X-Git-Url: https://jfr.im/git/yt-dlp.git/blobdiff_plain/b28f8d244a3cb901eff5be438ad8f0508bd8e024..b7da73eb19e00e4eab43ec7de129e9aa12f6d5d3:/yt_dlp/extractor/youtube.py diff --git a/yt_dlp/extractor/youtube.py b/yt_dlp/extractor/youtube.py index 3e19d18a0..6c1a5b881 100644 --- a/yt_dlp/extractor/youtube.py +++ b/yt_dlp/extractor/youtube.py @@ -2,6 +2,7 @@ from __future__ import unicode_literals +import calendar import hashlib import itertools import json @@ -27,6 +28,8 @@ bool_or_none, clean_html, dict_get, + datetime_from_str, + error_to_compat_str, ExtractorError, format_field, float_or_none, @@ -46,7 +49,7 @@ update_url_query, url_or_none, urlencode_postdata, - urljoin, + urljoin ) @@ -125,7 +128,7 @@ def req(url, f_req, note, errnote): }) def warn(message): - self._downloader.report_warning(message) + self.report_warning(message) lookup_req = [ username, @@ -282,21 +285,8 @@ def _real_initialize(self): if not self._login(): return - _YT_WEB_CLIENT_VERSION = '2.20210301.08.00' - _DEFAULT_API_DATA = { - 'context': { - 'client': { - 'clientName': 'WEB', - 'clientVersion': _YT_WEB_CLIENT_VERSION, - } - }, - } - - _DEFAULT_BASIC_API_HEADERS = { - 'X-YouTube-Client-Name': '1', - 'X-YouTube-Client-Version': _YT_WEB_CLIENT_VERSION - } - + _YT_WEB_CLIENT_VERSION = '2.20210407.08.00' + _YT_INNERTUBE_API_KEY = 'AIzaSyAO_FJ2SlqU8Q4STEHLGCilw_Y9_11qcW8' _YT_INITIAL_DATA_RE = r'(?:window\s*\[\s*["\']ytInitialData["\']\s*\]|ytInitialData)\s*=\s*({.+?})\s*;' _YT_INITIAL_PLAYER_RESPONSE_RE = r'ytInitialPlayerResponse\s*=\s*({.+?})\s*;' _YT_INITIAL_BOUNDARY_RE = r'(?:var\s+meta|= 2 and sync_ids[1]: # datasyncid is of the form "channel_syncid||user_syncid" for secondary channel # and just "user_syncid||" for primary channel. We only want the channel_syncid return sync_ids[0] + # ytcfg includes channel_syncid if on secondary channel + return data.get('DELEGATED_SESSION_ID') def _extract_ytcfg(self, video_id, webpage): return self._parse_json( self._search_regex( r'ytcfg\.set\s*\(\s*({.+?})\s*\)\s*;', webpage, 'ytcfg', - default='{}'), video_id, fatal=False) + default='{}'), video_id, fatal=False) or {} + + def __extract_client_version(self, ytcfg): + return try_get(ytcfg, lambda x: x['INNERTUBE_CLIENT_VERSION'], compat_str) or self._YT_WEB_CLIENT_VERSION + + def _extract_context(self, ytcfg=None): + context = try_get(ytcfg, lambda x: x['INNERTUBE_CONTEXT'], dict) + if context: + return context + + # Recreate the client context (required) + client_version = self.__extract_client_version(ytcfg) + client_name = try_get(ytcfg, lambda x: x['INNERTUBE_CLIENT_NAME'], compat_str) or 'WEB' + context = { + 'client': { + 'clientName': client_name, + 'clientVersion': client_version, + } + } + visitor_data = try_get(ytcfg, lambda x: x['VISITOR_DATA'], compat_str) + if visitor_data: + context['client']['visitorData'] = visitor_data + return context + + def _generate_api_headers(self, ytcfg=None, identity_token=None, account_syncid=None, visitor_data=None): + headers = { + 'X-YouTube-Client-Name': '1', + 'X-YouTube-Client-Version': self.__extract_client_version(ytcfg), + } + if identity_token: + headers['x-youtube-identity-token'] = identity_token + if account_syncid: + headers['X-Goog-PageId'] = account_syncid + headers['X-Goog-AuthUser'] = 0 + if visitor_data: + headers['x-goog-visitor-id'] = visitor_data + auth = self._generate_sapisidhash_header() + if auth is not None: + headers['Authorization'] = auth + headers['X-Origin'] = 'https://www.youtube.com' + return headers def _extract_video(self, renderer): video_id = renderer.get('videoId') @@ -379,7 +418,7 @@ def _extract_video(self, renderer): (lambda x: x['ownerText']['runs'][0]['text'], lambda x: x['shortBylineText']['runs'][0]['text']), compat_str) return { - '_type': 'url_transparent', + '_type': 'url', 'ie_key': YoutubeIE.ie_key(), 'id': video_id, 'url': video_id, @@ -1246,6 +1285,23 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'url': 'https://www.youtube.com/watch?v=nGC3D_FkCmg', 'only_matching': True, }, + { + # restricted location, https://github.com/ytdl-org/youtube-dl/issues/28685 + 'url': 'cBvYw8_A0vQ', + 'info_dict': { + 'id': 'cBvYw8_A0vQ', + 'ext': 'mp4', + 'title': '4K Ueno Okachimachi Street Scenes 上野御徒町歩き', + 'description': 'md5:ea770e474b7cd6722b4c95b833c03630', + 'upload_date': '20201120', + 'uploader': 'Walk around Japan', + 'uploader_id': 'UC3o_t8PzBmXf5S9b7GLx1Mw', + 'uploader_url': r're:https?://(?:www\.)?youtube\.com/channel/UC3o_t8PzBmXf5S9b7GLx1Mw', + }, + 'params': { + 'skip_download': True, + }, + }, ] def __init__(self, *args, **kwargs): @@ -1499,6 +1555,16 @@ def _extract_yt_initial_variable(self, webpage, regex, video_id, name): (r'%s\s*%s' % (regex, self._YT_INITIAL_BOUNDARY_RE), regex), webpage, name, default='{}'), video_id, fatal=False) + @staticmethod + def parse_time_text(time_text): + """ + Parse the comment time text + time_text is in the format 'X units ago (edited)' + """ + time_text_split = time_text.split(' ') + if len(time_text_split) >= 3: + return datetime_from_str('now-%s%s' % (time_text_split[0], time_text_split[1]), precision='auto') + @staticmethod def _join_text_entries(runs): text = None @@ -1521,7 +1587,7 @@ def _extract_comment(self, comment_renderer, parent=None): text = self._join_text_entries(comment_text_runs) or '' comment_time_text = try_get(comment_renderer, lambda x: x['publishedTimeText']['runs']) or [] time_text = self._join_text_entries(comment_time_text) - + timestamp = calendar.timegm(self.parse_time_text(time_text).timetuple()) author = try_get(comment_renderer, lambda x: x['authorText']['simpleText'], compat_str) author_id = try_get(comment_renderer, lambda x: x['authorEndpoint']['browseEndpoint']['browseId'], compat_str) @@ -1532,11 +1598,10 @@ def _extract_comment(self, comment_renderer, parent=None): author_is_uploader = try_get(comment_renderer, lambda x: x['authorIsChannelOwner'], bool) is_liked = try_get(comment_renderer, lambda x: x['isLiked'], bool) - return { 'id': comment_id, 'text': text, - # TODO: This should be parsed to timestamp + 'timestamp': timestamp, 'time_text': time_text, 'like_count': votes, 'is_favorited': is_liked, @@ -1548,7 +1613,7 @@ def _extract_comment(self, comment_renderer, parent=None): } def _comment_entries(self, root_continuation_data, identity_token, account_syncid, - session_token_list, parent=None, comment_counts=None): + ytcfg, session_token_list, parent=None, comment_counts=None): def extract_thread(parent_renderer): contents = try_get(parent_renderer, lambda x: x['contents'], list) or [] @@ -1574,7 +1639,7 @@ def extract_thread(parent_renderer): if comment_replies_renderer: comment_counts[2] += 1 comment_entries_iter = self._comment_entries( - comment_replies_renderer, identity_token, account_syncid, + comment_replies_renderer, identity_token, account_syncid, ytcfg, parent=comment.get('id'), session_token_list=session_token_list, comment_counts=comment_counts) @@ -1584,16 +1649,10 @@ def extract_thread(parent_renderer): if not comment_counts: # comment so far, est. total comments, current comment thread # comment_counts = [0, 0, 0] - headers = self._DEFAULT_BASIC_API_HEADERS.copy() # TODO: Generalize the download code with TabIE - if identity_token: - headers['x-youtube-identity-token'] = identity_token - - if account_syncid: - headers['X-Goog-PageId'] = account_syncid - headers['X-Goog-AuthUser'] = 0 - + context = self._extract_context(ytcfg) + visitor_data = try_get(context, lambda x: x['client']['visitorData'], compat_str) continuation = YoutubeTabIE._extract_continuation(root_continuation_data) # TODO first_continuation = False if parent is None: @@ -1602,6 +1661,7 @@ def extract_thread(parent_renderer): for page_num in itertools.count(0): if not continuation: break + headers = self._generate_api_headers(ytcfg, identity_token, account_syncid, visitor_data) retries = self._downloader.params.get('extractor_retries', 3) count = -1 last_error = None @@ -1624,12 +1684,12 @@ def extract_thread(parent_renderer): comment_prog_str = '(%d/%d)' % (comment_counts[0], comment_counts[1]) if page_num == 0: if first_continuation: - note_prefix = "Downloading initial comment continuation page" + note_prefix = 'Downloading initial comment continuation page' else: - note_prefix = " Downloading comment reply thread %d %s" % (comment_counts[2], comment_prog_str) + note_prefix = ' Downloading comment reply thread %d %s' % (comment_counts[2], comment_prog_str) else: - note_prefix = "%sDownloading comment%s page %d %s" % ( - " " if parent else "", + note_prefix = '%sDownloading comment%s page %d %s' % ( + ' ' if parent else '', ' replies' if parent else '', page_num, comment_prog_str) @@ -1644,13 +1704,13 @@ def extract_thread(parent_renderer): 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)") + 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)" + last_error = last_error + ' (this API is probably deprecated)' if count < retries: continue raise @@ -1668,7 +1728,7 @@ def extract_thread(parent_renderer): # YouTube sometimes gives reload: now json if something went wrong (e.g. bad auth) if browse.get('reload'): - raise ExtractorError("Invalid or missing params in continuation request", expected=False) + raise ExtractorError('Invalid or missing params in continuation request', expected=False) # TODO: not tested, merged from old extractor err_msg = browse.get('externalErrorMessage') @@ -1679,10 +1739,14 @@ def extract_thread(parent_renderer): # See: https://github.com/ytdl-org/youtube-dl/issues/28194 last_error = 'Incomplete data received' if count >= retries: - self._downloader.report_error(last_error) + raise ExtractorError(last_error) if not response: break + visitor_data = try_get( + response, + lambda x: x['responseContext']['webResponseContextExtensionData']['ytConfigData']['visitorData'], + compat_str) or visitor_data known_continuation_renderers = { 'itemSectionContinuation': extract_thread, @@ -1708,7 +1772,7 @@ def extract_thread(parent_renderer): 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)) + self.to_screen('Downloading ~%d comments' % str_to_int(expected_comment_count)) yield comment_counts[1] # TODO: cli arg. @@ -1724,7 +1788,7 @@ def extract_thread(parent_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')) + self.to_screen('Sorting comments by %s' % ('popular' if comment_sort_index == 0 else 'newest')) break for entry in known_continuation_renderers[key](continuation_renderer): @@ -1749,6 +1813,7 @@ def _extract_comments(self, ytcfg, video_id, contents, webpage, xsrf_token): renderer, identity_token=self._extract_identity_token(webpage, item_id=video_id), account_syncid=self._extract_account_syncid(ytcfg), + ytcfg=ytcfg, session_token_list=[xsrf_token]) for comment in comment_iter: @@ -1757,7 +1822,7 @@ def _extract_comments(self, ytcfg, video_id, contents, webpage, xsrf_token): continue comments.append(comment) break - self.to_screen("Downloaded %d/%d comments" % (len(comments), estimated_total)) + self.to_screen('Downloaded %d/%d comments' % (len(comments), estimated_total)) return { 'comments': comments, 'comment_count': len(comments), @@ -1776,9 +1841,11 @@ def _real_extract(self, url): player_response = self._extract_yt_initial_variable( webpage, self._YT_INITIAL_PLAYER_RESPONSE_RE, video_id, 'initial player response') + + ytcfg = self._extract_ytcfg(video_id, webpage) if not player_response: player_response = self._call_api( - 'player', {'videoId': video_id}, video_id) + 'player', {'videoId': video_id}, video_id, api_key=self._extract_api_key(ytcfg)) playability_status = player_response.get('playabilityStatus') or {} if playability_status.get('reason') == 'Sign in to confirm your age': @@ -1806,7 +1873,13 @@ def _real_extract(self, url): def get_text(x): if not x: return - return x.get('simpleText') or ''.join([r['text'] for r in x['runs']]) + text = x.get('simpleText') + if text and isinstance(text, compat_str): + return text + runs = x.get('runs') + if not isinstance(runs, list): + return + return ''.join([r['text'] for r in runs if isinstance(r.get('text'), compat_str)]) search_meta = ( lambda x: self._html_search_meta(x, webpage, default=None)) \ @@ -1977,7 +2050,7 @@ def feed_entry(name): if not formats: if not self._downloader.params.get('allow_unplayable_formats') and streaming_data.get('licenseInfos'): - raise ExtractorError( + self.raise_no_formats( 'This video is DRM protected.', expected=True) pemr = try_get( playability_status, @@ -1992,11 +2065,10 @@ def feed_entry(name): if not countries: regions_allowed = search_meta('regionsAllowed') countries = regions_allowed.split(',') if regions_allowed else None - self.raise_geo_restricted( - subreason, countries) + self.raise_geo_restricted(subreason, countries, metadata_available=True) reason += '\n' + subreason if reason: - raise ExtractorError(reason, expected=True) + self.raise_no_formats(reason, expected=True) self._sort_formats(formats) @@ -2007,8 +2079,10 @@ def feed_entry(name): for m in re.finditer(self._meta_regex('og:video:tag'), webpage)] for keyword in keywords: if keyword.startswith('yt:stretch='): - w, h = keyword.split('=')[1].split(':') - w, h = int(w), int(h) + stretch_ratio = map( + lambda x: int_or_none(x, default=0), + keyword.split('=')[1].split(':')) + w, h = (list(stretch_ratio) + [0])[:2] if w > 0 and h > 0: ratio = w / h for f in formats: @@ -2023,6 +2097,11 @@ def feed_entry(name): thumbnail_url = thumbnail.get('url') if not thumbnail_url: continue + # Sometimes youtube gives a wrong thumbnail URL. See: + # https://github.com/yt-dlp/yt-dlp/issues/233 + # https://github.com/ytdl-org/youtube-dl/issues/28023 + if 'maxresdefault' in thumbnail_url: + thumbnail_url = thumbnail_url.split('?')[0] thumbnails.append({ 'height': int_or_none(thumbnail.get('height')), 'url': thumbnail_url, @@ -2151,7 +2230,7 @@ def process_language(container, base_url, lang_code, query): 'yt initial data') if not initial_data: initial_data = self._call_api( - 'next', {'videoId': video_id}, video_id, fatal=False) + 'next', {'videoId': video_id}, video_id, fatal=False, api_key=self._extract_api_key(ytcfg)) if not is_live: try: @@ -2549,6 +2628,19 @@ class YoutubeTabIE(YoutubeBaseInfoExtractor): 'uploader_id': 'UCXw-G3eDE9trcvY2sBMM_aA', }, 'playlist_mincount': 21, + }, { + 'note': 'Playlist with "show unavailable videos" button', + 'url': 'https://www.youtube.com/playlist?list=UUTYLiWFZy8xtPwxFwX9rV7Q', + 'info_dict': { + 'title': 'Uploads from Phim Siêu Nhân Nhật Bản', + 'id': 'UUTYLiWFZy8xtPwxFwX9rV7Q', + 'uploader': 'Phim Siêu Nhân Nhật Bản', + 'uploader_id': 'UCTYLiWFZy8xtPwxFwX9rV7Q', + }, + 'playlist_mincount': 1400, + 'expected_warnings': [ + 'YouTube said: INFO - Unavailable videos are hidden', + ] }, { # https://github.com/ytdl-org/youtube-dl/issues/21844 'url': 'https://www.youtube.com/playlist?list=PLzH6n4zXuckpfMu_4Ff8E7Z1behQks5ba', @@ -2681,6 +2773,13 @@ class YoutubeTabIE(YoutubeBaseInfoExtractor): }, { 'url': 'https://www.youtube.com/TheYoungTurks/live', 'only_matching': True, + }, { + 'url': 'https://www.youtube.com/hashtag/cctv9', + 'info_dict': { + 'id': 'cctv9', + 'title': '#cctv9', + }, + 'playlist_mincount': 350, }] @classmethod @@ -2841,6 +2940,16 @@ def _post_thread_continuation_entries(self, post_thread_continuation): for entry in self._post_thread_entries(renderer): yield entry + r''' # unused + def _rich_grid_entries(self, contents): + for content in contents: + video_renderer = try_get(content, lambda x: x['richItemRenderer']['content']['videoRenderer'], dict) + if video_renderer: + entry = self._video_entry(video_renderer) + if entry: + yield entry + ''' + @staticmethod def _build_continuation_query(continuation, ctp=None): query = { @@ -2886,7 +2995,7 @@ def _extract_continuation(cls, renderer): ctp = continuation_ep.get('clickTrackingParams') return YoutubeTabIE._build_continuation_query(continuation, ctp) - def _entries(self, tab, item_id, identity_token, account_syncid): + 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 contents = try_get(parent_renderer, lambda x: x['contents'], list) or [] @@ -2938,61 +3047,26 @@ def extract_entries(parent_renderer): # this needs to called again for continua for entry in extract_entries(parent_renderer): yield entry continuation = continuation_list[0] - - headers = { - 'x-youtube-client-name': '1', - 'x-youtube-client-version': '2.20201112.04.01', - } - if identity_token: - headers['x-youtube-identity-token'] = identity_token - - if account_syncid: - headers['X-Goog-PageId'] = account_syncid - headers['X-Goog-AuthUser'] = 0 + context = self._extract_context(ytcfg) + visitor_data = try_get(context, lambda x: x['client']['visitorData'], compat_str) for page_num in itertools.count(1): if not continuation: break - retries = self._downloader.params.get('extractor_retries', 3) - count = -1 - last_error = None - while count < retries: - 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']}, - }, - 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) + query = { + 'continuation': continuation['continuation'], + 'clickTracking': {'clickTrackingParams': continuation['itct']} + } + headers = self._generate_api_headers(ytcfg, identity_token, account_syncid, visitor_data) + response = self._extract_response( + item_id='%s page %s' % (item_id, page_num), + query=query, headers=headers, ytcfg=ytcfg, + check_get_keys=('continuationContents', 'onResponseReceivedActions', 'onResponseReceivedEndpoints')) if not response: break + visitor_data = try_get( + response, lambda x: x['responseContext']['visitorData'], compat_str) or visitor_data known_continuation_renderers = { 'playlistVideoListContinuation': self._playlist_entries, @@ -3083,10 +3157,10 @@ def _extract_from_tabs(self, item_id, webpage, data, tabs): channel_name = renderer.get('title') channel_url = renderer.get('channelUrl') channel_id = renderer.get('externalId') - - if not renderer: + else: renderer = try_get( data, lambda x: x['metadata']['playlistMetadataRenderer'], dict) + if renderer: title = renderer.get('title') description = renderer.get('description', '') @@ -3112,11 +3186,12 @@ def _extract_from_tabs(self, item_id, webpage, data, tabs): 'width': int_or_none(t.get('width')), 'height': int_or_none(t.get('height')), }) - if playlist_id is None: playlist_id = item_id if title is None: - title = playlist_id + title = ( + try_get(data, lambda x: x['header']['hashtagHeaderRenderer']['hashtag']['simpleText']) + or playlist_id) title += format_field(selected_tab, 'title', ' - %s') metadata = { @@ -3139,11 +3214,17 @@ def _extract_from_tabs(self, item_id, webpage, data, tabs): self._entries( selected_tab, playlist_id, self._extract_identity_token(webpage, item_id), - self._extract_account_syncid(data)), + self._extract_account_syncid(data), + self._extract_ytcfg(item_id, webpage)), **metadata) - def _extract_mix_playlist(self, playlist, playlist_id): + def _extract_mix_playlist(self, playlist, playlist_id, data, webpage): 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): videos = list(self._playlist_entries(playlist)) if not videos: @@ -3158,14 +3239,25 @@ def _extract_mix_playlist(self, playlist, playlist_id): yield video first_id = first_id or videos[0]['id'] last_id = videos[-1]['id'] - - _, data = self._extract_webpage( - 'https://www.youtube.com/watch?list=%s&v=%s' % (playlist_id, last_id), - '%s page %d' % (playlist_id, page_num)) + watch_endpoint = try_get( + playlist, lambda x: x['contents'][-1]['playlistPanelVideoRenderer']['navigationEndpoint']['watchEndpoint']) + query = { + '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( - 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( data, lambda x: x['titleText']['simpleText'], compat_str) playlist_id = playlist.get('playlistId') or item_id @@ -3180,7 +3272,7 @@ def _extract_from_playlist(self, item_id, url, data, playlist): video_title=title) 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) def _extract_alerts(self, data, expected=False): @@ -3193,25 +3285,115 @@ def _real_extract_alerts(): alert_type = alert.get('type') if not alert_type: continue - message = try_get(alert, lambda x: x['text']['simpleText'], compat_str) + message = try_get(alert, lambda x: x['text']['simpleText'], compat_str) or '' if message: yield alert_type, message for run in try_get(alert, lambda x: x['text']['runs'], list) or []: - message = try_get(run, lambda x: x['text'], compat_str) - if message: - yield alert_type, message + message += try_get(run, lambda x: x['text'], compat_str) + if message: + yield alert_type, message - err_msg = None + errors = [] + warnings = [] for alert_type, alert_message in _real_extract_alerts(): if alert_type.lower() == 'error': - if err_msg: - self._downloader.report_warning('YouTube said: %s - %s' % ('ERROR', err_msg)) - err_msg = alert_message + errors.append([alert_type, alert_message]) else: - self._downloader.report_warning('YouTube said: %s - %s' % (alert_type, alert_message)) + warnings.append([alert_type, alert_message]) + + for alert_type, alert_message in (warnings + errors[:-1]): + self.report_warning('YouTube said: %s - %s' % (alert_type, alert_message)) + if errors: + raise ExtractorError('YouTube said: %s' % errors[-1][1], expected=expected) + + def _reload_with_unavailable_videos(self, item_id, data, webpage): + """ + Get playlist with unavailable videos if the 'show unavailable videos' button exists. + """ + sidebar_renderer = try_get( + data, lambda x: x['sidebar']['playlistSidebarRenderer']['items'], list) or [] + for item in sidebar_renderer: + if not isinstance(item, dict): + continue + renderer = item.get('playlistSidebarPrimaryInfoRenderer') + menu_renderer = try_get( + renderer, lambda x: x['menu']['menuRenderer']['items'], list) or [] + for menu_item in menu_renderer: + if not isinstance(menu_item, dict): + continue + nav_item_renderer = menu_item.get('menuNavigationItemRenderer') + text = try_get( + nav_item_renderer, lambda x: x['text']['simpleText'], compat_str) + if not text or text.lower() != 'show unavailable videos': + continue + browse_endpoint = try_get( + nav_item_renderer, lambda x: x['navigationEndpoint']['browseEndpoint'], dict) or {} + browse_id = browse_endpoint.get('browseId') + params = browse_endpoint.get('params') + if not browse_id or not params: + return + ytcfg = self._extract_ytcfg(item_id, webpage) + headers = self._generate_api_headers( + ytcfg, account_syncid=self._extract_account_syncid(ytcfg), + identity_token=self._extract_identity_token(webpage, item_id=item_id), + visitor_data=try_get( + self._extract_context(ytcfg), lambda x: x['client']['visitorData'], compat_str)) + query = { + 'params': params, + 'browseId': browse_id + } + return self._extract_response( + item_id=item_id, headers=headers, query=query, + check_get_keys='contents', fatal=False, + note='Downloading API JSON with unavailable videos') + + def _extract_response(self, item_id, query, note='Downloading API JSON', headers=None, + ytcfg=None, check_get_keys=None, ep='browse', fatal=True): + 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 + if fatal: + raise + else: + self.report_warning(error_to_compat_str(e)) + return - if err_msg: - raise ExtractorError('YouTube said: %s' % err_msg, expected=expected) + 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: + if fatal: + raise ExtractorError(last_error) + else: + self.report_warning(last_error) + return + return response def _extract_webpage(self, url, item_id): retries = self._downloader.params.get('extractor_retries', 3) @@ -3231,7 +3413,7 @@ def _extract_webpage(self, url, item_id): if data.get('contents') or data.get('currentVideoEndpoint'): break if count >= retries: - self._downloader.report_error(last_error) + raise ExtractorError(last_error) return webpage, data def _real_extract(self, url): @@ -3243,7 +3425,7 @@ def _real_extract(self, url): mobj = re.match(r'(?P
%s)(?P/?(?![^#?]).*$)' % self._VALID_URL, url)
         mobj = mobj.groupdict() if mobj else {}
         if mobj and not mobj.get('not_channel'):
-            self._downloader.report_warning(
+            self.report_warning(
                 'A channel/user page was given. All the channel\'s videos will be downloaded. '
                 'To download only the videos in the home page, add a "/featured" to the URL')
             url = '%s/videos%s' % (mobj.get('pre'), mobj.get('post') or '')
@@ -3258,7 +3440,7 @@ def _real_extract(self, url):
                 # If there is neither video or playlist ids,
                 # youtube redirects to home page, which is undesirable
                 raise ExtractorError('Unable to recognize tab page')
-            self._downloader.report_warning('A video URL was given without video ID. Trying to download playlist %s' % playlist_id)
+            self.report_warning('A video URL was given without video ID. Trying to download playlist %s' % playlist_id)
             url = 'https://www.youtube.com/playlist?list=%s' % playlist_id
 
         if video_id and playlist_id:
@@ -3269,6 +3451,9 @@ def _real_extract(self, url):
 
         webpage, data = self._extract_webpage(url, item_id)
 
+        # YouTube sometimes provides a button to reload playlist with unavailable videos.
+        data = self._reload_with_unavailable_videos(item_id, data, webpage) or data
+
         tabs = try_get(
             data, lambda x: x['contents']['twoColumnBrowseResultsRenderer']['tabs'], list)
         if tabs:
@@ -3277,13 +3462,13 @@ def _real_extract(self, url):
         playlist = try_get(
             data, lambda x: x['contents']['twoColumnWatchNextResults']['playlist']['playlist'], dict)
         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(
             data, lambda x: x['currentVideoEndpoint']['watchEndpoint']['videoId'],
             compat_str) or video_id
         if video_id:
-            self._downloader.report_warning('Unable to recognize playlist. Downloading just video %s' % video_id)
+            self.report_warning('Unable to recognize playlist. Downloading just video %s' % video_id)
             return self.url_result(video_id, ie=YoutubeIE.ie_key(), video_id=video_id)
 
         raise ExtractorError('Unable to recognize tab page')
@@ -3439,7 +3624,7 @@ def _real_extract(self, url):
             ie=YoutubeTabIE.ie_key())
 
 
-class YoutubeSearchIE(SearchInfoExtractor, YoutubeBaseInfoExtractor):
+class YoutubeSearchIE(SearchInfoExtractor, YoutubeTabIE):
     IE_DESC = 'YouTube.com searches, "ytsearch" keyword'
     # there doesn't appear to be a real limit, for example if you search for
     # 'python' you get more than 8.000.000 results
@@ -3455,9 +3640,10 @@ def _entries(self, query, n):
             data['params'] = self._SEARCH_PARAMS
         total = 0
         for page_num in itertools.count(1):
-            search = self._call_api(
-                ep='search', video_id='query "%s"' % query, fatal=False,
-                note='Downloading page %s' % page_num, query=data)
+            search = self._extract_response(
+                item_id='query "%s" page %s' % (query, page_num), ep='search', query=data,
+                check_get_keys=('contents', 'onResponseReceivedCommands')
+            )
             if not search:
                 break
             slr_contents = try_get(