yt_dlp/extractor/beatport.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..compat import compat_str
   5 from ..utils import int_or_none
   6
   7
   8 class BeatportIE(InfoExtractor):
   9     _VALID_URL = r'https?://(?:www\.|pro\.)?beatport\.com/track/(?P<display_id>[^/]+)/(?P<id>[0-9]+)'
  10     _TESTS = [{
  11         'url': 'https://beatport.com/track/synesthesia-original-mix/5379371',
  12         'md5': 'b3c34d8639a2f6a7f734382358478887',
  13         'info_dict': {
  14             'id': '5379371',
  15             'display_id': 'synesthesia-original-mix',
  16             'ext': 'mp4',
  17             'title': 'Froxic - Synesthesia (Original Mix)',
  18         },
  19     }, {
  20         'url': 'https://beatport.com/track/love-and-war-original-mix/3756896',
  21         'md5': 'e44c3025dfa38c6577fbaeb43da43514',
  22         'info_dict': {
  23             'id': '3756896',
  24             'display_id': 'love-and-war-original-mix',
  25             'ext': 'mp3',
  26             'title': 'Wolfgang Gartner - Love & War (Original Mix)',
  27         },
  28     }, {
  29         'url': 'https://beatport.com/track/birds-original-mix/4991738',
  30         'md5': 'a1fd8e8046de3950fd039304c186c05f',
  31         'info_dict': {
  32             'id': '4991738',
  33             'display_id': 'birds-original-mix',
  34             'ext': 'mp4',
  35             'title': "Tos, Middle Milk, Mumblin' Johnsson - Birds (Original Mix)",
  36         }
  37     }]
  38
  39     def _real_extract(self, url):
  40         mobj = self._match_valid_url(url)
  41         track_id = mobj.group('id')
  42         display_id = mobj.group('display_id')
  43
  44         webpage = self._download_webpage(url, display_id)
  45
  46         playables = self._parse_json(
  47             self._search_regex(
  48                 r'window\.Playables\s*=\s*({.+?});', webpage,
  49                 'playables info', flags=re.DOTALL),
  50             track_id)
  51
  52         track = next(t for t in playables['tracks'] if t['id'] == int(track_id))
  53
  54         title = ', '.join((a['name'] for a in track['artists'])) + ' - ' + track['name']
  55         if track['mix']:
  56             title += ' (' + track['mix'] + ')'
  57
  58         formats = []
  59         for ext, info in track['preview'].items():
  60             if not info['url']:
  61                 continue
  62             fmt = {
  63                 'url': info['url'],
  64                 'ext': ext,
  65                 'format_id': ext,
  66                 'vcodec': 'none',
  67             }
  68             if ext == 'mp3':
  69                 fmt['acodec'] = 'mp3'
  70                 fmt['abr'] = 96
  71                 fmt['asr'] = 44100
  72             elif ext == 'mp4':
  73                 fmt['acodec'] = 'aac'
  74                 fmt['abr'] = 96
  75                 fmt['asr'] = 44100
  76             formats.append(fmt)
  77
  78         images = []
  79         for name, info in track['images'].items():
  80             image_url = info.get('url')
  81             if name == 'dynamic' or not image_url:
  82                 continue
  83             image = {
  84                 'id': name,
  85                 'url': image_url,
  86                 'height': int_or_none(info.get('height')),
  87                 'width': int_or_none(info.get('width')),
  88             }
  89             images.append(image)
  90
  91         return {
  92             'id': compat_str(track.get('id')) or track_id,
  93             'display_id': track.get('slug') or display_id,
  94             'title': title,
  95             'formats': formats,
  96             'thumbnails': images,
  97         }