yt_dlp/extractor/jtbc.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     int_or_none,
   6     parse_duration,
   7     url_or_none,
   8 )
   9 from ..utils.traversal import traverse_obj
  10
  11
  12 class JTBCIE(InfoExtractor):
  13     IE_DESC = 'jtbc.co.kr'
  14     _VALID_URL = r'''(?x)
  15         https?://(?:
  16             vod\.jtbc\.co\.kr/player/(?:program|clip)
  17             |tv\.jtbc\.co\.kr/(?:replay|trailer|clip)/pr\d+/pm\d+
  18         )/(?P<id>(?:ep|vo)\d+)'''
  19     _GEO_COUNTRIES = ['KR']
  20
  21     _TESTS = [{
  22         'url': 'https://tv.jtbc.co.kr/replay/pr10011629/pm10067930/ep20216321/view',
  23         'md5': 'e6ade71d8c8685bbfd6e6ce4167c6a6c',
  24         'info_dict': {
  25             'id': 'VO10721192',
  26             'display_id': 'ep20216321',
  27             'ext': 'mp4',
  28             'title': '힘쎈여자 강남순 2회 다시보기',
  29             'description': 'md5:043c1d9019100ce271dba09995dbd1e2',
  30             'duration': 3770.0,
  31             'release_date': '20231008',
  32             'age_limit': 15,
  33             'thumbnail': 'https://fs.jtbc.co.kr//joydata/CP00000001/prog/drama/stronggirlnamsoon/img/20231008_163541_522_1.jpg',
  34             'series': '힘쎈여자 강남순',
  35         },
  36     }, {
  37         'url': 'https://vod.jtbc.co.kr/player/program/ep20216733',
  38         'md5': '217a6d190f115a75e4bda0ceaa4cd7f4',
  39         'info_dict': {
  40             'id': 'VO10721429',
  41             'display_id': 'ep20216733',
  42             'ext': 'mp4',
  43             'title': '헬로 마이 닥터 친절한 진료실 149회 다시보기',
  44             'description': 'md5:1d70788a982dd5de26874a92fcffddb8',
  45             'duration': 2720.0,
  46             'release_date': '20231009',
  47             'age_limit': 15,
  48             'thumbnail': 'https://fs.jtbc.co.kr//joydata/CP00000001/prog/culture/hellomydoctor/img/20231009_095002_528_1.jpg',
  49             'series': '헬로 마이 닥터 친절한 진료실',
  50         },
  51     }, {
  52         'url': 'https://vod.jtbc.co.kr/player/clip/vo10721270',
  53         'md5': '05782e2dc22a9c548aebefe62ae4328a',
  54         'info_dict': {
  55             'id': 'VO10721270',
  56             'display_id': 'vo10721270',
  57             'ext': 'mp4',
  58             'title': '뭉쳐야 찬다3 2회 예고편 - A매치로 향하는 마지막 관문💥',
  59             'description': 'md5:d48b51a8655c84843b4ed8d0c39aae68',
  60             'duration': 46.0,
  61             'release_date': '20231015',
  62             'age_limit': 15,
  63             'thumbnail': 'https://fs.jtbc.co.kr//joydata/CP00000001/prog/enter/soccer3/img/20231008_210957_775_1.jpg',
  64             'series': '뭉쳐야 찬다3',
  65         },
  66     }, {
  67         'url': 'https://tv.jtbc.co.kr/trailer/pr10010392/pm10032526/vo10720912/view',
  68         'md5': '367d480eb3ef54a9cd7a4b4d69c4b32d',
  69         'info_dict': {
  70             'id': 'VO10720912',
  71             'display_id': 'vo10720912',
  72             'ext': 'mp4',
  73             'title': '아는 형님 404회 예고편 | 10월 14일(토) 저녁 8시 50분 방송!',
  74             'description': 'md5:2743bb1079ceb85bb00060f2ad8f0280',
  75             'duration': 148.0,
  76             'release_date': '20231014',
  77             'age_limit': 15,
  78             'thumbnail': 'https://fs.jtbc.co.kr//joydata/CP00000001/prog/enter/jtbcbros/img/20231006_230023_802_1.jpg',
  79             'series': '아는 형님',
  80         },
  81     }]
