2 from __future__ import unicode_literals
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 from ..utils import compat_urlparse
16 urlunparse = compat_urlparse.urlunparse
17 urldefrag = compat_urlparse.urldefrag
20 class GroovesharkHtmlParser(compat_html_parser.HTMLParser):
22 self._current_object = None
24 compat_html_parser.HTMLParser.__init__(self)
26 def handle_starttag(self, tag, attrs):
27 attrs = dict((k, v) for k, v in attrs)
29 self._current_object = {'attrs': attrs, 'params': []}
31 self._current_object['params'].append(attrs)
33 def handle_endtag(self, tag):
35 self.objects.append(self._current_object)
36 self._current_object = None
39 def extract_object_tags(cls, html):
46 class GroovesharkIE(InfoExtractor):
47 _VALID_URL = r'https?://(www\.)?grooveshark\.com/#!/s/([^/]+)/([^/]+)'
49 'url': 'http://grooveshark.com/#!/s/Jolene+Tenth+Key+Remix+Ft+Will+Sessions/6SS1DW?src=5',
50 'md5': 'bbccc50b19daca23b8f961152c1dc95b',
53 'title': 'Jolene (Tenth Key Remix ft. Will Sessions)',
59 do_playerpage_request = True
60 do_bootstrap_request = True
62 def _parse_target(self, target):
63 uri = compat_urlparse.urlparse(target)
64 hash = uri.fragment[1:].split('?')[0]
65 token = basename(hash.rstrip('/'))
66 return (uri, hash, token)
68 def _build_bootstrap_url(self, target):
69 (uri, hash, token) = self._parse_target(target)
70 query = 'getCommunicationToken=1&hash=%s&%d' % (quote(hash, safe=''), self.ts)
71 return (urlunparse((uri.scheme, uri.netloc, '/preload.php', None, query, None)), token)
73 def _build_meta_url(self, target):
74 (uri, hash, token) = self._parse_target(target)
75 query = 'hash=%s&%d' % (quote(hash, safe=''), self.ts)
76 return (urlunparse((uri.scheme, uri.netloc, '/preload.php', None, query, None)), token)
78 def _build_stream_url(self, meta):
79 return urlunparse(('http', meta['streamKey']['ip'], '/stream.php', None, None, None))
81 def _build_swf_referer(self, target, obj):
82 (uri, _, _) = self._parse_target(target)
83 return urlunparse((uri.scheme, uri.netloc, obj['attrs']['data'], None, None, None))
85 def _transform_bootstrap(self, js):
86 return re.split('(?m)^\s*try\s*{', js)[0] \
87 .split(' = ', 1)[1].strip().rstrip(';')
89 def _transform_meta(self, js):
90 return js.split('\n')[0].split('=')[1].rstrip(';')
92 def _get_meta(self, target):
93 (meta_url, token) = self._build_meta_url(target)
94 self.to_screen('Metadata URL: %s' % meta_url)
96 headers = {'Referer': urldefrag(target)[0]}
97 req = compat_urllib_request.Request(meta_url, headers=headers)
98 res = self._download_json(req, token,
99 transform_source=self._transform_meta)
101 if 'getStreamKeyWithSong' not in res:
102 raise ExtractorError(
103 'Metadata not found. URL may be malformed, or Grooveshark API may have changed.')
105 if res['getStreamKeyWithSong'] is None:
106 raise ExtractorError(
107 'Metadata download failed, probably due to Grooveshark anti-abuse throttling. Wait at least an hour before retrying from this IP.',
110 return res['getStreamKeyWithSong']
112 def _get_bootstrap(self, target):
113 (bootstrap_url, token) = self._build_bootstrap_url(target)
115 headers = {'Referer': urldefrag(target)[0]}
116 req = compat_urllib_request.Request(bootstrap_url, headers=headers)
117 res = self._download_json(req, token, fatal=False,
118 note='Downloading player bootstrap data',
119 errnote='Unable to download player bootstrap data',
120 transform_source=self._transform_bootstrap)
123 def _get_playerpage(self, target):
124 (_, _, token) = self._parse_target(target)
127 webpage = self._download_webpage(
129 note='Downloading player page',
130 errnote='Unable to download player page',
133 if webpage is not None:
134 # Search (for example German) error message
135 error_msg = self._html_search_regex(
136 r'<div id="content">\s*<h2>(.*?)</h2>', webpage,
137 'error message', default=None)
138 if error_msg is not None:
139 error_msg = error_msg.replace('\n', ' ')
140 raise ExtractorError('Grooveshark said: %s' % error_msg)
142 if webpage is not None:
143 o = GroovesharkHtmlParser.extract_object_tags(webpage)
144 return (webpage, [x for x in o if x['attrs']['id'] == 'jsPlayerEmbed'])
146 return (webpage, None)
148 def _real_initialize(self):
149 self.ts = int(time.time() * 1000) # timestamp in millis
151 def _real_extract(self, url):
152 (target_uri, _, token) = self._parse_target(url)
154 # 1. Fill cookiejar by making a request to the player page
156 if self.do_playerpage_request:
157 (_, player_objs) = self._get_playerpage(url)
158 if player_objs is not None:
159 swf_referer = self._build_swf_referer(url, player_objs[0])
160 self.to_screen('SWF Referer: %s' % swf_referer)
162 # 2. Ask preload.php for swf bootstrap data to better mimic webapp
163 if self.do_bootstrap_request:
164 bootstrap = self._get_bootstrap(url)
165 self.to_screen('CommunicationToken: %s' % bootstrap['getCommunicationToken'])
167 # 3. Ask preload.php for track metadata.
168 meta = self._get_meta(url)
170 # 4. Construct stream request for track.
171 stream_url = self._build_stream_url(meta)
172 duration = int(math.ceil(float(meta['streamKey']['uSecs']) / 1000000))
173 post_dict = {'streamKey': meta['streamKey']['streamKey']}
174 post_data = urlencode(post_dict).encode('utf-8')
176 'Content-Length': len(post_data),
177 'Content-Type': 'application/x-www-form-urlencoded'
179 if swf_referer is not None:
180 headers['Referer'] = swf_referer
184 'title': meta['song']['Name'],
185 'http_method': 'POST',
188 'format': 'mp3 audio',
189 'duration': duration,
190 'http_post_data': post_data,
191 'http_headers': headers,