yt_dlp/extractor/ondemandkorea.py

   1 import functools
   2 import re
   3 import uuid
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     ExtractorError,
   8     OnDemandPagedList,
   9     float_or_none,
  10     int_or_none,
  11     join_nonempty,
  12     parse_age_limit,
  13     parse_qs,
  14     str_or_none,
  15     unified_strdate,
  16     url_or_none,
  17 )
  18 from ..utils.traversal import traverse_obj
  19
  20
  21 class OnDemandKoreaIE(InfoExtractor):
  22     _VALID_URL = r'https?://(?:www\.)?ondemandkorea\.com/(?:en/)?player/vod/[a-z0-9-]+\?(?:[^#]+&)?contentId=(?P<id>\d+)'
  23     _GEO_COUNTRIES = ['US', 'CA']
  24
  25     _TESTS = [{
  26         'url': 'https://www.ondemandkorea.com/player/vod/ask-us-anything?contentId=686471',
  27         'md5': 'e2ff77255d989e3135bde0c5889fbce8',
  28         'info_dict': {
  29             'id': '686471',
  30             'ext': 'mp4',
  31             'title': 'Ask Us Anything: Jung Sung-ho, Park Seul-gi, Kim Bo-min, Yang Seung-won',
  32             'thumbnail': r're:^https?://.*\.(jpg|jpeg|png)',
  33             'duration': 5486.955,
  34             'release_date': '20220924',
  35             'series': 'Ask Us Anything',
  36             'series_id': '11790',
  37             'episode_number': 351,
  38             'episode': 'Jung Sung-ho, Park Seul-gi, Kim Bo-min, Yang Seung-won',
  39         },
  40     }, {
  41         'url': 'https://www.ondemandkorea.com/player/vod/breakup-probation-a-week?contentId=1595796',
  42         'md5': '57266c720006962be7ff415b24775caa',
  43         'info_dict': {
  44             'id': '1595796',
  45             'ext': 'mp4',
  46             'title': 'Breakup Probation, A Week: E08',
  47             'thumbnail': r're:^https?://.*\.(jpg|jpeg|png)',
  48             'duration': 1586.0,
  49             'release_date': '20231001',
  50             'series': 'Breakup Probation, A Week',
  51             'series_id': '22912',
  52             'episode_number': 8,
  53             'episode': 'E08',
  54         },
  55     }, {
  56         'url': 'https://www.ondemandkorea.com/player/vod/the-outlaws?contentId=369531',
  57         'md5': 'fa5523b87aa1f6d74fc622a97f2b47cd',
  58         'info_dict': {
  59             'id': '369531',
  60             'ext': 'mp4',
  61             'release_date': '20220519',
  62             'duration': 7267.0,
  63             'title': 'The Outlaws: Main Movie',
  64             'thumbnail': r're:^https?://.*\.(jpg|jpeg|png)',
  65             'age_limit': 18,
  66         },
  67     }, {
  68         'url': 'https://www.ondemandkorea.com/en/player/vod/capture-the-moment-how-is-that-possible?contentId=1605006',
  69         'only_matching': True,
  70     }]
  71
  72     def _real_extract(self, url):
  73         video_id = self._match_id(url)
  74
  75         data = self._download_json(
  76             f'https://odkmedia.io/odx/api/v3/playback/{video_id}/', video_id, fatal=False,
  77             headers={'service-name': 'odk'}, query={'did': str(uuid.uuid4())}, expected_status=(403, 404))
  78         if not traverse_obj(data, ('result', {dict})):
  79             msg = traverse_obj(data, ('messages', '__default'), 'title', expected_type=str)
  80             raise ExtractorError(msg or 'Got empty response from playback API', expected=True)
  81
  82         data = data['result']
  83
  84         def try_geo_bypass(url):
  85             return traverse_obj(url, ({parse_qs}, 'stream_url', 0, {url_or_none})) or url
  86
  87         formats = []
  88         for m3u8_url in traverse_obj(data, (('sources', 'manifest'), ..., 'url', {url_or_none}, {try_geo_bypass})):
  89             mod_url = re.sub(r'_720(p?)\.m3u8', r'_1080\1.m3u8', m3u8_url)
  90             if mod_url != m3u8_url:
