[youtube:comments] Move comment extraction to new API (#466)

Closes #438, #481, #485 

Authored by: colethedj
This commit is contained in:
coletdjnz 2021-07-12 11:18:40 +12:00 committed by GitHub
parent c5370857b3
commit 2d6659b9ea
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 247 additions and 222 deletions

View File

@ -1335,6 +1335,9 @@ The following extractors use this feature:
* `skip`: `hls` or `dash` (or both) to skip download of the respective manifests * `skip`: `hls` or `dash` (or both) to skip download of the respective manifests
* `player_client`: `web` (default) or `android` (force use the android client fallbacks for video extraction) * `player_client`: `web` (default) or `android` (force use the android client fallbacks for video extraction)
* `player_skip`: `configs` - skip requests if applicable for client configs and use defaults * `player_skip`: `configs` - skip requests if applicable for client configs and use defaults
* `comment_sort`: `top` or `new` (default) - choose comment sorting mode (on YouTube's side).
* `max_comments`: maximum amount of comments to download (default all).
* `max_comment_depth`: maximum depth for nested comments. YouTube supports depths 1 or 2 (default).
* **funimation** * **funimation**
* `language`: Languages to extract. Eg: `funimation:language=english,japanese` * `language`: Languages to extract. Eg: `funimation:language=english,japanese`

View File

@ -2,6 +2,7 @@
from __future__ import unicode_literals from __future__ import unicode_literals
import base64
import calendar import calendar
import copy import copy
import hashlib import hashlib
@ -27,6 +28,7 @@ from ..compat import (
from ..jsinterp import JSInterpreter from ..jsinterp import JSInterpreter
from ..utils import ( from ..utils import (
bool_or_none, bool_or_none,
bytes_to_intlist,
clean_html, clean_html,
dict_get, dict_get,
datetime_from_str, datetime_from_str,
@ -35,6 +37,7 @@ from ..utils import (
format_field, format_field,
float_or_none, float_or_none,
int_or_none, int_or_none,
intlist_to_bytes,
mimetype2ext, mimetype2ext,
parse_codecs, parse_codecs,
parse_duration, parse_duration,
@ -98,8 +101,9 @@ class YoutubeBaseInfoExtractor(InfoExtractor):
if username: if username:
warn('Logging in using username and password is broken. %s' % self._LOGIN_HINTS['cookies']) warn('Logging in using username and password is broken. %s' % self._LOGIN_HINTS['cookies'])
return return
# Everything below this is broken!
# Everything below this is broken!
r'''
# No authentication to be performed # No authentication to be performed
if username is None: if username is None:
if self._LOGIN_REQUIRED and self.get_param('cookiefile') is None: if self._LOGIN_REQUIRED and self.get_param('cookiefile') is None:
@ -272,6 +276,7 @@ class YoutubeBaseInfoExtractor(InfoExtractor):
return False return False
return True return True
'''
def _initialize_consent(self): def _initialize_consent(self):
cookies = self._get_cookies('https://www.youtube.com/') cookies = self._get_cookies('https://www.youtube.com/')
@ -520,6 +525,9 @@ class YoutubeBaseInfoExtractor(InfoExtractor):
'X-YouTube-Client-Version': self._extract_client_version(ytcfg, client), 'X-YouTube-Client-Version': self._extract_client_version(ytcfg, client),
'Origin': origin 'Origin': origin
} }
if not visitor_data and ytcfg:
visitor_data = try_get(
self._extract_context(ytcfg, client), lambda x: x['client']['visitorData'], compat_str)
if identity_token: if identity_token:
headers['X-Youtube-Identity-Token'] = identity_token headers['X-Youtube-Identity-Token'] = identity_token
if account_syncid: if account_syncid:
@ -533,6 +541,75 @@ class YoutubeBaseInfoExtractor(InfoExtractor):
headers['X-Origin'] = origin headers['X-Origin'] = origin
return headers return headers
@staticmethod
def _build_api_continuation_query(continuation, ctp=None):
