[Youtube] Improve comment API requests

co-authored by bbepis
This commit is contained in:
Bepis 2020-12-30 19:25:24 +11:00 committed by pukkandan
parent 0748b3317b
commit 8d0ea5f955

View File

@ -2485,17 +2485,15 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
try: try:
ncd = next(search_dict(yt_initial_data, 'nextContinuationData')) ncd = next(search_dict(yt_initial_data, 'nextContinuationData'))
continuations = [(ncd['continuation'], ncd['clickTrackingParams'])] continuations = [ncd['continuation']]
# Handle videos where comments have been disabled entirely # Handle videos where comments have been disabled entirely
except StopIteration: except StopIteration:
continuations = [] continuations = []
def get_continuation(continuation, itct, session_token, replies=False): def get_continuation(continuation, session_token, replies=False):
query = { query = {
'pbj': 1, 'pbj': 1,
'ctoken': continuation, 'ctoken': continuation,
'continuation': continuation,
'itct': itct,
} }
if replies: if replies:
query['action_get_comment_replies'] = 1 query['action_get_comment_replies'] = 1
@ -2523,23 +2521,26 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
response_code = handle.getcode() response_code = handle.getcode()
if (response_code == 200): if (response_code == 200):
return self._parse_json(content, video_id) return self._parse_json(content, video_id)
if (response_code == 413): # Sometimes google makes continuations that are too big to be accepted by themselves. Grade A engineering if (response_code == 413):
# self.to_screen(json.dumps(query))
# self.to_screen('Google API rate limit detected; waiting 30 seconds before continuing')
# time.sleep(30)
# continue
return None return None
raise ExtractorError('Unexpected HTTP error code: %s' % response_code) raise ExtractorError('Unexpected HTTP error code: %s' % response_code)
first_continuation = True first_continuation = True
while continuations: while continuations:
continuation, itct = continuations.pop() continuation, itct = continuations.pop()
comment_response = get_continuation(continuation, itct, xsrf_token) comment_response = get_continuation(continuation, xsrf_token)
if not comment_response: if not comment_response:
continue continue
if list(search_dict(comment_response, 'externalErrorMessage')): if list(search_dict(comment_response, 'externalErrorMessage')):
raise ExtractorError('Error returned from server: ' + next(search_dict(comment_response, 'externalErrorMessage'))) raise ExtractorError('Error returned from server: ' + next(search_dict(comment_response, 'externalErrorMessage')))
if 'continuationContents' not in comment_response['response']:
# Something is wrong here. Youtube won't accept this continuation token for some reason and responds with a user satisfaction dialog (error?)
continue
# not sure if this actually helps
if 'xsrf_token' in comment_response:
xsrf_token = comment_response['xsrf_token']
item_section = comment_response['response']['continuationContents']['itemSectionContinuation'] item_section = comment_response['response']['continuationContents']['itemSectionContinuation']
if first_continuation: if first_continuation:
expected_video_comment_count = int(item_section['header']['commentsHeaderRenderer']['countText']['runs'][0]['text'].replace(' Comments', '').replace('1 Comment', '1').replace(',', '')) expected_video_comment_count = int(item_section['header']['commentsHeaderRenderer']['countText']['runs'][0]['text'].replace(' Comments', '').replace('1 Comment', '1').replace(',', ''))
@ -2554,7 +2555,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
video_comments.append({ video_comments.append({
'id': comment['commentId'], 'id': comment['commentId'],
'text': ''.join([c['text'] for c in comment['contentText']['runs']]), 'text': ''.join([c['text'] for c in comment['contentText']['runs']]),
'time_text': comment['publishedTimeText']['runs'][0]['text'], 'time_text': ''.join([c['text'] for c in comment['publishedTimeText']['runs']]),
'author': comment.get('authorText', {}).get('simpleText', ''), 'author': comment.get('authorText', {}).get('simpleText', ''),
