2 from __future__ import unicode_literals
7 from .common import InfoExtractor
14 class VierIE(InfoExtractor):
16 IE_DESC = 'vier.be and vijf.be'
17 _VALID_URL = r'https?://(?:www\.)?(?P<site>vier|vijf)\.be/(?:[^/]+/videos/(?P<display_id>[^/]+)(?:/(?P<id>\d+))?|video/v3/embed/(?P<embed_id>\d+))'
18 _NETRC_MACHINE = 'vier'
20 'url': 'http://www.vier.be/planb/videos/het-wordt-warm-de-moestuin/16129',
23 'display_id': 'het-wordt-warm-de-moestuin',
25 'title': 'Het wordt warm in De Moestuin',
26 'description': 'De vele uren werk eisen hun tol. Wim droomt van assistentie...',
30 'skip_download': True,
32 # 'skip': 'Requires account credentials',
34 'url': 'http://www.vijf.be/temptationisland/videos/zo-grappig-temptation-island-hosts-moeten-kiezen-tussen-onmogelijke-dilemmas/2561614',
37 'display_id': 'zo-grappig-temptation-island-hosts-moeten-kiezen-tussen-onmogelijke-dilemmas',
39 'title': 'EXTRA: Temptation Island hosts moeten kiezen tussen onmogelijke dilemma\'s',
40 'description': 'Het spel is simpel: Annelien Coorevits en Rick Brandsteder krijgen telkens 2 dilemma\'s voorgeschoteld en ze MOETEN een keuze maken.',
44 'skip_download': True,
47 'url': 'http://www.vier.be/janigaat/videos/jani-gaat-naar-tokio-aflevering-4/2674839',
50 'display_id': 'jani-gaat-naar-tokio-aflevering-4',
52 'title': 'Jani gaat naar Tokio - Aflevering 4',
53 'description': 'Bekijk hier de volledige vierde aflevering van het 2de seizoen van Jani gaat...',
57 'skip_download': True,
59 'skip': 'Requires account credentials',
61 'url': 'http://www.vier.be/janigaat/videos/jani-gaat-naar-tokio-aflevering-4/2674839',
64 'display_id': 'jani-gaat-naar-tokio-aflevering-4',
66 'title': 'jani-gaat-naar-tokio-aflevering-4',
70 'skip_download': True,
72 'expected_warnings': ['Log in to extract metadata'],
74 'url': 'http://www.vier.be/planb/videos/mieren-herders-van-de-bladluizen',
75 'only_matching': True,
77 'url': 'http://www.vier.be/video/v3/embed/16129',
78 'only_matching': True,
81 def _real_initialize(self):
82 self._logged_in = False
84 def _login(self, site):
85 username, password = self._get_login_info()
86 if username is None or password is None:
89 login_page = self._download_webpage(
90 'http://www.%s.be/user/login' % site,
91 None, note='Logging in', errnote='Unable to log in',
92 data=urlencode_postdata({
93 'form_id': 'user_login',
97 headers={'Content-Type': 'application/x-www-form-urlencoded'})
99 login_error = self._html_search_regex(
100 r'(?s)<div class="messages error">\s*<div>\s*<h2.+?</h2>(.+?)<',
101 login_page, 'login error', default=None)
103 self.report_warning('Unable to log in: %s' % login_error)
105 self._logged_in = True
107 def _real_extract(self, url):
108 mobj = re.match(self._VALID_URL, url)
109 embed_id = mobj.group('embed_id')
110 display_id = mobj.group('display_id') or embed_id
111 video_id = mobj.group('id') or embed_id
112 site = mobj.group('site')
114 if not self._logged_in:
117 webpage = self._download_webpage(url, display_id)
119 if r'id="user-login"' in webpage:
121 'Log in to extract metadata', video_id=display_id)
122 webpage = self._download_webpage(
123 'http://www.%s.be/video/v3/embed/%s' % (site, video_id),
126 video_id = self._search_regex(
127 [r'data-nid="(\d+)"', r'"nid"\s*:\s*"(\d+)"'],
128 webpage, 'video id', default=video_id)
129 application = self._search_regex(
130 [r'data-application="([^"]+)"', r'"application"\s*:\s*"([^"]+)"'],
131 webpage, 'application', default=site + '_vod')
132 filename = self._search_regex(
133 [r'data-filename="([^"]+)"', r'"filename"\s*:\s*"([^"]+)"'],
136 playlist_url = 'http://vod.streamcloud.be/%s/_definst_/mp4:%s.mp4/playlist.m3u8' % (application, filename)
137 formats = self._extract_wowza_formats(playlist_url, display_id, skip_protocols=['dash'])
138 self._sort_formats(formats)
140 title = self._og_search_title(webpage, default=display_id)
141 description = self._og_search_description(webpage, default=None)
142 thumbnail = self._og_search_thumbnail(webpage, default=None)
146 'display_id': display_id,
148 'description': description,
149 'thumbnail': thumbnail,
154 class VierVideosIE(InfoExtractor):
155 IE_NAME = 'vier:videos'
156 _VALID_URL = r'https?://(?:www\.)?(?P<site>vier|vijf)\.be/(?P<program>[^/]+)/videos(?:\?.*\bpage=(?P<page>\d+)|$)'
158 'url': 'http://www.vier.be/demoestuin/videos',
162 'playlist_mincount': 153,
164 'url': 'http://www.vijf.be/temptationisland/videos',
166 'id': 'temptationisland',
168 'playlist_mincount': 159,
170 'url': 'http://www.vier.be/demoestuin/videos?page=6',
172 'id': 'demoestuin-page6',
174 'playlist_mincount': 20,
176 'url': 'http://www.vier.be/demoestuin/videos?page=7',
178 'id': 'demoestuin-page7',
180 'playlist_mincount': 13,
183 def _real_extract(self, url):
184 mobj = re.match(self._VALID_URL, url)
185 program = mobj.group('program')
186 site = mobj.group('site')
188 page_id = mobj.group('page')
190 page_id = int(page_id)
192 playlist_id = '%s-page%d' % (program, page_id)
195 playlist_id = program
198 for current_page_id in itertools.count(start_page):
199 current_page = self._download_webpage(
200 'http://www.%s.be/%s/videos?page=%d' % (site, program, current_page_id),
202 'Downloading page %d' % (current_page_id + 1))
204 self.url_result('http://www.' + site + '.be' + video_url, 'Vier')
205 for video_url in re.findall(
206 r'<h[23]><a href="(/[^/]+/videos/[^/]+(?:/\d+)?)">', current_page)]
207 entries.extend(page_entries)
208 if page_id or '>Meer<' not in current_page:
211 return self.playlist_result(entries, playlist_id)