description}
  22
  23 ───────────────────────────────────────────────────────────────────────
  24
  25 Manage notifications: https://www.youtube.com/subscription_manager
  26 """
  27
  28
  29 def parse_file(fd):
  30     msg = message_from_file(fd)
  31
  32     payload = ""
  33     if msg.is_multipart():
  34         for part in msg.get_payload():
  35             if part.get_content_type() != 'text/plain':
  36                 continue
  37             charset = part.get_content_charset()
  38             payload = part.get_payload(decode=True).decode(charset)
  39     else:
  40         payload = msg.get_payload(decode=True).decode(charset)
  41
  42     url = re.search(r'^https?://www.youtube.com/watch\?v=[\w-]{11}', payload, re.MULTILINE).group()
  43     # print(url)
  44     with youtube_dl.YoutubeDL({'quiet' : True}) as ytdl:
  45         info = ytdl.extract_info(url, download=False, force_generic_extractor=True)
  46
  47     if info is None:
  48         return False
  49
  50     info['duration'] = timedelta(seconds=info['duration'])
  51
  52     del msg['From']
  53     msg['From'] = FROM.format(**info)
  54
  55     del msg['Subject']
  56     msg['Subject'] = SUBJECT.format(**info)
  57
  58     msg.set_payload(CONTENT.format(**info))
  59
  60     return msg
  61
  62
  63 if __name__ == "__main__":
  64     if len(sys.argv) == 1:
  65         msg = parse_file(sys.stdin)
  66     else:
  67         with open(sys.argv[1], 'r') as fd:
  68             msg = parse_file(fd)
  69     print(msg)