diff options
Diffstat (limited to 'youtube_dl/extractor/youtube.py')
| -rw-r--r-- | youtube_dl/extractor/youtube.py | 519 |
1 files changed, 281 insertions, 238 deletions
diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 52909b0da..3d8b31f98 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, @@ -28,11 +30,12 @@ from ..utils import ( get_element_by_attribute, get_element_by_id, int_or_none, - OnDemandPagedList, orderedSet, + str_to_int, unescapeHTML, unified_strdate, uppercase_escape, + ISO3166Utils, ) @@ -50,6 +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. @@ -230,6 +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 @@ -512,6 +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 '<meta itemprop="datePublished"', but in other places it's 20150510 + 'uploader': 'Airtek', + 'description': 'Retransmisión en directo de la XVIII media maratón de Zaragoza.', + 'uploader_id': 'UCzTzUmjXxxacNnL8I3m4LnQ', + 'title': 'Retransmisión XVIII Media maratón Zaragoza 2015', + }, + 'params': { + 'youtube_include_dash_manifest': True, + 'format': '135', # bestvideo + } + }, ] def __init__(self, *args, **kwargs): @@ -776,16 +818,20 @@ class YoutubeIE(YoutubeBaseInfoExtractor): return self._download_webpage(url, video_id, note='Searching for annotations.', errnote='Unable to download video annotations.') def _parse_dash_manifest( - self, video_id, dash_manifest_url, player_url, age_gate): + self, video_id, dash_manifest_url, player_url, age_gate, fatal=True): def decrypt_sig(mobj): s = mobj.group(1) dec_s = self._decrypt_signature(s, video_id, player_url, age_gate) return '/signature/%s' % dec_s - dash_manifest_url = re.sub(r'/s/([\w\.]+)', decrypt_sig, dash_manifest_url) + dash_manifest_url = re.sub(r'/s/([a-fA-F0-9\.]+)', decrypt_sig, dash_manifest_url) dash_doc = self._download_xml( dash_manifest_url, video_id, note='Downloading DASH manifest', - errnote='Could not download DASH manifest') + errnote='Could not download DASH manifest', + fatal=fatal) + + if dash_doc is False: + return [] formats = [] for a in dash_doc.findall('.//{urn:mpeg:DASH:schema:MPD:2011}AdaptationSet'): @@ -798,6 +844,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): # TODO implement WebVTT downloading pass elif mime_type.startswith('audio/') or mime_type.startswith('video/'): + segment_list = r.find('{urn:mpeg:DASH:schema:MPD:2011}SegmentList') format_id = r.attrib['id'] video_url = url_el.text filesize = int_or_none(url_el.attrib.get('{http://youtube.com/yt/2012/10/10}contentLength')) @@ -811,6 +858,12 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'filesize': filesize, 'fps': int_or_none(r.attrib.get('frameRate')), } + if segment_list is not None: + f.update({ + 'initialization_url': segment_list.find('{urn:mpeg:DASH:schema:MPD:2011}Initialization').attrib['sourceURL'], + 'segment_urls': [segment.attrib.get('media') for segment in segment_list.findall('{urn:mpeg:DASH:schema:MPD:2011}SegmentURL')], + 'protocol': 'http_dash_segments', + }) try: existing_format = next( fo for fo in formats @@ -818,6 +871,12 @@ class YoutubeIE(YoutubeBaseInfoExtractor): except StopIteration: full_info = self._formats.get(format_id, {}).copy() full_info.update(f) + codecs = r.attrib.get('codecs') + if codecs: + if full_info.get('acodec') == 'none' and 'vcodec' not in full_info: + full_info['vcodec'] = codecs + elif full_info.get('vcodec') == 'none' and 'acodec' not in full_info: + full_info['acodec'] = codecs formats.append(full_info) else: existing_format.update(f) @@ -833,7 +892,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): # Extract original video URL from URL with redirection, like age verification, using next_url parameter mobj = re.search(self._NEXT_URL_RE, url) if mobj: - url = proto + '://www.youtube.com/' + compat_urllib_parse.unquote(mobj.group(1)).lstrip('/') + url = proto + '://www.youtube.com/' + compat_urllib_parse_unquote(mobj.group(1)).lstrip('/') video_id = self.extract_id(url) # Get video webpage @@ -847,8 +906,16 @@ class YoutubeIE(YoutubeBaseInfoExtractor): else: player_url = None + dash_mpds = [] + + def add_dash_mpd(video_info): + dash_mpd = video_info.get('dashmpd') + if dash_mpd and dash_mpd[0] not in dash_mpds: + dash_mpds.append(dash_mpd[0]) + # Get video info embed_webpage = None + is_live = None if re.search(r'player-age-gate-content">', video_webpage) is not None: age_gate = True # We simulate the access to the video from www.youtube.com/v/{video_id} @@ -867,24 +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)) @@ -892,11 +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) @@ -920,7 +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 @@ -947,18 +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.*?>(.*?)</span>', video_webpage) - if mobj is None: - mobj = re.search( - r'(?s)id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live) on (.*?)</strong>', - 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.*?>(.*?)</span>', + r'id="watch-uploader-info".*?>.*?(?:Published|Uploaded|Streamed live|Started) on (.+?)</strong>'], + 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)<h4[^>]*>\s*Category\s*</h4>\s*<ul[^>]*>(.*?)</ul>', @@ -992,12 +1076,11 @@ class YoutubeIE(YoutubeBaseInfoExtractor): video_description = '' def _extract_count(count_name): - count = self._search_regex( - r'id="watch-%s"[^>]*>.*?([\d,]+)\s*</span>' % 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[^>]+><span[^>]+class="yt-uix-button-content"[^>]*>([\d,]+)</span>' + % re.escape(count_name), + video_webpage, count_name, default=None)) + like_count = _extract_count('like') dislike_count = _extract_count('dislike') @@ -1009,7 +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 @@ -1112,23 +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( @@ -1162,6 +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, } @@ -1262,11 +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 @@ -1290,46 +1378,55 @@ 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 - # Check if the playlist exists or is private - if re.search(r'<div class="yt-alert-message">[^<]*?(The|This) playlist (does not exist|is private)[^<]*?</div>', page) is not None: - raise ExtractorError( - 'The playlist doesn\'t exist or is private, use --username or ' - '--netrc to access it.', - expected=True) + for match in re.findall(r'<div class="yt-alert-message">([^<]+)</div>', page): + match = match.strip() + # Check if the playlist exists or is private + if re.match(r'[^<]*(The|This) playlist (does not exist|is private)[^<]*', match): + raise ExtractorError( + 'The playlist doesn\'t exist or is private, use --username or ' + '--netrc to access it.', + expected=True) + elif re.match(r'[^<]*Invalid parameters[^<]*', match): + raise ExtractorError( + 'Invalid parameters. Maybe URL is incorrect.', + expected=True) + elif re.match(r'[^<]*Choose your language[^<]*', match): + continue + else: + 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>[^"]+)"', 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>[^"]+)"', 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)<h1 class="pl-header-title[^"]*">\s*(.*?)\s*</h1>', 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 @@ -1358,6 +1455,7 @@ class YoutubePlaylistIE(YoutubeBaseInfoExtractor): class YoutubeChannelIE(InfoExtractor): IE_DESC = 'YouTube.com channels' _VALID_URL = r'https?://(?:youtu\.be|(?:\w+\.)?youtube(?:-nocookie)?