yt_dlp/extractor/trtworld.py

   1 from .common import InfoExtractor
   2 from ..utils import ExtractorError, determine_ext, parse_iso8601, url_or_none
   3 from ..utils.traversal import traverse_obj
   4
   5
   6 class TrtWorldIE(InfoExtractor):
   7     _VALID_URL = r'https?://www\.trtworld\.com/video/[\w-]+/[\w-]+-(?P<id>\d+)'
   8
   9     _TESTS = [{
  10         'url': 'https://www.trtworld.com/video/news/turkiye-switches-to-sustainable-tourism-16067690',
  11         'info_dict': {
  12             'id': '16067690',
  13             'ext': 'mp4',
  14             'title': 'Türkiye switches to sustainable tourism',
  15             'release_timestamp': 1701529569,
  16             'release_date': '20231202',
  17             'thumbnail': 'https://cdn-i.pr.trt.com.tr/trtworld/17647563_0-0-1920-1080.jpeg',
  18             'description': 'md5:0a975c04257fb529c8f99c7b76a2cf12',
  19         },
  20     }, {
  21         'url': 'https://www.trtworld.com/video/one-offs/frames-from-anatolia-recreating-a-james-bond-scene-in-istanbuls-grand-bazaar-14541780',
  22         'info_dict': {
  23             'id': '14541780',
  24             'ext': 'mp4',
  25             'title': 'Frames From Anatolia: Recreating a ‘James Bond’ Scene in Istanbul’s Grand Bazaar',
  26             'release_timestamp': 1692440844,
  27             'release_date': '20230819',
  28             'thumbnail': 'https://cdn-i.pr.trt.com.tr/trtworld/16939810_0-0-1920-1080.jpeg',
  29             'description': 'md5:4050e21570cc3c40b6c9badae800a94f',
  30         },
  31     }, {
  32         'url': 'https://www.trtworld.com/video/the-newsmakers/can-sudan-find-peace-amidst-failed-transition-to-democracy-12904760',
  33         'info_dict': {
  34             'id': '12904760',
  35             'ext': 'mp4',
  36             'title': 'Can Sudan find peace amidst failed transition to democracy?',
  37             'release_timestamp': 1681972747,
  38             'release_date': '20230420',
  39             'thumbnail': 'http://cdni0.trtworld.com/w768/q70/154214_NMYOUTUBETEMPLATE1_1681833018736.jpg',
  40         },
  41     }, {
  42         'url': 'https://www.trtworld.com/video/africa-matters/locals-learning-to-cope-with-rising-tides-of-kenyas-great-lakes-16059545',
  43         'info_dict': {
  44             'id': 'zEns2dWl00w',
  45             'ext': 'mp4',
  46             'title': "Locals learning to cope with rising tides of Kenya's Great Lakes",
  47             'thumbnail': 'https://i.ytimg.com/vi/zEns2dWl00w/maxresdefault.jpg',
  48             'description': 'md5:3ad9d7c5234d752a4ead4340c79c6b8d',
  49             'channel_id': 'UC7fWeaHhqgM4Ry-RMpM2YYw',
  50             'channel_url': 'https://www.youtube.com/channel/UC7fWeaHhqgM4Ry-RMpM2YYw',
  51             'duration': 210,
  52             'view_count': int,
  53             'age_limit': 0,
  54             'webpage_url': 'https://www.youtube.com/watch?v=zEns2dWl00w',
  55             'categories': ['News & Politics'],
  56             'channel': 'TRT World',
  57             'channel_follower_count': int,
  58             'channel_is_verified': True,
  59             'uploader': 'TRT World',
  60             'uploader_id': '@trtworld',
  61             'uploader_url': 'https://www.youtube.com/@trtworld',
  62             'upload_date': '20231202',
  63             'availability': 'public',
  64             'comment_count': int,
  65             'playable_in_embed': True,
  66             'tags': [],
  67             'live_status': 'not_live',
  68             'like_count': int,
  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         nuxtjs_data = self._search_nuxt_data(webpage, display_id)['videoData']['content']['platforms']
  76         formats = []
  77         for media_url in traverse_obj(nuxtjs_data, (
  78                 ('website', 'ott'), 'metadata', ('hls_url', 'url'), {url_or_none})):
  79             # NB: Website sometimes serves mp4 files under `hls_url` key
  80             if determine_ext(media_url) == 'm3u8':
  81                 formats.extend(self._extract_m3u8_formats(media_url, display_id, fatal=False))
  82             else:
  83                 formats.append({
  84                     'format_id': 'http',
  85                     'url': media_url,
  86                 })
  87         if not formats:
  88             if youtube_id := traverse_obj(nuxtjs_data, ('youtube', 'metadata', 'youtubeId')):
  89                 return self.url_result(youtube_id, 'Youtube')
  90             raise ExtractorError('No video found', expected=True)
  91
  92         return {
  93             'id': display_id,
  94             'formats': formats,
  95             **traverse_obj(nuxtjs_data, (('website', 'ott'), {
  96                 'title': ('fields', 'title', 'text', {str}),
  97                 'description': ('fields', 'description', 'text', {str}),
  98                 'thumbnail': ('fields', 'thumbnail', 'url', {url_or_none}),
  99                 'release_timestamp': ('published', 'date', {parse_iso8601}),
 100             }), get_all=False),
 101         }