Merge branch 'patch-1' of https://github.com/tuexss/youtube-dl into tuexss-patch-1

[youtube-dl] / youtube_dl / extractor / mixcloud.py

diff --git a/youtube_dl/extractor/mixcloud.py b/youtube_dl/extractor/mixcloud.py
index 2fe55179aac2b737cbf1fc68797fe48976c23270..84f29155841007f3088a86470040407073726067 100644 (file)
--- a/youtube_dl/extractor/mixcloud.py
+++ b/youtube_dl/extractor/mixcloud.py
@@ -1,6 +1,7 @@
 from __future__ import unicode_literals
 
 import re
+import itertools
 
 from .common import InfoExtractor
 from ..compat import (
@@ -45,20 +46,19 @@ class MixcloudIE(InfoExtractor):
         },
     }]
 
-    def _get_url(self, track_id, template_url):
-        server_count = 30
-        for i in range(server_count):
-            url = template_url % i
+    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 ...' % (i + 1, server_count + 1))
+                    'Checking URL %d/%d ...' % (nr, boundaries[-1]))
                 return url
             except ExtractorError:
                 pass
-
         return None
 
     def _real_extract(self, url):
@@ -72,12 +72,13 @@ class MixcloudIE(InfoExtractor):
         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)
+        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)
+            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')
 
@@ -96,7 +97,7 @@ class MixcloudIE(InfoExtractor):
             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="([^"]+)"',
+            r'\bbutton-favorite\b[^>]+m-ajax-toggle-count="([^"]+)"',
             webpage, 'like count', fatal=False))
         view_count = str_to_int(self._search_regex(
             [r'<meta itemprop="interactionCount" content="UserPlays:([0-9]+)"',
@@ -114,3 +115,35 @@ class MixcloudIE(InfoExtractor):
             'view_count': view_count,
             'like_count': like_count,
         }
+
+
+def server_numbers(first, boundaries):
+    """ Server numbers to try in descending order of probable availability.
+    Starting from first (i.e. the number of the server hosting the preview file)
+    and going further and further up to the higher boundary and down to the
+    lower one in an alternating fashion. Namely:
+
+        server_numbers(2, (1, 5))
+
+        # Where the preview server is 2, min number is 1 and max is 5.
+        # Yields: 2, 3, 1, 4, 5
+
+    Why not random numbers or increasing sequences? Since from what I've seen,
+    full length files seem to be hosted on servers whose number is closer to
+    that of the preview; to be confirmed.
+    """
+    zip_longest = getattr(itertools, 'zip_longest', None)
+    if zip_longest is None:
+        # python 2.x
+        zip_longest = itertools.izip_longest
+
+    if len(boundaries) != 2:
+        raise ValueError("boundaries should be a two-element tuple")
+    min, max = boundaries
+    highs = range(first + 1, max + 1)
+    lows = range(first - 1, min - 1, -1)
+    rest = filter(
+        None, itertools.chain.from_iterable(zip_longest(highs, lows)))
+    yield first
+    for n in rest:
+        yield n