[yt-dlp.git] / yt_dlp / extractor / ndr.py

# coding: utf-8
from __future__ import unicode_literals

import re

from .common import InfoExtractor
from ..utils import (
    determine_ext,
    int_or_none,
    merge_dicts,
    parse_iso8601,
    qualities,
    try_get,
    urljoin,
)


class NDRBaseIE(InfoExtractor):
    def _real_extract(self, url):
        mobj = re.match(self._VALID_URL, url)
        display_id = next(group for group in mobj.groups() if group)
        id = mobj.group('id')
        webpage = self._download_webpage(url, display_id)
        return self._extract_embed(webpage, display_id, id)


class NDRIE(NDRBaseIE):
    IE_NAME = 'ndr'
    IE_DESC = 'NDR.de - Norddeutscher Rundfunk'
    _VALID_URL = r'https?://(?:www\.)?(?:daserste\.)?ndr\.de/(?:[^/]+/)*(?P<display_id>[^/?#]+),(?P<id>[\da-z]+)\.html'
    _TESTS = [{
        # httpVideo, same content id
        'url': 'http://www.ndr.de/fernsehen/Party-Poette-und-Parade,hafengeburtstag988.html',
        'md5': '6515bc255dc5c5f8c85bbc38e035a659',
        'info_dict': {
            'id': 'hafengeburtstag988',
            'display_id': 'Party-Poette-und-Parade',
            'ext': 'mp4',
            'title': 'Party, Pötte und Parade',
            'description': 'md5:ad14f9d2f91d3040b6930c697e5f6b4c',
            'uploader': 'ndrtv',
            'timestamp': 1431108900,
            'upload_date': '20150510',
            'duration': 3498,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        # httpVideo, different content id
        'url': 'http://www.ndr.de/sport/fussball/40-Osnabrueck-spielt-sich-in-einen-Rausch,osna270.html',
        'md5': '1043ff203eab307f0c51702ec49e9a71',
        'info_dict': {
            'id': 'osna272',
            'display_id': '40-Osnabrueck-spielt-sich-in-einen-Rausch',
            'ext': 'mp4',
            'title': 'Osnabrück - Wehen Wiesbaden: Die Highlights',
            'description': 'md5:32e9b800b3d2d4008103752682d5dc01',
            'uploader': 'ndrtv',
            'timestamp': 1442059200,
            'upload_date': '20150912',
            'duration': 510,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        # httpAudio, same content id
        'url': 'http://www.ndr.de/info/La-Valette-entgeht-der-Hinrichtung,audio51535.html',
        'md5': 'bb3cd38e24fbcc866d13b50ca59307b8',
        'info_dict': {
            'id': 'audio51535',
            'display_id': 'La-Valette-entgeht-der-Hinrichtung',
            'ext': 'mp3',
            'title': 'La Valette entgeht der Hinrichtung',
            'description': 'md5:22f9541913a40fe50091d5cdd7c9f536',
            'uploader': 'ndrinfo',
            'timestamp': 1290626100,
            'upload_date': '20140729',
            'duration': 884,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        # with subtitles
        'url': 'https://www.ndr.de/fernsehen/sendungen/extra_3/extra-3-Satiremagazin-mit-Christian-Ehring,sendung1091858.html',
        'info_dict': {
            'id': 'extra18674',
            'display_id': 'extra-3-Satiremagazin-mit-Christian-Ehring',
            'ext': 'mp4',
            'title': 'Extra 3 vom 11.11.2020 mit Christian Ehring',
            'description': 'md5:42ee53990a715eaaf4dc7f13a3bd56c6',
            'uploader': 'ndrtv',
            'upload_date': '20201113',
            'duration': 1749,
            'subtitles': {
                'de': [{
                    'ext': 'ttml',
                    'url': r're:^https://www\.ndr\.de.+',
                }],
            },
        },
        'params': {
            'skip_download': True,
        },
        'expected_warnings': ['Unable to download f4m manifest'],
    }, {
        'url': 'https://www.ndr.de/Fettes-Brot-Ferris-MC-und-Thees-Uhlmann-live-on-stage,festivalsommer116.html',
        'only_matching': True,
    }]

    def _extract_embed(self, webpage, display_id, id):
        embed_url = self._html_search_meta(
            'embedURL', webpage, 'embed URL',
            default=None) or self._search_regex(
            r'\bembedUrl["\']\s*:\s*(["\'])(?P<url>(?:(?!\1).)+)\1', webpage,
            'embed URL', fatal=False, group='url')
        if embed_url is None:
            return self.url_result('ndr:%s' % id, ie=NDREmbedBaseIE.ie_key())
        description = self._search_regex(
            r'<p[^>]+itemprop="description">([^<]+)</p>',
            webpage, 'description', default=None) or self._og_search_description(webpage)
        timestamp = parse_iso8601(
            self._search_regex(
                r'<span[^>]+itemprop="(?:datePublished|uploadDate)"[^>]+content="([^"]+)"',
                webpage, 'upload date', default=None))
        info = self._search_json_ld(webpage, display_id, default={})
        return merge_dicts({
            '_type': 'url_transparent',
            'url': embed_url,
            'display_id': display_id,
            'description': description,
            'timestamp': timestamp,
        }, info)


