yt_dlp/extractor/tf1.py

   1 import json
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     int_or_none,
   6     parse_iso8601,
   7     try_get,
   8 )
   9
  10
  11 class TF1IE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:www\.)?tf1\.fr/[^/]+/(?P<program_slug>[^/]+)/videos/(?P<id>[^/?&#]+)\.html'
  13     _TESTS = [{
  14         'url': 'https://www.tf1.fr/tmc/quotidien-avec-yann-barthes/videos/quotidien-premiere-partie-11-juin-2019.html',
  15         'info_dict': {
  16             'id': '13641379',
  17             'ext': 'mp4',
  18             'title': 'md5:f392bc52245dc5ad43771650c96fb620',
  19             'description': 'md5:a02cdb217141fb2d469d6216339b052f',
  20             'upload_date': '20190611',
  21             'timestamp': 1560273989,
  22             'duration': 1738,
  23             'series': 'Quotidien avec Yann Barthès',
  24             'tags': ['intégrale', 'quotidien', 'Replay'],
  25         },
  26         'params': {
  27             # Sometimes wat serves the whole file with the --test option
  28             'skip_download': True,
  29         },
  30     }, {
  31         'url': 'https://www.tf1.fr/tmc/burger-quiz/videos/burger-quiz-du-19-aout-2023-s03-episode-21-85585666.html',
  32         'info_dict': {
  33             'id': '14010600',
  34             'ext': 'mp4',
  35             'title': 'Burger Quiz - S03 EP21 avec Eye Haidara, Anne Depétrini, Jonathan Zaccaï et Pio Marmaï',
  36             'thumbnail': 'https://photos.tf1.fr/1280/720/burger-quiz-11-9adb79-0@1x.jpg',
  37             'description': 'Manu Payet recevra Eye Haidara, Anne Depétrini, Jonathan Zaccaï et Pio Marmaï.',
  38             'upload_date': '20230819',
  39             'timestamp': 1692469471,
  40             'season_number': 3,
  41             'series': 'Burger Quiz',
  42             'episode_number': 21,
  43             'season': 'Season 3',
  44             'tags': 'count:13',
  45             'episode': 'Episode 21',
  46             'duration': 2312,
  47         },
  48         'params': {'skip_download': 'm3u8'},
  49     }, {
  50         'url': 'http://www.tf1.fr/tf1/koh-lanta/videos/replay-koh-lanta-22-mai-2015.html',
  51         'only_matching': True,
  52     }, {
  53         'url': 'http://www.tf1.fr/hd1/documentaire/videos/mylene-farmer-d-une-icone.html',
  54         'only_matching': True,
  55     }]
  56
  57     def _real_extract(self, url):
  58         program_slug, slug = self._match_valid_url(url).groups()
  59         video = self._download_json(
  60             'https://www.tf1.fr/graphql/web', slug, query={
  61                 'id': '9b80783950b85247541dd1d851f9cc7fa36574af015621f853ab111a679ce26f',
  62                 'variables': json.dumps({
  63                     'programSlug': program_slug,
  64                     'slug': slug,
  65                 }),
  66             })['data']['videoBySlug']
  67         wat_id = video['streamId']
  68
  69         tags = []
  70         for tag in (video.get('tags') or []):
  71             label = tag.get('label')
  72             if not label:
  73                 continue
  74             tags.append(label)
  75
  76         decoration = video.get('decoration') or {}
  77
  78         thumbnails = []
  79         for source in (try_get(decoration, lambda x: x['image']['sources'], list) or []):
  80             source_url = source.get('url')
  81             if not source_url:
  82                 continue
  83             thumbnails.append({
  84                 'url': source_url,
  85                 'width': int_or_none(source.get('width')),
  86             })
  87
  88         return {
  89             '_type': 'url_transparent',
  90             'id': wat_id,
  91             'url': 'wat:' + wat_id,
  92             'title': video.get('title'),
  93             'thumbnails': thumbnails,
  94             'description': decoration.get('description'),
  95             'timestamp': parse_iso8601(video.get('date')),
  96             'duration': int_or_none(try_get(video, lambda x: x['publicPlayingInfos']['duration'])),
  97             'tags': tags,
  98             'series': decoration.get('programLabel'),
  99             'season_number': int_or_none(video.get('season')),
 100             'episode_number': int_or_none(video.get('episode')),
 101         }