yt_dlp/extractor/awaan.py

   1 import base64
   2
   3 from .common import InfoExtractor
   4 from ..compat import (
   5     compat_str,
   6     compat_urllib_parse_urlencode,
   7 )
   8 from ..utils import (
   9     format_field,
  10     int_or_none,
  11     parse_iso8601,
  12     smuggle_url,
  13     unsmuggle_url,
  14     urlencode_postdata,
  15 )
  16
  17
  18 class AWAANIE(InfoExtractor):
  19     _VALID_URL = r'https?://(?:www\.)?(?:awaan|dcndigital)\.ae/(?:#/)?show/(?P<show_id>\d+)/[^/]+(?:/(?P<id>\d+)/(?P<season_id>\d+))?'
  20
  21     def _real_extract(self, url):
  22         show_id, video_id, season_id = self._match_valid_url(url).groups()
  23         if video_id and int(video_id) > 0:
  24             return self.url_result(
  25                 'http://awaan.ae/media/%s' % video_id, 'AWAANVideo')
  26         elif season_id and int(season_id) > 0:
  27             return self.url_result(smuggle_url(
  28                 'http://awaan.ae/program/season/%s' % season_id,
  29                 {'show_id': show_id}), 'AWAANSeason')
  30         else:
  31             return self.url_result(
  32                 'http://awaan.ae/program/%s' % show_id, 'AWAANSeason')
  33
  34
  35 class AWAANBaseIE(InfoExtractor):
  36     def _parse_video_data(self, video_data, video_id, is_live):
  37         title = video_data.get('title_en') or video_data['title_ar']
  38         img = video_data.get('img')
  39
  40         return {
  41             'id': video_id,
  42             'title': title,
  43             'description': video_data.get('description_en') or video_data.get('description_ar'),
  44             'thumbnail': format_field(img, None, 'http://admin.mangomolo.com/analytics/%s'),
  45             'duration': int_or_none(video_data.get('duration')),
  46             'timestamp': parse_iso8601(video_data.get('create_time'), ' '),
  47             'is_live': is_live,
  48             'uploader_id': video_data.get('user_id'),
  49         }
  50
  51
  52 class AWAANVideoIE(AWAANBaseIE):
  53     IE_NAME = 'awaan:video'
  54     _VALID_URL = r'https?://(?:www\.)?(?:awaan|dcndigital)\.ae/(?:#/)?(?:video(?:/[^/]+)?|media|catchup/[^/]+/[^/]+)/(?P<id>\d+)'
  55     _TESTS = [{
  56         'url': 'http://www.dcndigital.ae/#/video/%D8%B1%D8%AD%D9%84%D8%A9-%D8%A7%D9%84%D8%B9%D9%85%D8%B1-%D8%A7%D9%84%D8%AD%D9%84%D9%82%D8%A9-1/17375',
  57         'md5': '5f61c33bfc7794315c671a62d43116aa',
  58         'info_dict':
  59         {
  60             'id': '17375',
  61             'ext': 'mp4',
  62             'title': 'رحلة العمر : الحلقة 1',
  63             'description': 'md5:0156e935d870acb8ef0a66d24070c6d6',
  64             'duration': 2041,
  65             'timestamp': 1227504126,
  66             'upload_date': '20081124',
  67             'uploader_id': '71',
  68         },
  69     }, {
  70         'url': 'http://awaan.ae/video/26723981/%D8%AF%D8%A7%D8%B1-%D8%A7%D9%84%D8%B3%D9%84%D8%A7%D9%85:-%D8%AE%D9%8A%D8%B1-%D8%AF%D9%88%D8%B1-%D8%A7%D9%84%D8%A3%D9%86%D8%B5%D8%A7%D8%B1',
  71         'only_matching': True,
  72     }]
  73
  74     def _real_extract(self, url):
  75         video_id = self._match_id(url)
  76
  77         video_data = self._download_json(
  78             'http://admin.mangomolo.com/analytics/index.php/plus/video?id=%s' % video_id,
  79             video_id, headers={'Origin': 'http://awaan.ae'})
  80         info = self._parse_video_data(video_data, video_id, False)
  81
  82         embed_url = 'http://admin.mangomolo.com/analytics/index.php/customers/embed/video?' + compat_urllib_parse_urlencode({
  83             'id': video_data['id'],
  84             'user_id': video_data['user_id'],
  85             'signature': video_data['signature'],
  86             'countries': 'Q0M=',
  87             'filter': 'DENY',
  88         })
  89         info.update({
  90             '_type': 'url_transparent',
  91             'url': embed_url,
  92             'ie_key': 'MangomoloVideo',
  93         })
