From 950505370416e79073acdeaa73a1023950363266 Mon Sep 17 00:00:00 2001 From: Aidan Rowe Date: Mon, 8 Feb 2016 20:57:07 +1000 Subject: [PATCH] [dplay] add support for it.dplay.com and dplay.dk --- youtube_dl/extractor/dplay.py | 84 +++++++++++++++++++++++++++-------- 1 file changed, 65 insertions(+), 19 deletions(-) diff --git a/youtube_dl/extractor/dplay.py b/youtube_dl/extractor/dplay.py index 6cda56a7fd..adcfc1f0ef 100644 --- a/youtube_dl/extractor/dplay.py +++ b/youtube_dl/extractor/dplay.py @@ -5,40 +5,83 @@ import time from .common import InfoExtractor from ..utils import int_or_none - +from ..compat import compat_urlparse class DPlayIE(InfoExtractor): - _VALID_URL = r'http://www\.dplay\.se/[^/]+/(?P[^/?#]+)' + _VALID_URL = r'(?Phttp://(?:it|www)\.dplay\.(?:com|dk|se))/[^/]+/(?P[^/?#]+)' - _TEST = { - 'url': 'http://www.dplay.se/nugammalt-77-handelser-som-format-sverige/season-1-svensken-lar-sig-njuta-av-livet/', - 'info_dict': { - 'id': '3172', - 'ext': 'mp4', - 'display_id': 'season-1-svensken-lar-sig-njuta-av-livet', - 'title': 'Svensken lär sig njuta av livet', - 'duration': 2650, + _TESTS = [ + { + 'url': 'http://it.dplay.com/take-me-out/stagione-1-episodio-25/', + 'info_dict': { + 'id': '1255600', + 'ext': 'mp4', + 'display_id': 'stagione-1-episodio-25', + 'title': 'Episodio 25', + 'duration': 2761, + 'description': "Gabriele Corsi conduce un nuovo provocante e divertente dating show. 30 ragazze single hanno l'opportunità di conoscere un ragazzo e decidere se tenerlo in gioco oppure no accendendo o spegnendo le luci.", + 'season_number': 1, + 'episode_number': 25, + }, }, - } + { + 'url': 'http://www.dplay.se/nugammalt-77-handelser-som-format-sverige/season-1-svensken-lar-sig-njuta-av-livet/', + 'info_dict': { + 'id': '3172', + 'ext': 'mp4', + 'display_id': 'season-1-svensken-lar-sig-njuta-av-livet', + 'title': 'Svensken lär sig njuta av livet', + 'duration': 2650, + 'description': "\"Svensken lär sig njuta av livet\". Införandet av systembolaget, industrisemestern och Skarastadgan. Med hjälp av arkivmaterial, experter och fakta ska händelserna dissekeras, analyseras och dras till sin absoluta underhållningsspets.", + 'season_number': 1, + 'episode_number': 1, + }, + }, + { + 'url': 'http://www.dplay.dk/mig-og-min-mor/season-6-episode-12/', + 'info_dict': { + 'id': '70816', + 'ext': 'mp4', + 'display_id': 'season-6-episode-12', + 'title': 'Episode 12', + 'duration': 2563, + 'description': " I sæsonafslutningen sker der store ting for mor og datter.\nDagen er endelig kommet for den højgravide Irina - hun skal føde! Men det bliver en lang og sej kamp for Irina, som selvfølgelig har mor Mila med som støtte hele vejen.\nMor Jette og Jessica er igen hjemme i Danmark efter deres store USA-eventyr. Og for at holde fast i den amerikanske ånd, tager pigerne i dag til gospel-undervisning. \nOg så skal Joy og mor Mia under kniven - de skal nemlig have gjort lårene mindre og ballerne større. \n ", + 'season_number': 6, + 'episode_number': 12, + }, + } + ] def _real_extract(self, url): + #this extrator works with it.dplay.com, www.dplay.se and www.dplay.dk + # so we need to determine the domain to send the requests to + domain = self._search_regex(self._VALID_URL, url, 'domain') display_id = self._match_id(url) webpage = self._download_webpage(url, display_id) video_id = self._search_regex( r'data-video-id="(\d+)"', webpage, 'video id') - info = self._download_json( - 'http://www.dplay.se/api/v2/ajax/videos?video_id=' + video_id, + video_url = compat_urlparse.urljoin(domain, 'api/v2/ajax/videos?video_id=') + + info = self._download_json(video_url + video_id, video_id)['data'][0] - self._set_cookie( - 'secure.dplay.se', 'dsc-geo', - '{"countryCode":"NL","expiry":%d}' % ((time.time() + 20 * 60) * 1000)) # TODO: consider adding support for 'stream_type=hds', it seems to # require setting some cookies - manifest_url = self._download_json( - 'https://secure.dplay.se/secure/api/v2/user/authorization/stream/%s?stream_type=hls' % video_id, - video_id, 'Getting manifest url for hls stream')['hls'] + # get url's TLD to determine which cookie and url to use + domain_tld = domain.split('.')[-1] + if domain_tld == 'se' or domain_tld == 'dk': + self._set_cookie( + 'secure.dplay.%s' % domain_tld, 'dsc-geo', + '{"countryCode":"%s","expiry":%d}' % (domain_tld.upper(), ((time.time() + 20 * 60) * 1000))) + + manifest_url = self._download_json( + 'https://secure.dplay.%s/secure/api/v2/user/authorization/stream/%s?stream_type=hls' % (domain_tld, video_id), + video_id, 'Getting manifest url for hls stream')['hls'] + else: + #.it requires no cookies at this point + manifest_url = info['hls'] + formats = self._extract_m3u8_formats( manifest_url, video_id, ext='mp4', entry_protocol='m3u8_native') @@ -48,4 +91,7 @@ class DPlayIE(InfoExtractor): 'title': info['title'], 'formats': formats, 'duration': int_or_none(info.get('video_metadata_length'), scale=1000), + 'description': info.get('video_metadata_longDescription'), + 'season_number': int_or_none(info.get('season')), + 'episode_number': int_or_none(info.get('episode')), }