1
0
Fork 0
mirror of https://gitlab.com/dstftw/youtube-dl.git synced 2020-11-16 09:42:26 +00:00

[nrk] Improve extraction and update tests (closes #11571)

This commit is contained in:
Sergey M․ 2017-02-01 00:29:29 +07:00
parent 8fd65faece
commit 7c5329e6f4
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D

View file

@ -134,11 +134,15 @@ class NRKBaseIE(InfoExtractor):
_season_episode = data.get('scoresStatistics', {}).get('springStreamStream') or \ _season_episode = data.get('scoresStatistics', {}).get('springStreamStream') or \
data.get('relativeOriginUrl', '') data.get('relativeOriginUrl', '')
EPISODENUM_RE = [ EPISODENUM_RE = [
r'/s(?P<season>\d+)e(?P<episode>\d+)\.', r'/s(?P<season>\d{,2})e(?P<episode>\d{,2})\.',
r'/sesong-(?P<season>\d+)/episode-(?P<episode>\d+)', r'/sesong-(?P<season>\d{,2})/episode-(?P<episode>\d{,2})',
] ]
season_number = int_or_none(self._search_regex(EPISODENUM_RE, _season_episode, "S##E##", fatal=False, group='season')) season_number = int_or_none(self._search_regex(
episode_number = int_or_none(self._search_regex(EPISODENUM_RE, _season_episode, "S##E##", fatal=False, group='episode')) EPISODENUM_RE, _season_episode, 'season number',
default=None, group='season'))
episode_number = int_or_none(self._search_regex(
EPISODENUM_RE, _season_episode, 'episode number',
default=None, group='episode'))
thumbnails = None thumbnails = None
images = data.get('images') images = data.get('images')
@ -243,54 +247,102 @@ class NRKTVIE(NRKBaseIE):
'title': '20 spørsmål 23.05.2014', 'title': '20 spørsmål 23.05.2014',
'description': 'md5:bdea103bc35494c143c6a9acdd84887a', 'description': 'md5:bdea103bc35494c143c6a9acdd84887a',
'duration': 1741, 'duration': 1741,
'series': '20 spørsmål - TV',
'episode': '23.05.2014',
}, },
}, { }, {
'url': 'https://tv.nrk.no/program/mdfp15000514', 'url': 'https://tv.nrk.no/program/mdfp15000514',
'md5': '43d0be26663d380603a9cf0c24366531',
'info_dict': { 'info_dict': {
'id': 'MDFP15000514CA', 'id': 'MDFP15000514CA',
'ext': 'mp4', 'ext': 'mp4',
'title': 'Grunnlovsjubiléet - Stor ståhei for ingenting 24.05.2014', 'title': 'Grunnlovsjubiléet - Stor ståhei for ingenting 24.05.2014',
'description': 'md5:89290c5ccde1b3a24bb8050ab67fe1db', 'description': 'md5:89290c5ccde1b3a24bb8050ab67fe1db',
'duration': 4605, 'duration': 4605,
'series': 'Kunnskapskanalen',
'episode': '24.05.2014',
},
'params': {
'skip_download': True,
}, },
}, { }, {
# single playlist video # single playlist video
'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015#del=2', 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015#del=2',
'md5': 'adbd1dbd813edaf532b0a253780719c2',
'info_dict': { 'info_dict': {
'id': 'MSPO40010515-part2', 'id': 'MSPO40010515-part2',
'ext': 'flv', 'ext': 'flv',
'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 2:2)', 'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 2:2)',
'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26', 'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26',
}, },
'skip': 'Only works from Norway', 'params': {
'skip_download': True,
},
'expected_warnings': ['Video is geo restricted'],
'skip': 'particular part is not supported currently',
}, { }, {
'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015', 'url': 'https://tv.nrk.no/serie/tour-de-ski/MSPO40010515/06-01-2015',
'playlist': [{ 'playlist': [{
'md5': '9480285eff92d64f06e02a5367970a7a',
'info_dict': { 'info_dict': {
'id': 'MSPO40010515-part1', 'id': 'MSPO40010515AH',
'ext': 'flv', 'ext': 'mp4',
'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 1:2)', 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 1)',
'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26', 'description': 'md5:c03aba1e917561eface5214020551b7a',
'duration': 772,
'series': 'Tour de Ski',
'episode': '06.01.2015',
},
'params': {
'skip_download': True,
}, },
}, { }, {
'md5': 'adbd1dbd813edaf532b0a253780719c2',
'info_dict': { 'info_dict': {
'id': 'MSPO40010515-part2', 'id': 'MSPO40010515BH',
'ext': 'flv', 'ext': 'mp4',
'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn 06.01.2015 (del 2:2)', 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015 (Part 2)',
'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26', 'description': 'md5:c03aba1e917561eface5214020551b7a',
'duration': 6175,
'series': 'Tour de Ski',
'episode': '06.01.2015',
},
'params': {
'skip_download': True,
}, },
}], }],
'info_dict': { 'info_dict': {
'id': 'MSPO40010515', 'id': 'MSPO40010515',
'title': 'Tour de Ski: Sprint fri teknikk, kvinner og menn', 'title': 'Sprint fri teknikk, kvinner og menn 06.01.2015',
'description': 'md5:238b67b97a4ac7d7b4bf0edf8cc57d26', 'description': 'md5:c03aba1e917561eface5214020551b7a',
'duration': 6947.52, },
'expected_warnings': ['Video is geo restricted'],
}, {
'url': 'https://tv.nrk.no/serie/anno/KMTE50001317/sesong-3/episode-13',
'info_dict': {
'id': 'KMTE50001317AA',
'ext': 'mp4',
'title': 'Anno 13:30',
'description': 'md5:11d9613661a8dbe6f9bef54e3a4cbbfa',
'duration': 2340,
'series': 'Anno',
'episode': '13:30',
'season_number': 3,
'episode_number': 13,
},
'params': {
'skip_download': True,
},
}, {
'url': 'https://tv.nrk.no/serie/nytt-paa-nytt/MUHH46000317/27-01-2017',
'info_dict': {
'id': 'MUHH46000317AA',
'ext': 'mp4',
'title': 'Nytt på Nytt 27.01.2017',
'description': 'md5:5358d6388fba0ea6f0b6d11c48b9eb4b',
'duration': 1796,
'series': 'Nytt på nytt',
'episode': '27.01.2017',
},
'params': {
'skip_download': True,
}, },
'skip': 'Only works from Norway',
}, { }, {
'url': 'https://radio.nrk.no/serie/dagsnytt/NPUB21019315/12-07-2015#', 'url': 'https://radio.nrk.no/serie/dagsnytt/NPUB21019315/12-07-2015#',
'only_matching': True, 'only_matching': True,
@ -377,36 +429,61 @@ class NRKTVEpisodesIE(NRKPlaylistBaseIE):
class NRKTVSeriesIE(InfoExtractor): class NRKTVSeriesIE(InfoExtractor):
_VALID_URL = r'https?://tv\.nrk\.no/serie/(?P<id>[^/]+)/?' _VALID_URL = r'https?://(?:tv|radio)\.nrk(?:super)?\.no/serie/(?P<id>[^/]+)'
_ITEM_RE = r'data-season=["\'](?P<id>\d+)["\']' _ITEM_RE = r'(?:data-season=["\']|id=["\']season-)(?P<id>\d+)'
_TESTS = [{ _TESTS = [{
'url': 'https://tv.nrk.no/serie/groenn-glede',
'info_dict': {
'id': 'groenn-glede',
'title': 'Grønn glede',
'description': 'md5:7576e92ae7f65da6993cf90ee29e4608',
},
'playlist_mincount': 9,
}, {
'url': 'http://tv.nrksuper.no/serie/labyrint',
'info_dict': {
'id': 'labyrint',
'title': 'Labyrint',
'description': 'md5:58afd450974c89e27d5a19212eee7115',
},
'playlist_mincount': 3,
}, {
'url': 'https://tv.nrk.no/serie/broedrene-dal-og-spektralsteinene', 'url': 'https://tv.nrk.no/serie/broedrene-dal-og-spektralsteinene',
'playlist_count': 1, 'only_matching': True,
}, { }, {
'url': 'https://tv.nrk.no/serie/saving-the-human-race', 'url': 'https://tv.nrk.no/serie/saving-the-human-race',
'playlist_count': 1, 'only_matching': True,
}, { }, {
'url': 'https://tv.nrk.no/serie/postmann-pat', 'url': 'https://tv.nrk.no/serie/postmann-pat',
'playlist_count': 3, 'only_matching': True,
}, {
'url': 'https://tv.nrk.no/serie/groenn-glede',
'playlist_count': 9,
}] }]
@classmethod
def suitable(cls, url):
return False if NRKTVIE.suitable(url) else super(NRKTVSeriesIE, cls).suitable(url)
def _real_extract(self, url): def _real_extract(self, url):
series_id = self._match_id(url) series_id = self._match_id(url)
webpage = self._download_webpage(url, series_id) webpage = self._download_webpage(url, series_id)
entries = [ entries = [
self.url_result('https://tv.nrk.no/program/Episodes/{series}/{season}'.format( self.url_result(
series=series_id, 'https://tv.nrk.no/program/Episodes/{series}/{season}'.format(
season=season_id series=series_id, season=season_id))
))
for season_id in re.findall(self._ITEM_RE, webpage) for season_id in re.findall(self._ITEM_RE, webpage)
] ]
return self.playlist_result(entries) title = self._html_search_meta(
'seriestitle', webpage,
'title', default=None) or self._og_search_title(
webpage, fatal=False)
description = self._html_search_meta(
'series_description', webpage,
'description', default=None) or self._og_search_description(webpage)
return self.playlist_result(entries, series_id, title, description)
class NRKSkoleIE(InfoExtractor): class NRKSkoleIE(InfoExtractor):