\.com)/channel/(?P<id>[0-9A-Za-z_-]+)' + _TEMPLATE_URL = 'https://www.youtube.com/channel/%s/videos' IE_NAME = 'youtube:channel' _TESTS = [{ 'note': 'paginated channel', @@ -1368,7 +1466,8 @@ class YoutubeChannelIE(InfoExtractor): } }] - def extract_videos_from_page(self, page): + @staticmethod + def extract_videos_from_page(page): ids_in_page = [] titles_in_page = [] for mobj in re.finditer(r'(?:title="(?P<title>[^"]+)"[^>]+)?href="/watch\?v=(?P<id>[0-9A-Za-z_-]+)&?', page): @@ -1386,8 +1485,26 @@ class YoutubeChannelIE(InfoExtractor): def _real_extract(self, url): channel_id = self._match_id(url) - url = 'https://www.youtube.com/channel/%s/videos' % channel_id - channel_page = self._download_webpage(url, channel_id) + 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="[^"]*?(?: channel-header-autogenerated-label| @@ -1429,12 +1546,10 @@ class YoutubeChannelIE(InfoExtractor): return self.playlist_result(_entries(), channel_id) -class YoutubeUserIE(InfoExtractor): +class YoutubeUserIE(YoutubeChannelIE): IE_DESC = 'YouTube.com user videos (URL or "ytuser" keyword)' _VALID_URL = r'(?:(?:(?:https?://)?(?:\w+\.)?youtube\.com/(?:user/)?(?!(?:attribution_link|watch|results)(?:$|[^a-z_A-Z0-9-])))|ytuser:)(?!feed/)(?P<id>[A-Za-z0-9_-]+)' - _TEMPLATE_URL = 'https://gdata.youtube.com/feeds/api/users/%s' - _GDATA_PAGE_SIZE = 50 - _GDATA_URL = 'https://gdata.youtube.com/feeds/api/users/%s/uploads?max-results=%d&start-index=%d&alt=json' + _TEMPLATE_URL = 'https://www.youtube.com/user/%s/videos' IE_NAME = 'youtube:user' _TESTS = [{ @@ -1458,95 +1573,57 @@ class YoutubeUserIE(InfoExtractor): else: return super(YoutubeUserIE, cls).suitable(url) - def _real_extract(self, url): - username = self._match_id(url) - - # Download video ids using YouTube Data API. Result size per - # query is limited (currently to 50 videos) so we need to query - # page by page until there are no video ids - it means we got - # all of them. - - def download_page(pagenum): - start_index = pagenum * self._GDATA_PAGE_SIZE + 1 - - gdata_url = self._GDATA_URL % (username, self._GDATA_PAGE_SIZE, start_index) - page = self._download_webpage( - gdata_url, username, - 'Downloading video ids from %d to %d' % ( - start_index, start_index + self._GDATA_PAGE_SIZE)) - - try: - response = json.loads(page) - except ValueError as err: - raise ExtractorError('Invalid JSON in API response: ' + compat_str(err)) - if 'entry' not in response['feed']: - return - - # Extract video identifiers - entries = response['feed']['entry'] - for entry in entries: - title = entry['title']['$t'] - video_id = entry['id']['$t'].split('/')[-1] - yield { - '_type': 'url', - 'url': video_id, - 'ie_key': 'Youtube', - 'id': video_id, - 'title': title, - } - url_results = OnDemandPagedList(download_page, self._GDATA_PAGE_SIZE) - - return self.playlist_result(url_results, playlist_title=username) - -class YoutubeSearchIE(SearchInfoExtractor): +class YoutubeSearchIE(SearchInfoExtractor, YoutubePlaylistIE): IE_DESC = 'YouTube.com searches' - _API_URL = 'https://gdata.youtube.com/feeds/api/videos?q=%s&start-index=%i&max-results=50&v=2&alt=jsonc' - _MAX_RESULTS = 1000 + # there doesn't appear to be a real limit, for example if you search for + # 'python' you get more than 8.000.000 results + _MAX_RESULTS = float('inf') IE_NAME = 'youtube:search' _SEARCH_KEY = 'ytsearch' + _EXTRA_QUERY_ARGS = {} + _TESTS = [] def _get_n_results(self, query, n): """Get a specified number of results for a query""" - video_ids = [] - pagenum = 0 + videos = [] limit = n - PAGE_SIZE = 50 - while (PAGE_SIZE * pagenum) < limit: - result_url = self._API_URL % ( - compat_urllib_parse.quote_plus(query.encode('utf-8')), - (PAGE_SIZE * pagenum) + 