class NJoyIE(NDRBaseIE):
    IE_NAME = 'njoy'
    IE_DESC = 'N-JOY'
    _VALID_URL = r'https?://(?:www\.)?n-joy\.de/(?:[^/]+/)*(?:(?P<display_id>[^/?#]+),)?(?P<id>[\da-z]+)\.html'
    _TESTS = [{
        # httpVideo, same content id
        'url': 'http://www.n-joy.de/entertainment/comedy/comedy_contest/Benaissa-beim-NDR-Comedy-Contest,comedycontest2480.html',
        'md5': 'cb63be60cd6f9dd75218803146d8dc67',
        'info_dict': {
            'id': 'comedycontest2480',
            'display_id': 'Benaissa-beim-NDR-Comedy-Contest',
            'ext': 'mp4',
            'title': 'Benaissa beim NDR Comedy Contest',
            'description': 'md5:f057a6c4e1c728b10d33b5ffd36ddc39',
            'uploader': 'ndrtv',
            'upload_date': '20141129',
            'duration': 654,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        # httpVideo, different content id
        'url': 'http://www.n-joy.de/musik/Das-frueheste-DJ-Set-des-Nordens-live-mit-Felix-Jaehn-,felixjaehn168.html',
        'md5': '417660fffa90e6df2fda19f1b40a64d8',
        'info_dict': {
            'id': 'dockville882',
            'display_id': 'Das-frueheste-DJ-Set-des-Nordens-live-mit-Felix-Jaehn-',
            'ext': 'mp4',
            'title': '"Ich hab noch nie" mit Felix Jaehn',
            'description': 'md5:85dd312d53be1b99e1f998a16452a2f3',
            'uploader': 'njoy',
            'upload_date': '20150822',
            'duration': 211,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        'url': 'http://www.n-joy.de/radio/webradio/morningshow209.html',
        'only_matching': True,
    }]

    def _extract_embed(self, webpage, display_id, id):
        video_id = self._search_regex(
            r'<iframe[^>]+id="pp_([\da-z]+)"', webpage, 'embed id')
        description = self._search_regex(
            r'<div[^>]+class="subline"[^>]*>[^<]+</div>\s*<p>([^<]+)</p>',
            webpage, 'description', fatal=False)
        return {
            '_type': 'url_transparent',
            'ie_key': 'NDREmbedBase',
            'url': 'ndr:%s' % video_id,
            'display_id': display_id,
            'description': description,
        }


class NDREmbedBaseIE(InfoExtractor):
    IE_NAME = 'ndr:embed:base'
    _VALID_URL = r'(?:ndr:(?P<id_s>[\da-z]+)|https?://www\.ndr\.de/(?P<id>[\da-z]+)-ppjson\.json)'
    _TESTS = [{
        'url': 'ndr:soundcheck3366',
        'only_matching': True,
    }, {
        'url': 'http://www.ndr.de/soundcheck3366-ppjson.json',
        'only_matching': True,
    }]

    def _real_extract(self, url):
        mobj = re.match(self._VALID_URL, url)
        video_id = mobj.group('id') or mobj.group('id_s')

        ppjson = self._download_json(
            'http://www.ndr.de/%s-ppjson.json' % video_id, video_id)

        playlist = ppjson['playlist']

        formats = []
        quality_key = qualities(('xs', 's', 'm', 'l', 'xl'))

        for format_id, f in playlist.items():
            src = f.get('src')
            if not src:
                continue
            ext = determine_ext(src, None)
            if ext == 'f4m':
                formats.extend(self._extract_f4m_formats(
                    src + '?hdcore=3.7.0&plugin=aasp-3.7.0.39.44', video_id,
                    f4m_id='hds', fatal=False))
            elif ext == 'm3u8':
                formats.extend(self._extract_m3u8_formats(
                    src, video_id, 'mp4', m3u8_id='hls',
                    entry_protocol='m3u8_native', fatal=False))
            else:
                quality = f.get('quality')
                ff = {
                    'url': src,
                    'format_id': quality or format_id,
                    'quality': quality_key(quality),
                }
                type_ = f.get('type')
                if type_ and type_.split('/')[0] == 'audio':
                    ff['vcodec'] = 'none'
                    ff['ext'] = ext or 'mp3'
                formats.append(ff)
        self._sort_formats(formats)

        config = playlist['config']

        live = playlist.get('config', {}).get('streamType') in ['httpVideoLive', 'httpAudioLive']
        title = config['title']
        if live:
            title = self._live_title(title)
        uploader = ppjson.get('config', {}).get('branding')
        upload_date = ppjson.get('config', {}).get('publicationDate')
        duration = int_or_none(config.get('duration'))

