from ..utils import (
NO_DEFAULT,
ExtractorError,
+ UserNotLive,
bug_reports_message,
classproperty,
clean_html,
datetime_from_str,
dict_get,
- error_to_compat_str,
float_or_none,
format_field,
get_first,
parse_iso8601,
parse_qs,
qualities,
- remove_end,
remove_start,
smuggle_url,
str_or_none,
def _extract_response(self, item_id, query, note='Downloading API JSON', headers=None,
ytcfg=None, check_get_keys=None, ep='browse', fatal=True, api_hostname=None,
default_client='web'):
- response = None
- last_error = None
- count = -1
- retries = self.get_param('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 ...' % remove_end(last_error, '.'))
+ for retry in self.RetryManager():
try:
response = self._call_api(
ep=ep, fatal=True, headers=headers,
- video_id=item_id, query=query,
+ video_id=item_id, query=query, note=note,
context=self._extract_context(ytcfg, default_client),
api_key=self._extract_api_key(ytcfg, default_client),
- api_hostname=api_hostname, default_client=default_client,
- note='%s%s' % (note, ' (retry #%d)' % count if count else ''))
+ api_hostname=api_hostname, default_client=default_client)
except ExtractorError as e:
- if isinstance(e.cause, network_exceptions):
- if isinstance(e.cause, urllib.error.HTTPError):
- first_bytes = e.cause.read(512)
- if not is_html(first_bytes):
- yt_error = try_get(
- self._parse_json(
- self._webpage_read_content(e.cause, None, item_id, prefix=first_bytes) or '{}', item_id, fatal=False),
- lambda x: x['error']['message'], str)
- if yt_error:
- self._report_alerts([('ERROR', yt_error)], fatal=False)
- # Downloading page may result in intermittent 5xx HTTP error
- # Sometimes a 404 is also received. See: https://github.com/ytdl-org/youtube-dl/issues/28289
- # We also want to catch all other network exceptions since errors in later pages can be troublesome
- # See https://github.com/yt-dlp/yt-dlp/issues/507#issuecomment-880188210
- if not isinstance(e.cause, urllib.error.HTTPError) or e.cause.code not in (403, 429):
- last_error = error_to_compat_str(e.cause or e.msg)
- if count < retries:
- continue
- if fatal:
- raise
- else:
- self.report_warning(error_to_compat_str(e))
- return
+ if not isinstance(e.cause, network_exceptions):
+ return self._error_or_warning(e, fatal=fatal)
+ elif not isinstance(e.cause, urllib.error.HTTPError):
+ retry.error = e
+ continue
- else:
- try:
- self._extract_and_report_alerts(response, only_once=True)
- except ExtractorError as e:
- # YouTube servers may return errors we want to retry on in a 200 OK response
- # See: https://github.com/yt-dlp/yt-dlp/issues/839
- if 'unknown error' in e.msg.lower():
- last_error = e.msg
- continue
- if fatal:
- raise
- self.report_warning(error_to_compat_str(e))
- return
- 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:
- if fatal:
- raise ExtractorError(last_error)
- else:
- self.report_warning(last_error)
- return
- return response
+ first_bytes = e.cause.read(512)
+ if not is_html(first_bytes):
+ yt_error = try_get(
+ self._parse_json(
+ self._webpage_read_content(e.cause, None, item_id, prefix=first_bytes) or '{}', item_id, fatal=False),
+ lambda x: x['error']['message'], str)
+ if yt_error:
+ self._report_alerts([('ERROR', yt_error)], fatal=False)
+ # 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
+ # We also want to catch all other network exceptions since errors in later pages can be troublesome
+ # See https://github.com/yt-dlp/yt-dlp/issues/507#issuecomment-880188210
