From 6aeba407db84a636fc2522b4f2344eac9e0c1fdb Mon Sep 17 00:00:00 2001 From: remitamine Date: Fri, 25 Sep 2015 10:52:48 +0100 Subject: [jukebox] remove extractor and handle it using generic extractor --- youtube_dl/extractor/ultimedia.py | 97 ++++++++++++++++----------------------- 1 file changed, 39 insertions(+), 58 deletions(-) (limited to 'youtube_dl/extractor/ultimedia.py') diff --git a/youtube_dl/extractor/ultimedia.py b/youtube_dl/extractor/ultimedia.py index c4751050e..45201332d 100644 --- a/youtube_dl/extractor/ultimedia.py +++ b/youtube_dl/extractor/ultimedia.py @@ -4,102 +4,83 @@ from __future__ import unicode_literals import re from .common import InfoExtractor -from ..compat import compat_urllib_parse_urlparse -from ..utils import ( - ExtractorError, - qualities, - unified_strdate, - clean_html, -) +from ..utils import int_or_none class UltimediaIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?ultimedia\.com/default/index/video[^/]+/id/(?P[\d+a-z]+)' + _VALID_URL = r'https?://(?:www\.)?ultimedia\.com/deliver/(?Pgeneric|musique)(?:/[^/]+)*/(?:src|article)/(?P[\d+a-z]+)' _TESTS = [{ # news - 'url': 'https://www.ultimedia.com/default/index/videogeneric/id/s8uk0r', + 'url': 'https://www.ultimedia.com/deliver/generic/iframe/mdtk/01601930/zone/1/src/s8uk0r/autoplay/yes/ad/no/width/714/height/435', 'md5': '276a0e49de58c7e85d32b057837952a2', 'info_dict': { 'id': 's8uk0r', 'ext': 'mp4', 'title': 'Loi sur la fin de vie: le texte prévoit un renforcement des directives anticipées', - 'description': 'md5:3e5c8fd65791487333dda5db8aed32af', 'thumbnail': 're:^https?://.*\.jpg', + 'duration': 74, 'upload_date': '20150317', + 'timestamp': 1426604939, + 'uploader_id': '3fszv', }, }, { # music - 'url': 'https://www.ultimedia.com/default/index/videomusic/id/xvpfp8', + 'url': 'https://www.ultimedia.com/deliver/musique/iframe/mdtk/01601930/zone/1/article/xvpfp8/autoplay/yes/ad/no/width/714/height/435', 'md5': '2ea3513813cf230605c7e2ffe7eca61c', 'info_dict': { 'id': 'xvpfp8', 'ext': 'mp4', - 'title': "Two - C'est la vie (Clip)", - 'description': 'Two', + 'title': 'Two - C\'est La Vie (clip)', 'thumbnail': 're:^https?://.*\.jpg', + 'duration': 233, 'upload_date': '20150224', + 'timestamp': 1424760500, + 'uploader_id': '3rfzk', }, }] - def _real_extract(self, url): - video_id = self._match_id(url) - webpage = self._download_webpage(url, video_id) + @staticmethod + def _extract_url(webpage): + mobj = re.search( + r'<(?:iframe|script)[^>]+src=["\'](?P(?:https?:)?//(?:www\.)?ultimedia\.com/deliver/(?:generic|musique)(?:/[^/]+)*/(?:src|article)/[\d+a-z]+)', + webpage) + if mobj: + return mobj.group('url') - deliver_url = self._proto_relative_url(self._search_regex( - r']+src="((?:https?:)?//(?:www\.)?ultimedia\.com/deliver/[^"]+)"', - webpage, 'deliver URL'), compat_urllib_parse_urlparse(url).scheme + ':') + def _real_extract(self, url): + video_type, video_id = re.match(self._VALID_URL, url).groups() - deliver_page = self._download_webpage( - deliver_url, video_id, 'Downloading iframe page') + deliver_info = self._download_json( + 'http://www.ultimedia.com/deliver/video?video=%s&topic=%s' % (video_id, video_type), + video_id) - if '>This video is currently not available' in deliver_page: - raise ExtractorError( - 'Video %s is currently not available' % video_id, expected=True) + yt_id = deliver_info.get('yt_id') + if yt_id: + return self.url_result(yt_id, 'Youtube') - player = self._parse_json( - self._search_regex( - r"jwplayer\('player(?:_temp)?'\)\.setup\(({.+?})\)\.on", - deliver_page, 'player'), - video_id) + jwconf = deliver_info['jwconf'] - quality = qualities(['flash', 'html5']) formats = [] - for mode in player['modes']: - video_url = mode.get('config', {}).get('file') - if not video_url: - continue - if re.match(r'https?://www\.youtube\.com/.+?', video_url): - return self.url_result(video_url, 'Youtube') + for source in jwconf['playlist'][0]['sources']: formats.append({ - 'url': video_url, - 'format_id': mode.get('type'), - 'quality': quality(mode.get('type')), + 'url': source['file'], + 'format_id': source.get('label'), }) - self._sort_formats(formats) - - thumbnail = player.get('image') - title = clean_html(( - self._html_search_regex( - r'(?s).+?(.+?)', - webpage, 'title', default=None) or - self._search_regex( - r"var\s+nameVideo\s*=\s*'([^']+)'", - deliver_page, 'title'))) - - description = clean_html(self._html_search_regex( - r'(?s)Description(.+?)

', webpage, - 'description', fatal=False)) + self._sort_formats(formats) - upload_date = unified_strdate(self._search_regex( - r'Ajouté le\s*([^<]+)', webpage, - 'upload date', fatal=False)) + title = deliver_info['title'] + thumbnail = jwconf.get('image') + duration = int_or_none(deliver_info.get('duration')) + timestamp = int_or_none(deliver_info.get('release_time')) + uploader_id = deliver_info.get('owner_id') return { 'id': video_id, 'title': title, - 'description': description, 'thumbnail': thumbnail, - 'upload_date': upload_date, + 'duration': duration, + 'timestamp': timestamp, + 'uploader_id': uploader_id, 'formats': formats, } -- cgit v1.2.3