1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-11-22 10:11:52 +00:00
youtube-dl/youtube_dl/extractor/orf.py

765 lines
31 KiB
Python

# coding: utf-8
from __future__ import unicode_literals
import base64
import functools
import re
from .common import InfoExtractor
from .youtube import YoutubeIE
from ..utils import (
clean_html,
determine_ext,
ExtractorError,
float_or_none,
int_or_none,
merge_dicts,
mimetype2ext,
parse_age_limit,
parse_iso8601,
strip_jsonp,
txt_or_none,
unified_strdate,
update_url_query,
url_or_none,
)
from ..traversal import T, traverse_obj
k_float_or_none = functools.partial(float_or_none, scale=1000)
class ORFRadioBase(InfoExtractor):
STATION_INFO = {
'fm4': ('fm4', 'fm4', 'orffm4'),
'noe': ('noe', 'oe2n', 'orfnoe'),
'wien': ('wie', 'oe2w', 'orfwie'),
'burgenland': ('bgl', 'oe2b', 'orfbgl'),
'ooe': ('ooe', 'oe2o', 'orfooe'),
'steiermark': ('stm', 'oe2st', 'orfstm'),
'kaernten': ('ktn', 'oe2k', 'orfktn'),
'salzburg': ('sbg', 'oe2s', 'orfsbg'),
'tirol': ('tir', 'oe2t', 'orftir'),
'vorarlberg': ('vbg', 'oe2v', 'orfvbg'),
'oe3': ('oe3', 'oe3', 'orfoe3'),
'oe1': ('oe1', 'oe1', 'orfoe1'),
}
_ID_NAMES = ('id', 'guid', 'program')
@classmethod
def _get_item_id(cls, data):
return traverse_obj(data, *cls._ID_NAMES, expected_type=txt_or_none)
@classmethod
def _get_api_payload(cls, data, expected_id, in_payload=False):
if expected_id not in traverse_obj(data, ('payload',)[:1 if in_payload else 0] + (cls._ID_NAMES, T(txt_or_none))):
raise ExtractorError('Unexpected API data result', video_id=expected_id)
return data['payload']
@staticmethod
def _extract_podcast_upload(data):
return traverse_obj(data, {
'url': ('enclosures', 0, 'url'),
'ext': ('enclosures', 0, 'type', T(mimetype2ext)),
'filesize': ('enclosures', 0, 'length', T(int_or_none)),
'title': ('title', T(txt_or_none)),
'description': ('description', T(clean_html)),
'timestamp': (('published', 'postDate'), T(parse_iso8601)),
'duration': ('duration', T(k_float_or_none)),
'series': ('podcast', 'title'),
'uploader': ((('podcast', 'author'), 'station'), T(txt_or_none)),
'uploader_id': ('podcast', 'channel', T(txt_or_none)),
}, get_all=False)
@classmethod
def _entries(cls, data, station, item_type=None):
if item_type in ('upload', 'podcast-episode'):
yield merge_dicts({
'id': cls._get_item_id(data),
'ext': 'mp3',
'vcodec': 'none',
}, cls._extract_podcast_upload(data), rev=True)
return
loop_station = cls.STATION_INFO[station][1]
for info in traverse_obj(data, ((('streams', Ellipsis), 'stream'), T(lambda v: v if v['loopStreamId'] else None))):
item_id = info['loopStreamId']
host = info.get('host') or 'loopstream01.apa.at'
yield merge_dicts({
'id': item_id.replace('.mp3', ''),
'ext': 'mp3',
'url': update_url_query('https://{0}/'.format(host), {
'channel': loop_station,
'id': item_id,
}),
'vcodec': 'none',
# '_old_archive_ids': [make_archive_id(old_ie, video_id)],
}, traverse_obj(data, {
'title': ('title', T(txt_or_none)),
'description': ('subtitle', T(clean_html)),
