detik.py 7.9 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159
  1. from .common import InfoExtractor
  2. from ..utils import int_or_none, merge_dicts, try_call, url_basename
  3. class DetikEmbedIE(InfoExtractor):
  4. _VALID_URL = False
  5. _WEBPAGE_TESTS = [{
  6. # cnn embed
  7. 'url': 'https://www.cnnindonesia.com/embed/video/846189',
  8. 'info_dict': {
  9. 'id': '846189',
  10. 'ext': 'mp4',
  11. 'description': 'md5:ece7b003b3ee7d81c6a5cfede7d5397d',
  12. 'thumbnail': r're:https?://akcdn\.detik\.net\.id/visual/2022/09/11/thumbnail-video-1_169.jpeg',
  13. 'title': 'Video CNN Indonesia - VIDEO: Momen Charles Disambut Meriah usai Dilantik jadi Raja Inggris',
  14. 'age_limit': 0,
  15. 'tags': ['raja charles', ' raja charles iii', ' ratu elizabeth', ' ratu elizabeth meninggal dunia', ' raja inggris', ' inggris'],
  16. 'release_timestamp': 1662869995,
  17. 'release_date': '20220911',
  18. 'uploader': 'REUTERS',
  19. },
  20. }, {
  21. # 20.detik
  22. 'url': 'https://20.detik.com/otobuzz/20220704-220704093/mulai-rp-10-jutaan-ini-skema-kredit-mitsubishi-pajero-sport',
  23. 'info_dict': {
  24. 'display_id': 'mulai-rp-10-jutaan-ini-skema-kredit-mitsubishi-pajero-sport',
  25. 'id': '220704093',
  26. 'ext': 'mp4',
  27. 'description': 'md5:9b2257341b6f375cdcf90106146d5ffb',
  28. 'thumbnail': r're:https?://cdnv\.detik\.com/videoservice/AdminTV/2022/07/04/5d6187e402ec4a91877755a5886ff5b6-20220704161859-0s.jpg',
  29. 'title': 'Mulai Rp 10 Jutaan! Ini Skema Kredit Mitsubishi Pajero Sport',
  30. 'timestamp': 1656951521,
  31. 'upload_date': '20220704',
  32. 'duration': 83.0,
  33. 'tags': ['cicilan mobil', 'mitsubishi pajero sport', 'mitsubishi', 'pajero sport'],
  34. 'release_timestamp': 1656926321,
  35. 'release_date': '20220704',
  36. 'age_limit': 0,
  37. 'uploader': 'Ridwan Arifin ', # TODO: strip trailling whitespace at uploader
  38. },
  39. }, {
  40. # pasangmata.detik
  41. 'url': 'https://pasangmata.detik.com/contribution/366649',
  42. 'info_dict': {
  43. 'id': '366649',
  44. 'ext': 'mp4',
  45. 'title': 'Saling Dorong Aparat dan Pendemo di Aksi Tolak Kenaikan BBM',
  46. 'description': 'md5:7a6580876c8381c454679e028620bea7',
  47. 'age_limit': 0,
  48. 'tags': 'count:17',
  49. 'thumbnail': 'https://akcdn.detik.net.id/community/data/media/thumbs-pasangmata/2022/09/08/366649-16626229351533009620.mp4-03.jpg',
  50. },
  51. }, {
  52. # insertlive embed
  53. 'url': 'https://www.insertlive.com/embed/video/290482',
  54. 'info_dict': {
  55. 'id': '290482',
  56. 'ext': 'mp4',
  57. 'release_timestamp': 1663063704,
  58. 'thumbnail': 'https://akcdn.detik.net.id/visual/2022/09/13/leonardo-dicaprio_169.png?w=600&q=90',
  59. 'age_limit': 0,
  60. 'description': 'Aktor Leonardo DiCaprio memang baru saja putus dari kekasihnya yang bernama Camilla Morrone.',
  61. 'release_date': '20220913',
  62. 'title': 'Diincar Leonardo DiCaprio, Gigi Hadid Ngaku Tertarik Tapi Belum Cinta',
  63. 'tags': ['leonardo dicaprio', ' gigi hadid', ' hollywood'],
  64. 'uploader': '!nsertlive',
  65. },
  66. }, {
  67. # beautynesia embed
  68. 'url': 'https://www.beautynesia.id/embed/video/261636',
  69. 'info_dict': {
  70. 'id': '261636',
  71. 'ext': 'mp4',
  72. 'age_limit': 0,
  73. 'release_timestamp': 1662375600,
  74. 'description': 'Menurut ramalan astrologi, tiga zodiak ini bakal hoki sepanjang September 2022.',
  75. 'title': '3 Zodiak Paling Beruntung Selama September 2022',
  76. 'release_date': '20220905',
  77. 'tags': ['zodiac update', ' zodiak', ' ramalan bintang', ' zodiak beruntung 2022', ' zodiak hoki september 2022', ' zodiak beruntung september 2022'],
  78. 'thumbnail': 'https://akcdn.detik.net.id/visual/2022/09/05/3-zodiak-paling-beruntung-selama-september-2022_169.jpeg?w=600&q=90',
