yt_dlp/extractor/beatport.py

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