query = {
'continuation': continuation
}
# TODO: Inconsistency with clickTrackingParams.
# Currently we have a fixed ctp contained within context (from ytcfg)
# and a ctp in root query for continuation.
if ctp:
query['clickTracking'] = {'clickTrackingParams': ctp}
return query
@classmethod
def _continuation_query_ajax_to_api(cls, continuation_query):
continuation = dict_get(continuation_query, ('continuation', 'ctoken'))
return cls._build_api_continuation_query(continuation, continuation_query.get('itct'))
@staticmethod
def _build_continuation_query(continuation, ctp=None):
query = {
'ctoken': continuation,
'continuation': continuation,
}
if ctp:
query['itct'] = ctp
return query
@classmethod
def _extract_next_continuation_data(cls, renderer):
next_continuation = try_get(
renderer, (lambda x: x['continuations'][0]['nextContinuationData'],
lambda x: x['continuation']['reloadContinuationData']), dict)
if not next_continuation:
return
continuation = next_continuation.get('continuation')
if not continuation:
return
ctp = next_continuation.get('clickTrackingParams')
return cls._build_continuation_query(continuation, ctp)
@classmethod
def _extract_continuation_ep_data(cls, continuation_ep: dict):
if isinstance(continuation_ep, dict):
continuation = try_get(
continuation_ep, lambda x: x['continuationCommand']['token'], compat_str)
if not continuation:
return
ctp = continuation_ep.get('clickTrackingParams')
return cls._build_continuation_query(continuation, ctp)
@classmethod
def _extract_continuation(cls, renderer):
next_continuation = cls._extract_next_continuation_data(renderer)
if next_continuation:
return next_continuation
contents = []
for key in ('contents', 'items'):
contents.extend(try_get(renderer, lambda x: x[key], list) or [])
for content in contents:
if not isinstance(content, dict):
continue
continuation_ep = try_get(
content, (lambda x: x['continuationItemRenderer']['continuationEndpoint'],
lambda x: x['continuationItemRenderer']['button']['buttonRenderer']['command']),
dict)
continuation = cls._extract_continuation_ep_data(continuation_ep)
if continuation:
return continuation
@staticmethod @staticmethod
def _extract_alerts(data): def _extract_alerts(data):
for alert_dict in try_get(data, lambda x: x['alerts'], list) or []: for alert_dict in try_get(data, lambda x: x['alerts'], list) or []:
@ -1941,10 +2018,43 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
} }
def _comment_entries(self, root_continuation_data, identity_token, account_syncid, def _comment_entries(self, root_continuation_data, identity_token, account_syncid,
ytcfg, session_token_list, parent=None, comment_counts=None): ytcfg, video_id, parent=None, comment_counts=None):
def extract_thread(parent_renderer): def extract_header(contents):
contents = try_get(parent_renderer, lambda x: x['contents'], list) or [] _total_comments = 0
_continuation = None
for content in contents:
comments_header_renderer = try_get(content, lambda x: x['commentsHeaderRenderer'])
expected_comment_count = try_get(comments_header_renderer,
(lambda x: x['countText']['runs'][0]['text'],
lambda x: x['commentsCount']['runs'][0]['text']),
compat_str)
if expected_comment_count:
comment_counts[1] = str_to_int(expected_comment_count)
self.to_screen('Downloading ~%d comments' % str_to_int(expected_comment_count))
_total_comments = comment_counts[1]
sort_mode_str = self._configuration_arg('comment_sort', [''])[0]
comment_sort_index = int(sort_mode_str != 'top') # 1 = new, 0 = top
sort_menu_item = try_get(
comments_header_renderer,
lambda x: x['sortMenu']['sortFilterSubMenuRenderer']['subMenuItems'][comment_sort_index], dict) or {}
sort_continuation_ep = sort_menu_item.get('serviceEndpoint') or {}
_continuation = self._extract_continuation_ep_data(sort_continuation_ep) or self._extract_continuation(sort_menu_item)
if not _continuation:
continue
sort_text = sort_menu_item.get('title')
if isinstance(sort_text, compat_str):
sort_text = sort_text.lower()
else:
sort_text = 'top comments' if comment_sort_index == 0 else 'newest first'
self.to_screen('Sorting comments by %s' % sort_text)
break
return _total_comments, _continuation
