from __future__ import unicode_literals
+import base64
import functools
+import itertools
import re
from .common import InfoExtractor
from ..compat import (
+ compat_chr,
+ compat_ord,
compat_urllib_parse_unquote,
compat_urlparse,
)
from ..utils import (
clean_html,
ExtractorError,
- HEADRequest,
OnDemandPagedList,
- NO_DEFAULT,
parse_count,
str_to_int,
)
class MixcloudIE(InfoExtractor):
- _VALID_URL = r'^(?:https?://)?(?:www\.)?mixcloud\.com/([^/]+)/(?!stream|uploads|favorites|listens|playlists)([^/]+)'
+ _VALID_URL = r'https?://(?:(?:www|beta|m)\.)?mixcloud\.com/([^/]+)/(?!stream|uploads|favorites|listens|playlists)([^/]+)'
IE_NAME = 'mixcloud'
_TESTS = [{
'description': 'After quite a long silence from myself, finally another Drum\'n\'Bass mix with my favourite current dance floor bangers.',
'uploader': 'Daniel Holbach',
'uploader_id': 'dholbach',
- 'thumbnail': 're:https?://.*\.jpg',
+ 'thumbnail': r're:https?://.*\.jpg',
'view_count': int,
'like_count': int,
},
'description': 'md5:2b8aec6adce69f9d41724647c65875e8',
'uploader': 'Gilles Peterson Worldwide',
'uploader_id': 'gillespeterson',
- 'thumbnail': 're:https?://.*/images/',
+ 'thumbnail': 're:https?://.*',
'view_count': int,
'like_count': int,
},
+ }, {
+ 'url': 'https://beta.mixcloud.com/RedLightRadio/nosedrip-15-red-light-radio-01-18-2016/',
+ 'only_matching': True,
}]
- def _check_url(self, url, track_id, ext):
- try:
- # We only want to know if the request succeed
- # don't download the whole file
- self._request_webpage(
- HEADRequest(url), track_id,
- 'Trying %s URL' % ext)
- return True
- except ExtractorError:
- return False
+ # See https://www.mixcloud.com/media/js2/www_js_2.9e23256562c080482435196ca3975ab5.js
+ @staticmethod
+ def _decrypt_play_info(play_info):
+ KEY = 'pleasedontdownloadourmusictheartistswontgetpaid'
+
+ play_info = base64.b64decode(play_info.encode('ascii'))
+
+ return ''.join([
+ compat_chr(compat_ord(ch) ^ compat_ord(KEY[idx % len(KEY)]))
+ for idx, ch in enumerate(play_info)])
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
r'(?s)<div[^>]+class="global-message cloudcast-disabled-notice-light"[^>]*>(.+?)<(?:a|/div)',
webpage, 'error message', default=None)
- preview_url = self._search_regex(
- r'\s(?:data-preview-url|m-preview)="([^"]+)"',
- webpage, 'preview url', default=None if message else NO_DEFAULT)
+ encrypted_play_info = self._search_regex(
+ r'm-play-info="([^"]+)"', webpage, 'play info')
+ play_info = self._parse_json(
+ self._decrypt_play_info(encrypted_play_info), track_id)
- if message:
+ if message and 'stream_url' not in play_info:
raise ExtractorError('%s said: %s' % (self.IE_NAME, message), expected=True)
- song_url = re.sub(r'audiocdn(\d+)', r'stream\1', preview_url)
- song_url = song_url.replace('/previews/', '/c/originals/')
- if not self._check_url(song_url, track_id, 'mp3'):
- song_url = song_url.replace('.mp3', '.m4a').replace('originals/', 'm4a/64/')
- if not self._check_url(song_url, track_id, 'm4a'):
- raise ExtractorError('Unable to extract track url')
+ song_url = play_info['stream_url']
PREFIX = (
r'm-play-on-spacebar[^>]+'
description = self._og_search_description(webpage)
like_count = parse_count(self._search_regex(
r'\bbutton-favorite[^>]+>.*?<span[^>]+class=["\']toggle-number[^>]+>\s*([^<]+)',
- webpage, 'like count', fatal=False))
+ webpage, 'like count', default=None))
view_count = str_to_int(self._search_regex(