  91                 mod_format = self._extract_m3u8_formats(
  92                     mod_url, video_id, note='Checking for higher quality format',
  93                     errnote='No higher quality format found', fatal=False)
  94                 if mod_format:
  95                     formats.extend(mod_format)
  96                     continue
  97             formats.extend(self._extract_m3u8_formats(m3u8_url, video_id, fatal=False))
  98
  99         subtitles = {}
 100         for track in traverse_obj(data, ('text_tracks', lambda _, v: url_or_none(v['url']))):
 101             subtitles.setdefault(track.get('language', 'und'), []).append({
 102                 'url': track['url'],
 103                 'ext': track.get('codec'),
 104                 'name': track.get('label'),
 105             })
 106
 107         def if_series(key=None):
 108             return lambda obj: obj[key] if key and obj['kind'] == 'series' else None
 109
 110         return {
 111             'id': video_id,
 112             'title': join_nonempty(
 113                 ('episode', 'program', 'title'),
 114                 ('episode', 'title'), from_dict=data, delim=': '),
 115             **traverse_obj(data, {
 116                 'thumbnail': ('episode', 'images', 'thumbnail', {url_or_none}),
 117                 'release_date': ('episode', 'release_date', {lambda x: x.replace('-', '')}, {unified_strdate}),
 118                 'duration': ('duration', {functools.partial(float_or_none, scale=1000)}),
 119                 'age_limit': ('age_rating', 'name', {lambda x: x.replace('R', '')}, {parse_age_limit}),
 120                 'series': ('episode', {if_series(key='program')}, 'title'),
 121                 'series_id': ('episode', {if_series(key='program')}, 'id', {str_or_none}),
 122                 'episode': ('episode', {if_series(key='title')}),
 123                 'episode_number': ('episode', {if_series(key='number')}, {int_or_none}),
 124             }, get_all=False),
 125             'formats': formats,
 126             'subtitles': subtitles,
 127         }
 128
 129
 130 class OnDemandKoreaProgramIE(InfoExtractor):
 131     _VALID_URL = r'https?://(?:www\.)?ondemandkorea\.com/(?:en/)?player/vod/(?P<id>[a-z0-9-]+)(?:$|#)'
 132     _GEO_COUNTRIES = ['US', 'CA']
 133
 134     _TESTS = [{
 135         'url': 'https://www.ondemandkorea.com/player/vod/uskn-news',
 136         'info_dict': {
 137             'id': 'uskn-news',
 138         },
 139         'playlist_mincount': 755,
 140     }, {
 141         'url': 'https://www.ondemandkorea.com/en/player/vod/the-land',
 142         'info_dict': {
 143             'id': 'the-land',
 144         },
 145         'playlist_count': 52,
 146     }]
 147
 148     _PAGE_SIZE = 100
 149
 150     def _fetch_page(self, display_id, page):
 151         page += 1
 152         page_data = self._download_json(
 153             f'https://odkmedia.io/odx/api/v3/program/{display_id}/episodes/', display_id,
 154             headers={'service-name': 'odk'}, query={
 155                 'page': page,
 156                 'page_size': self._PAGE_SIZE,
 157             }, note=f'Downloading page {page}', expected_status=404)
 158         for episode in traverse_obj(page_data, ('result', 'results', ...)):
 159             yield self.url_result(
 160                 f'https://www.ondemandkorea.com/player/vod/{display_id}?contentId={episode["id"]}',
 161                 ie=OnDemandKoreaIE, video_title=episode.get('title'))
 162
 163     def _real_extract(self, url):
 164         display_id = self._match_id(url)
 165
 166         entries = OnDemandPagedList(functools.partial(
 167             self._fetch_page, display_id), self._PAGE_SIZE)
 168
 169         return self.playlist_result(entries, display_id)