1) - data_json = self._download_webpage( + for pagenum in itertools.count(1): + url_query = { + 'search_query': query.encode('utf-8'), + 'page': pagenum, + 'spf': 'navigate', + } + url_query.update(self._EXTRA_QUERY_ARGS) + result_url = 'https://www.youtube.com/results?' + compat_urllib_parse.urlencode(url_query) + data = self._download_json( result_url, video_id='query "%s"' % query, - note='Downloading page %s' % (pagenum + 1), + note='Downloading page %s' % pagenum, errnote='Unable to download API page') - data = json.loads(data_json) - api_response = data['data'] + html_content = data[1]['body']['content'] - if 'items' not in api_response: + if 'class="search-message' in html_content: raise ExtractorError( '[youtube] No video results', expected=True) - new_ids = list(video['id'] for video in api_response['items']) - video_ids += new_ids - - limit = min(n, api_response['totalItems']) - pagenum += 1 + new_videos = self._ids_to_results(orderedSet(re.findall( + r'href="/watch\?v=(.{11})', html_content))) + videos += new_videos + if not new_videos or len(videos) > limit: + break - if len(video_ids) > n: - video_ids = video_ids[:n] - videos = [self.url_result(video_id, 'Youtube', video_id=video_id) - for video_id in video_ids] + if len(videos) > n: + videos = videos[:n] return self.playlist_result(videos, query) class YoutubeSearchDateIE(YoutubeSearchIE): IE_NAME = YoutubeSearchIE.IE_NAME + ':date' - _API_URL = 'https://gdata.youtube.com/feeds/api/videos?q=%s&start-index=%i&max-results=50&v=2&alt=jsonc&orderby=published' _SEARCH_KEY = 'ytsearchdate' IE_DESC = 'YouTube.com searches, newest videos first' + _EXTRA_QUERY_ARGS = {'search_sort': 'video_date_uploaded'} class YoutubeSearchURLIE(InfoExtractor): @@ -1563,7 +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( @@ -1630,20 +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): @@ -1653,36 +1720,38 @@ 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<paging>\d+)', - load_more_widget_html) - if mobj is None: + 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) + + # '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 - paging = mobj.group('paging') - return self.playlist_result(feed_entries, playlist_title=self._PLAYLIST_TITLE) + ids.extend(new_ids) -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' + mobj = re.search(r'data-uix-load-more-href="/?(?P<more>[^"]+)"', more_widget_html) + if not mobj: + break + + more = self._download_json( + '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 self.playlist_result( + self._ids_to_results(ids), playlist_title=self._PLAYLIST_TITLE) class YoutubeWatchLaterIE(YoutubePlaylistIE): @@ -1696,15 +1765,6 @@ class YoutubeWatchLaterIE(YoutubePlaylistIE): return self._extract_playlist('WL') -class YoutubeHistoryIE(YoutubeFeedsInfoExtractor): - IE_NAME = 'youtube:history' - IE_DESC = 'Youtube watch history, ":ythistory" for short (requires authentication)' - _VALID_URL = 'https?://www\.youtube\.com/feed/history|:ythistory' - _FEED_NAME = 'history' - _PERSONAL_FEED = True - _PLAYLIST_TITLE = 'Youtube Watch History' - - class YoutubeFavouritesIE(YoutubeBaseInfoExtractor): IE_NAME = 'youtube:favorites' IE_DESC = 'YouTube.com favourite videos, ":ytfav" for short (requires authentication)' @@ -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) - - # 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 +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' - 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) - mobj = re.search(r'data-uix-load-more-href="/?(?P<more>[^"]+)"', more_widget_html) - if not mobj: - break +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' - 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): |