yt_dlp/extractor/detik.py

   1 from .common import InfoExtractor
   2 from ..utils import int_or_none, merge_dicts, try_call, url_basename
   3
   4
   5 class DetikEmbedIE(InfoExtractor):
   6     _VALID_URL = False
   7     _WEBPAGE_TESTS = [{
   8         # cnn embed
   9         'url': 'https://www.cnnindonesia.com/embed/video/846189',
  10         'info_dict': {
  11             'id': '846189',
  12             'ext': 'mp4',
  13             'description': 'md5:ece7b003b3ee7d81c6a5cfede7d5397d',
  14             'thumbnail': r're:https?://akcdn\.detik\.net\.id/visual/2022/09/11/thumbnail-video-1_169.jpeg',
  15             'title': 'Video CNN Indonesia - VIDEO: Momen Charles Disambut Meriah usai Dilantik jadi Raja Inggris',
  16             'age_limit': 0,
  17             'tags': ['raja charles', ' raja charles iii', ' ratu elizabeth', ' ratu elizabeth meninggal dunia', ' raja inggris', ' inggris'],
  18             'release_timestamp': 1662869995,
  19             'release_date': '20220911',
  20             'uploader': 'REUTERS'
  21         }
  22     }, {
  23         # 20.detik
  24         'url': 'https://20.detik.com/otobuzz/20220704-220704093/mulai-rp-10-jutaan-ini-skema-kredit-mitsubishi-pajero-sport',
  25         'info_dict': {
  26             'display_id': 'mulai-rp-10-jutaan-ini-skema-kredit-mitsubishi-pajero-sport',
  27             'id': '220704093',
  28             'ext': 'mp4',
  29             'description': 'md5:9b2257341b6f375cdcf90106146d5ffb',
  30             'thumbnail': r're:https?://cdnv\.detik\.com/videoservice/AdminTV/2022/07/04/5d6187e402ec4a91877755a5886ff5b6-20220704161859-0s.jpg',
  31             'title': 'Mulai Rp 10 Jutaan! Ini Skema Kredit Mitsubishi Pajero Sport',
  32             'timestamp': 1656951521,
  33             'upload_date': '20220704',
  34             'duration': 83.0,
  35             'tags': ['cicilan mobil', 'mitsubishi pajero sport', 'mitsubishi', 'pajero sport'],
  36             'release_timestamp': 1656926321,
  37             'release_date': '20220704',
  38             'age_limit': 0,
  39             'uploader': 'Ridwan Arifin '  # TODO: strip trailling whitespace at uploader
  40         }
  41     }, {
  42         # pasangmata.detik
  43         'url': 'https://pasangmata.detik.com/contribution/366649',
  44         'info_dict': {
  45             'id': '366649',
  46             'ext': 'mp4',
  47             'title': 'Saling Dorong Aparat dan Pendemo di Aksi Tolak Kenaikan BBM',
  48             'description': 'md5:7a6580876c8381c454679e028620bea7',
  49             'age_limit': 0,
  50             'tags': 'count:17',
  51             'thumbnail': 'https://akcdn.detik.net.id/community/data/media/thumbs-pasangmata/2022/09/08/366649-16626229351533009620.mp4-03.jpg',
  52         }
  53     }, {
  54         # insertlive embed
  55         'url': 'https://www.insertlive.com/embed/video/290482',
  56         'info_dict': {
  57             'id': '290482',
  58             'ext': 'mp4',
  59             'release_timestamp': 1663063704,
  60             'thumbnail': 'https://akcdn.detik.net.id/visual/2022/09/13/leonardo-dicaprio_169.png?w=600&q=90',
  61             'age_limit': 0,
  62             'description': 'Aktor Leonardo DiCaprio memang baru saja putus dari kekasihnya yang bernama Camilla Morrone.',
  63             'release_date': '20220913',
  64             'title': 'Diincar Leonardo DiCaprio, Gigi Hadid Ngaku Tertarik Tapi Belum Cinta',
  65             'tags': ['leonardo dicaprio', ' gigi hadid', ' hollywood'],
  66             'uploader': '!nsertlive',
  67         }
  68     }, {
  69         # beautynesia embed
  70         'url': 'https://www.beautynesia.id/embed/video/261636',
  71         'info_dict': {
  72             'id': '261636',
  73             'ext': 'mp4',
  74             'age_limit': 0,
  75             'release_timestamp': 1662375600,
  76             'description': 'Menurut ramalan astrologi, tiga zodiak ini bakal hoki sepanjang September 2022.',
  77             'title': '3 Zodiak Paling Beruntung Selama September 2022',
  78             'release_date': '20220905',
  79             'tags': ['zodiac update', ' zodiak', ' ramalan bintang', ' zodiak beruntung 2022', ' zodiak hoki september 2022', ' zodiak beruntung september 2022'],
  80             'thumbnail': 'https://akcdn.detik.net.id/visual/2022/09/05/3-zodiak-paling-beruntung-selama-september-2022_169.jpeg?w=600&q=90',