[r'<meta itemprop="interactionCount" content="UserPlays:([0-9]+)"',
r'/listeners/?">([0-9,.]+)</a>'],
- webpage, 'play count', fatal=False))
+ webpage, 'play count', default=None))
return {
'id': track_id,
class MixcloudPlaylistBaseIE(InfoExtractor):
_PAGE_SIZE = 24
- def _fetch_tracks_page(self, path, video_id, page_name, current_page):
- resp = self._download_webpage(
+ def _find_urls_in_page(self, page):
+ for url in re.findall(r'm-play-button m-url="(?P<url>[^"]+)"', page):
+ yield self.url_result(
+ compat_urlparse.urljoin('https://www.mixcloud.com', clean_html(url)),
+ MixcloudIE.ie_key())
+
+ def _fetch_tracks_page(self, path, video_id, page_name, current_page, real_page_number=None):
+ real_page_number = real_page_number or current_page + 1
+ return self._download_webpage(
'https://www.mixcloud.com/%s/' % path, video_id,
note='Download %s (page %d)' % (page_name, current_page + 1),
errnote='Unable to download %s' % page_name,
- query={'page': (current_page + 1), 'list': 'main', '_ajax': '1'},
+ query={'page': real_page_number, 'list': 'main', '_ajax': '1'},
headers={'X-Requested-With': 'XMLHttpRequest'})
- for url in re.findall(r'm-play-button m-url="(?P<url>[^"]+)"', resp):
- yield self.url_result(
- compat_urlparse.urljoin('https://www.mixcloud.com', clean_html(url)),
- MixcloudIE.ie_key())
+ def _tracks_page_func(self, page, video_id, page_name, current_page):
+ resp = self._fetch_tracks_page(page, video_id, page_name, current_page)
+
+ for item in self._find_urls_in_page(resp):
+ yield item
def _get_user_description(self, page_content):
return self._html_search_regex(
entries = OnDemandPagedList(
functools.partial(
- self._fetch_tracks_page,
+ self._tracks_page_func,
'%s/%s' % (user_id, list_type), video_id, 'list of %s' % list_type),
self._PAGE_SIZE, use_cache=True)
entries = OnDemandPagedList(
functools.partial(
- self._fetch_tracks_page,
+ self._tracks_page_func,
'%s/playlists/%s' % (user_id, playlist_id), video_id, 'tracklist'),
self._PAGE_SIZE)
return self.playlist_result(entries, video_id, playlist_title, description)
+
+
+class MixcloudStreamIE(MixcloudPlaylistBaseIE):
+ _VALID_URL = r'^(?:https?://)?(?:www\.)?mixcloud\.com/(?P<id>[^/]+)/stream/?$'
+ IE_NAME = 'mixcloud:stream'
+
+ _TEST = {
+ 'url': 'https://www.mixcloud.com/FirstEar/stream/',
+ 'info_dict': {
+ 'id': 'FirstEar',
+ 'title': 'First Ear',
+ 'description': 'Curators of good music\nfirstearmusic.com',
+ },
+ 'playlist_mincount': 192,
+ }
+
+ def _real_extract(self, url):
+ user_id = self._match_id(url)
+
+ webpage = self._download_webpage(url, user_id)
+
+ entries = []
+ prev_page_url = None
+
+ def _handle_page(page):
+ entries.extend(self._find_urls_in_page(page))
+ return self._search_regex(
+ r'm-next-page-url="([^"]+)"', page,
+ 'next page URL', default=None)
+
+ next_page_url = _handle_page(webpage)
+
+ for idx in itertools.count(0):
+ if not next_page_url or prev_page_url == next_page_url:
+ break
+
+ prev_page_url = next_page_url
+ current_page = int(self._search_regex(
+ r'\?page=(\d+)', next_page_url, 'next page number'))
+
+ next_page_url = _handle_page(self._fetch_tracks_page(
+ '%s/stream' % user_id, user_id, 'stream', idx,
+ real_page_number=current_page))
+
+ username = self._og_search_title(webpage)
+ description = self._get_user_description(webpage)
+
+ return self.playlist_result(entries, user_id, username, description)