def extract_thread(contents):
if not parent: if not parent:
comment_counts[2] = 0 comment_counts[2] = 0
for content in contents: for content in contents:
@ -1968,117 +2078,48 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
comment_counts[2] += 1 comment_counts[2] += 1
comment_entries_iter = self._comment_entries( comment_entries_iter = self._comment_entries(
comment_replies_renderer, identity_token, account_syncid, ytcfg, comment_replies_renderer, identity_token, account_syncid, ytcfg,
parent=comment.get('id'), session_token_list=session_token_list, video_id, parent=comment.get('id'), comment_counts=comment_counts)
comment_counts=comment_counts)
for reply_comment in comment_entries_iter: for reply_comment in comment_entries_iter:
yield reply_comment yield reply_comment
# YouTube comments have a max depth of 2
max_depth = int_or_none(self._configuration_arg('max_comment_depth', [''])[0]) or float('inf')
if max_depth == 1 and parent:
return
if not comment_counts: if not comment_counts:
# comment so far, est. total comments, current comment thread # # comment so far, est. total comments, current comment thread #
comment_counts = [0, 0, 0] comment_counts = [0, 0, 0]
# TODO: Generalize the download code with TabIE continuation = self._extract_continuation(root_continuation_data)
context = self._extract_context(ytcfg) if continuation and len(continuation['ctoken']) < 27:
visitor_data = try_get(context, lambda x: x['client']['visitorData'], compat_str) self.write_debug('Detected old API continuation token. Generating new API compatible token.')
continuation = YoutubeTabIE._extract_continuation(root_continuation_data) # TODO continuation_token = self._generate_comment_continuation(video_id)
first_continuation = False continuation = self._build_continuation_query(continuation_token, None)
if parent is None:
first_continuation = True visitor_data = None
is_first_continuation = parent is None
for page_num in itertools.count(0): for page_num in itertools.count(0):
if not continuation: if not continuation:
break break
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.get_param('extractor_retries', 3)
count = -1
last_error = None
while count < retries:
count += 1
if last_error:
self.report_warning('%s. Retrying ...' % last_error)
try:
query = {
'ctoken': continuation['ctoken'],
'pbj': 1,
'type': 'next',
}
if 'itct' in continuation:
query['itct'] = continuation['itct']
if parent:
query['action_get_comment_replies'] = 1
else:
query['action_get_comments'] = 1
comment_prog_str = '(%d/%d)' % (comment_counts[0], comment_counts[1]) comment_prog_str = '(%d/%d)' % (comment_counts[0], comment_counts[1])
if page_num == 0: if page_num == 0:
if first_continuation: if is_first_continuation:
note_prefix = 'Downloading initial comment continuation page' note_prefix = 'Downloading comment section API JSON'
else: else:
note_prefix = ' Downloading comment reply thread %d %s' % (comment_counts[2], comment_prog_str) note_prefix = ' Downloading comment API JSON reply thread %d %s' % (
comment_counts[2], comment_prog_str)
else: else:
note_prefix = '%sDownloading comment%s page %d %s' % ( note_prefix = '%sDownloading comment%s API JSON page %d %s' % (
' ' if parent else '', ' ' if parent else '', ' replies' if parent else '',
' replies' if parent else '', page_num, comment_prog_str)
page_num,
comment_prog_str)
browse = self._download_json(
'https://www.youtube.com/comment_service_ajax', None,
'%s %s' % (note_prefix, '(retry #%d)' % count if count else ''),
headers=headers, query=query,
data=urlencode_postdata({
'session_token': session_token_list[0]
}))
except ExtractorError as e:
if isinstance(e.cause, compat_HTTPError) and e.cause.code in (500, 503, 404, 413):
if e.cause.code == 413:
self.report_warning('Assumed end of comments (received HTTP Error 413)')
return
# 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 e.cause.code == 404:
last_error = last_error + ' (this API is probably deprecated)'
if count < retries:
continue
raise
else:
session_token = try_get(browse, lambda x: x['xsrf_token'], compat_str)
if session_token:
session_token_list[0] = session_token