+ if e.cause.code not in (403, 429):
+ retry.error = e
+ continue
+ return self._error_or_warning(e, fatal=fatal)
+
+ try:
+ self._extract_and_report_alerts(response, only_once=True)
+ except ExtractorError as e:
+ # YouTube servers may return errors we want to retry on in a 200 OK response
+ # See: https://github.com/yt-dlp/yt-dlp/issues/839
+ if 'unknown error' in e.msg.lower():
+ retry.error = e
+ continue
+ return self._error_or_warning(e, fatal=fatal)
+ # Youtube sometimes sends incomplete data
+ # See: https://github.com/ytdl-org/youtube-dl/issues/28194
+ if not traverse_obj(response, *variadic(check_get_keys)):
+ retry.error = ExtractorError('Incomplete data received')
+ continue
+
+ return response
@staticmethod
def is_music_url(url):
(?:\#|$)""" % {
'invidious': '|'.join(YoutubeBaseInfoExtractor._INVIDIOUS_SITES),
}
+ _EMBED_REGEX = [r'''(?x)
+ (?:
+ <iframe[^>]+?src=|
+ data-video-url=|
+ <embed[^>]+?src=|
+ embedSWF\(?:\s*|
+ <object[^>]+data=|
+ new\s+SWFObject\(
+ )
+ (["\'])
+ (?P<url>(?:https?:)?//(?:www\.)?youtube(?:-nocookie)?\.com/
+ (?:embed|v|p)/[0-9A-Za-z_-]{11}.*?)
+ \1''']
_PLAYER_INFO_RE = (
r'/s/player/(?P<id>[a-zA-Z0-9_-]{8,})/player',
r'/(?P<id>[a-zA-Z0-9_-]{8,})/player(?:_ias\.vflset(?:/[a-zA-Z]{2,3}_[a-zA-Z]{2,3})?|-plasma-ias-(?:phone|tablet)-[a-z]{2}_[A-Z]{2}\.vflset)/base\.js$',
}
]
+ _WEBPAGE_TESTS = [
+ # YouTube <object> embed
+ {
+ 'url': 'http://www.improbable.com/2017/04/03/untrained-modern-youths-and-ancient-masters-in-selfie-portraits/',
+ 'md5': '873c81d308b979f0e23ee7e620b312a3',
+ 'info_dict': {
+ 'id': 'msN87y-iEx0',
+ 'ext': 'mp4',
+ 'title': 'Feynman: Mirrors FUN TO IMAGINE 6',
+ 'upload_date': '20080526',
+ 'description': 'md5:873c81d308b979f0e23ee7e620b312a3',
+ 'uploader': 'Christopher Sykes',
+ 'uploader_id': 'ChristopherJSykes',
+ 'age_limit': 0,
+ 'tags': ['feynman', 'mirror', 'science', 'physics', 'imagination', 'fun', 'cool', 'puzzle'],
+ 'channel_id': 'UCCeo--lls1vna5YJABWAcVA',
+ 'playable_in_embed': True,
+ 'thumbnail': 'https://i.ytimg.com/vi/msN87y-iEx0/hqdefault.jpg',
+ 'like_count': int,
+ 'comment_count': int,
+ 'channel': 'Christopher Sykes',
+ 'live_status': 'not_live',
+ 'channel_url': 'https://www.youtube.com/channel/UCCeo--lls1vna5YJABWAcVA',
+ 'availability': 'public',
+ 'duration': 195,
+ 'view_count': int,
+ 'categories': ['Science & Technology'],
+ 'channel_follower_count': int,
+ 'uploader_url': 'http://www.youtube.com/user/ChristopherJSykes',
+ },
+ 'params': {
+ 'skip_download': True,
+ }
+ },
+ ]
+
@classmethod
def suitable(cls, url):
from ..utils import parse_qs
microformats = traverse_obj(
prs, (..., 'microformat', 'playerMicroformatRenderer'),
expected_type=dict, default=[])
- _, is_live, _, formats = self._list_formats(video_id, microformats, video_details, prs, player_url)
+ _, is_live, _, formats, _ = self._list_formats(video_id, microformats, video_details, prs, player_url)
start_time = time.time()
def mpd_feed(format_id, delay):
url, video_id, f'Marking {label}watched',
'Unable to mark watched', fatal=False)
- @staticmethod
- def _extract_urls(webpage):
- # Embedded YouTube player
- entries = [
- unescapeHTML(mobj.group('url'))
- for mobj in re.finditer(r'''(?x)
- (?:
- <iframe[^>]+?src=|
- data-video-url=|
- <embed[^>]+?src=|
- embedSWF\(?:\s*|
- <object[^>]+data=|
- new\s+SWFObject\(
- )
- (["\'])
- (?P<url>(?:https?:)?//(?:www\.)?youtube(?:-nocookie)?\.com/
- (?:embed|v|p)/[0-9A-Za-z_-]{11}.*?)
