yt_dlp/postprocessor/sponsorblock.py

   1 from hashlib import sha256
   2 import itertools
   3 import json
   4 import re
   5 import time
   6
   7 from .ffmpeg import FFmpegPostProcessor
   8 from ..compat import compat_urllib_parse_urlencode, compat_HTTPError
   9 from ..utils import PostProcessingError, network_exceptions, sanitized_Request
  10
  11
  12 class SponsorBlockPP(FFmpegPostProcessor):
  13
  14     EXTRACTORS = {
  15         'Youtube': 'YouTube',
  16     }
  17     CATEGORIES = {
  18         'sponsor': 'Sponsor',
  19         'intro': 'Intermission/Intro Animation',
  20         'outro': 'Endcards/Credits',
  21         'selfpromo': 'Unpaid/Self Promotion',
  22         'interaction': 'Interaction Reminder',
  23         'preview': 'Preview/Recap',
  24         'music_offtopic': 'Non-Music Section'
  25     }
  26
  27     def __init__(self, downloader, categories=None, api='https://sponsor.ajay.app'):
  28         FFmpegPostProcessor.__init__(self, downloader)
  29         self._categories = tuple(categories or self.CATEGORIES.keys())
  30         self._API_URL = api if re.match('^https?://', api) else 'https://' + api
  31
  32     def run(self, info):
  33         extractor = info['extractor_key']
  34         if extractor not in self.EXTRACTORS:
  35             self.to_screen(f'SponsorBlock is not supported for {extractor}')
  36             return [], info
  37
  38         self.to_screen('Fetching SponsorBlock segments')
  39         info['sponsorblock_chapters'] = self._get_sponsor_chapters(info, info['duration'])
  40         return [], info
  41
  42     def _get_sponsor_chapters(self, info, duration):
  43         segments = self._get_sponsor_segments(info['id'], self.EXTRACTORS[info['extractor_key']])
  44
  45         def duration_filter(s):
  46             start_end = s['segment']
  47             # Ignore milliseconds difference at the start.
  48             if start_end[0] <= 1:
  49                 start_end[0] = 0
  50             # Ignore milliseconds difference at the end.
  51             # Never allow the segment to exceed the video.
  52             if duration and duration - start_end[1] <= 1:
  53                 start_end[1] = duration
  54             # SponsorBlock duration may be absent or it may deviate from the real one.
  55             return s['videoDuration'] == 0 or not duration or abs(duration - s['videoDuration']) <= 1
  56
  57         duration_match = [s for s in segments if duration_filter(s)]
  58         if len(duration_match) != len(segments):
  59             self.report_warning('Some SponsorBlock segments are from a video of different duration, maybe from an old version of this video')
  60
  61         def to_chapter(s):
  62             (start, end), cat = s['segment'], s['category']
  63             return {
  64                 'start_time': start,
  65                 'end_time': end,
  66                 'category': cat,
  67                 'title': self.CATEGORIES[cat],
  68                 '_categories': [(cat, start, end)]
  69             }
  70
  71         sponsor_chapters = [to_chapter(s) for s in duration_match]
  72         if not sponsor_chapters:
  73             self.to_screen('No segments were found in the SponsorBlock database')
  74         else:
  75             self.to_screen(f'Found {len(sponsor_chapters)} segments in the SponsorBlock database')
  76         return sponsor_chapters
  77
  78     def _get_sponsor_segments(self, video_id, service):
  79         hash = sha256(video_id.encode('ascii')).hexdigest()
  80         # SponsorBlock API recommends using first 4 hash characters.
  81         url = f'{self._API_URL}/api/skipSegments/{hash[:4]}?' + compat_urllib_parse_urlencode({
  82             'service': service,
  83             'categories': json.dumps(self._categories),
  84         })
  85         self.write_debug(f'SponsorBlock query: {url}')
  86         for d in self._get_json(url):
  87             if d['videoID'] == video_id:
  88                 return d['segments']
  89         return []
  90
  91     def _get_json(self, url):
  92         # While this is not an extractor, it behaves similar to one and
  93         # so obey extractor_retries and sleep_interval_requests
  94         max_retries = self.get_param('extractor_retries', 3)
  95         sleep_interval = self.get_param('sleep_interval_requests') or 0
  96         for retries in itertools.count():
  97             try:
  98                 rsp = self._downloader.urlopen(sanitized_Request(url))
  99                 return json.loads(rsp.read().decode(rsp.info().get_param('charset') or 'utf-8'))
 100             except network_exceptions as e:
 101                 if isinstance(e, compat_HTTPError) and e.code == 404:
 102                     return []
 103                 if retries < max_retries:
 104                     self.report_warning(f'{e}. Retrying...')
 105                     if sleep_interval > 0:
 106                         self.to_screen(f'Sleeping {sleep_interval} seconds ...')
 107                         time.sleep(sleep_interval)
 108                     continue
 109                 raise PostProcessingError(f'Unable to communicate with SponsorBlock API: {e}')