1 from __future__ import unicode_literals
5 from .common import InfoExtractor
7 compat_urllib_parse_unquote,
20 class MixcloudIE(InfoExtractor):
21 _VALID_URL = r'^(?:https?://)?(?:www\.)?mixcloud\.com/([^/]+)/(?!stream|uploads|favorites|listens|playlists)([^/]+)'
25 'url': 'http://www.mixcloud.com/dholbach/cryptkeeper/',
27 'id': 'dholbach-cryptkeeper',
29 'title': 'Cryptkeeper',
30 'description': 'After quite a long silence from myself, finally another Drum\'n\'Bass mix with my favourite current dance floor bangers.',
31 'uploader': 'Daniel Holbach',
32 'uploader_id': 'dholbach',
33 'thumbnail': 're:https?://.*\.jpg',
38 'url': 'http://www.mixcloud.com/gillespeterson/caribou-7-inch-vinyl-mix-chat/',
40 'id': 'gillespeterson-caribou-7-inch-vinyl-mix-chat',
42 'title': 'Caribou 7 inch Vinyl Mix & Chat',
43 'description': 'md5:2b8aec6adce69f9d41724647c65875e8',
44 'uploader': 'Gilles Peterson Worldwide',
45 'uploader_id': 'gillespeterson',
46 'thumbnail': 're:https?://.*/images/',
52 def _check_url(self, url, track_id, ext):
54 # We only want to know if the request succeed
55 # don't download the whole file
56 self._request_webpage(
57 HEADRequest(url), track_id,
58 'Trying %s URL' % ext)
60 except ExtractorError:
63 def _real_extract(self, url):
64 mobj = re.match(self._VALID_URL, url)
65 uploader = mobj.group(1)
66 cloudcast_name = mobj.group(2)
67 track_id = compat_urllib_parse_unquote('-'.join((uploader, cloudcast_name)))
69 webpage = self._download_webpage(url, track_id)
71 message = self._html_search_regex(
72 r'(?s)<div[^>]+class="global-message cloudcast-disabled-notice-light"[^>]*>(.+?)<(?:a|/div)',
73 webpage, 'error message', default=None)
75 preview_url = self._search_regex(
76 r'\s(?:data-preview-url|m-preview)="([^"]+)"',
77 webpage, 'preview url', default=None if message else NO_DEFAULT)
80 raise ExtractorError('%s said: %s' % (self.IE_NAME, message), expected=True)
82 song_url = re.sub(r'audiocdn(\d+)', r'stream\1', preview_url)
83 song_url = song_url.replace('/previews/', '/c/originals/')
84 if not self._check_url(song_url, track_id, 'mp3'):
85 song_url = song_url.replace('.mp3', '.m4a').replace('originals/', 'm4a/64/')
86 if not self._check_url(song_url, track_id, 'm4a'):
87 raise ExtractorError('Unable to extract track url')
90 r'm-play-on-spacebar[^>]+'
91 r'(?:\s+[a-zA-Z0-9-]+(?:="[^"]+")?)*?\s+')
92 title = self._html_search_regex(
93 PREFIX + r'm-title="([^"]+)"', webpage, 'title')
94 thumbnail = self._proto_relative_url(self._html_search_regex(
95 PREFIX + r'm-thumbnail-url="([^"]+)"', webpage, 'thumbnail',
97 uploader = self._html_search_regex(
98 PREFIX + r'm-owner-name="([^"]+)"',
99 webpage, 'uploader', fatal=False)
100 uploader_id = self._search_regex(
101 r'\s+"profile": "([^"]+)",', webpage, 'uploader id', fatal=False)
102 description = self._og_search_description(webpage)
103 like_count = parse_count(self._search_regex(
104 r'\bbutton-favorite[^>]+>.*?<span[^>]+class=["\']toggle-number[^>]+>\s*([^<]+)',
105 webpage, 'like count', fatal=False))
106 view_count = str_to_int(self._search_regex(
107 [r'<meta itemprop="interactionCount" content="UserPlays:([0-9]+)"',
108 r'/listeners/?">([0-9,.]+)</a>'],
109 webpage, 'play count', fatal=False))
115 'description': description,
116 'thumbnail': thumbnail,
117 'uploader': uploader,
118 'uploader_id': uploader_id,
119 'view_count': view_count,
120 'like_count': like_count,
124 class MixcloudUserIE(InfoExtractor):
