[youtube] Standardize API calls for tabs, mixes and search (#245)

Authored by: colethedj
This commit is contained in:
coletdjnz 2021-04-15 11:22:59 +00:00 committed by GitHub
parent 46fff7105e
commit 79360d99d3
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23

View File

@ -3040,46 +3040,15 @@ def extract_entries(parent_renderer): # this needs to called again for continua
for page_num in itertools.count(1): for page_num in itertools.count(1):
if not continuation: if not continuation:
break break
query = {
'continuation': continuation['continuation'],
'clickTracking': {'clickTrackingParams': continuation['itct']}
}
headers = self._generate_api_headers(ytcfg, identity_token, account_syncid, visitor_data) headers = self._generate_api_headers(ytcfg, identity_token, account_syncid, visitor_data)
retries = self._downloader.params.get('extractor_retries', 3) response = self._extract_response(
count = -1 item_id='%s page %s' % (item_id, page_num),
last_error = None query=query, headers=headers, ytcfg=ytcfg,
while count < retries: check_get_keys=('continuationContents', 'onResponseReceivedActions', 'onResponseReceivedEndpoints'))
count += 1
if last_error:
self.report_warning('%s. Retrying ...' % last_error)
try:
response = self._call_api(
ep='browse', fatal=True, headers=headers,
video_id='%s page %s' % (item_id, page_num),
query={
'continuation': continuation['continuation'],
'clickTracking': {'clickTrackingParams': continuation['itct']},
},
context=context,
api_key=self._extract_api_key(ytcfg),
note='Downloading API JSON%s' % (' (retry #%d)' % count if count else ''))
except ExtractorError as e:
if isinstance(e.cause, compat_HTTPError) and e.cause.code in (500, 503, 404):
# Downloading page may result in intermittent 5xx HTTP error
# Sometimes a 404 is also recieved. See: https://github.com/ytdl-org/youtube-dl/issues/28289
last_error = 'HTTP Error %s' % e.cause.code
if count < retries:
continue
raise
else:
# Youtube sometimes sends incomplete data
# See: https://github.com/ytdl-org/youtube-dl/issues/28194
if dict_get(response,
('continuationContents', 'onResponseReceivedActions', 'onResponseReceivedEndpoints')):
break
# Youtube may send alerts if there was an issue with the continuation page
self._extract_alerts(response, expected=False)
last_error = 'Incomplete data received'
if count >= retries:
self._downloader.report_error(last_error)
if not response: if not response:
break break
@ -3236,8 +3205,13 @@ def _extract_from_tabs(self, item_id, webpage, data, tabs):
self._extract_ytcfg(item_id, webpage)), self._extract_ytcfg(item_id, webpage)),
**metadata) **metadata)
def _extract_mix_playlist(self, playlist, playlist_id): def _extract_mix_playlist(self, playlist, playlist_id, data, webpage):
first_id = last_id = None first_id = last_id = None
ytcfg = self._extract_ytcfg(playlist_id, webpage)
headers = self._generate_api_headers(
ytcfg, account_syncid=self._extract_account_syncid(data),
identity_token=self._extract_identity_token(webpage, item_id=playlist_id),
visitor_data=try_get(self._extract_context(ytcfg), lambda x: x['client']['visitorData'], compat_str))
for page_num in itertools.count(1): for page_num in itertools.count(1):
videos = list(self._playlist_entries(playlist)) videos = list(self._playlist_entries(playlist))
if not videos: if not videos:
@ -3252,14 +3226,25 @@ def _extract_mix_playlist(self, playlist, playlist_id):
yield video yield video
first_id = first_id or videos[0]['id'] first_id = first_id or videos[0]['id']
last_id = videos[-1]['id'] last_id = videos[-1]['id']
watch_endpoint = try_get(
_, data = self._extract_webpage( playlist, lambda x: x['contents'][-1]['playlistPanelVideoRenderer']['navigationEndpoint']['watchEndpoint'])
'https://www.youtube.com/watch?list=%s&v=%s' % (playlist_id, last_id), query = {
'%s page %d' % (playlist_id, page_num)) 'playlistId': playlist_id,
'videoId': watch_endpoint.get('videoId') or last_id,
'index': watch_endpoint.get('index') or len(videos),
'params': watch_endpoint.get('params') or 'OAE%3D'
}
response = self._extract_response(
item_id='%s page %d' % (playlist_id, page_num),
query=query,
ep='next',
headers=headers,
check_get_keys='contents'
)
playlist = try_get( playlist = try_get(
data, lambda x: x['contents']['twoColumnWatchNextResults']['playlist']['playlist'], dict) response, lambda x: x['contents']['twoColumnWatchNextResults']['playlist']['playlist'], dict)
def _extract_from_playlist(self, item_id, url, data, playlist): def _extract_from_playlist(self, item_id, url, data, playlist, webpage):