- \1''', webpage)]
+ @classmethod
+ def _extract_from_webpage(cls, url, webpage):
+ # Invidious Instances
+ # https://github.com/yt-dlp/yt-dlp/issues/195
+ # https://github.com/iv-org/invidious/pull/1730
+ mobj = re.search(
+ r'<link rel="alternate" href="(?P<url>https://www\.youtube\.com/watch\?v=[0-9A-Za-z_-]{11})"',
+ webpage)
+ if mobj:
+ yield cls.url_result(mobj.group('url'), cls)
+ raise cls.StopExtraction()
+
+ yield from super()._extract_from_webpage(url, webpage)
# lazyYT YouTube embed
- entries.extend(list(map(
- unescapeHTML,
- re.findall(r'class="lazyYT" data-youtube-id="([^"]+)"', webpage))))
+ for id_ in re.findall(r'class="lazyYT" data-youtube-id="([^"]+)"', webpage):
+ yield cls.url_result(unescapeHTML(id_), cls, id_)
# Wordpress "YouTube Video Importer" plugin
- matches = re.findall(r'''(?x)<div[^>]+
- class=(?P<q1>[\'"])[^\'"]*\byvii_single_video_player\b[^\'"]*(?P=q1)[^>]+
- data-video_id=(?P<q2>[\'"])([^\'"]+)(?P=q2)''', webpage)
- entries.extend(m[-1] for m in matches)
-
- return entries
-
- @staticmethod
- def _extract_url(webpage):
- urls = YoutubeIE._extract_urls(webpage)
- return urls[0] if urls else None
+ for m in re.findall(r'''(?x)<div[^>]+
+ class=(?P<q1>[\'"])[^\'"]*\byvii_single_video_player\b[^\'"]*(?P=q1)[^>]+
+ data-video_id=(?P<q2>[\'"])([^\'"]+)(?P=q2)''', webpage):
+ yield cls.url_result(m[-1], cls, m[-1])
@classmethod
def extract_id(cls, url):
self.report_warning(last_error)
return prs, player_url
- def _extract_formats(self, streaming_data, video_id, player_url, is_live, duration):
+ def _extract_formats_and_subtitles(self, streaming_data, video_id, player_url, is_live, duration):
itags, stream_ids = {}, []
itag_qualities, res_qualities = {}, {}
q = qualities([
if val in qdict), -1)
return True
+ subtitles = {}
for sd in streaming_data:
hls_manifest_url = get_hls and sd.get('hlsManifestUrl')
if hls_manifest_url:
- for f in self._extract_m3u8_formats(hls_manifest_url, video_id, 'mp4', fatal=False):
+ fmts, subs = self._extract_m3u8_formats_and_subtitles(hls_manifest_url, video_id, 'mp4', fatal=False, live=is_live)
+ subtitles = self._merge_subtitles(subs, subtitles)
+ for f in fmts:
if process_manifest_format(f, 'hls', self._search_regex(
r'/itag/(\d+)', f['url'], 'itag', default=None)):
yield f
dash_manifest_url = get_dash and sd.get('dashManifestUrl')
if dash_manifest_url:
- for f in self._extract_mpd_formats(dash_manifest_url, video_id, fatal=False):
+ formats, subs = self._extract_mpd_formats_and_subtitles(dash_manifest_url, video_id, fatal=False)
+ subtitles = self._merge_subtitles(subs, subtitles) # Prioritize HLS subs over DASH
+ for f in formats:
if process_manifest_format(f, 'dash', f['format_id']):
f['filesize'] = int_or_none(self._search_regex(
r'/clen/(\d+)', f.get('fragment_base_url') or f['url'], 'file size', default=None))
f['is_from_start'] = True
yield f
+ yield subtitles
def _extract_storyboard(self, player_responses, duration):
spec = get_first(
'url': url,
'width': width,
'height': height,
+ 'fps': frame_count / duration,
+ 'rows': rows,
+ 'columns': cols,
'fragments': [{
'url': url.replace('$M', str(j)),
