yt_dlp/extractor/telewebion.py

   1 from __future__ import annotations
   2
   3 import functools
   4 import json
   5 import textwrap
   6
   7 from .common import InfoExtractor
   8 from ..utils import ExtractorError, format_field, int_or_none, parse_iso8601
   9 from ..utils.traversal import traverse_obj
  10
  11
  12 def _fmt_url(url):
  13     return format_field(template=url, default=None)
  14
  15
  16 class TelewebionIE(InfoExtractor):
  17     _WORKING = False
  18     _VALID_URL = r'https?://(?:www\.)?telewebion\.com/episode/(?P<id>(?:0x[a-fA-F\d]+|\d+))'
  19     _TESTS = [{
  20         'url': 'http://www.telewebion.com/episode/0x1b3139c/',
  21         'info_dict': {
  22             'id': '0x1b3139c',
  23             'ext': 'mp4',
  24             'title': 'قرعه‌کشی لیگ قهرمانان اروپا',
  25             'series': '+ فوتبال',
  26             'series_id': '0x1b2505c',
  27             'channel': 'شبکه 3',
  28             'channel_id': '0x1b1a761',
  29             'channel_url': 'https://telewebion.com/live/tv3',
  30             'timestamp': 1425522414,
  31             'upload_date': '20150305',
  32             'release_timestamp': 1425517020,
  33             'release_date': '20150305',
  34             'duration': 420,
  35             'view_count': int,
  36             'tags': ['ورزشی', 'لیگ اروپا', 'اروپا'],
  37             'thumbnail': 'https://static.telewebion.com/episodeImages/YjFhM2MxMDBkMDNiZTU0MjE5YjQ3ZDY0Mjk1ZDE0ZmUwZWU3OTE3OWRmMDAyODNhNzNkNjdmMWMzMWIyM2NmMA/default',
  38         },
  39         'skip_download': 'm3u8',
  40     }, {
  41         'url': 'https://telewebion.com/episode/162175536',
  42         'info_dict': {
  43             'id': '0x9aa9a30',
  44             'ext': 'mp4',
  45             'title': 'کارما یعنی این !',
  46             'series': 'پاورقی',
  47             'series_id': '0x29a7426',
  48             'channel': 'شبکه 2',
  49             'channel_id': '0x1b1a719',
  50             'channel_url': 'https://telewebion.com/live/tv2',
  51             'timestamp': 1699979968,
  52             'upload_date': '20231114',
  53             'release_timestamp': 1699991638,
  54             'release_date': '20231114',
  55             'duration': 78,
  56             'view_count': int,
  57             'tags': ['کلیپ های منتخب', ' کلیپ طنز ', ' کلیپ سیاست ', 'پاورقی', 'ویژه فلسطین'],
  58             'thumbnail': 'https://static.telewebion.com/episodeImages/871e9455-7567-49a5-9648-34c22c197f5f/default',
  59         },
  60         'skip_download': 'm3u8',
  61     }]
  62
  63     def _call_graphql_api(
  64         self, operation, video_id, query,
  65         variables: dict[str, tuple[str, str]] | None = None,
  66         note='Downloading GraphQL JSON metadata',
  67     ):
  68         parameters = ''
  69         if variables:
  70             parameters = ', '.join(f'${name}: {type_}' for name, (type_, _) in variables.items())
  71             parameters = f'({parameters})'
  72
  73         result = self._download_json('https://graph.telewebion.com/graphql', video_id, note, data=json.dumps({
  74             'operationName': operation,
  75             'query': f'query {operation}{parameters} @cacheControl(maxAge: 60) {{{query}\n}}\n',
  76             'variables': {name: value for name, (_, value) in (variables or {}).items()},
  77         }, separators=(',', ':')).encode(), headers={
  78             'Content-Type': 'application/json',
  79             'Accept': 'application/json',
  80         })
  81         if not result or traverse_obj(result, 'errors'):
  82             message = ', '.join(traverse_obj(result, ('errors', ..., 'message', {str})))
  83             raise ExtractorError(message or 'Unknown GraphQL API error')
  84
  85         return result['data']
  86
  87     def _real_extract(self, url):
  88         video_id = self._match_id(url)
  89         if not video_id.startswith('0x'):
  90             video_id = hex(int(video_id))
  91
  92         episode_data = self._call_graphql_api('getEpisodeDetail', video_id, textwrap.dedent('''
  93             queryEpisode(filter: {EpisodeID: $EpisodeId}, first: 1) {
  94               title
  95               program {
  96                 ProgramID
  97                 title
  98               }
  99               image
 100               view_count
 101               duration
 102               started_at
 103               created_at
 104               channel {
 105                 ChannelID
 106                 name
 107                 descriptor
 108               }
 109               tags {
 110                 name
 111               }
 112             }
 113         '''), {'EpisodeId': ('[ID!]', video_id)})
 114
 115         info_dict = traverse_obj(episode_data, ('queryEpisode', 0, {
 116             'title': ('title', {str}),
 117             'view_count': ('view_count', {int_or_none}),
 118             'duration': ('duration', {int_or_none}),
 119             'tags': ('tags', ..., 'name', {str}),
 120             'release_timestamp': ('started_at', {parse_iso8601}),
 121             'timestamp': ('created_at', {parse_iso8601}),
 122             'series': ('program', 'title', {str}),
 123             'series_id': ('program', 'ProgramID', {str}),
 124             'channel': ('channel', 'name', {str}),
 125             'channel_id': ('channel', 'ChannelID', {str}),
 126             'channel_url': ('channel', 'descriptor', {_fmt_url('https://telewebion.com/live/%s')}),
 127             'thumbnail': ('image', {_fmt_url('https://static.telewebion.com/episodeImages/%s/default')}),
 128             'formats': (
 129                 'channel', 'descriptor', {str},
 130                 {_fmt_url(f'https://cdna.telewebion.com/%s/episode/{video_id}/playlist.m3u8')},
 131                 {functools.partial(self._extract_m3u8_formats, video_id=video_id, ext='mp4', m3u8_id='hls')}),
 132         }))
 133         info_dict['id'] = video_id
 134         return info_dict