title = playlist.get('title') or try_get( title = playlist.get('title') or try_get(
data, lambda x: x['titleText']['simpleText'], compat_str) data, lambda x: x['titleText']['simpleText'], compat_str)
playlist_id = playlist.get('playlistId') or item_id playlist_id = playlist.get('playlistId') or item_id
@ -3274,7 +3259,7 @@ def _extract_from_playlist(self, item_id, url, data, playlist):
video_title=title) video_title=title)
return self.playlist_result( return self.playlist_result(
self._extract_mix_playlist(playlist, playlist_id), self._extract_mix_playlist(playlist, playlist_id, data, webpage),
playlist_id=playlist_id, playlist_title=title) playlist_id=playlist_id, playlist_title=title)
def _extract_alerts(self, data, expected=False): def _extract_alerts(self, data, expected=False):
@ -3308,6 +3293,46 @@ def _real_extract_alerts():
if errors: if errors:
raise ExtractorError('YouTube said: %s' % errors[-1][1], expected=expected) raise ExtractorError('YouTube said: %s' % errors[-1][1], expected=expected)
def _extract_response(self, item_id, query, note='Downloading API JSON', headers=None,
ytcfg=None, check_get_keys=None, ep='browse'):
response = None
last_error = None
count = -1
retries = self._downloader.params.get('extractor_retries', 3)
if check_get_keys is None:
check_get_keys = []
while count < retries:
count += 1
if last_error:
self.report_warning('%s. Retrying ...' % last_error)
try:
response = self._call_api(
ep=ep, fatal=True, headers=headers,
video_id=item_id,
query=query,
context=self._extract_context(ytcfg),
api_key=self._extract_api_key(ytcfg),
note='%s%s' % (note, ' (retry #%d)' % count if count else ''))
except ExtractorError as e:
if isinstance(e.cause, compat_HTTPError) and e.cause.code in (500, 503, 404):
# Downloading page may result in intermittent 5xx HTTP error
# Sometimes a 404 is also recieved. See: https://github.com/ytdl-org/youtube-dl/issues/28289
last_error = 'HTTP Error %s' % e.cause.code
if count < retries:
continue
raise
else:
# Youtube may send alerts if there was an issue with the continuation page
self._extract_alerts(response, expected=False)
if not check_get_keys or dict_get(response, check_get_keys):
break
# Youtube sometimes sends incomplete data
# See: https://github.com/ytdl-org/youtube-dl/issues/28194
last_error = 'Incomplete data received'
if count >= retries:
self._downloader.report_error(last_error)
return response
def _extract_webpage(self, url, item_id): def _extract_webpage(self, url, item_id):
retries = self._downloader.params.get('extractor_retries', 3) retries = self._downloader.params.get('extractor_retries', 3)
count = -1 count = -1
@ -3372,7 +3397,7 @@ def _real_extract(self, url):
playlist = try_get( playlist = try_get(
data, lambda x: x['contents']['twoColumnWatchNextResults']['playlist']['playlist'], dict) data, lambda x: x['contents']['twoColumnWatchNextResults']['playlist']['playlist'], dict)
if playlist: if playlist:
return self._extract_from_playlist(item_id, url, data, playlist) return self._extract_from_playlist(item_id, url, data, playlist, webpage)
video_id = try_get( video_id = try_get(
data, lambda x: x['currentVideoEndpoint']['watchEndpoint']['videoId'], data, lambda x: x['currentVideoEndpoint']['watchEndpoint']['videoId'],
@ -3534,7 +3559,7 @@ def _real_extract(self, url):
ie=YoutubeTabIE.ie_key()) ie=YoutubeTabIE.ie_key())
class YoutubeSearchIE(SearchInfoExtractor, YoutubeBaseInfoExtractor): class YoutubeSearchIE(SearchInfoExtractor, YoutubeTabIE):
IE_DESC = 'YouTube.com searches, "ytsearch" keyword' IE_DESC = 'YouTube.com searches, "ytsearch" keyword'
# there doesn't appear to be a real limit, for example if you search for # there doesn't appear to be a real limit, for example if you search for
# 'python' you get more than 8.000.000 results # 'python' you get more than 8.000.000 results
@ -3550,9 +3575,10 @@ def _entries(self, query, n):
data['params'] = self._SEARCH_PARAMS data['params'] = self._SEARCH_PARAMS
total = 0 total = 0
for page_num in itertools.count(1): for page_num in itertools.count(1):
search = self._call_api( search = self._extract_response(
ep='search', video_id='query "%s"' % query, fatal=False, item_id='query "%s" page %s' % (query, page_num), ep='search', query=data,
note='Downloading page %s' % page_num, query=data) check_get_keys=('contents', 'onResponseReceivedCommands')
)
if not search: if not search:
break break
slr_contents = try_get( slr_contents = try_get(