  81             'uploader': 'amh',
  82         }
  83     }, {
  84         # cnbcindonesia embed
  85         'url': 'https://www.cnbcindonesia.com/embed/video/371839',
  86         'info_dict': {
  87             'id': '371839',
  88             'ext': 'mp4',
  89             'title': 'Puluhan Pejabat Rusia Tuntut Putin Mundur',
  90             'tags': ['putin'],
  91             'age_limit': 0,
  92             'thumbnail': 'https://awsimages.detik.net.id/visual/2022/09/13/cnbc-indonesia-tv-3_169.png?w=600&q=80',
  93             'description': 'md5:8b9111e37555fcd95fe549a9b4ae6fdc',
  94         }
  95     }, {
  96         # detik shortlink (we can get it from https://dtk.id/?<url>)
  97         'url': 'https://dtk.id/NkISKr',
  98         'info_dict': {
  99             'id': '220914049',
 100             'ext': 'mp4',
 101             'release_timestamp': 1663114488,
 102             'uploader': 'Tim 20Detik',
 103             'title': 'Pakar Bicara soal Tim Khusus Jokowi dan Mereka yang Pro ke Bjorka',
 104             'age_limit': 0,
 105             'thumbnail': 'https://cdnv.detik.com/videoservice/AdminTV/2022/09/14/f15cae71d7b640c58e75b254ecbb1ce1-20220914071613-0s.jpg?w=400&q=80',
 106             'display_id': 'pakar-bicara-soal-tim-khusus-jokowi-dan-mereka-yang-pro-ke-bjorka',
 107             'upload_date': '20220914',
 108             'release_date': '20220914',
 109             'description': 'md5:5eb03225f7ee40207dd3a1e18a73f1ff',
 110             'timestamp': 1663139688,
 111             'duration': 213.0,
 112             'tags': ['hacker bjorka', 'bjorka', 'hacker bjorka bocorkan data rahasia presiden jokowi', 'jokowi'],
 113         }
 114     }]
 115
 116     def _extract_from_webpage(self, url, webpage):
 117         player_type, video_data = self._search_regex(
 118             r'<script\s*[^>]+src="https?://(aws)?cdn\.detik\.net\.id/(?P<type>flowplayer|detikVideo)[^>]+>\s*(?P<video_data>{[^}]+})',
 119             webpage, 'playerjs', group=('type', 'video_data'), default=(None, ''))
 120         if not player_type:
 121             return
 122
 123         display_id, extra_info_dict = url_basename(url), {}
 124
 125         if player_type == 'flowplayer':
 126             video_json_data = self._parse_json(video_data.replace('\'', '"'), display_id)
 127             video_url = video_json_data['videoUrl']
 128
 129             extra_info_dict = {
 130                 'id': self._search_regex(r'identifier\s*:\s*\'([^\']+)', webpage, 'identifier'),
 131                 'thumbnail': video_json_data.get('imageUrl'),
 132             }
 133
 134         elif player_type == 'detikVideo':
 135             video_url = self._search_regex(
 136                 r'videoUrl\s*:\s*[\'"]?([^"\']+)', video_data, 'videoUrl')
 137             extra_info_dict = {
 138                 'id': self._html_search_meta(['video_id', 'dtk:video_id'], webpage),
 139                 'thumbnail': self._search_regex(r'imageUrl\s*:\s*[\'"]?([^"\']+)', video_data, 'videoUrl'),
 140                 'duration': int_or_none(self._html_search_meta('duration', webpage, fatal=False, default=None)),
 141                 'release_timestamp': int_or_none(self._html_search_meta('dtk:publishdateunix', webpage, fatal=False, default=None), 1000),
 142                 'timestamp': int_or_none(self._html_search_meta('dtk:createdateunix', webpage, fatal=False, default=None), 1000),
 143                 'uploader': self._search_regex(
 144                     r'([^-]+)', self._html_search_meta('dtk:author', webpage, default='').strip(), 'uploader',
 145                     default=None)
 146             }
 147
 148         formats, subtitles = self._extract_m3u8_formats_and_subtitles(video_url, display_id)
 149
 150         json_ld_data = self._search_json_ld(webpage, display_id, default={})
 151         yield merge_dicts(json_ld_data, extra_info_dict, {
 152             'display_id': display_id,
 153             'title': self._html_search_meta(['og:title', 'originalTitle'], webpage) or self._html_extract_title(webpage),
 154             'description': self._html_search_meta(['og:description', 'twitter:description', 'description'], webpage),
 155             'formats': formats,
 156             'subtitles': subtitles,
 157             'tags': try_call(lambda: self._html_search_meta(
 158                 ['keywords', 'keyword', 'dtk:keywords'], webpage).split(',')),
 159         })