yt_dlp/extractor/bfmtv.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import extract_attributes
   5
   6
   7 class BFMTVBaseIE(InfoExtractor):
   8     _VALID_URL_BASE = r'https?://(?:www\.|rmc\.)?bfmtv\.com/'
   9     _VALID_URL_TMPL = _VALID_URL_BASE + r'(?:[^/]+/)*[^/?&#]+_%s[A-Z]-(?P<id>\d{12})\.html'
  10     _VIDEO_BLOCK_REGEX = r'(<div[^>]+class="video_block[^"]*"[^>]*>)'
  11     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/%s_default/index.html?videoId=%s'
  12
  13     def _brightcove_url_result(self, video_id, video_block):
  14         account_id = video_block.get('accountid') or '876450612001'
  15         player_id = video_block.get('playerid') or 'I2qBTln4u'
  16         return self.url_result(
  17             self.BRIGHTCOVE_URL_TEMPLATE % (account_id, player_id, video_id),
  18             'BrightcoveNew', video_id)
  19
  20
  21 class BFMTVIE(BFMTVBaseIE):
  22     IE_NAME = 'bfmtv'
  23     _VALID_URL = BFMTVBaseIE._VALID_URL_TMPL % 'V'
  24     _TESTS = [{
  25         'url': 'https://www.bfmtv.com/politique/emmanuel-macron-l-islam-est-une-religion-qui-vit-une-crise-aujourd-hui-partout-dans-le-monde_VN-202010020146.html',
  26         'info_dict': {
  27             'id': '6196747868001',
  28             'ext': 'mp4',
  29             'title': 'Emmanuel Macron: "L\'Islam est une religion qui vit une crise aujourd’hui, partout dans le monde"',
  30             'description': 'Le Président s\'exprime sur la question du séparatisme depuis les Mureaux, dans les Yvelines.',
  31             'uploader_id': '876450610001',
  32             'upload_date': '20201002',
  33             'timestamp': 1601629620,
  34             'duration': 44.757,
  35             'tags': ['bfmactu', 'politique'],
  36             'thumbnail': 'https://cf-images.eu-west-1.prod.boltdns.net/v1/static/876450610001/5041f4c1-bc48-4af8-a256-1b8300ad8ef0/cf2f9114-e8e2-4494-82b4-ab794ea4bc7d/1920x1080/match/image.jpg',
  37         },
  38     }]
  39
  40     def _real_extract(self, url):
  41         bfmtv_id = self._match_id(url)
  42         webpage = self._download_webpage(url, bfmtv_id)
  43         video_block = extract_attributes(self._search_regex(
  44             self._VIDEO_BLOCK_REGEX, webpage, 'video block'))
  45         return self._brightcove_url_result(video_block['videoid'], video_block)
  46
  47
  48 class BFMTVLiveIE(BFMTVIE):  # XXX: Do not subclass from concrete IE
  49     IE_NAME = 'bfmtv:live'
  50     _VALID_URL = BFMTVBaseIE._VALID_URL_BASE + '(?P<id>(?:[^/]+/)?en-direct)'
  51     _TESTS = [{
  52         'url': 'https://www.bfmtv.com/en-direct/',
  53         'info_dict': {
  54             'id': '5615950982001',
  55             'ext': 'mp4',
  56             'title': r're:^le direct BFMTV WEB \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
  57             'uploader_id': '876450610001',
  58             'upload_date': '20220926',
  59             'timestamp': 1664207191,
  60             'live_status': 'is_live',
  61             'thumbnail': r're:https://.+/image\.jpg',
  62             'tags': [],
  63         },
  64         'params': {
  65             'skip_download': True,
  66         },
  67     }, {
  68         'url': 'https://www.bfmtv.com/economie/en-direct/',
  69         'only_matching': True,
  70     }]
  71
  72
  73 class BFMTVArticleIE(BFMTVBaseIE):
  74     IE_NAME = 'bfmtv:article'
  75     _VALID_URL = BFMTVBaseIE._VALID_URL_TMPL % 'A'
  76     _TESTS = [{
  77         'url': 'https://www.bfmtv.com/sante/covid-19-un-responsable-de-l-institut-pasteur-se-demande-quand-la-france-va-se-reconfiner_AV-202101060198.html',
  78         'info_dict': {
  79             'id': '202101060198',
  80             'title': 'Covid-19: un responsable de l\'Institut Pasteur se demande "quand la France va se reconfiner"',
  81             'description': 'md5:947974089c303d3ac6196670ae262843',
  82         },
  83         'playlist_count': 2,
  84     }, {
  85         'url': 'https://www.bfmtv.com/international/pour-bolsonaro-le-bresil-est-en-faillite-mais-il-ne-peut-rien-faire_AD-202101060232.html',
  86         'only_matching': True,
  87     }, {
  88         'url': 'https://www.bfmtv.com/sante/covid-19-oui-le-vaccin-de-pfizer-distribue-en-france-a-bien-ete-teste-sur-des-personnes-agees_AN-202101060275.html',
  89         'only_matching': True,
  90     }, {
  91         'url': 'https://rmc.bfmtv.com/actualites/societe/transports/ce-n-est-plus-tout-rentable-le-bioethanol-e85-depasse-1eu-le-litre-des-automobilistes-regrettent_AV-202301100268.html',
  92         'info_dict': {
  93             'id': '6318445464112',
  94             'ext': 'mp4',
  95             'title': 'Le plein de bioéthanol fait de plus en plus mal à la pompe',
  96             'uploader_id': '876630703001',
  97             'upload_date': '20230110',
  98             'timestamp': 1673341692,
  99             'duration': 109.269,
 100             'tags': ['rmc', 'show', 'apolline de malherbe', 'info', 'talk', 'matinale', 'radio'],
 101             'thumbnail': 'https://cf-images.eu-west-1.prod.boltdns.net/v1/static/876630703001/5bef74b8-9d5e-4480-a21f-60c2e2480c46/96c88b74-f9db-45e1-8040-e199c5da216c/1920x1080/match/image.jpg'
 102         }
 103     }]
 104
 105     def _real_extract(self, url):
 106         bfmtv_id = self._match_id(url)
 107         webpage = self._download_webpage(url, bfmtv_id)
 108
 109         entries = []
 110         for video_block_el in re.findall(self._VIDEO_BLOCK_REGEX, webpage):
 111             video_block = extract_attributes(video_block_el)
 112             video_id = video_block.get('videoid')
 113             if not video_id:
 114                 continue
 115             entries.append(self._brightcove_url_result(video_id, video_block))
 116
 117         return self.playlist_result(
 118             entries, bfmtv_id, self._og_search_title(webpage, fatal=False),
 119             self._html_search_meta(['og:description', 'description'], webpage))