Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/thesixtyone.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import unified_strdate
   6
   7
   8 class TheSixtyOneIE(InfoExtractor):
   9     _VALID_URL = r'''(?x)https?://(?:www\.)?thesixtyone\.com/
  10         (?:.*?/)*
  11         (?:
  12             s|
  13             song/comments/list|
  14             song
  15         )/(?P<id>[A-Za-z0-9]+)/?$'''
  16     _SONG_URL_TEMPLATE = 'http://thesixtyone.com/s/{0:}'
  17     _SONG_FILE_URL_TEMPLATE = 'http://{audio_server:}/thesixtyone_production/audio/{0:}_stream'
  18     _THUMBNAIL_URL_TEMPLATE = '{photo_base_url:}_desktop'
  19     _TESTS = [
  20         {
  21             'url': 'http://www.thesixtyone.com/s/SrE3zD7s1jt/',
  22             'md5': '821cc43b0530d3222e3e2b70bb4622ea',
  23             'info_dict': {
  24                 'id': 'SrE3zD7s1jt',
  25                 'ext': 'mp3',
  26                 'title': 'CASIO - Unicorn War Mixtape',
  27                 'thumbnail': 're:^https?://.*_desktop$',
  28                 'upload_date': '20071217',
  29                 'duration': 3208,
  30             }
  31         },
  32         {
  33             'url': 'http://www.thesixtyone.com/song/comments/list/SrE3zD7s1jt',
  34             'only_matching': True,
  35         },
  36         {
  37             'url': 'http://www.thesixtyone.com/s/ULoiyjuJWli#/s/SrE3zD7s1jt/',
  38             'only_matching': True,
  39         },
  40         {
  41             'url': 'http://www.thesixtyone.com/#/s/SrE3zD7s1jt/',
  42             'only_matching': True,
  43         },
  44         {
  45             'url': 'http://www.thesixtyone.com/song/SrE3zD7s1jt/',
  46             'only_matching': True,
  47         },
  48     ]
  49
  50     _DECODE_MAP = {
  51         'x': 'a',
  52         'm': 'b',
  53         'w': 'c',
  54         'q': 'd',
  55         'n': 'e',
  56         'p': 'f',
  57         'a': '0',
  58         'h': '1',
  59         'e': '2',
  60         'u': '3',
  61         's': '4',
  62         'i': '5',
  63         'o': '6',
  64         'y': '7',
  65         'r': '8',
  66         'c': '9'
  67     }
  68
  69     def _real_extract(self, url):
  70         song_id = self._match_id(url)
  71
  72         webpage = self._download_webpage(
  73             self._SONG_URL_TEMPLATE.format(song_id), song_id)
  74
  75         song_data = self._parse_json(self._search_regex(
  76             r'"%s":\s(\{.*?\})' % song_id, webpage, 'song_data'), song_id)
  77
  78         if self._search_regex(r'(t61\.s3_audio_load\s*=\s*1\.0;)', webpage, 's3_audio_load marker', default=None):
  79             song_data['audio_server'] = 's3.amazonaws.com'
  80         else:
  81             song_data['audio_server'] = song_data['audio_server'] + '.thesixtyone.com'
  82
  83         keys = [self._DECODE_MAP.get(s, s) for s in song_data['key']]
  84         url = self._SONG_FILE_URL_TEMPLATE.format(
  85             "".join(reversed(keys)), **song_data)
  86
  87         formats = [{
  88             'format_id': 'sd',
  89             'url': url,
  90             'ext': 'mp3',
  91         }]
  92
  93         return {
  94             'id': song_id,
  95             'title': '{artist:} - {name:}'.format(**song_data),
  96             'formats': formats,
  97             'comment_count': song_data.get('comments_count'),
  98             'duration': song_data.get('play_time'),
  99             'like_count': song_data.get('score'),
 100             'thumbnail': self._THUMBNAIL_URL_TEMPLATE.format(**song_data),
 101             'upload_date': unified_strdate(song_data.get('publish_date')),
 102         }