X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fcommon.py;h=641e50f3c51fc5d5ae086ce327a82439fdc9fbef;hb=c790e93ab5db5f318fb094b8a45f9160cdf4bd9b;hp=8eab5947f8ff2566628ac49b8813d33e18c302a5;hpb=6f1f59f39cf934d5ccae3ca2e6fd4eaab726137f;p=youtube-dl diff --git a/youtube_dl/extractor/common.py b/youtube_dl/extractor/common.py index 8eab5947f..641e50f3c 100644 --- a/youtube_dl/extractor/common.py +++ b/youtube_dl/extractor/common.py @@ -69,6 +69,7 @@ from ..utils import ( update_url_query, urljoin, url_basename, + url_or_none, xpath_element, xpath_text, xpath_with_ns, @@ -101,10 +102,20 @@ class InfoExtractor(object): from worst to best quality. Potential fields: - * url Mandatory. The URL of the video file + * url The mandatory URL representing the media: + for plain file media - HTTP URL of this file, + for RTMP - RTMP URL, + for HLS - URL of the M3U8 media playlist, + for HDS - URL of the F4M manifest, + for DASH - URL of the MPD manifest, + for MSS - URL of the ISM manifest. * manifest_url The URL of the manifest file in case of - fragmented media (DASH, hls, hds) + fragmented media: + for HLS - URL of the M3U8 master playlist, + for HDS - URL of the F4M manifest, + for DASH - URL of the MPD manifest, + for MSS - URL of the ISM manifest. * ext Will be calculated from URL if missing * format A human-readable description of the format ("mp4 container with h264/opus"). @@ -212,7 +223,7 @@ class InfoExtractor(object): uploader_id: Nickname or id of the video uploader. uploader_url: Full URL to a personal webpage of the video uploader. channel: Full name of the channel the video is uploaded on. - Note that channel fields may or may noy repeat uploader + Note that channel fields may or may not repeat uploader fields. This depends on a particular extractor. channel_id: Id of the channel. channel_url: Full URL to a channel webpage. @@ -605,6 +616,11 @@ class InfoExtractor(object): except (compat_urllib_error.URLError, compat_http_client.HTTPException, socket.error) as err: if isinstance(err, compat_urllib_error.HTTPError): if self.__can_accept_status_code(err, expected_status): + # Retain reference to error to prevent file object from + # being closed before it can be read. Works around the + # effects of + # introduced in Python 3.4.1. + err.fp._error = err return err.fp if errnote is False: @@ -1052,7 +1068,7 @@ class InfoExtractor(object): @staticmethod def _og_regexes(prop): content_re = r'content=(?:"([^"]+?)"|\'([^\']+?)\'|\s*([^\s"\'=<>`]+?))' - property_re = (r'(?:name|property)=(?:\'og:%(prop)s\'|"og:%(prop)s"|\s*og:%(prop)s\b)' + property_re = (r'(?:name|property)=(?:\'og[:-]%(prop)s\'|"og[:-]%(prop)s"|\s*og[:-]%(prop)s\b)' % {'prop': re.escape(prop)}) template = r']+?%s[^>]+?%s' return [ @@ -1213,10 +1229,10 @@ class InfoExtractor(object): def extract_video_object(e): assert e['@type'] == 'VideoObject' info.update({ - 'url': e.get('contentUrl'), + 'url': url_or_none(e.get('contentUrl')), 'title': unescapeHTML(e.get('name')), 'description': unescapeHTML(e.get('description')), - 'thumbnail': e.get('thumbnailUrl') or e.get('thumbnailURL'), + 'thumbnail': url_or_none(e.get('thumbnailUrl') or e.get('thumbnailURL')), 'duration': parse_duration(e.get('duration')), 'timestamp': unified_timestamp(e.get('uploadDate')), 'filesize': float_or_none(e.get('contentSize')), @@ -1233,17 +1249,30 @@ class InfoExtractor(object): if expected_type is not None and expected_type != item_type: return info if item_type in ('TVEpisode', 'Episode'): + episode_name = unescapeHTML(e.get('name')) info.update({ - 'episode': unescapeHTML(e.get('name')), + 'episode': episode_name, 