X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fdrtv.py;h=e966d7483cdc2193cfc96d2bdd808c90e515f821;hb=c73e330e7adc9c0c15ac51aeea8fbb7dad95351a;hp=e210cb6106020feb4b48b8429f2ee391f4a10e81;hpb=5e9e3d0f6bf2055c557f360758d6d7eb146edcba;p=youtube-dl diff --git a/youtube_dl/extractor/drtv.py b/youtube_dl/extractor/drtv.py index e210cb610..e966d7483 100644 --- a/youtube_dl/extractor/drtv.py +++ b/youtube_dl/extractor/drtv.py @@ -4,30 +4,36 @@ from __future__ import unicode_literals from .common import InfoExtractor from ..utils import ( ExtractorError, + int_or_none, + float_or_none, + mimetype2ext, parse_iso8601, remove_end, + update_url_query, ) class DRTVIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?dr\.dk/(?:tv/se|nyheder)/(?:[^/]+/)*(?P[\da-z-]+)(?:[/#?]|$)' - + _VALID_URL = r'https?://(?:www\.)?dr\.dk/(?:tv/se|nyheder|radio/ondemand)/(?:[^/]+/)*(?P[\da-z-]+)(?:[/#?]|$)' + IE_NAME = 'drtv' _TESTS = [{ - 'url': 'https://www.dr.dk/tv/se/boern/ultra/panisk-paske/panisk-paske-5', - 'md5': 'dc515a9ab50577fa14cc4e4b0265168f', + 'url': 'https://www.dr.dk/tv/se/boern/ultra/klassen-ultra/klassen-darlig-taber-10', + 'md5': '25e659cccc9a2ed956110a299fdf5983', 'info_dict': { - 'id': 'panisk-paske-5', + 'id': 'klassen-darlig-taber-10', 'ext': 'mp4', - 'title': 'Panisk PÃ¥ske (5)', - 'description': 'md5:ca14173c5ab24cd26b0fcc074dff391c', - 'timestamp': 1426984612, - 'upload_date': '20150322', - 'duration': 1455, + 'title': 'Klassen - DÃ¥rlig taber (10)', + 'description': 'md5:815fe1b7fa656ed80580f31e8b3c79aa', + 'timestamp': 1471991907, + 'upload_date': '20160823', + 'duration': 606.84, + }, + 'params': { + 'skip_download': True, }, - 'skip': 'Video is no longer available', }, { 'url': 'https://www.dr.dk/nyheder/indland/live-christianias-rydning-af-pusher-street-er-i-gang', - 'md5': '2ada5074f9e79afc0d324a8e9784d850', + 'md5': '2c37175c718155930f939ef59952474a', 'info_dict': { 'id': 'christiania-pusher-street-ryddes-drdkrjpo', 'ext': 'mp4', @@ -36,7 +42,7 @@ class DRTVIE(InfoExtractor): 'timestamp': 1472800279, 'upload_date': '20160902', 'duration': 131.4, - } + }, }] def _real_extract(self, url): @@ -58,10 +64,12 @@ class DRTVIE(InfoExtractor): video_id, 'Downloading video JSON') data = programcard['Data'][0] - title = remove_end(self._og_search_title(webpage), ' | TV | DR') or data['Title'] - description = self._og_search_description(webpage) or data['Description'] + title = remove_end(self._og_search_title( + webpage, default=None), ' | TV | DR') or data['Title'] + description = self._og_search_description( + webpage, default=None) or data.get('Description') - timestamp = parse_iso8601(data['CreatedTime']) + timestamp = parse_iso8601(data.get('CreatedTime')) thumbnail = None duration = None @@ -72,28 +80,35 @@ class DRTVIE(InfoExtractor): subtitles = {} for asset in data['Assets']: - if asset['Kind'] == 'Image': - thumbnail = asset['Uri'] - elif asset['Kind'] == 'VideoResource': - duration = asset['DurationInMilliseconds'] / 1000.0 - restricted_to_denmark = asset['RestrictedToDenmark'] - spoken_subtitles = asset['Target'] == 'SpokenSubtitles' - for link in asset['Links']: - uri = link['Uri'] - target = link['Target'] - format_id = target + kind = asset.get('Kind') + if kind == 'Image': + thumbnail = asset.get('Uri') + elif kind in ('VideoResource', 'AudioResource'): + duration = float_or_none(asset.get('DurationInMilliseconds'), 1000) + restricted_to_denmark = asset.get('RestrictedToDenmark') + spoken_subtitles = asset.get('Target') == 'SpokenSubtitles' + for link in asset.get('Links', []): + uri = link.get('Uri') + if not uri: + continue + target = link.get('Target') + format_id = target or '' preference = None if spoken_subtitles: preference = -1 format_id += '-spoken-subtitles' if target == 'HDS': - formats.extend(self._extract_f4m_formats( + f4m_formats = self._extract_f4m_formats( uri + '?hdcore=3.3.0&plugin=aasp-3.3.0.99.43', - video_id, preference, f4m_id=format_id)) + video_id, preference, f4m_id=format_id) + if kind == 'AudioResource': + for f in f4m_formats: + f['vcodec'] = 'none' + formats.extend(f4m_formats) elif target == 'HLS': formats.extend(self._extract_m3u8_formats( - uri, video_id, 'mp4', preference=preference, - m3u8_id=format_id)) + uri, video_id, 'mp4', entry_protocol='m3u8_native', + preference=preference, m3u8_id=format_id)) else: bitrate = link.get('Bitrate') if bitrate: @@ -101,8 +116,9 @@ class DRTVIE(InfoExtractor): formats.append({ 'url': uri, 'format_id': format_id, - 'tbr': bitrate, + 'tbr': int_or_none(bitrate), 'ext': link.get('FileFormat'), + 'vcodec': 'none' if kind == 'AudioResource' else None, }) subtitles_list = asset.get('SubtitlesList') if isinstance(subtitles_list, list): @@ -110,12 +126,18 @@ class DRTVIE(InfoExtractor): 'Danish': 'da', } for subs in subtitles_list: - lang = subs['Language'] - subtitles[LANGS.get(lang, lang)] = [{'url': subs['Uri'], 'ext': 'vtt'}] + if not subs.get('Uri'): + continue + lang = subs.get('Language') or 'da' + subtitles.setdefault(LANGS.get(lang, lang), []).append({ + 'url': subs['Uri'], + 'ext': mimetype2ext(subs.get('MimeType')) or 'vtt' + }) if not formats and restricted_to_denmark: - raise ExtractorError( - 'Unfortunately, DR is not allowed to show this program outside Denmark.', expected=True) + self.raise_geo_restricted( + 'Unfortunately, DR is not allowed to show this program outside Denmark.', + expected=True) self._sort_formats(formats) @@ -129,3 +151,58 @@ class DRTVIE(InfoExtractor): 'formats': formats, 'subtitles': subtitles, } + + +class DRTVLiveIE(InfoExtractor): + IE_NAME = 'drtv:live' + _VALID_URL = r'https?://(?:www\.)?dr\.dk/(?:tv|TV)/live/(?P[\da-z-]+)' + _TEST = { + 'url': 'https://www.dr.dk/tv/live/dr1', + 'info_dict': { + 'id': 'dr1', + 'ext': 'mp4', + 'title': 're:^DR1 [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$', + }, + 'params': { + # m3u8 download + 'skip_download': True, + }, + } + + def _real_extract(self, url): + channel_id = self._match_id(url) + channel_data = self._download_json( + 'https://www.dr.dk/mu-online/api/1.0/channel/' + channel_id, + channel_id) + title = self._live_title(channel_data['Title']) + + formats = [] + for streaming_server in channel_data.get('StreamingServers', []): + server = streaming_server.get('Server') + if not server: + continue + link_type = streaming_server.get('LinkType') + for quality in streaming_server.get('Qualities', []): + for stream in quality.get('Streams', []): + stream_path = stream.get('Stream') + if not stream_path: + continue + stream_url = update_url_query( + '%s/%s' % (server, stream_path), {'b': ''}) + if link_type == 'HLS': + formats.extend(self._extract_m3u8_formats( + stream_url, channel_id, 'mp4', + m3u8_id=link_type, fatal=False, live=True)) + elif link_type == 'HDS': + formats.extend(self._extract_f4m_formats(update_url_query( + '%s/%s' % (server, stream_path), {'hdcore': '3.7.0'}), + channel_id, f4m_id=link_type, fatal=False)) + self._sort_formats(formats) + + return { + 'id': channel_id, + 'title': title, + 'thumbnail': channel_data.get('PrimaryImageUri'), + 'formats': formats, + 'is_live': True, + }