        thumbnails = []
        poster = try_get(config, lambda x: x['poster'], dict) or {}
        for thumbnail_id, thumbnail in poster.items():
            thumbnail_url = urljoin(url, thumbnail.get('src'))
            if not thumbnail_url:
                continue
            thumbnails.append({
                'id': thumbnail.get('quality') or thumbnail_id,
                'url': thumbnail_url,
                'preference': quality_key(thumbnail.get('quality')),
            })

        subtitles = {}
        tracks = config.get('tracks')
        if tracks and isinstance(tracks, list):
            for track in tracks:
                if not isinstance(track, dict):
                    continue
                track_url = urljoin(url, track.get('src'))
                if not track_url:
                    continue
                subtitles.setdefault(track.get('srclang') or 'de', []).append({
                    'url': track_url,
                    'ext': 'ttml',
                })

        return {
            'id': video_id,
            'title': title,
            'is_live': live,
            'uploader': uploader if uploader != '-' else None,
            'upload_date': upload_date[0:8] if upload_date else None,
            'duration': duration,
            'thumbnails': thumbnails,
            'formats': formats,
            'subtitles': subtitles,
        }


class NDREmbedIE(NDREmbedBaseIE):
    IE_NAME = 'ndr:embed'
    _VALID_URL = r'https?://(?:www\.)?(?:daserste\.)?ndr\.de/(?:[^/]+/)*(?P<id>[\da-z]+)-(?:player|externalPlayer)\.html'
    _TESTS = [{
        'url': 'http://www.ndr.de/fernsehen/sendungen/ndr_aktuell/ndraktuell28488-player.html',
        'md5': '8b9306142fe65bbdefb5ce24edb6b0a9',
        'info_dict': {
            'id': 'ndraktuell28488',
            'ext': 'mp4',
            'title': 'Norddeutschland begrüßt Flüchtlinge',
            'is_live': False,
            'uploader': 'ndrtv',
            'upload_date': '20150907',
            'duration': 132,
        },
    }, {
        'url': 'http://www.ndr.de/ndr2/events/soundcheck/soundcheck3366-player.html',
        'md5': '002085c44bae38802d94ae5802a36e78',
        'info_dict': {
            'id': 'soundcheck3366',
            'ext': 'mp4',
            'title': 'Ella Henderson braucht Vergleiche nicht zu scheuen',
            'is_live': False,
            'uploader': 'ndr2',
            'upload_date': '20150912',
            'duration': 3554,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        'url': 'http://www.ndr.de/info/audio51535-player.html',
        'md5': 'bb3cd38e24fbcc866d13b50ca59307b8',
        'info_dict': {
            'id': 'audio51535',
            'ext': 'mp3',
            'title': 'La Valette entgeht der Hinrichtung',
            'is_live': False,
            'uploader': 'ndrinfo',
            'upload_date': '20140729',
            'duration': 884,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        'url': 'http://www.ndr.de/fernsehen/sendungen/visite/visite11010-externalPlayer.html',
        'md5': 'ae57f80511c1e1f2fd0d0d3d31aeae7c',
        'info_dict': {
            'id': 'visite11010',
            'ext': 'mp4',
            'title': 'Visite - die ganze Sendung',
            'is_live': False,
            'uploader': 'ndrtv',
            'upload_date': '20150902',
            'duration': 3525,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        # httpVideoLive
        'url': 'http://www.ndr.de/fernsehen/livestream/livestream217-externalPlayer.html',
        'info_dict': {
            'id': 'livestream217',
            'ext': 'flv',
            'title': r're:^NDR Fernsehen Niedersachsen \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
            'is_live': True,
            'upload_date': '20150910',
        },
        'params': {
            'skip_download': True,
        },
    }, {
        'url': 'http://www.ndr.de/ndrkultur/audio255020-player.html',
        'only_matching': True,
    }, {
        'url': 'http://www.ndr.de/fernsehen/sendungen/nordtour/nordtour7124-player.html',
        'only_matching': True,
    }, {
        'url': 'http://www.ndr.de/kultur/film/videos/videoimport10424-player.html',
        'only_matching': True,
    }, {
        'url': 'http://www.ndr.de/fernsehen/sendungen/hamburg_journal/hamj43006-player.html',
        'only_matching': True,
    }, {
        'url': 'http://www.ndr.de/fernsehen/sendungen/weltbilder/weltbilder4518-player.html',
        'only_matching': True,
    }, {
        'url': 'http://www.ndr.de/fernsehen/doku952-player.html',
        'only_matching': True,
    }]