response = try_get(browse,
(lambda x: x['response'],
lambda x: x[1]['response']), dict) or {}
if response.get('continuationContents'):
break
# YouTube sometimes gives reload: now json if something went wrong (e.g. bad auth)
if isinstance(browse, dict):
if browse.get('reload'):
raise ExtractorError('Invalid or missing params in continuation request', expected=False)
# TODO: not tested, merged from old extractor
err_msg = browse.get('externalErrorMessage')
if err_msg:
last_error = err_msg
continue
response_error = try_get(response, lambda x: x['responseContext']['errors']['error'][0], dict) or {}
err_msg = response_error.get('externalErrorMessage')
if err_msg:
last_error = err_msg
continue
# Youtube sometimes sends incomplete data
# See: https://github.com/ytdl-org/youtube-dl/issues/28194
last_error = 'Incomplete data received'
if count >= retries:
raise ExtractorError(last_error)
response = self._extract_response(
item_id=None, query=self._continuation_query_ajax_to_api(continuation),
ep='next', ytcfg=ytcfg, headers=headers, note=note_prefix,
check_get_keys=('onResponseReceivedEndpoints', 'continuationContents'))
if not response: if not response:
break break
visitor_data = try_get( visitor_data = try_get(
@ -2086,80 +2127,107 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
lambda x: x['responseContext']['webResponseContextExtensionData']['ytConfigData']['visitorData'], lambda x: x['responseContext']['webResponseContextExtensionData']['ytConfigData']['visitorData'],
compat_str) or visitor_data compat_str) or visitor_data
known_continuation_renderers = { continuation_contents = dict_get(response, ('onResponseReceivedEndpoints', 'continuationContents'))
'itemSectionContinuation': extract_thread,
'commentRepliesContinuation': extract_thread
}
# extract next root continuation from the results continuation = None
continuation_contents = try_get( if isinstance(continuation_contents, list):
response, lambda x: x['continuationContents'], dict) or {} for continuation_section in continuation_contents:
if not isinstance(continuation_section, dict):
continue
continuation_items = try_get(
continuation_section,
(lambda x: x['reloadContinuationItemsCommand']['continuationItems'],
lambda x: x['appendContinuationItemsAction']['continuationItems']),
list) or []
if is_first_continuation:
total_comments, continuation = extract_header(continuation_items)
if total_comments:
yield total_comments
is_first_continuation = False
if continuation:
break
continue
count = 0
for count, entry in enumerate(extract_thread(continuation_items)):
yield entry
continuation = self._extract_continuation({'contents': continuation_items})
if continuation:
# Sometimes YouTube provides a continuation without any comments
# In most cases we end up just downloading these with very little comments to come.
if count == 0:
if not parent:
self.report_warning('No comments received - assuming end of comments')
continuation = None
break
for key, value in continuation_contents.items(): # Deprecated response structure
elif isinstance(continuation_contents, dict):
known_continuation_renderers = ('itemSectionContinuation', 'commentRepliesContinuation')
for key, continuation_renderer in continuation_contents.items():
if key not in known_continuation_renderers: if key not in known_continuation_renderers:
continue continue
continuation_renderer = value if not isinstance(continuation_renderer, dict):
continue
if first_continuation: if is_first_continuation:
first_continuation = False header_continuation_items = [continuation_renderer.get('header') or {}]
expected_comment_count = try_get( total_comments, continuation = extract_header(header_continuation_items)
continuation_renderer, if total_comments:
(lambda x: x['header']['commentsHeaderRenderer']['countText']['runs'][0]['text'], yield total_comments
lambda x: x['header']['commentsHeaderRenderer']['commentsCount']['runs'][0]['text']), is_first_continuation = False
compat_str) if continuation:
if expected_comment_count:
comment_counts[1] = str_to_int(expected_comment_count)
self.to_screen('Downloading ~%d comments' % str_to_int(expected_comment_count))
yield comment_counts[1]
