yt_dlp/extractor/mzaalo.py

   1 from .common import InfoExtractor
   2 from ..utils import (
   3     parse_age_limit,
   4     parse_duration,
   5     traverse_obj,
   6     url_or_none,
   7 )
   8
   9
  10 class MzaaloIE(InfoExtractor):
  11     _VALID_URL = r'(?i)https?://(?:www\.)?mzaalo\.com/(?:play|watch)/(?P<type>movie|original|clip)/(?P<id>[a-f0-9-]+)/[\w-]+'
  12     _TESTS = [{
  13         # Movies
  14         'url': 'https://www.mzaalo.com/play/movie/c0958d9f-f90e-4503-a755-44358758921d/Jamun',
  15         'info_dict': {
  16             'id': 'c0958d9f-f90e-4503-a755-44358758921d',
  17             'title': 'Jamun',
  18             'ext': 'mp4',
  19             'description': 'md5:24fe9ebb9bbe5b36f7b54b90ab1e2f31',
  20             'thumbnails': 'count:15',
  21             'thumbnail': r're:^https?://.*\.jpg$',
  22             'duration': 5527.0,
  23             'language': 'hin',
  24             'categories': ['Drama'],
  25             'age_limit': 13,
  26         },
  27         'params': {'skip_download': 'm3u8'},
  28     }, {
  29         # Shows
  30         'url': 'https://www.mzaalo.com/play/original/93d42b2b-f373-4c2d-bca4-997412cb069d/Modi-Season-2-CM-TO-PM/Episode-1:Decision,-Not-Promises',
  31         'info_dict': {
  32             'id': '93d42b2b-f373-4c2d-bca4-997412cb069d',
  33             'title': 'Episode 1:Decision, Not Promises',
  34             'ext': 'mp4',
  35             'description': 'md5:16f76058432a54774fbb2561a1955652',
  36             'thumbnails': 'count:22',
  37             'thumbnail': r're:^https?://.*\.jpg$',
  38             'duration': 2040.0,
  39             'language': 'hin',
  40             'categories': ['Drama'],
  41             'age_limit': 13,
  42         },
  43         'params': {'skip_download': 'm3u8'},
  44     }, {
  45         # Streams/Clips
  46         'url': 'https://www.mzaalo.com/play/clip/83cdbcb5-400a-42f1-a1d2-459053cfbda5/Manto-Ki-Kahaaniya',
  47         'info_dict': {
  48             'id': '83cdbcb5-400a-42f1-a1d2-459053cfbda5',
  49             'title': 'Manto Ki Kahaaniya',
  50             'ext': 'mp4',
  51             'description': 'md5:c3c5f1d05f0fd1bfcb05b673d1cc9f2f',
  52             'thumbnails': 'count:3',
  53             'thumbnail': r're:^https?://.*\.jpg$',
  54             'duration': 1937.0,
  55             'language': 'hin',
  56         },
  57         'params': {'skip_download': 'm3u8'},
  58     }, {
  59         'url': 'https://mzaalo.com/watch/MOVIE/389c892d-0b65-4019-bf73-d4edcb1c014f/Chalo-Dilli',
  60         'only_matching': True,
  61     }]
  62
  63     def _real_extract(self, url):
  64         video_id, type_ = self._match_valid_url(url).group('id', 'type')
  65         path = (f'partner/streamurl?&assetId={video_id}&getClipDetails=YES' if type_ == 'clip'
  66                 else f'api/v2/player/details?assetType={type_.upper()}&assetId={video_id}')
  67         data = self._download_json(
  68             f'https://production.mzaalo.com/platform/{path}', video_id, headers={
  69                 'Ocp-Apim-Subscription-Key': '1d0caac2702049b89a305929fdf4cbae',
  70             })['data']
  71
  72         formats = self._extract_m3u8_formats(data['streamURL'], video_id)
  73
  74         subtitles = {}
  75         for subs_lang, subs_url in traverse_obj(data, ('subtitles', {dict.items}, ...)):
  76             if url_or_none(subs_url):
  77                 subtitles[subs_lang] = [{'url': subs_url, 'ext': 'vtt'}]
  78
  79         lang = traverse_obj(data, ('language', {str.lower}))
  80         for f in formats:
  81             f['language'] = lang
  82
  83         return {
  84             'id': video_id,
  85             'formats': formats,
  86             'subtitles': subtitles,
  87             **traverse_obj(data, {
  88                 'title': ('title', {str}),
  89                 'description': ('description', {str}),
  90                 'duration': ('duration', {parse_duration}),
  91                 'age_limit': ('maturity_rating', {parse_age_limit}),
  92                 'thumbnails': ('images', ..., {'url': {url_or_none}}),
  93                 'categories': ('genre', ..., {str}),
  94             }),
  95         }