X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/532a08904ffbacc5e5ccf99edb660c5f37ddb213..d0a866127feddafcff23f5eb8faa983b702f61c4:/youtube_dl/extractor/common.py diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index fec39da..fcdd0fd 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -27,8 +27,12 @@ from ..compat import ( compat_urllib_parse_urlencode, compat_urllib_request, compat_urlparse, + compat_xml_parse_error, +) +from ..downloader.f4m import ( + get_base_url, + remove_encrypted_media, ) -from ..downloader.f4m import remove_encrypted_media from ..utils import ( NO_DEFAULT, age_restricted, @@ -170,6 +174,8 @@ class InfoExtractor(object): width : height ratio as float. * no_resume The server does not support resuming the (HTTP or RTMP) download. Boolean. + * downloader_options A dictionary of downloader options as + described in FileDownloader url: Final video URL. ext: Video filename extension. @@ -297,8 +303,9 @@ class InfoExtractor(object): There must be a key "entries", which is a list, an iterable, or a PagedList object, each element of which is a valid dictionary by this specification. - Additionally, playlists can have "title", "description" and "id" attributes - with the same semantics as videos (see above). + Additionally, playlists can have "id", "title", "description", "uploader", + "uploader_id", "uploader_url" attributes with the same semantics as videos + (see above). _type "multi_video" indicates that there are multiple videos that @@ -376,7 +383,7 @@ class InfoExtractor(object): cls._VALID_URL_RE = re.compile(cls._VALID_URL) m = cls._VALID_URL_RE.match(url) assert m - return m.group('id') + return compat_str(m.group('id')) @classmethod def working(cls): @@ -420,7 +427,7 @@ class InfoExtractor(object): if country_code: self._x_forwarded_for_ip = GeoUtils.random_ipv4(country_code) if self._downloader.params.get('verbose', False): - self._downloader.to_stdout( + self._downloader.to_screen( '[debug] Using fake IP %s (%s) as X-Forwarded-For.' % (self._x_forwarded_for_ip, country_code.upper())) @@ -490,6 +497,16 @@ class InfoExtractor(object): self.to_screen('%s' % (note,)) else: self.to_screen('%s: %s' % (video_id, note)) + + # Some sites check X-Forwarded-For HTTP header in order to figure out + # the origin of the client behind proxy. This allows bypassing geo + # restriction by faking this header's value to IP that belongs to some + # geo unrestricted country. We will do so once we encounter any + # geo restriction error. + if self._x_forwarded_for_ip: + if 'X-Forwarded-For' not in headers: + headers['X-Forwarded-For'] = self._x_forwarded_for_ip + if isinstance(url_or_request, compat_urllib_request.Request): url_or_request = update_Request( url_or_request, data=data, headers=headers, query=query) @@ -519,15 +536,6 @@ class InfoExtractor(object): if isinstance(url_or_request, (compat_str, str)): url_or_request = url_or_request.partition('#')[0] - # Some sites check X-Forwarded-For HTTP header in order to figure out - # the origin of the client behind proxy. This allows bypassing geo - # restriction by faking this header's value to IP that belongs to some - # geo unrestricted country. We will do so once we encounter any - # geo restriction error. - if self._x_forwarded_for_ip: - if 'X-Forwarded-For' not in headers: - headers['X-Forwarded-For'] = self._x_forwarded_for_ip - urlh = self._request_webpage(url_or_request, video_id, note, errnote, fatal, data=data, headers=headers, query=query) if urlh is False: assert not fatal @@ -588,19 +596,11 @@ class InfoExtractor(object): if not encoding: encoding = self._guess_encoding_from_content(content_type, webpage_bytes) if self._downloader.params.get('dump_intermediate_pages', False): - try: - url = url_or_request.get_full_url() - except AttributeError: - url = url_or_request - self.to_screen('Dumping request to ' + url) + self.to_screen('Dumping request to ' + urlh.geturl()) dump = base64.b64encode(webpage_bytes).decode('ascii') self._downloader.to_screen(dump) if self._downloader.params.get('write_pages', False): - try: - url = url_or_request.get_full_url() - except AttributeError: - url = url_or_request - basen = '%s_%s' % (video_id, url) + basen = '%s_%s' % (video_id, urlh.geturl()) if len(basen) > 240: h = '___' + hashlib.md5(basen.encode('utf-8')).hexdigest() basen = basen[:240 - len(h)] + h @@ -646,15 +646,29 @@ class InfoExtractor(object): def _download_xml(self, url_or_request, video_id, note='Downloading XML', errnote='Unable to download XML', - transform_source=None, fatal=True, encoding=None, data=None, headers={}, query={}): + transform_source=None, fatal=True, encoding=None, + data=None, headers={}, query={}): """Return the xml as an xml.etree.ElementTree.Element""" xml_string = self._download_webpage( - url_or_request, video_id, note, errnote, fatal=fatal, encoding=encoding, data=data, headers=headers, query=query) + url_or_request, video_id, note, errnote, fatal=fatal, + encoding=encoding, data=data, headers=headers, query=query) if xml_string is False: return xml_string + return self._parse_xml( + xml_string, video_id, transform_source=transform_source, + fatal=fatal) + + def _parse_xml(self, xml_string, video_id, transform_source=None, fatal=True): if transform_source: xml_string = transform_source(xml_string) - return compat_etree_fromstring(xml_string.encode('utf-8')) + try: + return compat_etree_fromstring(xml_string.encode('utf-8')) + except compat_xml_parse_error as ve: + errmsg = '%s: Failed to parse XML ' % video_id + if fatal: + raise ExtractorError(errmsg, cause=ve) + else: + self.report_warning(errmsg + str(ve)) def _download_json(self, url_or_request, video_id, note='Downloading JSON metadata', @@ -730,12 +744,12 @@ class InfoExtractor(object): video_info['title'] = video_title return video_info - def playlist_from_matches(self, matches, video_id, video_title, getter=None, ie=None): - urlrs = orderedSet( + def playlist_from_matches(self, matches, playlist_id=None, playlist_title=None, getter=None, ie=None): + urls = orderedSet( self.url_result(self._proto_relative_url(getter(m) if getter else m), ie) for m in matches) return self.playlist_result( - urlrs, playlist_id=video_id, playlist_title=video_title) + urls, playlist_id=playlist_id, playlist_title=playlist_title) @staticmethod def playlist_result(entries, playlist_id=None, playlist_title=None, playlist_description=None): @@ -940,7 +954,8 @@ class InfoExtractor(object): def _family_friendly_search(self, html): # See http://schema.org/VideoObject - family_friendly = self._html_search_meta('isFamilyFriendly', html) + family_friendly = self._html_search_meta( + 'isFamilyFriendly', html, default=None) if not family_friendly: return None @@ -1002,19 +1017,19 @@ class InfoExtractor(object): item_type = e.get('@type') if expected_type is not None and expected_type != item_type: return info - if item_type == 'TVEpisode': + if item_type in ('TVEpisode', 'Episode'): info.update({ 'episode': unescapeHTML(e.get('name')), 'episode_number': int_or_none(e.get('episodeNumber')), 'description': unescapeHTML(e.get('description')), }) part_of_season = e.get('partOfSeason') - if isinstance(part_of_season, dict) and part_of_season.get('@type') == 'TVSeason': + if isinstance(part_of_season, dict) and part_of_season.get('@type') in ('TVSeason', 'Season', 'CreativeWorkSeason'): info['season_number'] = int_or_none(part_of_season.get('seasonNumber')) part_of_series = e.get('partOfSeries') or e.get('partOfTVSeries') - if isinstance(part_of_series, dict) and part_of_series.get('@type') == 'TVSeries': + if isinstance(part_of_series, dict) and part_of_series.get('@type') in ('TVSeries', 'Series', 'CreativeWorkSeries'): info['series'] = unescapeHTML(part_of_series.get('name')) - elif item_type == 'Article': + elif item_type in ('Article', 'NewsArticle'): info.update({ 'timestamp': parse_iso8601(e.get('datePublished')), 'title': unescapeHTML(e.get('headline')), @@ -1022,10 +1037,10 @@ class InfoExtractor(object): }) elif item_type == 'VideoObject': extract_video_object(e) - elif item_type == 'WebPage': - video = e.get('video') - if isinstance(video, dict) and video.get('@type') == 'VideoObject': - extract_video_object(video) + continue + video = e.get('video') + if isinstance(video, dict) and video.get('@type') == 'VideoObject': + extract_video_object(video) break return dict((k, v) for k, v in info.items() if v is not None) @@ -1223,11 +1238,8 @@ class InfoExtractor(object): media_nodes = remove_encrypted_media(media_nodes) if not media_nodes: return formats - base_url = xpath_text( - manifest, ['{http://ns.adobe.com/f4m/1.0}baseURL', '{http://ns.adobe.com/f4m/2.0}baseURL'], - 'base URL', default=None) - if base_url: - base_url = base_url.strip() + + manifest_base_url = get_base_url(manifest) bootstrap_info = xpath_element( manifest, ['{http://ns.adobe.com/f4m/1.0}bootstrapInfo', '{http://ns.adobe.com/f4m/2.0}bootstrapInfo'], @@ -1259,7 +1271,7 @@ class InfoExtractor(object): continue manifest_url = ( media_url if media_url.startswith('http://') or media_url.startswith('https://') - else ((base_url or '/'.join(manifest_url.split('/')[:-1])) + '/' + media_url)) + else ((manifest_base_url or '/'.join(manifest_url.split('/')[:-1])) + '/' + media_url)) # If media_url is itself a f4m manifest do the recursive extraction # since bitrates in parent manifest (this one) and media_url manifest # may differ leading to inability to resolve the format by requested @@ -1294,6 +1306,7 @@ class InfoExtractor(object): 'url': manifest_url, 'manifest_url': manifest_url, 'ext': 'flv' if bootstrap_info is not None else None, + 'protocol': 'f4m', 'tbr': tbr, 'width': width, 'height': height, @@ -1339,6 +1352,9 @@ class InfoExtractor(object): if '#EXT-X-FAXS-CM:' in m3u8_doc: # Adobe Flash Access return [] + if re.search(r'#EXT-X-SESSION-KEY:.*?URI="skd://', m3u8_doc): # Apple FairPlay + return [] + formats = [] format_url = lambda u: ( @@ -1385,7 +1401,7 @@ class InfoExtractor(object): media_url = media.get('URI') if media_url: format_id = [] - for v in (group_id, name): + for v in (m3u8_id, group_id, name): if v: format_id.append(v) f = { @@ -1785,7 +1801,7 @@ class InfoExtractor(object): ms_info['timescale'] = int(timescale) segment_duration = source.get('duration') if segment_duration: - ms_info['segment_duration'] = int(segment_duration) + ms_info['segment_duration'] = float(segment_duration) def extract_Initialization(source): initialization = source.find(_add_ns('Initialization')) @@ -1866,6 +1882,7 @@ class InfoExtractor(object): 'language': lang if lang not in ('mul', 'und', 'zxx', 'mis') else None, 'format_note': 'DASH %s' % content_type, 'filesize': filesize, + 'container': mimetype2ext(mime_type) + '_dash', } f.update(parse_codecs(representation_attrib.get('codecs'))) representation_ms_info = extract_multisegment_info(representation, adaption_set_ms_info) @@ -1892,19 +1909,23 @@ class InfoExtractor(object): 'Bandwidth': bandwidth, } + def location_key(location): + return 'url' if re.match(r'^https?://', location) else 'path' + if 'segment_urls' not in representation_ms_info and 'media' in representation_ms_info: media_template = prepare_template('media', ('Number', 'Bandwidth', 'Time')) + media_location_key = location_key(media_template) # As per [1, 5.3.9.4.4, Table 16, page 55] $Number$ and $Time$ # can't be used at the same time if '%(Number' in media_template and 's' not in representation_ms_info: segment_duration = None - if 'total_number' not in representation_ms_info and 'segment_duration': + if 'total_number' not in representation_ms_info and 'segment_duration' in representation_ms_info: segment_duration = float_or_none(representation_ms_info['segment_duration'], representation_ms_info['timescale']) representation_ms_info['total_number'] = int(math.ceil(float(period_duration) / segment_duration)) representation_ms_info['fragments'] = [{ - 'url': media_template % { + media_location_key: media_template % { 'Number': segment_number, 'Bandwidth': bandwidth, }, @@ -1928,7 +1949,7 @@ class InfoExtractor(object): 'Number': segment_number, } representation_ms_info['fragments'].append({ - 'url': segment_url, + media_location_key: segment_url, 'duration': float_or_none(segment_d, representation_ms_info['timescale']), }) @@ -1952,16 +1973,34 @@ class InfoExtractor(object): for s in representation_ms_info['s']: duration = float_or_none(s['d'], timescale) for r in range(s.get('r', 0) + 1): + segment_uri = representation_ms_info['segment_urls'][segment_index] fragments.append({ - 'url': representation_ms_info['segment_urls'][segment_index], + location_key(segment_uri): segment_uri, 'duration': duration, }) segment_index += 1 representation_ms_info['fragments'] = fragments + elif 'segment_urls' in representation_ms_info: + # Segment URLs with no SegmentTimeline + # Example: https://www.seznam.cz/zpravy/clanek/cesko-zasahne-vitr-o-sile-vichrice-muze-byt-i-zivotu-nebezpecny-39091 + # https://github.com/rg3/youtube-dl/pull/14844 + fragments = [] + segment_duration = float_or_none( + representation_ms_info['segment_duration'], + representation_ms_info['timescale']) if 'segment_duration' in representation_ms_info else None + for segment_url in representation_ms_info['segment_urls']: + fragment = { + location_key(segment_url): segment_url, + } + if segment_duration: + fragment['duration'] = segment_duration + fragments.append(fragment) + representation_ms_info['fragments'] = fragments # NB: MPD manifest may contain direct URLs to unfragmented media. # No fragments key is present in this case. if 'fragments' in representation_ms_info: f.update({ + 'fragment_base_url': base_url, 'fragments': [], 'protocol': 'http_dash_segments', }) @@ -1969,20 +2008,16 @@ class InfoExtractor(object): initialization_url = representation_ms_info['initialization_url'] if not f.get('url'): f['url'] = initialization_url - f['fragments'].append({'url': initialization_url}) + f['fragments'].append({location_key(initialization_url): initialization_url}) f['fragments'].extend(representation_ms_info['fragments']) - for fragment in f['fragments']: - fragment['url'] = urljoin(base_url, fragment['url']) - try: - existing_format = next( - fo for fo in formats - if fo['format_id'] == representation_id) - except StopIteration: - full_info = formats_dict.get(representation_id, {}).copy() - full_info.update(f) - formats.append(full_info) - else: - existing_format.update(f) + # According to [1, 5.3.5.2, Table 7, page 35] @id of Representation + # is not necessarily unique within a Period thus formats with + # the same `format_id` are quite possible. There are numerous examples + # of such manifests (see https://github.com/rg3/youtube-dl/issues/15111, + # https://github.com/rg3/youtube-dl/issues/13919) + full_info = formats_dict.get(representation_id, {}).copy() + full_info.update(f) + formats.append(full_info) else: self.report_warning('Unknown MIME type %s in DASH manifest' % mime_type) return formats @@ -2022,7 +2057,7 @@ class InfoExtractor(object): stream_timescale = int_or_none(stream.get('TimeScale')) or timescale stream_name = stream.get('Name') for track in stream.findall('QualityLevel'): - fourcc = track.get('FourCC') + fourcc = track.get('FourCC', 'AACL' if track.get('AudioTag') == '255' else None) # TODO: add support for WVC1 and WMAP if fourcc not in ('H264', 'AVC1', 'AACL'): self.report_warning('%s is not a supported codec' % fourcc) @@ -2110,19 +2145,19 @@ class InfoExtractor(object): return f return {} - def _media_formats(src, cur_media_type): + def _media_formats(src, cur_media_type, type_info={}): full_url = absolute_url(src) - ext = determine_ext(full_url) + ext = type_info.get('ext') or determine_ext(full_url) if ext == 'm3u8': is_plain_url = False formats = self._extract_m3u8_formats( full_url, video_id, ext='mp4', entry_protocol=m3u8_entry_protocol, m3u8_id=m3u8_id, - preference=preference) + preference=preference, fatal=False) elif ext == 'mpd': is_plain_url = False formats = self._extract_mpd_formats( - full_url, video_id, mpd_id=mpd_id) + full_url, video_id, mpd_id=mpd_id, fatal=False) else: is_plain_url = True formats = [{ @@ -2132,15 +2167,18 @@ class InfoExtractor(object): return is_plain_url, formats entries = [] + # amp-video and amp-audio are very similar to their HTML5 counterparts + # so we wll include them right here (see + # https://www.ampproject.org/docs/reference/components/amp-video) media_tags = [(media_tag, media_type, '') for media_tag, media_type - in re.findall(r'(?s)(<(video|audio)[^>]*/>)', webpage)] + in re.findall(r'(?s)(<(?:amp-)?(video|audio)[^>]*/>)', webpage)] media_tags.extend(re.findall( # We only allow video|audio followed by a whitespace or '>'. # Allowing more characters may end up in significant slow down (see # https://github.com/rg3/youtube-dl/issues/11979, example URL: # http://www.porntrex.com/maps/videositemap.xml). - r'(?s)(<(?Pvideo|audio)(?:\s+[^>]*)?>)(.*?)', webpage)) + r'(?s)(<(?P(?:amp-)?(?:video|audio))(?:\s+[^>]*)?>)(.*?)', webpage)) for media_tag, media_type, media_content in media_tags: media_info = { 'formats': [], @@ -2158,9 +2196,15 @@ class InfoExtractor(object): src = source_attributes.get('src') if not src: continue - is_plain_url, formats = _media_formats(src, media_type) + f = parse_content_type(source_attributes.get('type')) + is_plain_url, formats = _media_formats(src, media_type, f) if is_plain_url: - f = parse_content_type(source_attributes.get('type')) + # res attribute is not standard but seen several times + # in the wild + f.update({ + 'height': int_or_none(source_attributes.get('res')), + 'format_id': source_attributes.get('label'), + }) f.update(formats[0]) media_info['formats'].append(f) else: @@ -2204,27 +2248,36 @@ class InfoExtractor(object): return formats def _extract_wowza_formats(self, url, video_id, m3u8_entry_protocol='m3u8_native', skip_protocols=[]): + query = compat_urlparse.urlparse(url).query url = re.sub(r'/(?:manifest|playlist|jwplayer)\.(?:m3u8|f4m|mpd|smil)', '', url) - url_base = self._search_regex( - r'(?:(?:https?|rtmp|rtsp):)?(//[^?]+)', url, 'format url') - http_base_url = '%s:%s' % ('http', url_base) + mobj = re.search( + r'(?:(?:http|rtmp|rtsp)(?Ps)?:)?(?P//[^?]+)', url) + url_base = mobj.group('url') + http_base_url = '%s%s:%s' % ('http', mobj.group('s') or '', url_base) formats = [] + + def manifest_url(manifest): + m_url = '%s/%s' % (http_base_url, manifest) + if query: + m_url += '?%s' % query + return m_url + if 'm3u8' not in skip_protocols: formats.extend(self._extract_m3u8_formats( - http_base_url + '/playlist.m3u8', video_id, 'mp4', + manifest_url('playlist.m3u8'), video_id, 'mp4', m3u8_entry_protocol, m3u8_id='hls', fatal=False)) if 'f4m' not in skip_protocols: formats.extend(self._extract_f4m_formats( - http_base_url + '/manifest.f4m', + manifest_url('manifest.f4m'), video_id, f4m_id='hds', fatal=False)) if 'dash' not in skip_protocols: formats.extend(self._extract_mpd_formats( - http_base_url + '/manifest.mpd', + manifest_url('manifest.mpd'), video_id, mpd_id='dash', fatal=False)) if re.search(r'(?:/smil:|\.smil)', url_base): if 'smil' not in skip_protocols: rtmp_formats = self._extract_smil_formats( - http_base_url + '/jwplayer.smil', + manifest_url('jwplayer.smil'), video_id, fatal=False) for rtmp_format in rtmp_formats: rtsp_format = rtmp_format.copy() @@ -2293,13 +2346,17 @@ class InfoExtractor(object): formats = self._parse_jwplayer_formats( video_data['sources'], video_id=this_video_id, m3u8_id=m3u8_id, mpd_id=mpd_id, rtmp_params=rtmp_params, base_url=base_url) - self._sort_formats(formats) subtitles = {} tracks = video_data.get('tracks') if tracks and isinstance(tracks, list): for track in tracks: - if track.get('kind') != 'captions': + if not isinstance(track, dict): + continue + track_kind = track.get('kind') + if not track_kind or not isinstance(track_kind, compat_str): + continue + if track_kind.lower() not in ('captions', 'subtitles'): continue track_url = urljoin(base_url, track.get('file')) if not track_url: @@ -2308,16 +2365,25 @@ class InfoExtractor(object): 'url': self._proto_relative_url(track_url) }) - entries.append({ + entry = { 'id': this_video_id, - 'title': video_data['title'] if require_title else video_data.get('title'), + 'title': unescapeHTML(video_data['title'] if require_title else video_data.get('title')), 'description': video_data.get('description'), 'thumbnail': self._proto_relative_url(video_data.get('image')), 'timestamp': int_or_none(video_data.get('pubdate')), 'duration': float_or_none(jwplayer_data.get('duration') or video_data.get('duration')), 'subtitles': subtitles, - 'formats': formats, - }) + } + # https://github.com/jwplayer/jwplayer/blob/master/src/js/utils/validator.js#L32 + if len(formats) == 1 and re.search(r'^(?:http|//).*(?:youtube\.com|youtu\.be)/.+', formats[0]['url']): + entry.update({ + '_type': 'url_transparent', + 'url': formats[0]['url'], + }) + else: + self._sort_formats(formats) + entry['formats'] = formats + entries.append(entry) if len(entries) == 1: return entries[0] else: @@ -2328,6 +2394,8 @@ class InfoExtractor(object): urls = [] formats = [] for source in jwplayer_sources_data: + if not isinstance(source, dict): + continue source_url = self._proto_relative_url(source.get('file')) if not source_url: continue @@ -2342,7 +2410,7 @@ class InfoExtractor(object): formats.extend(self._extract_m3u8_formats( source_url, video_id, 'mp4', entry_protocol='m3u8_native', m3u8_id=m3u8_id, fatal=False)) - elif ext == 'mpd': + elif source_type == 'dash' or ext == 'mpd': formats.extend(self._extract_mpd_formats( source_url, video_id, mpd_id=mpd_id, fatal=False)) elif ext == 'smil': @@ -2416,10 +2484,12 @@ class InfoExtractor(object): self._downloader.report_warning(msg) return res - def _set_cookie(self, domain, name, value, expire_time=None): + def _set_cookie(self, domain, name, value, expire_time=None, port=None, + path='/', secure=False, discard=False, rest={}, **kwargs): cookie = compat_cookiejar.Cookie( - 0, name, value, None, None, domain, None, - None, '/', True, False, expire_time, '', None, None, None) + 0, name, value, port, port is not None, domain, True, + domain.startswith('.'), path, True, secure, expire_time, + discard, None, None, rest) self._downloader.cookiejar.set_cookie(cookie) def _get_cookies(self, url):