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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import json
   5 import re
   6
   7 from .common import InfoExtractor
   8 from ..utils import unified_strdate
   9
  10
  11 class TheSixtyOneIE(InfoExtractor):
  12     _VALID_URL = r'''(?x)https?://(?:www\.)?thesixtyone\.com/
  13         (?:.*?/)*
  14         (?:
  15             s|
  16             song/comments/list|
  17             song
  18         )/(?P<id>[A-Za-z0-9]+)/?$'''
  19     _SONG_URL_TEMPLATE = 'http://thesixtyone.com/s/{0:}'
  20     _SONG_FILE_URL_TEMPLATE = 'http://{audio_server:}.thesixtyone.com/thesixtyone_production/audio/{0:}_stream'
  21     _THUMBNAIL_URL_TEMPLATE = '{photo_base_url:}_desktop'
  22     _TESTS = [
  23         {
  24             'url': 'http://www.thesixtyone.com/s/SrE3zD7s1jt/',
  25             'md5': '821cc43b0530d3222e3e2b70bb4622ea',
  26             'info_dict': {
  27                 'id': 'SrE3zD7s1jt',
  28                 'ext': 'mp3',
  29                 'title': 'CASIO - Unicorn War Mixtape',
  30                 'thumbnail': 're:^https?://.*_desktop$',
  31                 'upload_date': '20071217',
  32                 'duration': 3208,
  33             }
  34         },
  35         {
  36             'url': 'http://www.thesixtyone.com/song/comments/list/SrE3zD7s1jt',
  37             'only_matching': True,
  38         },
  39         {
  40             'url': 'http://www.thesixtyone.com/s/ULoiyjuJWli#/s/SrE3zD7s1jt/',
  41             'only_matching': True,
  42         },
  43         {
  44             'url': 'http://www.thesixtyone.com/#/s/SrE3zD7s1jt/',
  45             'only_matching': True,
  46         },
  47         {
  48             'url': 'http://www.thesixtyone.com/song/SrE3zD7s1jt/',
  49             'only_matching': True,
  50         },
  51     ]
  52
  53     _DECODE_MAP = {
  54         "x": "a",
  55         "m": "b",
  56         "w": "c",
  57         "q": "d",
  58         "n": "e",
  59         "p": "f",
  60         "a": "0",
  61         "h": "1",
  62         "e": "2",
  63         "u": "3",
  64         "s": "4",
  65         "i": "5",
  66         "o": "6",
  67         "y": "7",
  68         "r": "8",
  69         "c": "9"
  70     }
  71
  72     def _real_extract(self, url):
  73         mobj = re.match(self._VALID_URL, url)
  74         song_id = mobj.group('id')
  75
  76         webpage = self._download_webpage(
  77             self._SONG_URL_TEMPLATE.format(song_id), song_id)
  78
  79         song_data = json.loads(self._search_regex(
  80             r'"%s":\s(\{.*?\})' % song_id, webpage, 'song_data'))
  81         keys = [self._DECODE_MAP.get(s, s) for s in song_data['key']]
  82         url = self._SONG_FILE_URL_TEMPLATE.format(
  83             "".join(reversed(keys)), **song_data)
  84
  85         formats = [{
  86             'format_id': 'sd',
  87             'url': url,
  88             'ext': 'mp3',
  89         }]
  90
  91         return {
  92             'id': song_id,
  93             'title': '{artist:} - {name:}'.format(**song_data),
  94             'formats': formats,
  95             'comment_count': song_data.get('comments_count'),
  96             'duration': song_data.get('play_time'),
  97             'like_count': song_data.get('score'),
  98             'thumbnail': self._THUMBNAIL_URL_TEMPLATE.format(**song_data),
  99             'upload_date': unified_strdate(song_data.get('publish_date')),
 100         }