'uploader': 'station',
'series': ('programTitle', T(txt_or_none)),
}), traverse_obj(info, {
'duration': (('duration',
(None, T(lambda x: x['end'] - x['start']))),
T(k_float_or_none), any),
'timestamp': (('start', 'startISO'), T(parse_iso8601), any),
}))
class ORFRadioIE(ORFRadioBase):
IE_NAME = 'orf:sound'
_STATION_RE = '|'.join(map(re.escape, ORFRadioBase.STATION_INFO.keys()))
_VALID_URL = (
r'https?://sound\.orf\.at/radio/(?P<station>{0})/sendung/(?P<id>\d+)(?:/(?P<show>\w+))?'.format(_STATION_RE),
r'https?://(?P<station>{0})\.orf\.at/player/(?P<date>\d{{8}})/(?P<id>\d+)'.format(_STATION_RE),
)
_TESTS = [{
'url': 'https://sound.orf.at/radio/ooe/sendung/37802/guten-morgen-oberoesterreich-am-feiertag',
'info_dict': {
'id': '37802',
'title': 'Guten Morgen Oberösterreich am Feiertag',
'description': 'Oberösterreichs meistgehörte regionale Frühsendung.\nRegionale Nachrichten zu jeder halben Stunde.\nModeration: Wolfgang Lehner\nNachrichten: Stephan Schnabl',
},
'playlist': [{
'md5': 'f9ff8517dd681b642a2c900e2c9e6085',
'info_dict': {
'id': '2024-05-30_0559_tl_66_7DaysThu1_443862',
'ext': 'mp3',
'title': 'Guten Morgen Oberösterreich am Feiertag',
'description': 'Oberösterreichs meistgehörte regionale Frühsendung.\nRegionale Nachrichten zu jeder halben Stunde.\nModeration: Wolfgang Lehner\nNachrichten: Stephan Schnabl',
'timestamp': 1717041587,
'upload_date': '20240530',
'uploader': 'ooe',
'duration': 14413.0,
}
}],
'skip': 'Shows from ORF Sound are only available for 30 days.'
}, {
'url': 'https://oe1.orf.at/player/20240531/758136',
'md5': '2397717aaf3ae9c22a4f090ee3b8d374',
'info_dict': {
'id': '2024-05-31_1905_tl_51_7DaysFri35_2413387',
'ext': 'mp3',
'title': '"Who Cares?"',
'description': 'Europas größte Netzkonferenz re:publica 2024',
'timestamp': 1717175100,
'upload_date': '20240531',
'uploader': 'oe1',
'duration': 1500,
},
'skip': 'Shows from ORF Sound are only available for 30 days.'
}]
def _real_extract(self, url):
m = self._match_valid_url(url)
station, show_id = m.group('station', 'id')
api_station, _, _ = self.STATION_INFO[station]
if 'date' in m.groupdict():
data = self._download_json(
'https://audioapi.orf.at/{0}/json/4.0/broadcast/{1}/{2}?_o={3}.orf.at'.format(
api_station, show_id, m.group('date'), station), show_id)
show_id = data['id']
else:
data = self._download_json(
'https://audioapi.orf.at/{0}/api/json/5.0/broadcast/{1}?_o=sound.orf.at'.format(
api_station, show_id), show_id)
data = self._get_api_payload(data, show_id, in_payload=True)
# site sends ISO8601 GMT date-times with separate TZ offset, ignored
# TODO: should `..._date` be calculated relative to TZ?
return merge_dicts(
{'_type': 'multi_video'},
self.playlist_result(
self._entries(data, station), show_id,
txt_or_none(data.get('title')),
clean_html(data.get('subtitle'))))
class ORFRadioCollectionIE(ORFRadioBase):
IE_NAME = 'orf:collection'
_VALID_URL = r'https?://sound\.orf\.at/collection/(?P<coll_id>\d+)(?:/(?P<item_id>\d+))?'