# TODO: cli arg.
# 1/True for newest, 0/False for popular (default)
comment_sort_index = int(True)
sort_continuation_renderer = try_get(
continuation_renderer,
lambda x: x['header']['commentsHeaderRenderer']['sortMenu']['sortFilterSubMenuRenderer']['subMenuItems']
[comment_sort_index]['continuation']['reloadContinuationData'], dict)
# If this fails, the initial continuation page
# starts off with popular anyways.
if sort_continuation_renderer:
continuation = YoutubeTabIE._build_continuation_query(
continuation=sort_continuation_renderer.get('continuation'),
ctp=sort_continuation_renderer.get('clickTrackingParams'))
self.to_screen('Sorting comments by %s' % ('popular' if comment_sort_index == 0 else 'newest'))
break break
for entry in known_continuation_renderers[key](continuation_renderer): # Sometimes YouTube provides a continuation without any comments
# In most cases we end up just downloading these with very little comments to come.
count = 0
for count, entry in enumerate(extract_thread(continuation_renderer.get('contents') or {})):
yield entry yield entry
continuation = self._extract_continuation(continuation_renderer)
continuation = YoutubeTabIE._extract_continuation(continuation_renderer) # TODO if count == 0:
if not parent:
self.report_warning('No comments received - assuming end of comments')
continuation = None
break break
def _extract_comments(self, ytcfg, video_id, contents, webpage, xsrf_token): @staticmethod
def _generate_comment_continuation(video_id):
"""
Generates initial comment section continuation token from given video id
"""
b64_vid_id = base64.b64encode(bytes(video_id.encode('utf-8')))
parts = ('Eg0SCw==', b64_vid_id, 'GAYyJyIRIgs=', b64_vid_id, 'MAB4AjAAQhBjb21tZW50cy1zZWN0aW9u')
new_continuation_intlist = list(itertools.chain.from_iterable(
[bytes_to_intlist(base64.b64decode(part)) for part in parts]))
return base64.b64encode(intlist_to_bytes(new_continuation_intlist)).decode('utf-8')
def _extract_comments(self, ytcfg, video_id, contents, webpage):
"""Entry for comment extraction""" """Entry for comment extraction"""
comments = [] def _real_comment_extract(contents):
known_entry_comment_renderers = ( if isinstance(contents, list):
'itemSectionRenderer',
)
estimated_total = 0
for entry in contents: for entry in contents:
for key, renderer in entry.items(): for key, renderer in entry.items():
if key not in known_entry_comment_renderers: if key not in known_entry_comment_renderers:
continue continue
yield from self._comment_entries(
comment_iter = self._comment_entries( renderer, video_id=video_id, ytcfg=ytcfg,
renderer,
identity_token=self._extract_identity_token(webpage, item_id=video_id), identity_token=self._extract_identity_token(webpage, item_id=video_id),
account_syncid=self._extract_account_syncid(ytcfg), account_syncid=self._extract_account_syncid(ytcfg))
ytcfg=ytcfg, break
session_token_list=[xsrf_token]) comments = []
known_entry_comment_renderers = ('itemSectionRenderer',)
estimated_total = 0
max_comments = int_or_none(self._configuration_arg('max_comments', [''])[0]) or float('inf')
for comment in comment_iter: try:
for comment in _real_comment_extract(contents):
if len(comments) >= max_comments:
break
if isinstance(comment, int): if isinstance(comment, int):
estimated_total = comment estimated_total = comment
continue continue
comments.append(comment) comments.append(comment)
break except KeyboardInterrupt:
self.to_screen('Interrupted by user')
self.to_screen('Downloaded %d/%d comments' % (len(comments), estimated_total)) self.to_screen('Downloaded %d/%d comments' % (len(comments), estimated_total))
return { return {
'comments': comments, 'comments': comments,
@ -2207,7 +2275,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
player_url = self._extract_player_url(ytcfg, webpage) player_url = self._extract_player_url(ytcfg, webpage)
player_client = (self._configuration_arg('player_client') or [''])[0] player_client = self._configuration_arg('player_client', [''])[0]
if player_client not in ('web', 'android', ''): if player_client not in ('web', 'android', ''):
self.report_warning(f'Invalid player_client {player_client} given. Falling back to WEB') self.report_warning(f'Invalid player_client {player_client} given. Falling back to WEB')