'duration': min(fragment_duration, duration - (j * fragment_duration)),
is_live = get_first(live_broadcast_details, 'isLiveNow')
streaming_data = traverse_obj(player_responses, (..., 'streamingData'), default=[])
- formats = list(self._extract_formats(streaming_data, video_id, player_url, is_live, duration))
+ *formats, subtitles = self._extract_formats_and_subtitles(streaming_data, video_id, player_url, is_live, duration)
- return live_broadcast_details, is_live, streaming_data, formats
+ return live_broadcast_details, is_live, streaming_data, formats, subtitles
def _real_extract(self, url):
url, smuggled_data = unsmuggle_url(url, {})
or get_first(microformats, 'lengthSeconds')
or parse_duration(search_meta('duration'))) or None
- if get_first(video_details, 'isPostLiveDvr'):
- self.write_debug('Video is in Post-Live Manifestless mode')
- if duration or 0 > 4 * 3600:
- self.report_warning(
- 'The livestream has not finished processing. Only 4 hours of the video can be currently downloaded. '
- 'This is a known issue and patches are welcome')
-
- live_broadcast_details, is_live, streaming_data, formats = self._list_formats(
- video_id, microformats, video_details, player_responses, player_url, duration)
+ live_broadcast_details, is_live, streaming_data, formats, automatic_captions = \
+ self._list_formats(video_id, microformats, video_details, player_responses, player_url)
if not formats:
if not self.get_param('allow_unplayable_formats') and traverse_obj(streaming_data, (..., 'licenseInfos')):
formats.extend(self._extract_storyboard(player_responses, duration))
- # Source is given priority since formats that throttle are given lower source_preference
- # When throttling issue is fully fixed, remove this
+ # source_preference is lower for throttled/potentially damaged formats
self._sort_formats(formats, ('quality', 'res', 'fps', 'hdr:12', 'source', 'codec:vp9.2', 'lang', 'proto'))
info = {
'release_timestamp': live_start_time,
}
+ if get_first(video_details, 'isPostLiveDvr'):
+ self.write_debug('Video is in Post-Live Manifestless mode')
+ info['live_status'] = 'post_live'
+ if (duration or 0) > 4 * 3600:
+ self.report_warning(
+ 'The livestream has not finished processing. Only 4 hours of the video can be currently downloaded. '
+ 'This is a known issue and patches are welcome')
+
+ subtitles = {}
pctr = traverse_obj(player_responses, (..., 'captions', 'playerCaptionsTracklistRenderer'), expected_type=dict)
if pctr:
def get_lang_code(track):
'name': sub_name,
})
- subtitles, automatic_captions = {}, {}
+ # NB: Constructing the full subtitle dictionary is slow
+ get_translated_subs = 'translated_subs' not in self._configuration_arg('skip') and (
+ self.get_param('writeautomaticsub', False) or self.get_param('listsubtitles'))
for lang_code, caption_track in captions.items():
base_url = caption_track.get('baseUrl')
orig_lang = parse_qs(base_url).get('lang', [None])[-1]
continue
orig_trans_code = trans_code
if caption_track.get('kind') != 'asr':
- if 'translated_subs' in self._configuration_arg('skip'):
+ if not get_translated_subs:
continue
trans_code += f'-{lang_code}'
trans_name += format_field(lang_name, None, ' from %s')