_TESTS = [{
'url': 'https://sound.orf.at/collection/4/61908/was-das-uberschreiten-des-15-limits-bedeutet',
'info_dict': {
'id': '2577582',
},
'playlist': [{
'md5': '5789cec7d75575ff58d19c0428c80eb3',
'info_dict': {
'id': '2024-06-06_1659_tl_54_7DaysThu6_153926',
'ext': 'mp3',
'title': 'Klimakrise: Was das Überschreiten des 1,5°-Limits bedeutet',
'timestamp': 1717686674,
'upload_date': '20240606',
'uploader': 'fm4',
},
}],
'skip': 'Shows from ORF Sound are only available for 30 days.'
}, {
# persistent playlist (FM4 Highlights)
'url': 'https://sound.orf.at/collection/4/',
'info_dict': {
'id': '4',
},
'playlist_mincount': 10,
'playlist_maxcount': 13,
}]
def _real_extract(self, url):
coll_id, item_id = self._match_valid_url(url).group('coll_id', 'item_id')
data = self._download_json(
'https://collector.orf.at/api/frontend/collections/{0}?_o=sound.orf.at'.format(
coll_id), coll_id)
data = self._get_api_payload(data, coll_id, in_payload=True)
def yield_items():
for item in traverse_obj(data, (
'content', 'items', lambda _, v: any(k in v['target']['params'] for k in self._ID_NAMES))):
if item_id is None or item_id == txt_or_none(item.get('id')):
target = item['target']
typed_item_id = self._get_item_id(target['params'])
station = target['params'].get('station')
item_type = target.get('type')
if typed_item_id and (station or item_type):
yield station, typed_item_id, item_type
if item_id is not None:
break
else:
if item_id is not None:
raise ExtractorError('Item not found in collection',
video_id=coll_id, expected=True)
def item_playlist(station, typed_item_id, item_type):
if item_type == 'upload':
item_data = self._download_json('https://audioapi.orf.at/radiothek/api/2.0/upload/{0}?_o=sound.orf.at'.format(
typed_item_id), typed_item_id)
elif item_type == 'podcast-episode':
item_data = self._download_json('https://audioapi.orf.at/radiothek/api/2.0/episode/{0}?_o=sound.orf.at'.format(
typed_item_id), typed_item_id)
else:
api_station, _, _ = self.STATION_INFO[station]
item_data = self._download_json(
'https://audioapi.orf.at/{0}/api/json/5.0/{1}/{2}?_o=sound.orf.at'.format(
api_station, item_type or 'broadcastitem', typed_item_id), typed_item_id)
item_data = self._get_api_payload(item_data, typed_item_id, in_payload=True)
return merge_dicts(
{'_type': 'multi_video'},
self.playlist_result(
self._entries(item_data, station, item_type), typed_item_id,
txt_or_none(data.get('title')),
clean_html(data.get('subtitle'))))
def yield_item_entries():
for station, typed_id, item_type in yield_items():
yield item_playlist(station, typed_id, item_type)
if item_id is not None:
# coll_id = '/'.join((coll_id, item_id))
return next(yield_item_entries())
return self.playlist_result(yield_item_entries(), coll_id, data.get('title'))
class ORFPodcastIE(ORFRadioBase):
IE_NAME = 'orf:podcast'
_STATION_RE = '|'.join(map(re.escape, (x[0] for x in ORFRadioBase.STATION_INFO.values()))) + '|tv'
_VALID_URL = r'https?://sound\.orf\.at/podcast/(?P<station>{0})/(?P<show>[\w-]+)/(?P<id>[\w-]+)'.format(_STATION_RE)
_TESTS = [{
'url': 'https://sound.orf.at/podcast/stm/der-kraeutertipp-von-christine-lackner/rotklee',
