_ 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<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         track_id = self._match_id(url)
  44         webpage = self._download_webpage(url, track_id)
  45
  46         playables = self._search_regex(
  47             r'window\.Playables\s*=\s*({.*?});', webpage,
  48             'playables info', flags=re.DOTALL)
  49         playables = json.loads(playables)
  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 info['url'] is None:
  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['preference'] = 0
  69                 fmt['acodec'] = 'mp3'
  70                 fmt['abr'] = 96
  71                 fmt['asr'] = 44100
  72             elif ext == 'mp4':
  73                 fmt['preference'] = 1
  74                 fmt['acodec'] = 'aac'
  75                 fmt['abr'] = 96
  76                 fmt['asr'] = 44100
  77             formats.append(fmt)
  78         self._sort_formats(formats)
  79
  80         images = []
  81         for name, info in track['images'].items():
  82             image_url = info.get('url')
  83             if name == 'dynamic' or not image_url:
  84                 continue
  85             img = {
  86                 'id': name,
  87                 'url': image_url,
  88                 'height': int_or_none(info.get('height')),
  89                 'width': int_or_none(info.get('width')),
  90             }
  91             images.append(img)
  92
  93         return {
  94             'id': track['id'],
  95             'display-id': track['slug'],
  96             'title': title,
  97             'formats': formats,
  98             'thumbnails': images,
  99         }