X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/415fdb62500dca2e22067a05008dfbf87c75b662..67afe88251ff8b1fcb43246c9a4bb1a0ea0185a2:/youtube_dl/extractor/trutube.py diff --git a/youtube_dl/extractor/trutube.py b/youtube_dl/extractor/trutube.py index 57f9566..e7b7924 100644 --- a/youtube_dl/extractor/trutube.py +++ b/youtube_dl/extractor/trutube.py @@ -1,13 +1,12 @@ from __future__ import unicode_literals -import re - from .common import InfoExtractor +from ..utils import xpath_text class TruTubeIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?trutube\.tv/video/(?P[0-9]+)/.*' - _TEST = { + _VALID_URL = r'https?://(?:www\.)?trutube\.tv/(?:video/|nuevo/player/embed\.php\?v=)(?P[0-9]+)' + _TESTS = [{ 'url': 'http://trutube.tv/video/14880/Ramses-II-Proven-To-Be-A-Red-Headed-Caucasoid-', 'md5': 'c5b6e301b0a2040b074746cbeaa26ca1', 'info_dict': { @@ -16,29 +15,26 @@ class TruTubeIE(InfoExtractor): 'title': 'Ramses II - Proven To Be A Red Headed Caucasoid', 'thumbnail': 're:^http:.*\.jpg$', } - } + }, { + 'url': 'https://trutube.tv/nuevo/player/embed.php?v=14880', + 'only_matching': True, + }] def _real_extract(self, url): - mobj = re.match(self._VALID_URL, url) - video_id = mobj.group('id') + video_id = self._match_id(url) - webpage = self._download_webpage(url, video_id) - video_title = self._og_search_title(webpage).strip() - thumbnail = self._search_regex( - r"var splash_img = '([^']+)';", webpage, 'thumbnail', fatal=False) + config = self._download_xml( + 'https://trutube.tv/nuevo/player/config.php?v=%s' % video_id, + video_id, transform_source=lambda s: s.strip()) - all_formats = re.finditer( - r"var (?P[a-z]+)_video_file\s*=\s*'(?P[^']+)';", webpage) - formats = [{ - 'format_id': m.group('key'), - 'quality': -i, - 'url': m.group('url'), - } for i, m in enumerate(all_formats)] - self._sort_formats(formats) + # filehd is always 404 + video_url = xpath_text(config, './file', 'video URL', fatal=True) + title = xpath_text(config, './title', 'title').strip() + thumbnail = xpath_text(config, './image', ' thumbnail') return { 'id': video_id, - 'title': video_title, - 'formats': formats, + 'url': video_url, + 'title': title, 'thumbnail': thumbnail, }