'md5': '1f2bab2ba90c2ce0c2754196ea78b35f',
'info_dict': {
'id': 'der-kraeutertipp-von-christine-lackner/rotklee',
'ext': 'mp3',
'title': 'Rotklee',
'description': 'In der Natur weit verbreitet - in der Medizin längst anerkennt: Rotklee. Dieser Podcast begleitet die Sendung "Radio Steiermark am Vormittag", Radio Steiermark, 28. Mai 2024.',
'timestamp': 1716891761,
'upload_date': '20240528',
'uploader_id': 'stm_kraeutertipp',
'uploader': 'ORF Radio Steiermark',
'duration': 101,
'series': 'Der Kräutertipp von Christine Lackner',
},
'skip': 'ORF podcasts are only available for a limited time'
}]
_ID_NAMES = ('slug', 'guid')
def _real_extract(self, url):
station, show, show_id = self._match_valid_url(url).group('station', 'show', 'id')
data = self._download_json(
'https://audioapi.orf.at/radiothek/api/2.0/podcast/{0}/{1}/{2}'.format(
station, show, show_id), show_id)
data = self._get_api_payload(data, show_id, in_payload=True)
return merge_dicts({
'id': '/'.join((show, show_id)),
'ext': 'mp3',
'vcodec': 'none',
}, self._extract_podcast_upload(data), rev=True)
class ORFIPTVBase(InfoExtractor):
_TITLE_STRIP_RE = ''
def _extract_video(self, video_id, webpage, fatal=False):
data = self._download_json(
'http://bits.orf.at/filehandler/static-api/json/current/data.json?file=%s' % video_id,
video_id)[0]
video = traverse_obj(data, (
'sources', ('default', 'q8c'),
T(lambda x: x if x['loadBalancerUrl'] else None),
any))
load_balancer_url = video['loadBalancerUrl']
try:
rendition = self._download_json(
load_balancer_url, video_id, transform_source=strip_jsonp)
except ExtractorError:
rendition = None
if not rendition:
rendition = {
'redirect': {
'smil': re.sub(
r'(/)jsonp(/.+\.)mp4$', r'\1dash\2smil/manifest.mpd',
load_balancer_url),
},
}
f = traverse_obj(video, {
'abr': ('audioBitrate', T(int_or_none)),
'vbr': ('bitrate', T(int_or_none)),
'fps': ('videoFps', T(int_or_none)),
'width': ('videoWidth', T(int_or_none)),
'height': ('videoHeight', T(int_or_none)),
})
formats = []
for format_id, format_url in traverse_obj(rendition, (
'redirect', T(dict.items), Ellipsis)):
if format_id == 'rtmp':
ff = f.copy()
ff.update({
'url': format_url,
'format_id': format_id,
})
formats.append(ff)
elif determine_ext(format_url) == 'f4m':
formats.extend(self._extract_f4m_formats(
format_url, video_id, f4m_id=format_id))
elif determine_ext(format_url) == 'm3u8':
formats.extend(self._extract_m3u8_formats(
format_url, video_id, 'mp4', m3u8_id=format_id,
entry_protocol='m3u8_native'))
elif determine_ext(format_url) == 'mpd':
formats.extend(self._extract_mpd_formats(
format_url, video_id, mpd_id=format_id))
if formats or fatal:
self._sort_formats(formats)
else:
return
return merge_dicts({
'id': video_id,
'title': re.sub(self._TITLE_STRIP_RE, '', self._og_search_title(webpage)),
'description': self._og_search_description(webpage),
'upload_date': unified_strdate(self._html_search_meta(
'dc.date', webpage, 'upload date', fatal=False)),
'formats': formats,
}, traverse_obj(data, {
'duration': ('duration', T(k_float_or_none)),
'thumbnail': ('sources', 'default', 'preview', T(url_or_none)),
}), rev=True)
class ORFIPTVIE(ORFIPTVBase):
IE_NAME = 'orf:iptv'
IE_DESC = 'iptv.ORF.at'
_WORKING = False # URLs redirect to orf.at/
_VALID_URL = r'https?://iptv\.orf\.at/(?:#/)?stories/(?P<id>\d+)'
_TITLE_STRIP_RE = r'\s+-\s+iptv\.ORF\.at\S*$'
_TEST = {
'url': 'http://iptv.orf.at/stories/2275236/',
'md5': 'c8b22af4718a4b4af58342529453e3e5',
'info_dict': {
'id': '350612',
'ext': 'flv',
'title': 'Weitere Evakuierungen um Vulkan Calbuco',
'description': 'md5:d689c959bdbcf04efeddedbf2299d633',
'duration': 68.197,
'thumbnail': r're:^https?://.*\.jpg$',
'upload_date': '20150425',
},
}
def _real_extract(self, url):
story_id = self._match_id(url)
webpage = self._download_webpage(
'http://iptv.orf.at/stories/%s' % story_id, story_id)
video_id = self._search_regex(
r'data-video(?:id)?="(\d+)"', webpage, 'video id')
return self._extract_video(video_id, webpage)
class ORFFM4StoryIE(ORFIPTVBase):
IE_NAME = 'orf:fm4:story'
IE_DESC = 'fm4.orf.at stories'
_VALID_URL = r'https?://fm4\.orf\.at/stories/(?P<id>\d+)'
_TITLE_STRIP_RE = r'\s+-\s+fm4\.ORF\.at\s*$'
_TESTS = [{
'url': 'https://fm4.orf.at/stories/3041554/',
'add_ie': ['Youtube'],
'info_dict': {
'id': '3041554',
'title': 'Is The EU Green Deal In Mortal Danger?',
},
'playlist_count': 4,
'params': {
'format': 'bestvideo',
},
}, {
'url': 'http://fm4.orf.at/stories/2865738/',
'info_dict': {
'id': '2865738',
'title': 'Manu Delago und Inner Tongue live',
},
'playlist': [{
'md5': 'e1c2c706c45c7b34cf478bbf409907ca',
'info_dict': {
'id': '547792',
'ext': 'flv',
'title': 'Manu Delago und Inner Tongue live',
'description': 'Manu Delago und Inner Tongue haben bei der FM4 Soundpark Session live alles gegeben. Hier gibt es Fotos und die gesamte Session als Video.',
'duration': 1748.52,
'thumbnail': r're:^https?://.*\.jpg$',
'upload_date': '20170913',
},
}, {
'md5': 'c6dd2179731f86f4f55a7b49899d515f',
'info_dict': {
'id': '547798',
'ext': 'flv',
'title': 'Manu Delago und Inner Tongue https://vod-ww.mdn.ors.at/cms-worldwide_episodes_nas/_definst_/nas/cms-worldwide_episodes/online/14228823_0005.smil/chunklist_b992000_vo.m3u8live (2)',
'duration': 1504.08,
'thumbnail': r're:^https?://.*\.jpg$',
'upload_date': '20170913',
'description': 'Manu Delago und Inner Tongue haben bei der FM4 Soundpark Session live alles gegeben. Hier gibt es Fotos und die gesamte Session als Video.',
},
}],
'skip': 'Videos gone',
}]
def _real_extract(self, url):
story_id = self._match_id(url)
webpage = self._download_webpage(url, story_id)
entries = []
seen_ids = set()
for idx, video_id in enumerate(re.findall(r'data-video(?:id)?="(\d+)"', webpage)):
if video_id in seen_ids:
continue
seen_ids.add(video_id)
entry = self._extract_video(video_id, webpage, fatal=False)
if not entry:
continue
if idx >= 1:
# Titles are duplicates, make them unique
entry['title'] = '%s (%d)' % (entry['title'], idx)
entries.append(entry)
seen_ids = set()
for yt_id in re.findall(
r'data-id\s*=\s*["\']([\w-]+)[^>]+\bclass\s*=\s*["\']youtube\b',
webpage):
if yt_id in seen_ids:
continue
seen_ids.add(yt_id)
if YoutubeIE.suitable(yt_id):
entries.append(self.url_result(yt_id, ie='Youtube', video_id=yt_id))
return self.playlist_result(
entries, story_id,
re.sub(self._TITLE_STRIP_RE, '', self._og_search_title(webpage, default='') or None))
class ORFONBase(InfoExtractor):
_ENC_PFX = '3dSlfek03nsLKdj4Jsd'
_API_PATH = 'episode'
def _call_api(self, video_id, **kwargs):
encrypted_id = base64.b64encode('{0}{1}'.format(
self._ENC_PFX, video_id).encode('utf-8')).decode('ascii')
return self._download_json(
'https://api-tvthek.orf.at/api/v4.3/public/{0}/encrypted/{1}'.format(
self._API_PATH, encrypted_id),
video_id, **kwargs)
@classmethod
def _parse_metadata(cls, api_json):
return traverse_obj(api_json, {
'id': ('id', T(int), T(txt_or_none)),
'age_limit': ('age_classification', T(parse_age_limit)),
'duration': ((('exact_duration', T(k_float_or_none)),
('duration_second', T(float_or_none))),),
'title': (('title', 'headline'), T(txt_or_none)),
'description': (('description', 'teaser_text'), T(txt_or_none)),
# 'media_type': ('video_type', T(txt_or_none)),
'thumbnail': ('_embedded', 'image', 'public_urls', 'highlight_teaser', 'url', T(url_or_none)),
'timestamp': (('date', 'episode_date'), T(parse_iso8601)),
'release_timestamp': ('release_date', T(parse_iso8601)),
# 'modified_timestamp': ('updated_at', T(parse_iso8601)),
}, get_all=False)
def _extract_video(self, video_id, segment_id):
# Not a segmented episode: return single video
# Segmented episode without valid segment id: return entire playlist
# Segmented episode with valid segment id and yes-playlist: return entire playlist
# Segmented episode with valid segment id and no-playlist: return single video corresponding to segment id
# If a multi_video playlist would be returned, but an unsegmented source exists, that source is chosen instead.
api_json = self._call_api(video_id)
if traverse_obj(api_json, 'is_drm_protected'):
self.report_drm(video_id)
# updates formats, subtitles
def extract_sources(src_json, video_id):
for manifest_type in traverse_obj(src_json, ('sources', T(dict.keys), Ellipsis)):
for manifest_url in traverse_obj(src_json, ('sources', manifest_type, Ellipsis, 'src', T(url_or_none))):
if manifest_type == 'hls':
fmts, subs = self._extract_m3u8_formats(
manifest_url, video_id, fatal=False, m3u8_id='hls',
ext='mp4', entry_protocol='m3u8_native'), {}
for f in fmts:
if '_vo.' in f['url']:
f['acodec'] = 'none'
elif manifest_type == 'dash':
fmts, subs = self._extract_mpd_formats_and_subtitles(
manifest_url, video_id, fatal=False, mpd_id='dash')
else:
continue
formats.extend(fmts)
self._merge_subtitles(subs, target=subtitles)
formats, subtitles = [], {}
if segment_id is None:
extract_sources(api_json, video_id)
if not formats:
segments = traverse_obj(api_json, (
'_embedded', 'segments', lambda _, v: v['id']))
if len(segments) > 1 and segment_id is not None:
if not self._yes_playlist(video_id, segment_id, playlist_label='collection', video_label='segment'):
segments = [next(s for s in segments if txt_or_none(s['id']) == segment_id)]
entries = []
for seg in segments:
formats, subtitles = [], {}
extract_sources(seg, segment_id)
self._sort_formats(formats)
entries.append(merge_dicts({
'formats': formats,
'subtitles': subtitles,
}, self._parse_metadata(seg), rev=True))
result = merge_dicts(
{'_type': 'multi_video' if len(entries) > 1 else 'playlist'},
self._parse_metadata(api_json),
self.playlist_result(entries, video_id))
# not yet processed in core for playlist/multi
self._downloader._fill_common_fields(result)
return result
else:
self._sort_formats(formats)
for sub_url in traverse_obj(api_json, (
'_embedded', 'subtitle',
('xml_url', 'sami_url', 'stl_url', 'ttml_url', 'srt_url', 'vtt_url'),
T(url_or_none))):
self._merge_subtitles({'de': [{'url': sub_url}]}, target=subtitles)
return merge_dicts({
'id': video_id,
'formats': formats,
'subtitles': subtitles,
# '_old_archive_ids': [self._downloader._make_archive_id({'ie_key': 'ORFTVthek', 'id': video_id})],
}, self._parse_metadata(api_json), rev=True)
def _real_extract(self, url):
video_id, segment_id = self._match_valid_url(url).group('id', 'segment')
webpage = self._download_webpage(url, video_id)
# ORF doesn't like 410 or 404
if self._search_regex(r'<div\b[^>]*>\s*(Nicht mehr verfügbar)\s*</div>', webpage, 'Availability', default=False):
raise ExtractorError('Content is no longer available', expected=True, video_id=video_id)
return merge_dicts({
'id': video_id,
'title': self._html_search_meta(['og:title', 'twitter:title'], webpage, default=None),
'description': self._html_search_meta(
['description', 'og:description', 'twitter:description'], webpage, default=None),
}, self._search_json_ld(webpage, video_id, default={}),
self._extract_video(video_id, segment_id),
rev=True)
class ORFONIE(ORFONBase):
IE_NAME = 'orf:on'
_VALID_URL = r'https?://on\.orf\.at/video/(?P<id>\d+)(?:/(?P<segment>\d+))?'
_TESTS = [{
'url': 'https://on.orf.at/video/14210000/school-of-champions-48',
'info_dict': {
'id': '14210000',
'ext': 'mp4',
'duration': 2651.08,
'thumbnail': 'https://api-tvthek.orf.at/assets/segments/0167/98/thumb_16697671_segments_highlight_teaser.jpeg',
'title': 'School of Champions (4/8)',
'description': r're:(?s)Luca hat sein ganzes Leben in den Bergen Südtirols verbracht und ist bei seiner Mutter aufgewachsen, .{1029} Leo$',
# 'media_type': 'episode',
'timestamp': 1706558922,
'upload_date': '20240129',
'release_timestamp': 1706472362,
'release_date': '20240128',
# 'modified_timestamp': 1712756663,
# 'modified_date': '20240410',
# '_old_archive_ids': ['orftvthek 14210000'],
},
'params': {
'format': 'bestvideo',
},
'skip': 'Available until 2024-08-12',
}, {
'url': 'https://on.orf.at/video/3220355',
'md5': '925a93b2b9a37da5c9b979d7cf71aa2e',
'info_dict': {
'id': '3220355',
'ext': 'mp4',
'duration': 445.04,
'thumbnail': 'https://api-tvthek.orf.at/assets/segments/0002/60/thumb_159573_segments_highlight_teaser.png',
'title': '50 Jahre Burgenland: Der Festumzug',
'description': r're:(?s)Aus allen Landesteilen zogen festlich geschmückte Wagen und Musikkapellen .{270} Jenakowitsch$',
# 'media_type': 'episode',
'timestamp': 52916400,
'upload_date': '19710905',
'release_timestamp': 52916400,
'release_date': '19710905',
# 'modified_timestamp': 1498536049,
# 'modified_date': '20170627',
# '_old_archive_ids': ['orftvthek 3220355'],
},
}, {
# Video with multiple segments selecting the second segment
'url': 'https://on.orf.at/video/14226549/15639808/jugendbande-einbrueche-aus-langeweile',
'md5': 'fc151bba8c05ea77ab5693617e4a33d3',
'info_dict': {
'id': '15639808',
'ext': 'mp4',
'duration': 97.707,
'thumbnail': 'https://api-tvthek.orf.at/assets/segments/0175/43/thumb_17442704_segments_highlight_teaser.jpg',
'title': 'Jugendbande: Einbrüche aus Langeweile',
'description': r're:Jugendbande: Einbrüche aus Langeweile \| Neuer Kinder- und .{259} Wanda$',
# 'media_type': 'segment',
'timestamp': 1715792400,
'upload_date': '20240515',
# 'modified_timestamp': 1715794394,
# 'modified_date': '20240515',
# '_old_archive_ids': ['orftvthek 15639808'],
},
'params': {
'noplaylist': True,
'format': 'bestvideo',
},
'skip': 'Available until 2024-06-14',
}, {
# Video with multiple segments and no combined version
'url': 'https://on.orf.at/video/14227864/formel-1-grosser-preis-von-monaco-2024',
'info_dict': {
'_type': 'multi_video',
'id': '14227864',
'duration': 18410.52,
'thumbnail': 'https://api-tvthek.orf.at/assets/segments/0176/04/thumb_17503881_segments_highlight_teaser.jpg',
'title': 'Formel 1: Großer Preis von Monaco 2024',
'description': 'md5:aeeb010710ccf70ce28ccb4482243d4f',
# 'media_type': 'episode',
'timestamp': 1716721200,
'upload_date': '20240526',
'release_timestamp': 1716721802,
'release_date': '20240526',
# 'modified_timestamp': 1716884702,
# 'modified_date': '20240528',
},
'playlist_count': 42,
'skip': 'Gone: Nicht mehr verfügbar',
}, {
# Video with multiple segments, but with combined version
'url': 'https://on.orf.at/video/14228172',
'info_dict': {
'id': '14228172',
'ext': 'mp4',
'duration': 3294.878,
'thumbnail': 'https://api-tvthek.orf.at/assets/segments/0176/29/thumb_17528242_segments_highlight_teaser.jpg',
'title': 'Willkommen Österreich mit Stermann & Grissemann',
'description': r're:Zum Saisonfinale freuen sich die urlaubsreifen Gastgeber Stermann und .{1863} Geschichten\.$',
# 'media_type': 'episode',
'timestamp': 1716926584,
'upload_date': '20240528',
'release_timestamp': 1716919202,
'release_date': '20240528',
# 'modified_timestamp': 1716968045,
# 'modified_date': '20240529',
# '_old_archive_ids': ['orftvthek 14228172'],
},
'params': {
'format': 'bestvideo',
},
'skip': 'Gone: Nicht mehr verfügbar',
}]
class ORFONLiveIE(ORFONBase):
_ENC_PFX = '8876324jshjd7293ktd'
_API_PATH = 'livestream'
_VALID_URL = r'https?://on\.orf\.at/livestream/(?P<id>\d+)(?:/(?P<segment>\d+))?'
_TESTS = [{
'url': 'https://on.orf.at/livestream/14320204/pressekonferenz-neos-zu-aktuellen-entwicklungen',
'info_dict': {
'id': '14320204',
'ext': 'mp4',
'title': 'Pressekonferenz: Neos zu aktuellen Entwicklungen',
'description': r're:(?s)Neos-Chefin Beate Meinl-Reisinger informi.{598}ng\."',
'timestamp': 1716886335,
'upload_date': '20240528',
# 'modified_timestamp': 1712756663,
# 'modified_date': '20240410',
# '_old_archive_ids': ['orftvthek 14210000'],
},
'params': {
'format': 'bestvideo',
},
}]
@classmethod
def _parse_metadata(cls, api_json):
return merge_dicts(
super(ORFONLiveIE, cls)._parse_metadata(api_json),
traverse_obj(api_json, {
'timestamp': ('updated_at', T(parse_iso8601)),
'release_timestamp': ('start', T(parse_iso8601)),
'is_live': True,
}))