X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Ftwitch.py;h=36ee1adff2288570fc39936640bacd3abafe9ed2;hb=d6712378e73951bede475569c887a1ac73f660a9;hp=891499a1f370d6b2484f75db07842afcc83431cc;hpb=fbd9f6ea804328d536aafd2b20a8afb72968e351;p=youtube-dl

diff --git a/youtube_dl/extractor/twitch.py b/youtube_dl/extractor/twitch.py
index 891499a1f..36ee1adff 100644
--- a/youtube_dl/extractor/twitch.py
+++ b/youtube_dl/extractor/twitch.py
@@ -9,17 +9,18 @@ from .common import InfoExtractor
 from ..compat import (
     compat_parse_qs,
     compat_str,
-    compat_urllib_parse,
+    compat_urllib_parse_urlencode,
     compat_urllib_parse_urlparse,
-    compat_urllib_request,
     compat_urlparse,
 )
 from ..utils import (
-    encode_dict,
     ExtractorError,
     int_or_none,
+    orderedSet,
     parse_duration,
     parse_iso8601,
+    sanitized_Request,
+    urlencode_postdata,
 )
 
 
@@ -48,7 +49,7 @@ class TwitchBaseIE(InfoExtractor):
         for cookie in self._downloader.cookiejar:
             if cookie.name == 'api_token':
                 headers['Twitch-Api-Token'] = cookie.value
-        request = compat_urllib_request.Request(url, headers=headers)
+        request = sanitized_Request(url, headers=headers)
         response = super(TwitchBaseIE, self)._download_json(request, video_id, note)
         self._handle_error(response)
         return response
@@ -80,8 +81,8 @@ class TwitchBaseIE(InfoExtractor):
         if not post_url.startswith('http'):
             post_url = compat_urlparse.urljoin(redirect_url, post_url)
 
-        request = compat_urllib_request.Request(
-            post_url, compat_urllib_parse.urlencode(encode_dict(login_form)).encode('utf-8'))
+        request = sanitized_Request(
+            post_url, urlencode_postdata(login_form))
         request.add_header('Referer', redirect_url)
         response = self._download_webpage(
             request, None, 'Logging in as %s' % username)
@@ -240,14 +241,25 @@ class TwitchVodIE(TwitchItemBaseIE):
 
     def _real_extract(self, url):
         item_id = self._match_id(url)
+
         info = self._download_info(self._ITEM_SHORTCUT, item_id)
         access_token = self._download_json(
             '%s/api/vods/%s/access_token' % (self._API_BASE, item_id), item_id,
             'Downloading %s access token' % self._ITEM_TYPE)
+
         formats = self._extract_m3u8_formats(
-            '%s/vod/%s?nauth=%s&nauthsig=%s&allow_source=true'
-            % (self._USHER_BASE, item_id, access_token['token'], access_token['sig']),
+            '%s/vod/%s?%s' % (
+                self._USHER_BASE, item_id,
+                compat_urllib_parse_urlencode({
+                    'allow_source': 'true',
+                    'allow_audio_only': 'true',
+                    'allow_spectre': 'true',
+                    'player': 'twitchweb',
+                    'nauth': access_token['token'],
+                    'nauthsig': access_token['sig'],
+                })),
             item_id, 'mp4')
+
         self._prefer_source(formats)
         info['formats'] = formats
 
@@ -271,17 +283,37 @@ class TwitchPlaylistBaseIE(TwitchBaseIE):
         entries = []
         offset = 0
         limit = self._PAGE_LIMIT
+        broken_paging_detected = False
+        counter_override = None
         for counter in itertools.count(1):
             response = self._download_json(
                 self._PLAYLIST_URL % (channel_id, offset, limit),
-                channel_id, 'Downloading %s videos JSON page %d' % (self._PLAYLIST_TYPE, counter))
+                channel_id,
+                'Downloading %s videos JSON page %s'
+                % (self._PLAYLIST_TYPE, counter_override or counter))
             page_entries = self._extract_playlist_page(response)
             if not page_entries:
                 break
+            total = int_or_none(response.get('_total'))
+            # Since the beginning of March 2016 twitch's paging mechanism
+            # is completely broken on the twitch side. It simply ignores
+            # a limit and returns the whole offset number of videos.
+            # Working around by just requesting all videos at once.
+            # Upd: pagination bug was fixed by twitch on 15.03.2016.
+            if not broken_paging_detected and total and len(page_entries) > limit:
+                self.report_warning(
+                    'Twitch pagination is broken on twitch side, requesting all videos at once',
+                    channel_id)
+                broken_paging_detected = True
+                offset = total
+                counter_override = '(all at once)'
+                continue
             entries.extend(page_entries)
+            if broken_paging_detected or total and len(page_entries) >= total:
+                break
             offset += limit
         return self.playlist_result(
-            [self.url_result(entry) for entry in set(entries)],
+            [self.url_result(entry) for entry in orderedSet(entries)],
             channel_id, channel_name)
 
     def _extract_playlist_page(self, response):
@@ -401,6 +433,7 @@ class TwitchStreamIE(TwitchBaseIE):
 
         query = {
             'allow_source': 'true',
+            'allow_audio_only': 'true',
             'p': random.randint(1000000, 10000000),
             'player': 'twitchweb',
             'segment_preference': '4',
@@ -409,7 +442,7 @@ class TwitchStreamIE(TwitchBaseIE):
         }
         formats = self._extract_m3u8_formats(
             '%s/api/channel/hls/%s.m3u8?%s'
-            % (self._USHER_BASE, channel_id, compat_urllib_parse.urlencode(query)),
+            % (self._USHER_BASE, channel_id, compat_urllib_parse_urlencode(query)),
             channel_id, 'mp4')
         self._prefer_source(formats)