[twitch:stream] Prefer the 'source' format (fixes #4972)

[youtube-dl.git] / youtube_dl / extractor / twitch.py

diff --git a/youtube_dl/extractor/twitch.py b/youtube_dl/extractor/twitch.py
index 2891c4680cfd16d6bf7611a6d0f210f02e8c76d4..8e296698ebaedd250c4c5f035f190f87b2b3c049 100644 (file)
--- a/youtube_dl/extractor/twitch.py
+++ b/youtube_dl/extractor/twitch.py
@@ -3,6 +3,7 @@ from __future__ import unicode_literals
 
 import itertools
 import re
+import random
 
 from .common import InfoExtractor
 from ..compat import (
@@ -150,14 +151,17 @@ class TwitchChapterIE(TwitchItemBaseIE):
     _ITEM_TYPE = 'chapter'
     _ITEM_SHORTCUT = 'c'
 
-    _TEST = {
+    _TESTS = [{
         'url': 'http://www.twitch.tv/acracingleague/c/5285812',
         'info_dict': {
             'id': 'c5285812',
             'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
         },
         'playlist_mincount': 3,
-    }
+    }, {
+        'url': 'http://www.twitch.tv/tsm_theoddone/c/2349361',
+        'only_matching': True,
+    }]
 
 
 class TwitchVodIE(TwitchItemBaseIE):
@@ -216,12 +220,18 @@ class TwitchPlaylistBaseIE(TwitchBaseIE):
             response = self._download_json(
                 self._PLAYLIST_URL % (channel_id, offset, limit),
                 channel_id, 'Downloading %s videos JSON page %d' % (self._PLAYLIST_TYPE, counter))
-            videos = response['videos']
-            if not videos:
+            page_entries = self._extract_playlist_page(response)
+            if not page_entries:
                 break
-            entries.extend([self.url_result(video['url']) for video in videos])
+            entries.extend(page_entries)
             offset += limit
-        return self.playlist_result(entries, channel_id, channel_name)
+        return self.playlist_result(
+            [self.url_result(entry) for entry in set(entries)],
+            channel_id, channel_name)
+
+    def _extract_playlist_page(self, response):
+        videos = response.get('videos')
+        return [video['url'] for video in videos] if videos else []
 
     def _real_extract(self, url):
         return self._extract_playlist(self._match_id(url))
@@ -258,6 +268,31 @@ class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE):
     }
 
 
+class TwitchBookmarksIE(TwitchPlaylistBaseIE):
+    IE_NAME = 'twitch:bookmarks'
+    _VALID_URL = r'%s/(?P<id>[^/]+)/profile/bookmarks/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
+    _PLAYLIST_URL = '%s/api/bookmark/?user=%%s&offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE
+    _PLAYLIST_TYPE = 'bookmarks'
+
+    _TEST = {
+        'url': 'http://www.twitch.tv/ognos/profile/bookmarks',
+        'info_dict': {
+            'id': 'ognos',
+            'title': 'Ognos',
+        },
+        'playlist_mincount': 3,
+    }
+
+    def _extract_playlist_page(self, response):
+        entries = []
+        for bookmark in response.get('bookmarks', []):
+            video = bookmark.get('video')
+            if not video:
+                continue
+            entries.append(video['url'])
+        return entries
+
+
 class TwitchStreamIE(TwitchBaseIE):
     IE_NAME = 'twitch:stream'
     _VALID_URL = r'%s/(?P<id>[^/]+)/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
@@ -302,7 +337,7 @@ class TwitchStreamIE(TwitchBaseIE):
 
         query = {
             'allow_source': 'true',
-            'p': '9386337',
+            'p': random.randint(1000000, 10000000),
             'player': 'twitchweb',
             'segment_preference': '4',
             'sig': access_token['sig'],
@@ -313,6 +348,12 @@ class TwitchStreamIE(TwitchBaseIE):
             '%s/api/channel/hls/%s.m3u8?%s'
             % (self._USHER_BASE, channel_id, compat_urllib_parse.urlencode(query).encode('utf-8')),
             channel_id, 'mp4')
+        # prefer the 'source' stream, the others are limited to 30 fps
+        def _sort_source(f):
+            if f.get('m3u8_media') is not None and f['m3u8_media'].get('NAME') == 'Source':
+                return 1
+            return 0
+        formats = sorted(formats, key=_sort_source)
 
         view_count = stream.get('viewers')
         timestamp = parse_iso8601(stream.get('created_at'))
@@ -344,4 +385,4 @@ class TwitchStreamIE(TwitchBaseIE):
             'view_count': view_count,
             'formats': formats,
             'is_live': True,
-        }
\ No newline at end of file
+        }