class NJoyEmbedIE(NDREmbedBaseIE):
    IE_NAME = 'njoy:embed'
    _VALID_URL = r'https?://(?:www\.)?n-joy\.de/(?:[^/]+/)*(?P<id>[\da-z]+)-(?:player|externalPlayer)_[^/]+\.html'
    _TESTS = [{
        # httpVideo
        'url': 'http://www.n-joy.de/events/reeperbahnfestival/doku948-player_image-bc168e87-5263-4d6d-bd27-bb643005a6de_theme-n-joy.html',
        'md5': '8483cbfe2320bd4d28a349d62d88bd74',
        'info_dict': {
            'id': 'doku948',
            'ext': 'mp4',
            'title': 'Zehn Jahre Reeperbahn Festival - die Doku',
            'is_live': False,
            'upload_date': '20150807',
            'duration': 1011,
        },
    }, {
        # httpAudio
        'url': 'http://www.n-joy.de/news_wissen/stefanrichter100-player_image-d5e938b1-f21a-4b9a-86b8-aaba8bca3a13_theme-n-joy.html',
        'md5': 'd989f80f28ac954430f7b8a48197188a',
        'info_dict': {
            'id': 'stefanrichter100',
            'ext': 'mp3',
            'title': 'Interview mit einem Augenzeugen',
            'is_live': False,
            'uploader': 'njoy',
            'upload_date': '20150909',
            'duration': 140,
        },
        'params': {
            'skip_download': True,
        },
    }, {
        # httpAudioLive, no explicit ext
        'url': 'http://www.n-joy.de/news_wissen/webradioweltweit100-player_image-3fec0484-2244-4565-8fb8-ed25fd28b173_theme-n-joy.html',
        'info_dict': {
            'id': 'webradioweltweit100',
            'ext': 'mp3',
            'title': r're:^N-JOY Weltweit \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
            'is_live': True,
            'uploader': 'njoy',
            'upload_date': '20150810',
        },
        'params': {
            'skip_download': True,
        },
    }, {
        'url': 'http://www.n-joy.de/musik/dockville882-player_image-3905259e-0803-4764-ac72-8b7de077d80a_theme-n-joy.html',
        'only_matching': True,
    }, {
        'url': 'http://www.n-joy.de/radio/sendungen/morningshow/urlaubsfotos190-player_image-066a5df1-5c95-49ec-a323-941d848718db_theme-n-joy.html',
        'only_matching': True,
    }, {
        'url': 'http://www.n-joy.de/entertainment/comedy/krudetv290-player_image-ab261bfe-51bf-4bf3-87ba-c5122ee35b3d_theme-n-joy.html',
        'only_matching': True,
    }]
Commit	Line	Data
2ffe3bc1	1	# coding: utf-8
e9ea0bf1 S	2	from __future__ import unicode_literals
e9ea0bf1 S	3
2ffe3bc1 S	4	import re
2ffe3bc1 S	5
e9ea0bf1	6	from .common import InfoExtractor
c1ed1f70	7	from ..utils import (
2ffe3bc1	8	determine_ext,
c1ed1f70	9	int_or_none,
4cbce88f	10	merge_dicts,
2ffe3bc1	11	parse_iso8601,
376e1ad0	12	qualities,
3fc56635 S	13	try_get,
3fc56635 S	14	urljoin,
c1ed1f70	15	)
e9ea0bf1 S	16
e9ea0bf1 S	17
1934f3a0	18	class NDRBaseIE(InfoExtractor):
64997815	19	def _real_extract(self, url):
7e0dc613 S	20	mobj = re.match(self._VALID_URL, url)
7e0dc613 S	21	display_id = next(group for group in mobj.groups() if group)
dc9d8f44	22	id = mobj.group('id')
2ffe3bc1	23	webpage = self._download_webpage(url, display_id)
dc9d8f44	24	return self._extract_embed(webpage, display_id, id)
64997815	25
1934f3a0 YCH	26
	27	class NDRIE(NDRBaseIE):
	28	IE_NAME = 'ndr'
2ffe3bc1	29	IE_DESC = 'NDR.de - Norddeutscher Rundfunk'
dc9d8f44	30	_VALID_URL = r'https?://(?:www\.)?(?:daserste\.)?ndr\.de/(?:[^/]+/)*(?P<display_id>[^/?#]+),(?P<id>[\da-z]+)\.html'
2ffe3bc1 S	31	_TESTS = [{
	32	# httpVideo, same content id
	33	'url': 'http://www.ndr.de/fernsehen/Party-Poette-und-Parade,hafengeburtstag988.html',
	34	'md5': '6515bc255dc5c5f8c85bbc38e035a659',
	35	'info_dict': {
	36	'id': 'hafengeburtstag988',
	37	'display_id': 'Party-Poette-und-Parade',
	38	'ext': 'mp4',
	39	'title': 'Party, Pötte und Parade',
	40	'description': 'md5:ad14f9d2f91d3040b6930c697e5f6b4c',
	41	'uploader': 'ndrtv',
	42	'timestamp': 1431108900,
	43	'upload_date': '20150510',
	44	'duration': 3498,
	45	},
	46	'params': {
	47	'skip_download': True,
	48	},
	49	}, {
	50	# httpVideo, different content id
	51	'url': 'http://www.ndr.de/sport/fussball/40-Osnabrueck-spielt-sich-in-einen-Rausch,osna270.html',
	52	'md5': '1043ff203eab307f0c51702ec49e9a71',
	53	'info_dict': {
	54	'id': 'osna272',
	55	'display_id': '40-Osnabrueck-spielt-sich-in-einen-Rausch',
	56	'ext': 'mp4',
	57	'title': 'Osnabrück - Wehen Wiesbaden: Die Highlights',
	58	'description': 'md5:32e9b800b3d2d4008103752682d5dc01',
	59	'uploader': 'ndrtv',
	60	'timestamp': 1442059200,
	61	'upload_date': '20150912',
	62	'duration': 510,
	63	},
	64	'params': {
	65	'skip_download': True,
	66	},
	67	}, {
	68	# httpAudio, same content id
	69	'url': 'http://www.ndr.de/info/La-Valette-entgeht-der-Hinrichtung,audio51535.html',
	70	'md5': 'bb3cd38e24fbcc866d13b50ca59307b8',
	71	'info_dict': {
	72	'id': 'audio51535',
	73	'display_id': 'La-Valette-entgeht-der-Hinrichtung',
	74	'ext': 'mp3',
	75	'title': 'La Valette entgeht der Hinrichtung',
	76	'description': 'md5:22f9541913a40fe50091d5cdd7c9f536',
	77	'uploader': 'ndrinfo',
	78	'timestamp': 1290626100,
	79	'upload_date': '20140729',
	80	'duration': 884,
	81	},
	82	'params': {
	83	'skip_download': True,
	84	},
8bdd16b4	85	}, {
	86	# with subtitles
	87	'url': 'https://www.ndr.de/fernsehen/sendungen/extra_3/extra-3-Satiremagazin-mit-Christian-Ehring,sendung1091858.html',
	88	'info_dict': {
	89	'id': 'extra18674',
	90	'display_id': 'extra-3-Satiremagazin-mit-Christian-Ehring',
	91	'ext': 'mp4',
	92	'title': 'Extra 3 vom 11.11.2020 mit Christian Ehring',
	93	'description': 'md5:42ee53990a715eaaf4dc7f13a3bd56c6',
	94	'uploader': 'ndrtv',
	95	'upload_date': '20201113',
	96	'duration': 1749,
	97	'subtitles': {
	98	'de': [{
	99	'ext': 'ttml',
	100	'url': r're:^https://www\.ndr\.de.+',
	101	}],
	102	},
	103	},
	104	'params': {
	105	'skip_download': True,
	106	},
	107	'expected_warnings': ['Unable to download f4m manifest'],
01003d07 S	108	}, {
	109	'url': 'https://www.ndr.de/Fettes-Brot-Ferris-MC-und-Thees-Uhlmann-live-on-stage,festivalsommer116.html',
	110	'only_matching': True,
2ffe3bc1 S	111	}]
2ffe3bc1 S	112
dc9d8f44	113	def _extract_embed(self, webpage, display_id, id):
2ffe3bc1	114	embed_url = self._html_search_meta(
4cbce88f S	115	'embedURL', webpage, 'embed URL',
	116	default=None) or self._search_regex(
	117	r'\bembedUrl["\']\s:\s(["\'])(?P<url>(?:(?!\1).)+)\1', webpage,
dc9d8f44	118	'embed URL', fatal=False, group='url')
921dc153	119	if embed_url is None:
dc9d8f44	120	return self.url_result('ndr:%s' % id, ie=NDREmbedBaseIE.ie_key())
2ffe3bc1 S	121	description = self._search_regex(
2ffe3bc1 S	122	r'<p[^>]+itemprop="description">([^<]+)</p>',
9796a9b2	123	webpage, 'description', default=None) or self._og_search_description(webpage)
2ffe3bc1 S	124	timestamp = parse_iso8601(
2ffe3bc1 S	125	self._search_regex(
9796a9b2	126	r'<span[^>]+itemprop="(?:datePublished\|uploadDate)"[^>]+content="([^"]+)"',
4cbce88f S	127	webpage, 'upload date', default=None))
	128	info = self._search_json_ld(webpage, display_id, default={})
	129	return merge_dicts({
2ffe3bc1 S	130	'_type': 'url_transparent',
	131	'url': embed_url,
	132	'display_id': display_id,
	133	'description': description,
	134	'timestamp': timestamp,
4cbce88f	135	}, info)
1934f3a0 YCH	136
	137
	138	class NJoyIE(NDRBaseIE):
2ffe3bc1 S	139	IE_NAME = 'njoy'
2ffe3bc1 S	140	IE_DESC = 'N-JOY'
92519402	141	_VALID_URL = r'https?://(?:www\.)?n-joy\.de/(?:[^/]+/)*(?:(?P<display_id>[^/?#]+),)?(?P<id>[\da-z]+)\.html'
2ffe3bc1 S	142	_TESTS = [{
2ffe3bc1 S	143	# httpVideo, same content id
1934f3a0 YCH	144	'url': 'http://www.n-joy.de/entertainment/comedy/comedy_contest/Benaissa-beim-NDR-Comedy-Contest,comedycontest2480.html',
	145	'md5': 'cb63be60cd6f9dd75218803146d8dc67',
	146	'info_dict': {
64997815	147	'id': 'comedycontest2480',
2ffe3bc1	148	'display_id': 'Benaissa-beim-NDR-Comedy-Contest',
1934f3a0 YCH	149	'ext': 'mp4',
1934f3a0 YCH	150	'title': 'Benaissa beim NDR Comedy Contest',
2ffe3bc1 S	151	'description': 'md5:f057a6c4e1c728b10d33b5ffd36ddc39',
	152	'uploader': 'ndrtv',
	153	'upload_date': '20141129',
1934f3a0	154	'duration': 654,
2ffe3bc1 S	155	},
	156	'params': {
	157	'skip_download': True,
	158	},
	159	}, {
	160	# httpVideo, different content id
	161	'url': 'http://www.n-joy.de/musik/Das-frueheste-DJ-Set-des-Nordens-live-mit-Felix-Jaehn-,felixjaehn168.html',
	162	'md5': '417660fffa90e6df2fda19f1b40a64d8',
	163	'info_dict': {
	164	'id': 'dockville882',
	165	'display_id': 'Das-frueheste-DJ-Set-des-Nordens-live-mit-Felix-Jaehn-',
	166	'ext': 'mp4',
	167	'title': '"Ich hab noch nie" mit Felix Jaehn',
	168	'description': 'md5:85dd312d53be1b99e1f998a16452a2f3',
	169	'uploader': 'njoy',
	170	'upload_date': '20150822',
	171	'duration': 211,
	172	},
	173	'params': {
	174	'skip_download': True,
	175	},
7e0dc613 S	176	}, {
	177	'url': 'http://www.n-joy.de/radio/webradio/morningshow209.html',
	178	'only_matching': True,
2ffe3bc1 S	179	}]
2ffe3bc1 S	180
dc9d8f44	181	def _extract_embed(self, webpage, display_id, id):
2ffe3bc1 S	182	video_id = self._search_regex(
	183	r'<iframe[^>]+id="pp_([\da-z]+)"', webpage, 'embed id')
	184	description = self._search_regex(
	185	r'<div[^>]+class="subline"[^>]>[^<]+</div>\s<p>([^<]+)</p>',
	186	webpage, 'description', fatal=False)
	187	return {
	188	'_type': 'url_transparent',
	189	'ie_key': 'NDREmbedBase',
	190	'url': 'ndr:%s' % video_id,
	191	'display_id': display_id,
	192	'description': description,
1934f3a0	193	}
64997815	194
64997815	195
2ffe3bc1 S	196	class NDREmbedBaseIE(InfoExtractor):
	197	IE_NAME = 'ndr:embed:base'
	198	_VALID_URL = r'(?:ndr:(?P<id_s>[\da-z]+)\|https?://www\.ndr\.de/(?P<id>[\da-z]+)-ppjson\.json)'
	199	_TESTS = [{
	200	'url': 'ndr:soundcheck3366',
	201	'only_matching': True,
	202	}, {
	203	'url': 'http://www.ndr.de/soundcheck3366-ppjson.json',
	204	'only_matching': True,
	205	}]
64997815	206
64997815	207	def _real_extract(self, url):
2ffe3bc1 S	208	mobj = re.match(self._VALID_URL, url)
	209	video_id = mobj.group('id') or mobj.group('id_s')
	210
	211	ppjson = self._download_json(
	212	'http://www.ndr.de/%s-ppjson.json' % video_id, video_id)
	213
	214	playlist = ppjson['playlist']
	215
	216	formats = []
	217	quality_key = qualities(('xs', 's', 'm', 'l', 'xl'))
	218
	219	for format_id, f in playlist.items():
	220	src = f.get('src')
	221	if not src:
	222	continue
	223	ext = determine_ext(src, None)
	224	if ext == 'f4m':
	225	formats.extend(self._extract_f4m_formats(
310ea466 S	226	src + '?hdcore=3.7.0&plugin=aasp-3.7.0.39.44', video_id,
310ea466 S	227	f4m_id='hds', fatal=False))
2ffe3bc1 S	228	elif ext == 'm3u8':
2ffe3bc1 S	229	formats.extend(self._extract_m3u8_formats(
310ea466 S	230	src, video_id, 'mp4', m3u8_id='hls',
310ea466 S	231	entry_protocol='m3u8_native', fatal=False))
2ffe3bc1 S	232	else:
	233	quality = f.get('quality')
	234	ff = {
	235	'url': src,
	236	'format_id': quality or format_id,
	237	'quality': quality_key(quality),
	238	}
	239	type_ = f.get('type')
	240	if type_ and type_.split('/')[0] == 'audio':
	241	ff['vcodec'] = 'none'
	242	ff['ext'] = ext or 'mp3'
	243	formats.append(ff)
	244	self._sort_formats(formats)
	245
	246	config = playlist['config']
	247
	248	live = playlist.get('config', {}).get('streamType') in ['httpVideoLive', 'httpAudioLive']
	249	title = config['title']
	250	if live:
	251	title = self._live_title(title)
	252	uploader = ppjson.get('config', {}).get('branding')
	253	upload_date = ppjson.get('config', {}).get('publicationDate')
	254	duration = int_or_none(config.get('duration'))
	255
3fc56635 S	256	thumbnails = []
	257	poster = try_get(config, lambda x: x['poster'], dict) or {}
	258	for thumbnail_id, thumbnail in poster.items():
	259	thumbnail_url = urljoin(url, thumbnail.get('src'))
	260	if not thumbnail_url:
	261	continue
	262	thumbnails.append({
	263	'id': thumbnail.get('quality') or thumbnail_id,
	264	'url': thumbnail_url,
	265	'preference': quality_key(thumbnail.get('quality')),
	266	})
2ffe3bc1	267
8bdd16b4	268	subtitles = {}
	269	tracks = config.get('tracks')
	270	if tracks and isinstance(tracks, list):
	271	for track in tracks:
	272	if not isinstance(track, dict):
	273	continue
	274	track_url = urljoin(url, track.get('src'))
	275	if not track_url:
	276	continue
	277	subtitles.setdefault(track.get('srclang') or 'de', []).append({
	278	'url': track_url,
	279	'ext': 'ttml',
	280	})
	281
2ffe3bc1 S	282	return {
	283	'id': video_id,
	284	'title': title,
	285	'is_live': live,
	286	'uploader': uploader if uploader != '-' else None,
	287	'upload_date': upload_date[0:8] if upload_date else None,
	288	'duration': duration,
	289	'thumbnails': thumbnails,
	290	'formats': formats,
8bdd16b4	291	'subtitles': subtitles,
2ffe3bc1	292	}
64997815	293
	294
	295	class NDREmbedIE(NDREmbedBaseIE):
	296	IE_NAME = 'ndr:embed'
dc9d8f44	297	_VALID_URL = r'https?://(?:www\.)?(?:daserste\.)?ndr\.de/(?:[^/]+/)*(?P<id>[\da-z]+)-(?:player\|externalPlayer)\.html'
2ffe3bc1	298	_TESTS = [{
64997815	299	'url': 'http://www.ndr.de/fernsehen/sendungen/ndr_aktuell/ndraktuell28488-player.html',
2ffe3bc1	300	'md5': '8b9306142fe65bbdefb5ce24edb6b0a9',
64997815	301	'info_dict': {
	302	'id': 'ndraktuell28488',
	303	'ext': 'mp4',
	304	'title': 'Norddeutschland begrüßt Flüchtlinge',
2ffe3bc1 S	305	'is_live': False,
	306	'uploader': 'ndrtv',
	307	'upload_date': '20150907',
64997815	308	'duration': 132,
2ffe3bc1 S	309	},
	310	}, {
	311	'url': 'http://www.ndr.de/ndr2/events/soundcheck/soundcheck3366-player.html',
	312	'md5': '002085c44bae38802d94ae5802a36e78',
	313	'info_dict': {
	314	'id': 'soundcheck3366',
	315	'ext': 'mp4',
	316	'title': 'Ella Henderson braucht Vergleiche nicht zu scheuen',
	317	'is_live': False,
	318	'uploader': 'ndr2',
	319	'upload_date': '20150912',
	320	'duration': 3554,
	321	},
	322	'params': {
	323	'skip_download': True,
	324	},
	325	}, {
	326	'url': 'http://www.ndr.de/info/audio51535-player.html',
	327	'md5': 'bb3cd38e24fbcc866d13b50ca59307b8',
	328	'info_dict': {
	329	'id': 'audio51535',
	330	'ext': 'mp3',
	331	'title': 'La Valette entgeht der Hinrichtung',
	332	'is_live': False,
	333	'uploader': 'ndrinfo',
	334	'upload_date': '20140729',
	335	'duration': 884,
	336	},
	337	'params': {
	338	'skip_download': True,
	339	},
	340	}, {
	341	'url': 'http://www.ndr.de/fernsehen/sendungen/visite/visite11010-externalPlayer.html',
	342	'md5': 'ae57f80511c1e1f2fd0d0d3d31aeae7c',
	343	'info_dict': {
	344	'id': 'visite11010',
	345	'ext': 'mp4',
	346	'title': 'Visite - die ganze Sendung',
	347	'is_live': False,
	348	'uploader': 'ndrtv',
	349	'upload_date': '20150902',
	350	'duration': 3525,
	351	},
	352	'params': {
	353	'skip_download': True,
	354	},
	355	}, {
	356	# httpVideoLive
	357	'url': 'http://www.ndr.de/fernsehen/livestream/livestream217-externalPlayer.html',
	358	'info_dict': {
	359	'id': 'livestream217',
	360	'ext': 'flv',
ec85ded8	361	'title': r're:^NDR Fernsehen Niedersachsen \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
2ffe3bc1 S	362	'is_live': True,
	363	'upload_date': '20150910',
	364	},
	365	'params': {
	366	'skip_download': True,
	367	},
	368	}, {
	369	'url': 'http://www.ndr.de/ndrkultur/audio255020-player.html',
	370	'only_matching': True,
	371	}, {
	372	'url': 'http://www.ndr.de/fernsehen/sendungen/nordtour/nordtour7124-player.html',
	373	'only_matching': True,
	374	}, {
	375	'url': 'http://www.ndr.de/kultur/film/videos/videoimport10424-player.html',
	376	'only_matching': True,
	377	}, {
	378	'url': 'http://www.ndr.de/fernsehen/sendungen/hamburg_journal/hamj43006-player.html',
	379	'only_matching': True,
	380	}, {
	381	'url': 'http://www.ndr.de/fernsehen/sendungen/weltbilder/weltbilder4518-player.html',
	382	'only_matching': True,
	383	}, {
	384	'url': 'http://www.ndr.de/fernsehen/doku952-player.html',
	385	'only_matching': True,
	386	}]
64997815	387
	388
	389	class NJoyEmbedIE(NDREmbedBaseIE):
2ffe3bc1	390	IE_NAME = 'njoy:embed'
92519402	391	_VALID_URL = r'https?://(?:www\.)?n-joy\.de/(?:[^/]+/)*(?P<id>[\da-z]+)-(?:player\|externalPlayer)_[^/]+\.html'
2ffe3bc1 S	392	_TESTS = [{
	393	# httpVideo
	394	'url': 'http://www.n-joy.de/events/reeperbahnfestival/doku948-player_image-bc168e87-5263-4d6d-bd27-bb643005a6de_theme-n-joy.html',
	395	'md5': '8483cbfe2320bd4d28a349d62d88bd74',
64997815	396	'info_dict': {
2ffe3bc1	397	'id': 'doku948',
64997815	398	'ext': 'mp4',
2ffe3bc1 S	399	'title': 'Zehn Jahre Reeperbahn Festival - die Doku',
	400	'is_live': False,
	401	'upload_date': '20150807',
	402	'duration': 1011,
	403	},
	404	}, {
	405	# httpAudio
	406	'url': 'http://www.n-joy.de/news_wissen/stefanrichter100-player_image-d5e938b1-f21a-4b9a-86b8-aaba8bca3a13_theme-n-joy.html',
	407	'md5': 'd989f80f28ac954430f7b8a48197188a',
	408	'info_dict': {
	409	'id': 'stefanrichter100',
	410	'ext': 'mp3',
	411	'title': 'Interview mit einem Augenzeugen',
	412	'is_live': False,
	413	'uploader': 'njoy',
	414	'upload_date': '20150909',
	415	'duration': 140,
	416	},
	417	'params': {
	418	'skip_download': True,
	419	},
	420	}, {
	421	# httpAudioLive, no explicit ext
	422	'url': 'http://www.n-joy.de/news_wissen/webradioweltweit100-player_image-3fec0484-2244-4565-8fb8-ed25fd28b173_theme-n-joy.html',
	423	'info_dict': {
	424	'id': 'webradioweltweit100',
	425	'ext': 'mp3',
ec85ded8	426	'title': r're:^N-JOY Weltweit \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
2ffe3bc1 S	427	'is_live': True,
	428	'uploader': 'njoy',
	429	'upload_date': '20150810',
	430	},
	431	'params': {
	432	'skip_download': True,
	433	},
	434	}, {
	435	'url': 'http://www.n-joy.de/musik/dockville882-player_image-3905259e-0803-4764-ac72-8b7de077d80a_theme-n-joy.html',
	436	'only_matching': True,
	437	}, {
	438	'url': 'http://www.n-joy.de/radio/sendungen/morningshow/urlaubsfotos190-player_image-066a5df1-5c95-49ec-a323-941d848718db_theme-n-joy.html',
	439	'only_matching': True,
	440	}, {
	441	'url': 'http://www.n-joy.de/entertainment/comedy/krudetv290-player_image-ab261bfe-51bf-4bf3-87ba-c5122ee35b3d_theme-n-joy.html',
	442	'only_matching': True,
	443	}]