_ Git - youtube-dl/blob - youtube_dl/extractor/beatportpro.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5 import json
   6
   7 from .common import InfoExtractor
   8 from ..utils import int_or_none
   9
  10
  11 class BeatportProIE(InfoExtractor):
  12     _VALID_URL = r'https?://pro\.beatport\.com/track/(?P<display_id>[^/]+)/(?P<id>[0-9]+)'
  13     _TESTS = [{
  14         'url': 'https://pro.beatport.com/track/synesthesia-original-mix/5379371',
  15         'md5': 'b3c34d8639a2f6a7f734382358478887',
  16         'info_dict': {
  17             'id': 5379371,
  18             'display-id': 'synesthesia-original-mix',
  19             'ext': 'mp4',
  20             'title': 'Froxic - Synesthesia (Original Mix)',
  21         },
  22     }, {
  23         'url': 'https://pro.beatport.com/track/love-and-war-original-mix/3756896',
  24         'md5': 'e44c3025dfa38c6577fbaeb43da43514',
  25         'info_dict': {
  26             'id': 3756896,
  27             'display-id': 'love-and-war-original-mix',
  28             'ext': 'mp3',
  29             'title': 'Wolfgang Gartner - Love & War (Original Mix)',
  30         },
  31     }, {
  32         'url': 'https://pro.beatport.com/track/birds-original-mix/4991738',
  33         'md5': 'a1fd8e8046de3950fd039304c186c05f',
  34         'info_dict': {
  35             'id': 4991738,
  36             'display-id': 'birds-original-mix',
  37             'ext': 'mp4',
  38             'title': "Tos, Middle Milk, Mumblin' Johnsson - Birds (Original Mix)",
  39         }
  40     }]
  41
  42     def _real_extract(self, url):
  43         mobj = re.match(self._VALID_URL, url)
  44         track_id = mobj.group('id')
  45         display_id = mobj.group('display_id')
  46
  47         webpage = self._download_webpage(url, display_id)
  48
  49         playables = self._search_regex(
  50             r'window\.Playables\s*=\s*({.*?});', webpage,
  51             'playables info', flags=re.DOTALL)
  52         playables = json.loads(playables)
  53
  54         track = next(t for t in playables['tracks'] if t['id'] == int(track_id))
  55
  56         title = ', '.join((a['name'] for a in track['artists'])) + ' - ' + track['name']
  57         if track['mix']:
  58             title += ' (' + track['mix'] + ')'
  59
  60         formats = []
  61         for ext, info in track['preview'].items():
  62             if info['url'] is None:
  63                 continue
  64             fmt = {
  65                 'url': info['url'],
  66                 'ext': ext,
  67                 'format_id': ext,
  68                 'vcodec': 'none',
  69             }
  70             if ext == 'mp3':
  71                 fmt['preference'] = 0
  72                 fmt['acodec'] = 'mp3'
  73                 fmt['abr'] = 96
  74                 fmt['asr'] = 44100
  75             elif ext == 'mp4':
  76                 fmt['preference'] = 1
  77                 fmt['acodec'] = 'aac'
  78                 fmt['abr'] = 96
  79                 fmt['asr'] = 44100
  80             formats.append(fmt)
  81         self._sort_formats(formats)
  82
  83         images = []
  84         for name, info in track['images'].items():
  85             image_url = info.get('url')
  86             if name == 'dynamic' or not image_url:
  87                 continue
  88             img = {
  89                 'id': name,
  90                 'url': image_url,
  91                 'height': int_or_none(info.get('height')),
  92                 'width': int_or_none(info.get('width')),
  93             }
  94             images.append(img)
  95
  96         return {
  97             'id': track['id'],
  98             'display_id': track.get('slug') or display_id,
  99             'title': title,
 100             'formats': formats,
 101             'thumbnails': images,
 102         }