# Setting tlang=lang returns damaged subtitles.
process_language(automatic_captions, base_url, trans_code, trans_name,
{} if orig_lang == orig_trans_code else {'tlang': trans_code})
- info['automatic_captions'] = automatic_captions
- info['subtitles'] = subtitles
+
+ info['automatic_captions'] = automatic_captions
+ info['subtitles'] = subtitles
parsed_url = urllib.parse.urlparse(url)
for component in [parsed_url.fragment, parsed_url.query]:
return 'webpage' in self._configuration_arg('skip', ie_key=YoutubeTabIE.ie_key())
def _extract_webpage(self, url, item_id, fatal=True):
- retries = self.get_param('extractor_retries', 3)
- count = -1
- webpage = data = last_error = None
- while count < retries:
- count += 1
- # Sometimes youtube returns a webpage with incomplete ytInitialData
- # See: https://github.com/yt-dlp/yt-dlp/issues/116
- if last_error:
- self.report_warning('%s. Retrying ...' % last_error)
+ webpage, data = None, None
+ for retry in self.RetryManager(fatal=fatal):
try:
- webpage = self._download_webpage(
- url, item_id,
- note='Downloading webpage%s' % (' (retry #%d)' % count if count else '',))
+ webpage = self._download_webpage(url, item_id, note='Downloading webpage')
data = self.extract_yt_initial_data(item_id, webpage or '', fatal=fatal) or {}
except ExtractorError as e:
if isinstance(e.cause, network_exceptions):
if not isinstance(e.cause, urllib.error.HTTPError) or e.cause.code not in (403, 429):
- last_error = error_to_compat_str(e.cause or e.msg)
- if count < retries:
- continue
- if fatal:
- raise
- self.report_warning(error_to_compat_str(e))
+ retry.error = e
+ continue
+ self._error_or_warning(e, fatal=fatal)
break
- else:
- try:
- self._extract_and_report_alerts(data)
- except ExtractorError as e:
- if fatal:
- raise
- self.report_warning(error_to_compat_str(e))
- break
- if dict_get(data, ('contents', 'currentVideoEndpoint', 'onResponseReceivedActions')):
- break
+ try:
+ self._extract_and_report_alerts(data)
+ except ExtractorError as e:
+ self._error_or_warning(e, fatal=fatal)
+ break
- last_error = 'Incomplete yt initial data received'
- if count >= retries:
- if fatal:
- raise ExtractorError(last_error)
- self.report_warning(last_error)
- break
+ # Sometimes youtube returns a webpage with incomplete ytInitialData
+ # See: https://github.com/yt-dlp/yt-dlp/issues/116
+ if not traverse_obj(data, 'contents', 'currentVideoEndpoint', 'onResponseReceivedActions'):
+ retry.error = ExtractorError('Incomplete yt initial data received')
+ continue
return webpage, data
selected_tab_name = 'featured'
requested_tab_name = mobj['tab'][1:]
if 'no-youtube-channel-redirect' not in compat_opts:
- if requested_tab_name == 'live':
- # Live tab should have redirected to the video
- raise ExtractorError('The channel is not currently live', expected=True)
+ if requested_tab_name == 'live': # Live tab should have redirected to the video
+ raise UserNotLive(video_id=mobj['id'])
if requested_tab_name not in ('', selected_tab_name):
redirect_warning = f'The channel does not have a {requested_tab_name} tab'
if not original_tab_name:
projects / yt-dlp.git / blobdiff