2 from __future__ import unicode_literals
7 from .common import InfoExtractor
8 from ..utils import urlencode_postdata
11 class VierIE(InfoExtractor):
13 IE_DESC = 'vier.be and vijf.be'
14 _VALID_URL = r'https?://(?:www\.)?(?P<site>vier|vijf)\.be/(?:[^/]+/videos/(?P<display_id>[^/]+)(?:/(?P<id>\d+))?|video/v3/embed/(?P<embed_id>\d+))'
15 _NETRC_MACHINE = 'vier'
17 'url': 'http://www.vier.be/planb/videos/het-wordt-warm-de-moestuin/16129',
18 'md5': 'e4ae2054a6b040ef1e289e20d111b46e',
21 'display_id': 'het-wordt-warm-de-moestuin',
23 'title': 'Het wordt warm in De Moestuin',
24 'description': 'De vele uren werk eisen hun tol. Wim droomt van assistentie...',
27 'url': 'http://www.vijf.be/temptationisland/videos/zo-grappig-temptation-island-hosts-moeten-kiezen-tussen-onmogelijke-dilemmas/2561614',
30 'display_id': 'zo-grappig-temptation-island-hosts-moeten-kiezen-tussen-onmogelijke-dilemmas',
32 'title': 'md5:84f45fe48b8c1fa296a7f6d208d080a7',
33 'description': 'md5:0356d4981e58b8cbee19355cbd51a8fe',
36 'skip_download': True,
39 'url': 'http://www.vier.be/janigaat/videos/jani-gaat-naar-tokio-aflevering-4/2674839',
42 'display_id': 'jani-gaat-naar-tokio-aflevering-4',
44 'title': 'Jani gaat naar Tokio - Aflevering 4',
45 'description': 'md5:2d169e8186ae4247e50c99aaef97f7b2',
48 'skip_download': True,
50 'skip': 'Requires account credentials',
52 # Requires account credentials but bypassed extraction via v3/embed page
54 'url': 'http://www.vier.be/janigaat/videos/jani-gaat-naar-tokio-aflevering-4/2674839',
57 'display_id': 'jani-gaat-naar-tokio-aflevering-4',
59 'title': 'jani-gaat-naar-tokio-aflevering-4',
62 'skip_download': True,
64 'expected_warnings': ['Log in to extract metadata'],
66 # Without video id in URL
67 'url': 'http://www.vier.be/planb/videos/dit-najaar-plan-b',
68 'only_matching': True,
70 'url': 'http://www.vier.be/video/v3/embed/16129',
71 'only_matching': True,
74 def _real_initialize(self):
75 self._logged_in = False
77 def _login(self, site):
78 username, password = self._get_login_info()
79 if username is None or password is None:
82 login_page = self._download_webpage(
83 'http://www.%s.be/user/login' % site,
84 None, note='Logging in', errnote='Unable to log in',
85 data=urlencode_postdata({
86 'form_id': 'user_login',
90 headers={'Content-Type': 'application/x-www-form-urlencoded'})
92 login_error = self._html_search_regex(
93 r'(?s)<div class="messages error">\s*<div>\s*<h2.+?</h2>(.+?)<',
94 login_page, 'login error', default=None)
96 self.report_warning('Unable to log in: %s' % login_error)
98 self._logged_in = True
100 def _real_extract(self, url):
101 mobj = re.match(self._VALID_URL, url)
102 embed_id = mobj.group('embed_id')
103 display_id = mobj.group('display_id') or embed_id
104 video_id = mobj.group('id') or embed_id
105 site = mobj.group('site')
107 if not self._logged_in:
110 webpage = self._download_webpage(url, display_id)
112 if r'id="user-login"' in webpage:
114 'Log in to extract metadata', video_id=display_id)
115 webpage = self._download_webpage(
116 'http://www.%s.be/video/v3/embed/%s' % (site, video_id),
119 video_id = self._search_regex(
120 [r'data-nid="(\d+)"', r'"nid"\s*:\s*"(\d+)"'],
121 webpage, 'video id', default=video_id or display_id)
122 application = self._search_regex(
123 [r'data-application="([^"]+)"', r'"application"\s*:\s*"([^"]+)"'],
124 webpage, 'application', default=site + '_vod')
125 filename = self._search_regex(
126 [r'data-filename="([^"]+)"', r'"filename"\s*:\s*"([^"]+)"'],
129 playlist_url = 'http://vod.streamcloud.be/%s/_definst_/mp4:%s.mp4/playlist.m3u8' % (application, filename)
130 formats = self._extract_wowza_formats(playlist_url, display_id, skip_protocols=['dash'])
131 self._sort_formats(formats)
133 title = self._og_search_title(webpage, default=display_id)
134 description = self._og_search_description(webpage, default=None)
135 thumbnail = self._og_search_thumbnail(webpage, default=None)
139 'display_id': display_id,
141 'description': description,
142 'thumbnail': thumbnail,
147 class VierVideosIE(InfoExtractor):
148 IE_NAME = 'vier:videos'
149 _VALID_URL = r'https?://(?:www\.)?(?P<site>vier|vijf)\.be/(?P<program>[^/]+)/videos(?:\?.*\bpage=(?P<page>\d+)|$)'
151 'url': 'http://www.vier.be/demoestuin/videos',
155 'playlist_mincount': 153,
157 'url': 'http://www.vijf.be/temptationisland/videos',
159 'id': 'temptationisland',
161 'playlist_mincount': 159,
163 'url': 'http://www.vier.be/demoestuin/videos?page=6',
165 'id': 'demoestuin-page6',
167 'playlist_mincount': 20,
169 'url': 'http://www.vier.be/demoestuin/videos?page=7',
171 'id': 'demoestuin-page7',
173 'playlist_mincount': 13,
176 def _real_extract(self, url):
177 mobj = re.match(self._VALID_URL, url)
178 program = mobj.group('program')
179 site = mobj.group('site')
181 page_id = mobj.group('page')
183 page_id = int(page_id)
185 playlist_id = '%s-page%d' % (program, page_id)
188 playlist_id = program
191 for current_page_id in itertools.count(start_page):
192 current_page = self._download_webpage(
193 'http://www.%s.be/%s/videos?page=%d' % (site, program, current_page_id),
195 'Downloading page %d' % (current_page_id + 1))
197 self.url_result('http://www.' + site + '.be' + video_url, 'Vier')
198 for video_url in re.findall(
199 r'<h[23]><a href="(/[^/]+/videos/[^/]+(?:/\d+)?)">', current_page)]
200 entries.extend(page_entries)
201 if page_id or '>Meer<' not in current_page:
204 return self.playlist_result(entries, playlist_id)