  82
  83     def _real_extract(self, url):
  84         display_id = self._match_id(url)
  85
  86         if display_id.startswith('vo'):
  87             video_id = display_id.upper()
  88         else:
  89             webpage = self._download_webpage(url, display_id)
  90             video_id = self._search_regex(r'data-vod="(VO\d+)"', webpage, 'vod id')
  91
  92         playback_data = self._download_json(
  93             f'https://api.jtbc.co.kr/vod/{video_id}', video_id, note='Downloading VOD playback data')
  94
  95         subtitles = {}
  96         for sub in traverse_obj(playback_data, ('tracks', lambda _, v: v['file'])):
  97             subtitles.setdefault(sub.get('label', 'und'), []).append({'url': sub['file']})
  98
  99         formats = []
 100         for stream_url in traverse_obj(playback_data, ('sources', 'HLS', ..., 'file', {url_or_none})):
 101             stream_url = re.sub(r'/playlist(?:_pd\d+)?\.m3u8', '/index.m3u8', stream_url)
 102             formats.extend(self._extract_m3u8_formats(stream_url, video_id, fatal=False))
 103
 104         metadata = self._download_json(
 105             'https://now-api.jtbc.co.kr/v1/vod/detail', video_id,
 106             note='Downloading mobile details', fatal=False, query={'vodFileId': video_id})
 107         return {
 108             'id': video_id,
 109             'display_id': display_id,
 110             **traverse_obj(metadata, ('vodDetail', {
 111                 'title': 'vodTitleView',
 112                 'series': 'programTitle',
 113                 'age_limit': ('watchAge', {int_or_none}),
 114                 'release_date': ('broadcastDate', {lambda x: re.match(r'\d{8}', x.replace('.', ''))}, 0),
 115                 'description': 'episodeContents',
 116                 'thumbnail': ('imgFileUrl', {url_or_none}),
 117             })),
 118             'duration': parse_duration(playback_data.get('playTime')),
 119             'formats': formats,
 120             'subtitles': subtitles,
 121         }
 122
 123
 124 class JTBCProgramIE(InfoExtractor):
 125     IE_NAME = 'JTBC:program'
 126     _VALID_URL = r'https?://(?:vod\.jtbc\.co\.kr/program|tv\.jtbc\.co\.kr/replay)/(?P<id>pr\d+)/(?:replay|pm\d+)/?(?:$|[?#])'
 127
 128     _TESTS = [{
 129         'url': 'https://tv.jtbc.co.kr/replay/pr10010392/pm10032710',
 130         'info_dict': {
 131             '_type': 'playlist',
 132             'id': 'pr10010392',
 133         },
 134         'playlist_count': 398,
 135     }, {
 136         'url': 'https://vod.jtbc.co.kr/program/pr10011491/replay',
 137         'info_dict': {
 138             '_type': 'playlist',
 139             'id': 'pr10011491',
 140         },
 141         'playlist_count': 59,
 142     }]
 143
 144     def _real_extract(self, url):
 145         program_id = self._match_id(url)
 146
 147         vod_list = self._download_json(
 148             'https://now-api.jtbc.co.kr/v1/vodClip/programHome/programReplayVodList', program_id,
 149             note='Downloading program replay list', query={
 150                 'programId': program_id,
 151                 'rowCount': '10000',
 152             })
 153
 154         entries = [self.url_result(f'https://vod.jtbc.co.kr/player/program/{video_id}', JTBCIE, video_id)
 155                    for video_id in traverse_obj(vod_list, ('programReplayVodList', ..., 'episodeId'))]
 156         return self.playlist_result(entries, program_id)