mediaworksnz.py 4.0 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103
  1. import re
  2. from .common import InfoExtractor
  3. from ..utils import (
  4. bug_reports_message,
  5. float_or_none,
  6. traverse_obj,
  7. unified_timestamp,
  8. )
  9. class MediaWorksNZVODIE(InfoExtractor):
  10. _VALID_URL_BASE_RE = r'https?://vodupload-api\.mediaworks\.nz/library/asset/published/'
  11. _VALID_URL_ID_RE = r'(?P<id>[A-Za-z0-9-]+)'
  12. _VALID_URL = rf'{_VALID_URL_BASE_RE}{_VALID_URL_ID_RE}'
  13. _TESTS = [{
  14. 'url': 'https://vodupload-api.mediaworks.nz/library/asset/published/VID00359',
  15. 'info_dict': {
  16. 'id': 'VID00359',
  17. 'ext': 'mp4',
  18. 'title': 'GRG Jacinda Ardern safe drug testing 1920x1080',
  19. 'description': 'md5:d4d7dc366742e86d8130b257dcb520ba',
  20. 'duration': 142.76,
  21. 'timestamp': 1604268608,
  22. 'upload_date': '20201101',
  23. 'thumbnail': r're:^https?://.*\.jpg$',
  24. 'channel': 'George FM',
  25. },
  26. }, {
  27. # has audio-only format
  28. 'url': 'https://vodupload-api.mediaworks.nz/library/asset/published/VID02627',
  29. 'info_dict': {
  30. 'id': 'VID02627',
  31. 'ext': 'mp3',
  32. 'title': 'Tova O\'Brien meets Ukraine President Volodymyr Zelensky',
  33. 'channel': 'Today FM',
  34. 'description': 'Watch in full the much anticipated interview of Volodymyr Zelensky',
  35. 'duration': 2061.16,
  36. 'thumbnail': r're:^https?://.*\.jpg$',
  37. 'upload_date': '20220822',
  38. 'timestamp': 1661152289,
  39. },
  40. 'params': {'format': 'ba[ext=mp3]'},
  41. }]
  42. _WEBPAGE_TESTS = [{
  43. 'url': 'https://www.rova.nz/home/podcasts/socrates-walks-into-a-bar/the-trolley-problem---episode-1.html',
  44. 'info_dict': {
  45. 'id': 'VID02494',
  46. 'ext': 'mp4',
  47. 'title': 'The Trolley Problem',
  48. 'duration': 2843.56,
  49. 'channel': 'Other',
  50. 'timestamp': 1658356489,
  51. 'thumbnail': r're:^https?://.*\.jpg$',
  52. 'description': 'Socrates Walks Into A Bar Podcast Episode 1',
  53. 'upload_date': '20220720',
  54. },
  55. }]
  56. @classmethod
  57. def _extract_embed_urls(cls, url, webpage):
  58. for mobj in re.finditer(
  59. rf'''(?x)<div\s+\bid=["']Player-Attributes-JWID[^>]+\b
  60. data-request-url=["']{cls._VALID_URL_BASE_RE}["'][^>]+\b
  61. data-asset-id=["']{cls._VALID_URL_ID_RE}["']''', webpage,
  62. ):
  63. yield f'https://vodupload-api.mediaworks.nz/library/asset/published/{mobj.group("id")}'
  64. def _real_extract(self, url):
  65. video_id = self._match_id(url)
  66. asset = self._download_json(url, video_id)['asset']
  67. if asset.get('drm') not in ('NonDRM', None):
  68. self.report_drm(video_id)
  69. content_type = asset.get('type')
  70. if content_type and content_type != 'video':
  71. self.report_warning(f'Unknown content type: {content_type}' + bug_reports_message(), video_id)
  72. formats, subtitles = self._extract_m3u8_formats_and_subtitles(asset['streamingUrl'], video_id)
  73. audio_streaming_url = traverse_obj(
  74. asset, 'palyoutPathAudio', 'playoutpathaudio', expected_type=str)
  75. if audio_streaming_url:
  76. audio_formats = self._extract_m3u8_formats(audio_streaming_url, video_id, fatal=False, ext='mp3')
  77. for audio_format in audio_formats:
  78. # all the audio streams appear to be aac
  79. audio_format.setdefault('vcodec', 'none')
  80. audio_format.setdefault('acodec', 'aac')
  81. formats.append(audio_format)
  82. return {
  83. 'id': video_id,
  84. 'title': asset.get('title'),
  85. 'description': asset.get('description'),
  86. 'duration': float_or_none(asset.get('duration')),
  87. 'timestamp': unified_timestamp(asset.get('dateadded')),
  88. 'channel': asset.get('brand'),
  89. 'thumbnails': [{'url': thumbnail_url} for thumbnail_url in asset.get('thumbnails') or []],
  90. 'formats': formats,
  91. 'subtitles': subtitles,
  92. }