'episode_number': int_or_none(e.get('episodeNumber')), 'description': unescapeHTML(e.get('description')), }) + if not info.get('title') and episode_name: + info['title'] = episode_name part_of_season = e.get('partOfSeason') 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')) + info.update({ + 'season': unescapeHTML(part_of_season.get('name')), + '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') in ('TVSeries', 'Series', 'CreativeWorkSeries'): info['series'] = unescapeHTML(part_of_series.get('name')) + elif item_type == 'Movie': + info.update({ + 'title': unescapeHTML(e.get('name')), + 'description': unescapeHTML(e.get('description')), + 'duration': parse_duration(e.get('duration')), + 'timestamp': unified_timestamp(e.get('dateCreated')), + }) elif item_type in ('Article', 'NewsArticle'): info.update({ 'timestamp': parse_iso8601(e.get('datePublished')), @@ -1580,6 +1609,7 @@ class InfoExtractor(object): # References: # 1. https://tools.ietf.org/html/draft-pantos-http-live-streaming-21 # 2. https://github.com/rg3/youtube-dl/issues/12211 + # 3. https://github.com/rg3/youtube-dl/issues/18923 # We should try extracting formats only from master playlists [1, 4.3.4], # i.e. playlists that describe available qualities. On the other hand @@ -1651,11 +1681,16 @@ class InfoExtractor(object): rendition = stream_group[0] return rendition.get('NAME') or stream_group_id + # parse EXT-X-MEDIA tags before EXT-X-STREAM-INF in order to have the + # chance to detect video only formats when EXT-X-STREAM-INF tags + # precede EXT-X-MEDIA tags in HLS manifest such as [3]. + for line in m3u8_doc.splitlines(): + if line.startswith('#EXT-X-MEDIA:'): + extract_media(line) + for line in m3u8_doc.splitlines(): if line.startswith('#EXT-X-STREAM-INF:'): last_stream_inf = parse_m3u8_attributes(line) - elif line.startswith('#EXT-X-MEDIA:'): - extract_media(line) elif line.startswith('#') or not line.strip(): continue else: @@ -1706,9 +1741,9 @@ class InfoExtractor(object): # However, this is not always respected, for example, [2] # contains EXT-X-STREAM-INF tag which references AUDIO # rendition group but does not have CODECS and despite - # referencing audio group an audio group, it represents - # a complete (with audio and video) format. So, for such cases - # we will ignore references to rendition groups and treat them + # referencing an audio group it represents a complete + # (with audio and video) format. So, for such cases we will + # ignore references to rendition groups and treat them # as complete formats. if audio_group_id and codecs and f.get('vcodec') != 'none': audio_group = groups.get(audio_group_id) @@ -2095,7 +2130,7 @@ class InfoExtractor(object): bandwidth = int_or_none(representation_attrib.get('bandwidth')) f = { 'format_id': '%s-%s' % (mpd_id, representation_id) if mpd_id else representation_id, - 'url': base_url, + 'url': mpd_url, 'manifest_url': mpd_url, 'ext': mimetype2ext(mime_type), 'width': int_or_none(representation_attrib.get('width')), @@ -2608,7 +2643,7 @@ class InfoExtractor(object): 'id': this_video_id, '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')), + 'thumbnail': urljoin(base_url, 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, @@ -2635,12 +2670,9 @@ class InfoExtractor(object): 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 - if base_url: - source_url = compat_urlparse.urljoin(base_url, source_url) - if source_url in urls: + source_url = urljoin( + base_url, self._proto_relative_url(source.get('file'))) + if not source_url or source_url in urls: continue urls.append(source_url) source_type = source.get('type') or ''