X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;ds=inline;f=youtube_dl%2Fextractor%2Fmixcloud.py;h=9638cc9e6d09a62ba6b8de278a6dc8e53dbabd51;hb=b9f2fdd37fe2085deb09710a2084c940e9920304;hp=84f29155841007f3088a86470040407073726067;hpb=ae0dd4b2988efc6252121ca2824ca8b3a8a18fba;p=youtube-dl
diff --git a/youtube_dl/extractor/mixcloud.py b/youtube_dl/extractor/mixcloud.py
index 84f291558..9638cc9e6 100644
--- a/youtube_dl/extractor/mixcloud.py
+++ b/youtube_dl/extractor/mixcloud.py
@@ -1,15 +1,14 @@
from __future__ import unicode_literals
import re
-import itertools
from .common import InfoExtractor
-from ..compat import (
- compat_urllib_parse,
-)
+from ..compat import compat_urllib_parse_unquote
from ..utils import (
ExtractorError,
HEADRequest,
+ NO_DEFAULT,
+ parse_count,
str_to_int,
)
@@ -46,41 +45,42 @@ class MixcloudIE(InfoExtractor):
},
}]
- def _get_url(self, track_id, template_url, server_number):
- boundaries = (1, 30)
- for nr in server_numbers(server_number, boundaries):
- url = template_url % nr
- try:
- # We only want to know if the request succeed
- # don't download the whole file
- self._request_webpage(
- HEADRequest(url), track_id,
- 'Checking URL %d/%d ...' % (nr, boundaries[-1]))
- return url
- except ExtractorError:
- pass
- return None
+ 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
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
uploader = mobj.group(1)
cloudcast_name = mobj.group(2)
- track_id = compat_urllib_parse.unquote('-'.join((uploader, cloudcast_name)))
+ track_id = compat_urllib_parse_unquote('-'.join((uploader, cloudcast_name)))
webpage = self._download_webpage(url, track_id)
+ message = self._html_search_regex(
+ r'(?s)
]+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')
- song_url = preview_url.replace('/previews/', '/c/originals/')
- server_number = int(self._search_regex(r'stream(\d+)', song_url, 'server number'))
- template_url = re.sub(r'(stream\d*)', 'stream%d', song_url)
- final_song_url = self._get_url(track_id, template_url, server_number)
- if final_song_url is None:
- self.to_screen('Trying with m4a extension')
- template_url = template_url.replace('.mp3', '.m4a').replace('originals/', 'm4a/64/')
- final_song_url = self._get_url(track_id, template_url, server_number)
- if final_song_url is None:
- raise ExtractorError('Unable to extract track url')
+ r'\s(?:data-preview-url|m-preview)="([^"]+)"',
+ webpage, 'preview url', default=None if message else NO_DEFAULT)
+
+ if message:
+ 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')
PREFIX = (
r'm-play-on-spacebar[^>]+'
@@ -96,8 +96,8 @@ class MixcloudIE(InfoExtractor):
uploader_id = self._search_regex(
r'\s+"profile": "([^"]+)",', webpage, 'uploader id', fatal=False)
description = self._og_search_description(webpage)
- like_count = str_to_int(self._search_regex(
- r'\bbutton-favorite\b[^>]+m-ajax-toggle-count="([^"]+)"',
+ like_count = parse_count(self._search_regex(
+ r'\bbutton-favorite[^>]+>.*?]+class=["\']toggle-number[^>]+>\s*([^<]+)',
webpage, 'like count', fatal=False))
view_count = str_to_int(self._search_regex(
[r'