force_mobile_client = player_client == 'android' force_mobile_client = player_client == 'android'
@ -2231,7 +2299,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
if sts and not force_mobile_client and 'configs' not in player_skip: if sts and not force_mobile_client and 'configs' not in player_skip:
ytm_webpage = self._download_webpage( ytm_webpage = self._download_webpage(
'https://music.youtube.com', 'https://music.youtube.com',
video_id, fatal=False, note="Downloading remix client config") video_id, fatal=False, note='Downloading remix client config')
ytm_cfg = self._extract_ytcfg(video_id, ytm_webpage) or {} ytm_cfg = self._extract_ytcfg(video_id, ytm_webpage) or {}
ytm_client = 'WEB_REMIX' ytm_client = 'WEB_REMIX'
@ -2254,8 +2322,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
ytcfg=ytm_cfg, headers=ytm_headers, fatal=False, ytcfg=ytm_cfg, headers=ytm_headers, fatal=False,
default_client=ytm_client, default_client=ytm_client,
note='Downloading %sremix player API JSON' % ('mobile ' if force_mobile_client else '')) note='Downloading %sremix player API JSON' % ('mobile ' if force_mobile_client else ''))
ytm_streaming_data = try_get(ytm_player_response, lambda x: x['streamingData'], dict) or {}
ytm_streaming_data = try_get(ytm_player_response, lambda x: x['streamingData']) or {}
player_response = None player_response = None
if webpage: if webpage:
player_response = self._extract_yt_initial_variable( player_response = self._extract_yt_initial_variable(
@ -2672,7 +2740,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
continue continue
process_language( process_language(
subtitles, base_url, lang_code, subtitles, base_url, lang_code,
try_get(caption_track, lambda x: x.get('name').get('simpleText')), try_get(caption_track, lambda x: x['name']['simpleText']),
{}) {})
continue continue
automatic_captions = {} automatic_captions = {}
@ -2928,7 +2996,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
data=urlencode_postdata({xsrf_field_name: xsrf_token})) data=urlencode_postdata({xsrf_field_name: xsrf_token}))
if get_comments: if get_comments:
info['__post_extractor'] = lambda: self._extract_comments(ytcfg, video_id, contents, webpage, xsrf_token) info['__post_extractor'] = lambda: self._extract_comments(ytcfg, video_id, contents, webpage)
self.mark_watched(video_id, player_response) self.mark_watched(video_id, player_response)
@ -3553,52 +3621,6 @@ class YoutubeTabIE(YoutubeBaseInfoExtractor):
if entry: if entry:
yield entry yield entry
''' '''
@staticmethod
def _build_continuation_query(continuation, ctp=None):
query = {
'ctoken': continuation,
'continuation': continuation,
}
if ctp:
query['itct'] = ctp
return query
@staticmethod
def _extract_next_continuation_data(renderer):
next_continuation = try_get(
renderer, lambda x: x['continuations'][0]['nextContinuationData'], dict)
if not next_continuation:
return
continuation = next_continuation.get('continuation')
if not continuation:
return
ctp = next_continuation.get('clickTrackingParams')
return YoutubeTabIE._build_continuation_query(continuation, ctp)
@classmethod
def _extract_continuation(cls, renderer):
next_continuation = cls._extract_next_continuation_data(renderer)
if next_continuation:
return next_continuation
contents = []
for key in ('contents', 'items'):
contents.extend(try_get(renderer, lambda x: x[key], list) or [])
for content in contents:
if not isinstance(content, dict):
continue
continuation_ep = try_get(
content, lambda x: x['continuationItemRenderer']['continuationEndpoint'],
dict)
if not continuation_ep:
continue
continuation = try_get(
continuation_ep, lambda x: x['continuationCommand']['token'], compat_str)
if not continuation:
continue
ctp = continuation_ep.get('clickTrackingParams')
return YoutubeTabIE._build_continuation_query(continuation, ctp)
def _entries(self, tab, item_id, identity_token, account_syncid, ytcfg): def _entries(self, tab, item_id, identity_token, account_syncid, ytcfg):
def extract_entries(parent_renderer): # this needs to called again for continuation to work with feeds def extract_entries(parent_renderer): # this needs to called again for continuation to work with feeds