'votes': comment.get('voteCount', {}).get('simpleText', '0'), 'votes': comment.get('voteCount', {}).get('simpleText', '0'),
'author_thumbnail': comment['authorThumbnail']['thumbnails'][-1]['url'], 'author_thumbnail': comment['authorThumbnail']['thumbnails'][-1]['url'],
@ -2563,14 +2564,13 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
if 'replies' not in meta_comment['commentThreadRenderer']: if 'replies' not in meta_comment['commentThreadRenderer']:
continue continue
reply_continuation = meta_comment['commentThreadRenderer']['replies']['commentRepliesRenderer']['continuations'][0]['nextContinuationData'] reply_continuations = [rcn['nextContinuationData']['continuation'] for rcn in meta_comment['commentThreadRenderer']['replies']['commentRepliesRenderer']['continuations']]
continuation = reply_continuation['continuation'] while reply_continuations:
itct = reply_continuation['clickTrackingParams']
while True:
time.sleep(1) time.sleep(1)
replies_data = get_continuation(continuation, itct, xsrf_token, True) continuation = reply_continuations.pop()
replies_data = get_continuation(continuation, xsrf_token, True)
if not replies_data or 'continuationContents' not in replies_data[1]['response']: if not replies_data or 'continuationContents' not in replies_data[1]['response']:
break continue
if self._downloader.params.get('verbose', False): if self._downloader.params.get('verbose', False):
self.to_screen('[debug] Comments downloaded (chain %s) %s of ~%s' % (comment['commentId'], len(video_comments), expected_video_comment_count)) self.to_screen('[debug] Comments downloaded (chain %s) %s of ~%s' % (comment['commentId'], len(video_comments), expected_video_comment_count))
@ -2580,25 +2580,21 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
video_comments.append({ video_comments.append({
'id': reply_comment['commentId'], 'id': reply_comment['commentId'],
'text': ''.join([c['text'] for c in reply_comment['contentText']['runs']]), 'text': ''.join([c['text'] for c in reply_comment['contentText']['runs']]),
'time_text': reply_comment['publishedTimeText']['runs'][0]['text'], 'time_text': ''.join([c['text'] for c in reply_comment['publishedTimeText']['runs']]),
'author': reply_comment.get('authorText', {}).get('simpleText', ''), 'author': reply_comment.get('authorText', {}).get('simpleText', ''),
'votes': reply_comment.get('voteCount', {}).get('simpleText', '0'), 'votes': reply_comment.get('voteCount', {}).get('simpleText', '0'),
'author_thumbnail': reply_comment['authorThumbnail']['thumbnails'][-1]['url'], 'author_thumbnail': reply_comment['authorThumbnail']['thumbnails'][-1]['url'],
'parent': comment['commentId'] 'parent': comment['commentId']
}) })
if 'continuations' not in reply_comment_meta or len(reply_comment_meta['continuations']) == 0: if 'continuations' not in reply_comment_meta or len(reply_comment_meta['continuations']) == 0:
break continue
continuation = reply_comment_meta['continuations'][0]['nextContinuationData']['continuation'] reply_continuations += [rcn['nextContinuationData']['continuation'] for rcn in reply_comment_meta['continuations']]
itct = reply_comment_meta['continuations'][0]['nextContinuationData']['clickTrackingParams']
self.to_screen('Comments downloaded %s of ~%s' % (len(video_comments), expected_video_comment_count)) self.to_screen('Comments downloaded %s of ~%s' % (len(video_comments), expected_video_comment_count))
if 'continuations' in item_section: if 'continuations' in item_section:
new_continuations = [ continuations += [ncd['nextContinuationData']['continuation'] for ncd in item_section['continuations']]
(ncd['nextContinuationData']['continuation'], ncd['nextContinuationData']['clickTrackingParams'])
for ncd in item_section['continuations']]
continuations += new_continuations
time.sleep(1) time.sleep(1)
self.to_screen('Total comments downloaded %s of ~%s' % (len(video_comments), expected_video_comment_count)) self.to_screen('Total comments downloaded %s of ~%s' % (len(video_comments), expected_video_comment_count))