  94         return info
  95
  96
  97 class AWAANLiveIE(AWAANBaseIE):
  98     IE_NAME = 'awaan:live'
  99     _VALID_URL = r'https?://(?:www\.)?(?:awaan|dcndigital)\.ae/(?:#/)?live/(?P<id>\d+)'
 100     _TEST = {
 101         'url': 'http://awaan.ae/live/6/dubai-tv',
 102         'info_dict': {
 103             'id': '6',
 104             'ext': 'mp4',
 105             'title': 're:Dubai Al Oula [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
 106             'upload_date': '20150107',
 107             'timestamp': 1420588800,
 108             'uploader_id': '71',
 109         },
 110         'params': {
 111             # m3u8 download
 112             'skip_download': True,
 113         },
 114     }
 115
 116     def _real_extract(self, url):
 117         channel_id = self._match_id(url)
 118
 119         channel_data = self._download_json(
 120             'http://admin.mangomolo.com/analytics/index.php/plus/getchanneldetails?channel_id=%s' % channel_id,
 121             channel_id, headers={'Origin': 'http://awaan.ae'})
 122         info = self._parse_video_data(channel_data, channel_id, True)
 123
 124         embed_url = 'http://admin.mangomolo.com/analytics/index.php/customers/embed/index?' + compat_urllib_parse_urlencode({
 125             'id': base64.b64encode(channel_data['user_id'].encode()).decode(),
 126             'channelid': base64.b64encode(channel_data['id'].encode()).decode(),
 127             'signature': channel_data['signature'],
 128             'countries': 'Q0M=',
 129             'filter': 'DENY',
 130         })
 131         info.update({
 132             '_type': 'url_transparent',
 133             'url': embed_url,
 134             'ie_key': 'MangomoloLive',
 135         })
 136         return info
 137
 138
 139 class AWAANSeasonIE(InfoExtractor):
 140     IE_NAME = 'awaan:season'
 141     _VALID_URL = r'https?://(?:www\.)?(?:awaan|dcndigital)\.ae/(?:#/)?program/(?:(?P<show_id>\d+)|season/(?P<season_id>\d+))'
 142     _TEST = {
 143         'url': 'http://dcndigital.ae/#/program/205024/%D9%85%D8%AD%D8%A7%D8%B6%D8%B1%D8%A7%D8%AA-%D8%A7%D9%84%D8%B4%D9%8A%D8%AE-%D8%A7%D9%84%D8%B4%D8%B9%D8%B1%D8%A7%D9%88%D9%8A',
 144         'info_dict':
 145         {
 146             'id': '7910',
 147             'title': 'محاضرات الشيخ الشعراوي',
 148         },
 149         'playlist_mincount': 27,
 150     }
 151
 152     def _real_extract(self, url):
 153         url, smuggled_data = unsmuggle_url(url, {})
 154         show_id, season_id = self._match_valid_url(url).groups()
 155
 156         data = {}
 157         if season_id:
 158             data['season'] = season_id
 159             show_id = smuggled_data.get('show_id')
 160             if show_id is None:
 161                 season = self._download_json(
 162                     'http://admin.mangomolo.com/analytics/index.php/plus/season_info?id=%s' % season_id,
 163                     season_id, headers={'Origin': 'http://awaan.ae'})
 164                 show_id = season['id']
 165         data['show_id'] = show_id
 166         show = self._download_json(
 167             'http://admin.mangomolo.com/analytics/index.php/plus/show',
 168             show_id, data=urlencode_postdata(data), headers={
 169                 'Origin': 'http://awaan.ae',
 170                 'Content-Type': 'application/x-www-form-urlencoded'
 171             })
 172         if not season_id:
 173             season_id = show['default_season']
 174         for season in show['seasons']:
 175             if season['id'] == season_id:
 176                 title = season.get('title_en') or season['title_ar']
 177
 178                 entries = []
 179                 for video in show['videos']:
 180                     video_id = compat_str(video['id'])
 181                     entries.append(self.url_result(
 182                         'http://awaan.ae/media/%s' % video_id, 'AWAANVideo', video_id))
 183
 184                 return self.playlist_result(entries, season_id, title)