X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/575f75429e2e479957d006a26940f4a5a51347ce..bc5a42ee389bd51b63eb8a6b914aa8139762bc10:/youtube_dl/extractor/viu.py diff --git a/youtube_dl/extractor/viu.py b/youtube_dl/extractor/viu.py index 3fd889c..5cf9359 100644 --- a/youtube_dl/extractor/viu.py +++ b/youtube_dl/extractor/viu.py @@ -4,7 +4,10 @@ from __future__ import unicode_literals import re from .common import InfoExtractor -from ..compat import compat_str +from ..compat import ( + compat_kwargs, + compat_str, +) from ..utils import ( ExtractorError, int_or_none, @@ -36,7 +39,8 @@ class ViuBaseIE(InfoExtractor): headers.update(kwargs.get('headers', {})) kwargs['headers'] = headers response = self._download_json( - 'https://www.viu.com/api/' + path, *args, **kwargs)['response'] + 'https://www.viu.com/api/' + path, *args, + **compat_kwargs(kwargs))['response'] if response.get('status') != 'success': raise ExtractorError('%s said: %s' % ( self.IE_NAME, response['message']), expected=True) @@ -44,7 +48,7 @@ class ViuBaseIE(InfoExtractor): class ViuIE(ViuBaseIE): - _VALID_URL = r'(?:viu:|https?://www\.viu\.com/[a-z]{2}/media/)(?P\d+)' + _VALID_URL = r'(?:viu:|https?://[^/]+\.viu\.com/[a-z]{2}/media/)(?P\d+)' _TESTS = [{ 'url': 'https://www.viu.com/en/media/1116705532?containerId=playlist-22168059', 'info_dict': { @@ -69,6 +73,9 @@ class ViuIE(ViuBaseIE): 'skip_download': 'm3u8 download', }, 'skip': 'Geo-restricted to Indonesia', + }, { + 'url': 'https://india.viu.com/en/media/1126286865', + 'only_matching': True, }] def _real_extract(self, url):