yt_dlp/extractor/lefigaro.py

   1 import json
   2 import math
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     InAdvancePagedList,
   7     traverse_obj,
   8 )
   9
  10
  11 class LeFigaroVideoEmbedIE(InfoExtractor):
  12     _VALID_URL = r'https?://video\.lefigaro\.fr/embed/[^?#]+/(?P<id>[\w-]+)'
  13
  14     _TESTS = [{
  15         'url': 'https://video.lefigaro.fr/embed/figaro/video/les-francais-ne-veulent-ils-plus-travailler-suivez-en-direct-le-club-le-figaro-idees/',
  16         'md5': 'a0c3069b7e4c4526abf0053a7713f56f',
  17         'info_dict': {
  18             'id': 'g9j7Eovo',
  19             'title': 'Les Français ne veulent-ils plus travailler ? Retrouvez Le Club Le Figaro Idées',
  20             'description': 'md5:862b8813148ba4bf10763a65a69dfe41',
  21             'upload_date': '20230216',
  22             'timestamp': 1676581615,
  23             'duration': 3076,
  24             'thumbnail': r're:^https?://[^?#]+\.(?:jpeg|jpg)',
  25             'ext': 'mp4',
  26         },
  27     }, {
  28         'url': 'https://video.lefigaro.fr/embed/figaro/video/intelligence-artificielle-faut-il-sen-mefier/',
  29         'md5': '319c662943dd777bab835cae1e2d73a5',
  30         'info_dict': {
  31             'id': 'LeAgybyc',
  32             'title': 'Intelligence artificielle : faut-il s’en méfier ?',
  33             'description': 'md5:249d136e3e5934a67c8cb704f8abf4d2',
  34             'upload_date': '20230124',
  35             'timestamp': 1674584477,
  36             'duration': 860,
  37             'thumbnail': r're:^https?://[^?#]+\.(?:jpeg|jpg)',
  38             'ext': 'mp4',
  39         },
  40     }]
  41
  42     _WEBPAGE_TESTS = [{
  43         'url': 'https://video.lefigaro.fr/figaro/video/suivez-en-direct-le-club-le-figaro-international-avec-philippe-gelie-9/',
  44         'md5': '6289f9489efb969e38245f31721596fe',
  45         'info_dict': {
  46             'id': 'QChnbPYA',
  47             'title': 'Où en est le couple franco-allemand ? Retrouvez Le Club Le Figaro International',
  48             'description': 'md5:6f47235b7e7c93b366fd8ebfa10572ac',
  49             'upload_date': '20230123',
  50             'timestamp': 1674503575,
  51             'duration': 3153,
  52             'thumbnail': r're:^https?://[^?#]+\.(?:jpeg|jpg)',
  53             'age_limit': 0,
  54             'ext': 'mp4',
  55         },
  56     }, {
  57         'url': 'https://video.lefigaro.fr/figaro/video/la-philosophe-nathalie-sarthou-lajus-est-linvitee-du-figaro-live/',
  58         'md5': 'f6df814cae53e85937621599d2967520',
  59         'info_dict': {
  60             'id': 'QJzqoNbf',
  61             'title': 'La philosophe Nathalie Sarthou-Lajus est l’invitée du Figaro Live',
  62             'description': 'md5:c586793bb72e726c83aa257f99a8c8c4',
  63             'upload_date': '20230217',
  64             'timestamp': 1676661986,
  65             'duration': 1558,
  66             'thumbnail': r're:^https?://[^?#]+\.(?:jpeg|jpg)',
  67             'age_limit': 0,
  68             'ext': 'mp4',
  69         },
  70     }]
  71
  72     def _real_extract(self, url):
  73         display_id = self._match_id(url)
  74         webpage = self._download_webpage(url, display_id)
  75
  76         player_data = self._search_nextjs_data(
  77             webpage, display_id)['props']['pageProps']['initialProps']['pageData']['playerData']
  78
  79         return self.url_result(
  80             f'jwplatform:{player_data["videoId"]}', title=player_data.get('title'),
  81             description=player_data.get('description'), thumbnail=player_data.get('poster'))
  82
  83
  84 class LeFigaroVideoSectionIE(InfoExtractor):
  85     _VALID_URL = r'https?://video\.lefigaro\.fr/figaro/(?P<id>[\w-]+)/?(?:[#?]|$)'
  86
  87     _TESTS = [{
  88         'url': 'https://video.lefigaro.fr/figaro/le-club-le-figaro-idees/',
  89         'info_dict': {
  90             'id': 'le-club-le-figaro-idees',
  91             'title': 'Le Club Le Figaro Idées',
  92         },
  93         'playlist_mincount': 14,
  94     }, {
  95         'url': 'https://video.lefigaro.fr/figaro/factu/',
  96         'info_dict': {
  97             'id': 'factu',
  98             'title': 'Factu',
  99         },
 100         'playlist_mincount': 519,
 101     }]
 102
 103     _PAGE_SIZE = 20
 104
 105     def _get_api_response(self, display_id, page_num, note=None):
 106         return self._download_json(
 107             'https://api-graphql.lefigaro.fr/graphql', display_id, note=note,
 108             query={
 109                 'id': 'flive-website_UpdateListPage_1fb260f996bca2d78960805ac382544186b3225f5bedb43ad08b9b8abef79af6',
 110                 'variables': json.dumps({
 111                     'slug': display_id,
 112                     'videosLimit': self._PAGE_SIZE,
 113                     'sort': 'DESC',
 114                     'order': 'PUBLISHED_AT',
 115                     'page': page_num,
 116                 }).encode(),
 117             })
 118
 119     def _real_extract(self, url):
 120         display_id = self._match_id(url)
 121         initial_response = self._get_api_response(display_id, page_num=1)['data']['playlist']
 122
 123         def page_func(page_num):
 124             api_response = self._get_api_response(display_id, page_num + 1, note=f'Downloading page {page_num + 1}')
 125
 126             return [self.url_result(
 127                 video['embedUrl'], LeFigaroVideoEmbedIE, **traverse_obj(video, {
 128                     'title': 'name',
 129                     'description': 'description',
 130                     'thumbnail': 'thumbnailUrl',
 131                 })) for video in api_response['data']['playlist']['jsonLd'][0]['itemListElement']]
 132
 133         entries = InAdvancePagedList(
 134             page_func, math.ceil(initial_response['videoCount'] / self._PAGE_SIZE), self._PAGE_SIZE)
 135
 136         return self.playlist_result(entries, playlist_id=display_id, playlist_title=initial_response.get('title'))