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
15 class GroovesharkHtmlParser(compat_html_parser.HTMLParser):
17 self._current_object = None
19 compat_html_parser.HTMLParser.__init__(self)
21 def handle_starttag(self, tag, attrs):
22 attrs = dict((k, v) for k, v in attrs)
24 self._current_object = {'attrs': attrs, 'params': []}
26 self._current_object['params'].append(attrs)
28 def handle_endtag(self, tag):
30 self.objects.append(self._current_object)
31 self._current_object = None
34 def extract_object_tags(cls, html):
41 class GroovesharkIE(InfoExtractor):
42 _VALID_URL = r'https?://(www\.)?grooveshark\.com/#!/s/([^/]+)/([^/]+)'
44 'url': 'http://grooveshark.com/#!/s/Jolene+Tenth+Key+Remix+Ft+Will+Sessions/6SS1DW?src=5',
45 'md5': 'bbccc50b19daca23b8f961152c1dc95b',
48 'title': 'Jolene (Tenth Key Remix ft. Will Sessions)',
54 do_playerpage_request = True
55 do_bootstrap_request = True
57 def _parse_target(self, target):
58 uri = urlparse(target)
59 hash = uri.fragment[1:].split('?')[0]
60 token = basename(hash.rstrip('/'))
61 return (uri, hash, token)
63 def _build_bootstrap_url(self, target):
64 (uri, hash, token) = self._parse_target(target)
65 query = 'getCommunicationToken=1&hash=%s&%d' % (quote(hash, safe=''), self.ts)
66 return (urlunparse((uri.scheme, uri.netloc, '/preload.php', None, query, None)), token)
68 def _build_meta_url(self, target):
69 (uri, hash, token) = self._parse_target(target)
70 query = 'hash=%s&%d' % (quote(hash, safe=''), self.ts)
71 return (urlunparse((uri.scheme, uri.netloc, '/preload.php', None, query, None)), token)
73 def _build_stream_url(self, meta):
74 return urlunparse(('http', meta['streamKey']['ip'], '/stream.php', None, None, None))
76 def _build_swf_referer(self, target, obj):
77 (uri, _, _) = self._parse_target(target)
78 return urlunparse((uri.scheme, uri.netloc, obj['attrs']['data'], None, None, None))
80 def _transform_bootstrap(self, js):
81 return re.split('^\s*try\s*{', js, flags=re.M)[0] \
82 .split(' = ', 1)[1].strip().rstrip(';')
84 def _transform_meta(self, js):
85 return js.split('\n')[0].split('=')[1].rstrip(';')
87 def _get_meta(self, target):
88 (meta_url, token) = self._build_meta_url(target)
89 self.to_screen('Metadata URL: %s' % meta_url)
91 headers = {'Referer': urldefrag(target)[0]}
92 req = compat_urllib_request.Request(meta_url, headers=headers)
93 res = self._download_json(req, token,
94 transform_source=self._transform_meta)
96 if 'getStreamKeyWithSong' not in res:
98 'Metadata not found. URL may be malformed, or Grooveshark API may have changed.')
100 if res['getStreamKeyWithSong'] is None:
101 raise ExtractorError(
102 'Metadata download failed, probably due to Grooveshark anti-abuse throttling. Wait at least an hour before retrying from this IP.',
105 return res['getStreamKeyWithSong']
107 def _get_bootstrap(self, target):
108 (bootstrap_url, token) = self._build_bootstrap_url(target)
110 headers = {'Referer': urldefrag(target)[0]}
111 req = compat_urllib_request.Request(bootstrap_url, headers=headers)
112 res = self._download_json(req, token, fatal=False,
113 note='Downloading player bootstrap data',
114 errnote='Unable to download player bootstrap data',
115 transform_source=self._transform_bootstrap)
118 def _get_playerpage(self, target):
119 (_, _, token) = self._parse_target(target)
121 res = self._download_webpage(
123 note='Downloading player page',
124 errnote='Unable to download player page',
128 o = GroovesharkHtmlParser.extract_object_tags(res)
129 return (res, [x for x in o if x['attrs']['id'] == 'jsPlayerEmbed'])
133 def _real_extract(self, url):
134 (target_uri, _, token) = self._parse_target(url)
136 # 1. Fill cookiejar by making a request to the player page
137 if self.do_playerpage_request:
138 (_, player_objs) = self._get_playerpage(url)
139 if player_objs is not None:
140 swf_referer = self._build_swf_referer(url, player_objs[0])
141 self.to_screen('SWF Referer: %s' % swf_referer)
143 # 2. Ask preload.php for swf bootstrap data to better mimic webapp
144 if self.do_bootstrap_request:
145 bootstrap = self._get_bootstrap(url)
146 self.to_screen('CommunicationToken: %s' % bootstrap['getCommunicationToken'])
148 # 3. Ask preload.php for track metadata.
149 meta = self._get_meta(url)
151 # 4. Construct stream request for track.
152 stream_url = self._build_stream_url(meta)
153 duration = int(math.ceil(float(meta['streamKey']['uSecs']) / 1000000))
154 post_dict = {'streamKey': meta['streamKey']['streamKey']}
155 post_data = urlencode(post_dict).encode('utf-8')
157 'Content-Length': len(post_data),
158 'Content-Type': 'application/x-www-form-urlencoded'
161 if 'swf_referer' in locals():
162 headers['Referer'] = swf_referer
164 req = compat_urllib_request.Request(stream_url, post_data, headers)
168 'title': meta['song']['Name'],
169 'http_method': 'POST',
172 'format': 'mp3 audio',
173 'duration': duration,
175 # various ways of supporting the download request.
176 # remove keys unnecessary to the eventual post implementation
177 'post_data': post_data,
178 'post_dict': post_dict,
183 if 'swf_referer' in locals():
184 info_dict['http_referer'] = swf_referer
188 def _real_initialize(self):
189 self.ts = int(time.time() * 1000) # timestamp in millis
191 def _download_json(self, url_or_request, video_id,
192 note=u'Downloading JSON metadata',
193 errnote=u'Unable to download JSON metadata',
195 transform_source=None):
197 out = super(GroovesharkIE, self)._download_json(
198 url_or_request, video_id, note, errnote, transform_source)
200 except ExtractorError as ee: