X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/9dc487f48b50767cf540fa36c3de2c386fd74c04..7b550e005da7fd498cfbf1b7c04b05e5540eb6fc:/youtube_dl/extractor/iprima.py?ds=sidebyside diff --git a/youtube_dl/extractor/iprima.py b/youtube_dl/extractor/iprima.py index 788bbe0..11bbeb5 100644 --- a/youtube_dl/extractor/iprima.py +++ b/youtube_dl/extractor/iprima.py @@ -8,12 +8,12 @@ from .common import InfoExtractor from ..utils import ( determine_ext, js_to_json, - sanitized_Request, ) class IPrimaIE(InfoExtractor): - _VALID_URL = r'https?://play\.iprima\.cz/(?:.+/)?(?P[^?#]+)' + _VALID_URL = r'https?://(?:[^/]+)\.iprima\.cz/(?:[^/]+/)*(?P[^/?#&]+)' + _GEO_BYPASS = False _TESTS = [{ 'url': 'http://play.iprima.cz/gondici-s-r-o-33', @@ -29,20 +29,57 @@ class IPrimaIE(InfoExtractor): }, { 'url': 'http://play.iprima.cz/particka/particka-92', 'only_matching': True, + }, { + # geo restricted + 'url': 'http://play.iprima.cz/closer-nove-pripady/closer-nove-pripady-iv-1', + 'only_matching': True, + }, { + # iframe api.play-backend.iprima.cz + 'url': 'https://prima.iprima.cz/my-little-pony/mapa-znameni-2-2', + 'only_matching': True, + }, { + # iframe prima.iprima.cz + 'url': 'https://prima.iprima.cz/porady/jak-se-stavi-sen/rodina-rathousova-praha', + 'only_matching': True, + }, { + 'url': 'http://www.iprima.cz/filmy/desne-rande', + 'only_matching': True, + }, { + 'url': 'https://zoom.iprima.cz/10-nejvetsich-tajemstvi-zahad/posvatna-mista-a-stavby', + 'only_matching': True, + }, { + 'url': 'https://krimi.iprima.cz/mraz-0/sebevrazdy', + 'only_matching': True, + }, { + 'url': 'https://cool.iprima.cz/derava-silnice-nevadi', + 'only_matching': True, + }, { + 'url': 'https://love.iprima.cz/laska-az-za-hrob/slib-dany-bratrovi', + 'only_matching': True, + }, { + 'url': 'https://autosalon.iprima.cz/motorsport/7-epizoda-1', + 'only_matching': True, }] def _real_extract(self, url): video_id = self._match_id(url) + self._set_cookie('play.iprima.cz', 'ott_adult_confirmed', '1') + webpage = self._download_webpage(url, video_id) - video_id = self._search_regex(r'data-product="([^"]+)">', webpage, 'real id') + video_id = self._search_regex( + (r']+\bsrc=["\'](?:https?:)?//(?:api\.play-backend\.iprima\.cz/prehravac/embedded|prima\.iprima\.cz/[^/]+/[^/]+)\?.*?\bid=(p\d+)', + r'data-product="([^"]+)">'), + webpage, 'real id') - req = sanitized_Request( - 'http://play.iprima.cz/prehravac/init?_infuse=1' - '&_ts=%s&productId=%s' % (round(time.time()), video_id)) - req.add_header('Referer', url) - playerpage = self._download_webpage(req, video_id, note='Downloading player') + playerpage = self._download_webpage( + 'http://play.iprima.cz/prehravac/init', + video_id, note='Downloading player', query={ + '_infuse': 1, + '_ts': round(time.time()), + 'productId': video_id, + }, headers={'Referer': url}) formats = [] @@ -65,7 +102,7 @@ class IPrimaIE(InfoExtractor): options = self._parse_json( self._search_regex( - r'(?s)var\s+playerOptions\s*=\s*({.+?});', + r'(?s)(?:TDIPlayerOptions|playerOptions)\s*=\s*({.+?});\s*\]\]', playerpage, 'player options', default='{}'), video_id, transform_source=js_to_json, fatal=False) if options: @@ -81,6 +118,9 @@ class IPrimaIE(InfoExtractor): for _, src in re.findall(r'src["\']\s*:\s*(["\'])(.+?)\1', playerpage): extract_formats(src) + if not formats and '>GEO_IP_NOT_ALLOWED<' in playerpage: + self.raise_geo_restricted(countries=['CZ']) + self._sort_formats(formats) return {