depau/youtube-dl
1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-12-22 16:57:40 +00:00
Code Issues Releases Wiki Activity

[twitch] Workaround broken paging (Closes #8740)

Browse source
This commit is contained in:
Sergey M․ 2016-03-03 22:10:36 +06:00
parent 1b0635aba3
commit 539a1641c6
1 changed files with 22 additions and 5 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

27
youtube_dl/extractor/twitch.py
View file

@ -271,7 +271,7 @@ class TwitchVodIE(TwitchItemBaseIE):
class TwitchPlaylistBaseIE(TwitchBaseIE): class TwitchPlaylistBaseIE(TwitchBaseIE):
_PLAYLIST_URL = '%s/kraken/channels/%%s/videos/?offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE _PLAYLIST_URL = '%s/kraken/channels/%%s/videos/?offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE
_PAGE_LIMIT = 100 _PAGE_LIMIT = 10
def _extract_playlist(self, channel_id): def _extract_playlist(self, channel_id):
info = self._download_json( info = self._download_json(
@ -281,14 +281,34 @@ class TwitchPlaylistBaseIE(TwitchBaseIE):
entries = [] entries = []
offset = 0 offset = 0
limit = self._PAGE_LIMIT limit = self._PAGE_LIMIT
broken_paging_detected = False
counter_override = None
for counter in itertools.count(1): for counter in itertools.count(1):
response = self._download_json( response = self._download_json(
self._PLAYLIST_URL % (channel_id, offset, limit), 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) page_entries = self._extract_playlist_page(response)
if not page_entries: if not page_entries:
break 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.
if not broken_paging_detected and total and len(page_entries) > limit:
self.report_warning(
'Twitch paging is broken on twitch side, requesting all videos at once',
channel_id)
broken_paging_detected = True
limit = total
offset = 0
counter_override = '(all at once)'
continue
entries.extend(page_entries) entries.extend(page_entries)
if broken_paging_detected or total and len(page_entries) >= total:
break
offset += limit offset += limit
return self.playlist_result( return self.playlist_result(
[self.url_result(entry) for entry in set(entries)], [self.url_result(entry) for entry in set(entries)],
@ -303,7 +323,6 @@ class TwitchPlaylistBaseIE(TwitchBaseIE):
class TwitchProfileIE(TwitchPlaylistBaseIE): class TwitchProfileIE(TwitchPlaylistBaseIE):
_WORKING = False
IE_NAME = 'twitch:profile' IE_NAME = 'twitch:profile'
_VALID_URL = r'%s/(?P<id>[^/]+)/profile/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE _VALID_URL = r'%s/(?P<id>[^/]+)/profile/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
_PLAYLIST_TYPE = 'profile' _PLAYLIST_TYPE = 'profile'
@ -319,7 +338,6 @@ class TwitchProfileIE(TwitchPlaylistBaseIE):
class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE): class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE):
_WORKING = False
IE_NAME = 'twitch:past_broadcasts' IE_NAME = 'twitch:past_broadcasts'
_VALID_URL = r'%s/(?P<id>[^/]+)/profile/past_broadcasts/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE _VALID_URL = r'%s/(?P<id>[^/]+)/profile/past_broadcasts/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
_PLAYLIST_URL = TwitchPlaylistBaseIE._PLAYLIST_URL + '&broadcasts=true' _PLAYLIST_URL = TwitchPlaylistBaseIE._PLAYLIST_URL + '&broadcasts=true'
@ -336,7 +354,6 @@ class TwitchPastBroadcastsIE(TwitchPlaylistBaseIE):
class TwitchBookmarksIE(TwitchPlaylistBaseIE): class TwitchBookmarksIE(TwitchPlaylistBaseIE):
_WORKING = False
IE_NAME = 'twitch:bookmarks' IE_NAME = 'twitch:bookmarks'
_VALID_URL = r'%s/(?P<id>[^/]+)/profile/bookmarks/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE _VALID_URL = r'%s/(?P<id>[^/]+)/profile/bookmarks/?(?:\#.*)?$' % TwitchBaseIE._VALID_URL_BASE
_PLAYLIST_URL = '%s/api/bookmark/?user=%%s&offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE _PLAYLIST_URL = '%s/api/bookmark/?user=%%s&offset=%%d&limit=%%d' % TwitchBaseIE._API_BASE