126 Information extractor for Mixcloud users.
127 It can retrieve a list of a user's uploads, favorites or listens.
130 _VALID_URL = r'^(?:https?://)?(?:www\.)?mixcloud\.com/(?P<user>[^/]+)/(?P<type>uploads|favorites|listens)?/?$'
131 IE_NAME = 'mixcloud:user'
134 'url': 'http://www.mixcloud.com/dholbach/',
136 'id': 'dholbach/uploads',
137 'title': 'Daniel Holbach (uploads)',
138 'description': 'md5:327af72d1efeb404a8216c27240d1370',
140 'playlist_mincount': 11
142 'url': 'http://www.mixcloud.com/dholbach/uploads/',
144 'id': 'dholbach/uploads',
145 'title': 'Daniel Holbach (uploads)',
146 'description': 'md5:327af72d1efeb404a8216c27240d1370',
148 'playlist_mincount': 11
150 'url': 'http://www.mixcloud.com/dholbach/favorites/',
152 'id': 'dholbach/favorites',
153 'title': 'Daniel Holbach (favorites)',
154 'description': 'md5:327af72d1efeb404a8216c27240d1370',
156 'playlist_mincount': 244
158 'url': 'http://www.mixcloud.com/dholbach/listens/',
160 'id': 'dholbach/listens',
161 'title': 'Daniel Holbach (listens)',
162 'description': 'md5:327af72d1efeb404a8216c27240d1370',
164 'playlist_mincount': 846
167 def _fetch_tracks(self, base_url, video_id, dl_note=None, dl_errnote=None):
168 # retrieve all fragments of a list of tracks with fake AJAX calls
172 # fake a AJAX request to retrieve a list fragment
173 page_url = base_url + "?page=%d&list=main&_ajax=1" % current_page
174 req = compat_urllib_request.Request(page_url, headers={"X-Requested-With": "XMLHttpRequest"}, method="GET")
175 resp = self._download_webpage(req, video_id, note=dl_note + " (page %d)" % current_page, errnote=dl_errnote)
177 # extract all track URLs from fragment
178 urls = re.findall(r'm-play-button m-url="(?P<url>[^"]+)"', resp)
180 urls = map(clean_html, urls)
181 # create absolute URLs
182 urls = map(lambda u: "https://www.mixcloud.com" + u, urls)
183 track_urls.extend(urls)
185 # advance to next fragment, if any
186 if " m-next-page-url=" in resp:
193 def _handle_track_urls(self, urls):
194 return map(lambda u: self.url_result(u, "Mixcloud"), urls)
196 def _get_user_description(self, page_content):
197 return self._html_search_regex(
198 r'<div class="description-text">.*?<p>(?P<description>.*?)</p></div></div></div>',
205 def _get_username(self, page_content):
206 return self._og_search_title(page_content)
208 def _real_extract(self, url):
209 mobj = re.match(self._VALID_URL, url)
210 user_id = mobj.group("user")
211 list_type = mobj.group("type")
213 # if only a profile URL was supplied, default to download all uploads
214 if list_type is None:
215 list_type = "uploads"
217 video_id = "%s/%s" % (user_id, list_type)
219 # download the user's profile to retrieve some metadata
220 profile = self._download_webpage("https://www.mixcloud.com/%s/" % user_id,
222 note="Downloading user profile",
223 errnote="Unable to download user profile")
225 username = self._get_username(profile)
226 description = self._get_user_description(profile)
228 # retrieve all page fragments of uploads, favorites or listens
229 track_urls = self._fetch_tracks(
230 "https://www.mixcloud.com/%s/%s/" % (user_id, list_type),
232 dl_note="Downloading list of %s" % list_type,
233 dl_errnote="Unable to download list of %s" % list_type)
235 # let MixcloudIE handle each track URL
236 entries = self._handle_track_urls(track_urls)
241 'title': "%s (%s)" % (username, list_type),
243 "description": description
247 class MixcloudPlaylistIE(MixcloudUserIE):
249 Information extractor for Mixcloud playlists.
252 _VALID_URL = r'^(?:https?://)?(?:www\.)?mixcloud\.com/(?P<user>[^/]+)/playlists/(?P<playlist>[^/]+)/?$'
253 IE_NAME = 'mixcloud:playlist'
256 'url': 'https://www.mixcloud.com/RedBullThre3style/playlists/tokyo-finalists-2015/',
258 'id': 'RedBullThre3style/playlists/tokyo-finalists-2015',
259 'title': 'National Champions 2015',
260 'description': 'md5:6ff5fb01ac76a31abc9b3939c16243a3',
262 'playlist_mincount': 16
264 'url': 'https://www.mixcloud.com/maxvibes/playlists/jazzcat-on-ness-radio/',
266 'id': 'maxvibes/playlists/jazzcat-on-ness-radio',
267 'title': 'Jazzcat on Ness Radio',
268 'description': 'md5:7bbbf0d6359a0b8cda85224be0f8f263',
270 'playlist_mincount': 23
273 def _get_playlist_title(self, page_content):
274 return self._html_search_regex(
275 r'<span class="main-list-title list-playlist-title ">(?P<title>.*?)</span>',
282 def _real_extract(self, url):
283 mobj = re.match(self._VALID_URL, url)
284 user_id = mobj.group("user")
285 playlist_id = mobj.group("playlist")
286 video_id = "%s/playlists/%s" % (user_id, playlist_id)
288 # download the playlist page to retrieve some metadata
289 profile = self._download_webpage(url,
291 note="Downloading playlist page",
292 errnote="Unable to download playlist page")
294 description = self._get_user_description(profile)
295 playlist_title = self._get_playlist_title(profile)
297 # retrieve all page fragments of playlist
298 track_urls = self._fetch_tracks(
299 "https://www.mixcloud.com/%s/playlists/%s/" % (user_id, playlist_id),
301 dl_note="Downloading tracklist of %s" % playlist_title,
302 dl_errnote="Unable to tracklist of %s" % playlist_title)
304 # let MixcloudIE handle each track
305 entries = self._handle_track_urls(track_urls)
310 'title': playlist_title,
312 "description": description