2 from __future__ import unicode_literals
7 from urlparse import urlparse, urlunparse, urldefrag
8 from urllib import quote, urlencode
9 from os.path import basename
11 from .common import InfoExtractor
12 from ..utils import ExtractorError, compat_urllib_request, compat_html_parser
14 class GroovesharkHtmlParser(compat_html_parser.HTMLParser):
16 self._current_object = None
18 compat_html_parser.HTMLParser.__init__(self)
20 def handle_starttag(self, tag, attrs):
21 attrs = dict((k, v) for k, v in attrs)
23 self._current_object = { 'attrs': attrs, 'params': [] }
25 self._current_object['params'].append(attrs)
27 def handle_endtag(self, tag):
29 self.objects.append(self._current_object)
30 self._current_object = None
33 def extract_object_tags(cls, html):
39 class GroovesharkIE(InfoExtractor):
40 _VALID_URL = r'https?://(www\.)?grooveshark\.com/#!/s/([^/]+)/([^/]+)'
42 'url': 'http://grooveshark.com/#!/s/Jolene+Tenth+Key+Remix+Ft+Will+Sessions/6SS1DW?src=5',
43 'md5': 'bbccc50b19daca23b8f961152c1dc95b',
46 'title': 'Jolene (Tenth Key Remix ft. Will Sessions)',
52 do_playerpage_request = True
53 do_bootstrap_request = True
55 def _parse_target(self, target):
56 uri = urlparse(target)
57 hash = uri.fragment[1:].split('?')[0]
58 token = basename(hash.rstrip('/'))
59 return (uri, hash, token)
61 def _build_bootstrap_url(self, target):
62 (uri, hash, token) = self._parse_target(target)
63 query = 'getCommunicationToken=1&hash=%s&%d' % (quote(hash, safe=''), self.ts)
64 return (urlunparse((uri.scheme, uri.netloc, '/preload.php', None, query, None)), token)
66 def _build_meta_url(self, target):
67 (uri, hash, token) = self._parse_target(target)
68 query = 'hash=%s&%d' % (quote(hash, safe=''), self.ts)
69 return (urlunparse((uri.scheme, uri.netloc, '/preload.php', None, query, None)), token)
71 def _build_stream_url(self, meta):
72 return urlunparse(('http', meta['streamKey']['ip'], '/stream.php', None, None, None))
74 def _build_swf_referer(self, target, obj):
75 (uri, _, _) = self._parse_target(target)
76 return urlunparse((uri.scheme, uri.netloc, obj['attrs']['data'], None, None, None))
78 def _transform_bootstrap(self, js):
79 return re.split('^\s*try\s*{', js, flags=re.M)[0] \
80 .split(' = ', 1)[1].strip().rstrip(';')
82 def _transform_meta(self, js):
83 return js.split('\n')[0].split('=')[1].rstrip(';')
85 def _get_meta(self, target):
86 (meta_url, token) = self._build_meta_url(target)
87 self.to_screen('Metadata URL: %s' % meta_url)
89 headers = {'Referer': urldefrag(target)[0]}
90 req = compat_urllib_request.Request(meta_url, headers=headers)
91 res = self._download_json(req, token,
92 transform_source=self._transform_meta)
94 if 'getStreamKeyWithSong' not in res:
96 'Metadata not found. URL may be malformed, or Grooveshark API may have changed.')
98 if res['getStreamKeyWithSong'] is None:
100 'Metadata download failed, probably due to Grooveshark anti-abuse throttling. Wait at least an hour before retrying from this IP.',
103 return res['getStreamKeyWithSong']
105 def _get_bootstrap(self, target):
106 (bootstrap_url, token) = self._build_bootstrap_url(target)
108 headers = {'Referer': urldefrag(target)[0]}
109 req = compat_urllib_request.Request(bootstrap_url, headers=headers)
110 res = self._download_json(req, token, fatal=False,
111 note='Downloading player bootstrap data',
112 errnote='Unable to download player bootstrap data',
113 transform_source=self._transform_bootstrap)
116 def _get_playerpage(self, target):
117 (_, _, token) = self._parse_target(target)
119 res = self._download_webpage(
121 note='Downloading player page',
122 errnote='Unable to download player page',
126 o = GroovesharkHtmlParser.extract_object_tags(res)
127 return (res, [x for x in o if x['attrs']['id'] == 'jsPlayerEmbed'])
131 def _real_extract(self, url):
132 (target_uri, _, token) = self._parse_target(url)
134 # 1. Fill cookiejar by making a request to the player page
135 if self.do_playerpage_request:
136 (_, player_objs) = self._get_playerpage(url)
137 if player_objs is not None:
138 swf_referer = self._build_swf_referer(url, player_objs[0])
139 self.to_screen('SWF Referer: %s' % swf_referer)
141 # 2. Ask preload.php for swf bootstrap data to better mimic webapp
142 if self.do_bootstrap_request:
143 bootstrap = self._get_bootstrap(url)
144 self.to_screen('CommunicationToken: %s' % bootstrap['getCommunicationToken'])
146 # 3. Ask preload.php for track metadata.
147 meta = self._get_meta(url)
149 # 4. Construct stream request for track.
150 stream_url = self._build_stream_url(meta)
151 duration = int(math.ceil(float(meta['streamKey']['uSecs']) / 1000000))
152 post_dict = {'streamKey': meta['streamKey']['streamKey']}
153 post_data = urlencode(post_dict).encode('utf-8')
155 'Content-Length': len(post_data),
156 'Content-Type': 'application/x-www-form-urlencoded'
159 if 'swf_referer' in locals():
160 headers['Referer'] = swf_referer
162 req = compat_urllib_request.Request(streamurl, post_data, headers)
166 'title': meta['song']['Name'],
167 'http_method': 'POST',
170 'format': 'mp3 audio',
171 'duration': duration,
173 'post_data': post_data,
174 'post_dict': post_dict,
179 if 'swf_referer' in locals():
180 info_dict['http_referer'] = swf_referer
184 def _real_initialize(self):
185 self.ts = int(time.time() * 1000) # timestamp in millis
187 def _download_json(self, url_or_request, video_id,
188 note=u'Downloading JSON metadata',
189 errnote=u'Unable to download JSON metadata',
191 transform_source=None):
193 out = super(GroovesharkIE, self)._download_json(
194 url_or_request, video_id, note, errnote, transform_source)
196 except ExtractorError as ee: