X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/540fd68c40df72763aee5d75598675c45cfa9aba..d317bf370d3cbd2231f3d6b1970e0a86fddc4e5b:/youtube_dl/extractor/youtube.py diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index e58184a..3d8b31f 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -17,6 +17,8 @@ from ..compat import ( compat_chr, compat_parse_qs, compat_urllib_parse, + compat_urllib_parse_unquote, + compat_urllib_parse_unquote_plus, compat_urllib_request, compat_urlparse, compat_str, @@ -29,9 +31,11 @@ from ..utils import ( get_element_by_id, int_or_none, orderedSet, + str_to_int, unescapeHTML, unified_strdate, uppercase_escape, + ISO3166Utils, ) @@ -49,6 +53,11 @@ class YoutubeBaseInfoExtractor(InfoExtractor): # YouTube sets the expire time to about two months expire_time=time.time() + 2 * 30 * 24 * 3600) + def _ids_to_results(self, ids): + return [ + self.url_result(vid_id, 'Youtube', video_id=vid_id) + for vid_id in ids] + def _login(self): """ Attempt to log in to YouTube. @@ -229,6 +238,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor): '44': {'ext': 'webm', 'width': 854, 'height': 480}, '45': {'ext': 'webm', 'width': 1280, 'height': 720}, '46': {'ext': 'webm', 'width': 1920, 'height': 1080}, + '59': {'ext': 'mp4', 'width': 854, 'height': 480}, + '78': {'ext': 'mp4', 'width': 854, 'height': 480}, # 3d videos @@ -511,6 +522,38 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'skip_download': 'requires avconv', } }, + # Extraction from multiple DASH manifests (https://github.com/rg3/youtube-dl/pull/6097) + { + 'url': 'https://www.youtube.com/watch?v=FIl7x6_3R5Y', + 'info_dict': { + 'id': 'FIl7x6_3R5Y', + 'ext': 'mp4', + 'title': 'md5:7b81415841e02ecd4313668cde88737a', + 'description': 'md5:116377fd2963b81ec4ce64b542173306', + 'upload_date': '20150625', + 'uploader_id': 'dorappi2000', + 'uploader': 'dorappi2000', + 'formats': 'mincount:33', + }, + }, + # DASH manifest with segment_list + { + 'url': 'https://www.youtube.com/embed/CsmdDsKjzN8', + 'md5': '8ce563a1d667b599d21064e982ab9e31', + 'info_dict': { + 'id': 'CsmdDsKjzN8', + 'ext': 'mp4', + 'upload_date': '20150501', # According to '', video_webpage) is not None: age_gate = True # We simulate the access to the video from www.youtube.com/v/{video_id} @@ -866,24 +934,31 @@ class YoutubeIE(YoutubeBaseInfoExtractor): note='Refetching age-gated info webpage', errnote='unable to download video info webpage') video_info = compat_parse_qs(video_info_webpage) + add_dash_mpd(video_info) else: age_gate = False - try: - # Try looking directly into the video webpage - mobj = re.search(r';ytplayer\.config\s*=\s*({.*?});', video_webpage) - if not mobj: - raise ValueError('Could not find ytplayer.config') # caught below + video_info = None + # Try looking directly into the video webpage + mobj = re.search(r';ytplayer\.config\s*=\s*({.*?});', video_webpage) + if mobj: json_code = uppercase_escape(mobj.group(1)) ytplayer_config = json.loads(json_code) args = ytplayer_config['args'] - # Convert to the same format returned by compat_parse_qs - video_info = dict((k, [v]) for k, v in args.items()) - if not args.get('url_encoded_fmt_stream_map'): - raise ValueError('No stream_map present') # caught below - except ValueError: - # We fallback to the get_video_info pages (used by the embed page) + if args.get('url_encoded_fmt_stream_map'): + # Convert to the same format returned by compat_parse_qs + video_info = dict((k, [v]) for k, v in args.items()) + add_dash_mpd(video_info) + if args.get('livestream') == '1' or args.get('live_playback') == 1: + is_live = True + if not video_info or self._downloader.params.get('youtube_include_dash_manifest', True): + # We also try looking in get_video_info since it may contain different dashmpd + # URL that points to a DASH manifest with possibly different itag set (some itags + # are missing from DASH manifest pointed by webpage's dashmpd, some - from DASH + # manifest pointed by get_video_info's dashmpd). + # The general idea is to take a union of itags of both DASH manifests (for example + # video with such 'manifest behavior' see https://github.com/rg3/youtube-dl/issues/6093) self.report_video_info_webpage_download(video_id) - for el_type in ['&el=embedded', '&el=detailpage', '&el=vevo', '']: + for el_type in ['&el=info', '&el=embedded', '&el=detailpage', '&el=vevo', '']: video_info_url = ( '%s://www.youtube.com/get_video_info?&video_id=%s%s&ps=default&eurl=&gl=US&hl=en' % (proto, video_id, el_type)) @@ -891,11 +966,20 @@ class YoutubeIE(YoutubeBaseInfoExtractor): video_info_url, video_id, note=False, errnote='unable to download video info webpage') - video_info = compat_parse_qs(video_info_webpage) - if 'token' in video_info: + get_video_info = compat_parse_qs(video_info_webpage) + add_dash_mpd(get_video_info) + if not video_info: + video_info = get_video_info + if 'token' in get_video_info: break if 'token' not in video_info: if 'reason' in video_info: + if 'The uploader has not made this video available in your country.' in video_info['reason']: + regions_allowed = self._html_search_meta('regionsAllowed', video_webpage, default=None) + if regions_allowed is not None: + raise ExtractorError('YouTube said: This video is available in %s only' % ( + ', '.join(map(ISO3166Utils.short2full, regions_allowed.split(',')))), + expected=True) raise ExtractorError( 'YouTube said: %s' % video_info['reason'][0], expected=True, video_id=video_id) @@ -919,7 +1003,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): # uploader if 'author' not in video_info: raise ExtractorError('Unable to extract uploader name') - video_uploader = compat_urllib_parse.unquote_plus(video_info['author'][0]) + video_uploader = compat_urllib_parse_unquote_plus(video_info['author'][0]) # uploader_id video_uploader_id = None @@ -946,18 +1030,19 @@ class YoutubeIE(YoutubeBaseInfoExtractor): self._downloader.report_warning('unable to extract video thumbnail') video_thumbnail = None else: # don't panic if we can't find it - video_thumbnail = compat_urllib_parse.unquote_plus(video_info['thumbnail_url'][0]) + video_thumbnail = compat_urllib_parse_unquote_plus(video_info['thumbnail_url'][0]) # upload date - upload_date = None - mobj = re.search(r'(?s)id="eow-date.*?>(.*?)', video_webpage) - if mobj is None: - mobj = re.search( - r'(?s)id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live) on (.*?)', - video_webpage) - if mobj is not None: - upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split()) - upload_date = unified_strdate(upload_date) + upload_date = self._html_search_meta( + 'datePublished', video_webpage, 'upload date', default=None) + if not upload_date: + upload_date = self._search_regex( + [r'(?s)id="eow-date.*?>(.*?)', + r'id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live|Started) on (.+?)'], + video_webpage, 'upload date', default=None) + if upload_date: + upload_date = ' '.join(re.sub(r'[/,-]', r' ', mobj.group(1)).split()) + upload_date = unified_strdate(upload_date) m_cat_container = self._search_regex( r'(?s)]*>\s*Category\s*\s*]*>(.*?)', @@ -991,12 +1076,11 @@ class YoutubeIE(YoutubeBaseInfoExtractor): video_description = '' def _extract_count(count_name): - count = self._search_regex( - r'id="watch-%s"[^>]*>.*?([\d,]+)\s*' % re.escape(count_name), - video_webpage, count_name, default=None) - if count is not None: - return int(count.replace(',', '')) - return None + return str_to_int(self._search_regex( + r'-%s-button[^>]+>]+class="yt-uix-button-content"[^>]*>([\d,]+)' + % re.escape(count_name), + video_webpage, count_name, default=None)) + like_count = _extract_count('like') dislike_count = _extract_count('dislike') @@ -1008,7 +1092,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): self._downloader.report_warning('unable to extract video duration') video_duration = None else: - video_duration = int(compat_urllib_parse.unquote_plus(video_info['length_seconds'][0])) + video_duration = int(compat_urllib_parse_unquote_plus(video_info['length_seconds'][0])) # annotations video_annotations = None @@ -1111,23 +1195,32 @@ class YoutubeIE(YoutubeBaseInfoExtractor): # Look for the DASH manifest if self._downloader.params.get('youtube_include_dash_manifest', True): - dash_mpd = video_info.get('dashmpd') - if dash_mpd: - dash_manifest_url = dash_mpd[0] + dash_mpd_fatal = True + for dash_manifest_url in dash_mpds: + dash_formats = {} try: - dash_formats = self._parse_dash_manifest( - video_id, dash_manifest_url, player_url, age_gate) + for df in self._parse_dash_manifest( + video_id, dash_manifest_url, player_url, age_gate, dash_mpd_fatal): + # Do not overwrite DASH format found in some previous DASH manifest + if df['format_id'] not in dash_formats: + dash_formats[df['format_id']] = df + # Additional DASH manifests may end up in HTTP Error 403 therefore + # allow them to fail without bug report message if we already have + # some DASH manifest succeeded. This is temporary workaround to reduce + # burst of bug reports until we figure out the reason and whether it + # can be fixed at all. + dash_mpd_fatal = False except (ExtractorError, KeyError) as e: self.report_warning( 'Skipping DASH manifest: %r' % e, video_id) - else: - # Hide the formats we found through non-DASH - dash_keys = set(df['format_id'] for df in dash_formats) - for f in formats: - if f['format_id'] in dash_keys: - f['format_id'] = 'nondash-%s' % f['format_id'] - f['preference'] = f.get('preference', 0) - 10000 - formats.extend(dash_formats) + if dash_formats: + # Remove the formats we found through non-DASH, they + # contain less info and it can be wrong, because we use + # fixed values (for example the resolution). See + # https://github.com/rg3/youtube-dl/issues/5774 for an + # example. + formats = [f for f in formats if f['format_id'] not in dash_formats.keys()] + formats.extend(dash_formats.values()) # Check for malformed aspect ratio stretched_m = re.search( @@ -1161,6 +1254,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'dislike_count': dislike_count, 'average_rating': float_or_none(video_info.get('avg_rating', [None])[0]), 'formats': formats, + 'is_live': is_live, } @@ -1261,11 +1355,6 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor): def _real_initialize(self): self._login() - def _ids_to_results(self, ids): - return [ - self.url_result(vid_id, 'Youtube', video_id=vid_id) - for vid_id in ids] - def _extract_mix(self, playlist_id): # The mixes are generated from a single video # the id of the playlist is just 'RD' + video_id @@ -1289,7 +1378,6 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor): def _extract_playlist(self, playlist_id): url = self._TEMPLATE_URL % playlist_id page = self._download_webpage(url, playlist_id) - more_widget_html = content_html = page for match in re.findall(r'
([^<]+)
', page): match = match.strip() @@ -1309,36 +1397,36 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor): self.report_warning('Youtube gives an alert message: ' + match) # Extract the video ids from the playlist pages - ids = [] - - for page_num in itertools.count(1): - matches = re.finditer(self._VIDEO_RE, content_html) - # We remove the duplicates and the link with index 0 - # (it's not the first video of the playlist) - new_ids = orderedSet(m.group('id') for m in matches if m.group('index') != '0') - ids.extend(new_ids) - - mobj = re.search(r'data-uix-load-more-href="/?(?P[^"]+)"', more_widget_html) - if not mobj: - break + def _entries(): + more_widget_html = content_html = page + for page_num in itertools.count(1): + matches = re.finditer(self._VIDEO_RE, content_html) + # We remove the duplicates and the link with index 0 + # (it's not the first video of the playlist) + new_ids = orderedSet(m.group('id') for m in matches if m.group('index') != '0') + for vid_id in new_ids: + yield self.url_result(vid_id, 'Youtube', video_id=vid_id) + + mobj = re.search(r'data-uix-load-more-href="/?(?P[^"]+)"', more_widget_html) + if not mobj: + break - more = self._download_json( - 'https://youtube.com/%s' % mobj.group('more'), playlist_id, - 'Downloading page #%s' % page_num, - transform_source=uppercase_escape) - content_html = more['content_html'] - if not content_html.strip(): - # Some webpages show a "Load more" button but they don't - # have more videos - break - more_widget_html = more['load_more_widget_html'] + more = self._download_json( + 'https://youtube.com/%s' % mobj.group('more'), playlist_id, + 'Downloading page #%s' % page_num, + transform_source=uppercase_escape) + content_html = more['content_html'] + if not content_html.strip(): + # Some webpages show a "Load more" button but they don't + # have more videos + break + more_widget_html = more['load_more_widget_html'] playlist_title = self._html_search_regex( r'(?s)

\s*(.*?)\s*

', page, 'title') - url_results = self._ids_to_results(ids) - return self.playlist_result(url_results, playlist_id, playlist_title) + return self.playlist_result(_entries(), playlist_id, playlist_title) def _real_extract(self, url): # Extract playlist id @@ -1398,6 +1486,24 @@ class YoutubeChannelIE(InfoExtractor): channel_id = self._match_id(url) url = self._TEMPLATE_URL % channel_id + + # Channel by page listing is restricted to 35 pages of 30 items, i.e. 1050 videos total (see #5778) + # Workaround by extracting as a playlist if managed to obtain channel playlist URL + # otherwise fallback on channel by page extraction + channel_page = self._download_webpage( + url + '?view=57', channel_id, + 'Downloading channel page', fatal=False) + channel_playlist_id = self._html_search_meta( + 'channelId', channel_page, 'channel id', default=None) + if not channel_playlist_id: + channel_playlist_id = self._search_regex( + r'data-channel-external-id="([^"]+)"', + channel_page, 'channel id', default=None) + if channel_playlist_id and channel_playlist_id.startswith('UC'): + playlist_id = 'UU' + channel_playlist_id[2:] + return self.url_result( + compat_urlparse.urljoin(url, '/playlist?list=%s' % playlist_id), 'YoutubePlaylist') + channel_page = self._download_webpage(url, channel_id, 'Downloading page #1') autogenerated = re.search(r'''(?x) class="[^"]*?(?: @@ -1486,7 +1592,7 @@ class YoutubeSearchIE(SearchInfoExtractor, YoutubePlaylistIE): for pagenum in itertools.count(1): url_query = { - 'search_query': query, + 'search_query': query.encode('utf-8'), 'page': pagenum, 'spf': 'navigate', } @@ -1534,7 +1640,7 @@ class YoutubeSearchURLIE(InfoExtractor): def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) - query = compat_urllib_parse.unquote_plus(mobj.group('query')) + query = compat_urllib_parse_unquote_plus(mobj.group('query')) webpage = self._download_webpage(url, query) result_code = self._search_regex( @@ -1601,20 +1707,10 @@ class YoutubeShowIE(InfoExtractor): class YoutubeFeedsInfoExtractor(YoutubeBaseInfoExtractor): """ - Base class for extractors that fetch info from - http://www.youtube.com/feed_ajax + Base class for feed extractors Subclasses must define the _FEED_NAME and _PLAYLIST_TITLE properties. """ _LOGIN_REQUIRED = True - # use action_load_personal_feed instead of action_load_system_feed - _PERSONAL_FEED = False - - @property - def _FEED_TEMPLATE(self): - action = 'action_load_system_feed' - if self._PERSONAL_FEED: - action = 'action_load_personal_feed' - return 'https://www.youtube.com/feed_ajax?%s=1&feed_name=%s&paging=%%s' % (action, self._FEED_NAME) @property def IE_NAME(self): @@ -1624,67 +1720,23 @@ class YoutubeFeedsInfoExtractor(YoutubeBaseInfoExtractor): self._login() def _real_extract(self, url): - feed_entries = [] - paging = 0 - for i in itertools.count(1): - info = self._download_json( - self._FEED_TEMPLATE % paging, - '%s feed' % self._FEED_NAME, - 'Downloading page %s' % i, - transform_source=uppercase_escape) - feed_html = info.get('feed_html') or info.get('content_html') - load_more_widget_html = info.get('load_more_widget_html') or feed_html - m_ids = re.finditer(r'"/watch\?v=(.*?)["&]', feed_html) - ids = orderedSet(m.group(1) for m in m_ids) - feed_entries.extend( - self.url_result(video_id, 'Youtube', video_id=video_id) - for video_id in ids) - mobj = re.search( - r'data-uix-load-more-href="/?[^"]+paging=(?P\d+)', - load_more_widget_html) - if mobj is None: - break - paging = mobj.group('paging') - return self.playlist_result(feed_entries, playlist_title=self._PLAYLIST_TITLE) - - -class YoutubeRecommendedIE(YoutubeFeedsInfoExtractor): - IE_NAME = 'youtube:recommended' - IE_DESC = 'YouTube.com recommended videos, ":ytrec" for short (requires authentication)' - _VALID_URL = r'https?://www\.youtube\.com/feed/recommended|:ytrec(?:ommended)?' - _FEED_NAME = 'recommended' - _PLAYLIST_TITLE = 'Youtube Recommended videos' - - -class YoutubeWatchLaterIE(YoutubePlaylistIE): - IE_NAME = 'youtube:watchlater' - IE_DESC = 'Youtube watch later list, ":ytwatchlater" for short (requires authentication)' - _VALID_URL = r'https?://www\.youtube\.com/(?:feed/watch_later|playlist\?list=WL)|:ytwatchlater' - - _TESTS = [] # override PlaylistIE tests - - def _real_extract(self, url): - return self._extract_playlist('WL') - - -class YoutubeHistoryIE(YoutubePlaylistIE): - IE_NAME = 'youtube:history' - IE_DESC = 'Youtube watch history, ":ythistory" for short (requires authentication)' - _VALID_URL = 'https?://www\.youtube\.com/feed/history|:ythistory' - _TESTS = [] - - def _real_extract(self, url): - title = 'Youtube History' - page = self._download_webpage('https://www.youtube.com/feed/history', title) + page = self._download_webpage( + 'https://www.youtube.com/feed/%s' % self._FEED_NAME, self._PLAYLIST_TITLE) # The extraction process is the same as for playlists, but the regex # for the video ids doesn't contain an index ids = [] more_widget_html = content_html = page - for page_num in itertools.count(1): matches = re.findall(r'href="\s*/watch\?v=([0-9A-Za-z_-]{11})', content_html) - new_ids = orderedSet(matches) + + # 'recommended' feed has infinite 'load more' and each new portion spins + # the same videos in (sometimes) slightly different order, so we'll check + # for unicity and break when portion has no new videos + new_ids = filter(lambda video_id: video_id not in ids, orderedSet(matches)) + if not new_ids: + break + ids.extend(new_ids) mobj = re.search(r'data-uix-load-more-href="/?(?P[^"]+)"', more_widget_html) @@ -1692,17 +1744,25 @@ class YoutubeHistoryIE(YoutubePlaylistIE): break more = self._download_json( - 'https://youtube.com/%s' % mobj.group('more'), title, + 'https://youtube.com/%s' % mobj.group('more'), self._PLAYLIST_TITLE, 'Downloading page #%s' % page_num, transform_source=uppercase_escape) content_html = more['content_html'] more_widget_html = more['load_more_widget_html'] - return { - '_type': 'playlist', - 'title': title, - 'entries': self._ids_to_results(ids), - } + return self.playlist_result( + self._ids_to_results(ids), playlist_title=self._PLAYLIST_TITLE) + + +class YoutubeWatchLaterIE(YoutubePlaylistIE): + IE_NAME = 'youtube:watchlater' + IE_DESC = 'Youtube watch later list, ":ytwatchlater" for short (requires authentication)' + _VALID_URL = r'https?://www\.youtube\.com/(?:feed/watch_later|playlist\?list=WL)|:ytwatchlater' + + _TESTS = [] # override PlaylistIE tests + + def _real_extract(self, url): + return self._extract_playlist('WL') class YoutubeFavouritesIE(YoutubeBaseInfoExtractor): @@ -1717,42 +1777,25 @@ class YoutubeFavouritesIE(YoutubeBaseInfoExtractor): return self.url_result(playlist_id, 'YoutubePlaylist') -class YoutubeSubscriptionsIE(YoutubePlaylistIE): - IE_NAME = 'youtube:subscriptions' - IE_DESC = 'YouTube.com subscriptions feed, "ytsubs" keyword (requires authentication)' - _VALID_URL = r'https?://www\.youtube\.com/feed/subscriptions|:ytsubs(?:criptions)?' - _TESTS = [] - - def _real_extract(self, url): - title = 'Youtube Subscriptions' - page = self._download_webpage('https://www.youtube.com/feed/subscriptions', title) +class YoutubeRecommendedIE(YoutubeFeedsInfoExtractor): + IE_DESC = 'YouTube.com recommended videos, ":ytrec" for short (requires authentication)' + _VALID_URL = r'https?://www\.youtube\.com/feed/recommended|:ytrec(?:ommended)?' + _FEED_NAME = 'recommended' + _PLAYLIST_TITLE = 'Youtube Recommended videos' - # The extraction process is the same as for playlists, but the regex - # for the video ids doesn't contain an index - ids = [] - more_widget_html = content_html = page - for page_num in itertools.count(1): - matches = re.findall(r'href="\s*/watch\?v=([0-9A-Za-z_-]{11})', content_html) - new_ids = orderedSet(matches) - ids.extend(new_ids) +class YoutubeSubscriptionsIE(YoutubeFeedsInfoExtractor): + IE_DESC = 'YouTube.com subscriptions feed, "ytsubs" keyword (requires authentication)' + _VALID_URL = r'https?://www\.youtube\.com/feed/subscriptions|:ytsubs(?:criptions)?' + _FEED_NAME = 'subscriptions' + _PLAYLIST_TITLE = 'Youtube Subscriptions' - mobj = re.search(r'data-uix-load-more-href="/?(?P[^"]+)"', more_widget_html) - if not mobj: - break - more = self._download_json( - 'https://youtube.com/%s' % mobj.group('more'), title, - 'Downloading page #%s' % page_num, - transform_source=uppercase_escape) - content_html = more['content_html'] - more_widget_html = more['load_more_widget_html'] - - return { - '_type': 'playlist', - 'title': title, - 'entries': self._ids_to_results(ids), - } +class YoutubeHistoryIE(YoutubeFeedsInfoExtractor): + IE_DESC = 'Youtube watch history, ":ythistory" for short (requires authentication)' + _VALID_URL = 'https?://www\.youtube\.com/feed/history|:ythistory' + _FEED_NAME = 'history' + _PLAYLIST_TITLE = 'Youtube History' class YoutubeTruncatedURLIE(InfoExtractor):