  79. 'uploader': 'amh',
  80. },
  81. }, {
  82. # cnbcindonesia embed
  83. 'url': 'https://www.cnbcindonesia.com/embed/video/371839',
  84. 'info_dict': {
  85. 'id': '371839',
  86. 'ext': 'mp4',
  87. 'title': 'Puluhan Pejabat Rusia Tuntut Putin Mundur',
  88. 'tags': ['putin'],
  89. 'age_limit': 0,
  90. 'thumbnail': 'https://awsimages.detik.net.id/visual/2022/09/13/cnbc-indonesia-tv-3_169.png?w=600&q=80',
  91. 'description': 'md5:8b9111e37555fcd95fe549a9b4ae6fdc',
  92. },
  93. }, {
  94. # detik shortlink (we can get it from https://dtk.id/?<url>)
  95. 'url': 'https://dtk.id/NkISKr',
  96. 'info_dict': {
  97. 'id': '220914049',
  98. 'ext': 'mp4',
  99. 'release_timestamp': 1663114488,
  100. 'uploader': 'Tim 20Detik',
  101. 'title': 'Pakar Bicara soal Tim Khusus Jokowi dan Mereka yang Pro ke Bjorka',
  102. 'age_limit': 0,
  103. 'thumbnail': 'https://cdnv.detik.com/videoservice/AdminTV/2022/09/14/f15cae71d7b640c58e75b254ecbb1ce1-20220914071613-0s.jpg?w=400&q=80',
  104. 'display_id': 'pakar-bicara-soal-tim-khusus-jokowi-dan-mereka-yang-pro-ke-bjorka',
  105. 'upload_date': '20220914',
  106. 'release_date': '20220914',
  107. 'description': 'md5:5eb03225f7ee40207dd3a1e18a73f1ff',
  108. 'timestamp': 1663139688,
  109. 'duration': 213.0,
  110. 'tags': ['hacker bjorka', 'bjorka', 'hacker bjorka bocorkan data rahasia presiden jokowi', 'jokowi'],
  111. },
  112. }]
  113. def _extract_from_webpage(self, url, webpage):
  114. player_type, video_data = self._search_regex(
  115. r'<script\s*[^>]+src="https?://(aws)?cdn\.detik\.net\.id/(?P<type>flowplayer|detikVideo)[^>]+>\s*(?P<video_data>{[^}]+})',
  116. webpage, 'playerjs', group=('type', 'video_data'), default=(None, ''))
  117. if not player_type:
  118. return
  119. display_id, extra_info_dict = url_basename(url), {}
  120. if player_type == 'flowplayer':
  121. video_json_data = self._parse_json(video_data.replace('\'', '"'), display_id)
  122. video_url = video_json_data['videoUrl']
  123. extra_info_dict = {
  124. 'id': self._search_regex(r'identifier\s*:\s*\'([^\']+)', webpage, 'identifier'),
  125. 'thumbnail': video_json_data.get('imageUrl'),
  126. }
  127. elif player_type == 'detikVideo':
  128. video_url = self._search_regex(
  129. r'videoUrl\s*:\s*[\'"]?([^"\']+)', video_data, 'videoUrl')
  130. extra_info_dict = {
  131. 'id': self._html_search_meta(['video_id', 'dtk:video_id'], webpage),
  132. 'thumbnail': self._search_regex(r'imageUrl\s*:\s*[\'"]?([^"\']+)', video_data, 'videoUrl'),
  133. 'duration': int_or_none(self._html_search_meta('duration', webpage, fatal=False, default=None)),
  134. 'release_timestamp': int_or_none(self._html_search_meta('dtk:publishdateunix', webpage, fatal=False, default=None), 1000),
  135. 'timestamp': int_or_none(self._html_search_meta('dtk:createdateunix', webpage, fatal=False, default=None), 1000),
  136. 'uploader': self._search_regex(
  137. r'([^-]+)', self._html_search_meta('dtk:author', webpage, default='').strip(), 'uploader',
  138. default=None),
  139. }
  140. formats, subtitles = self._extract_m3u8_formats_and_subtitles(video_url, display_id)
  141. json_ld_data = self._search_json_ld(webpage, display_id, default={})
  142. yield merge_dicts(json_ld_data, extra_info_dict, {
  143. 'display_id': display_id,
  144. 'title': self._html_search_meta(['og:title', 'originalTitle'], webpage) or self._html_extract_title(webpage),
  145. 'description': self._html_search_meta(['og:description', 'twitter:description', 'description'], webpage),
  146. 'formats': formats,
  147. 'subtitles': subtitles,
  148. 'tags': try_call(lambda: self._html_search_meta(
  149. ['keywords', 'keyword', 'dtk:keywords'], webpage).split(',')),
  150. })