2 from __future__ import unicode_literals
7 from .common import InfoExtractor
15 class VierIE(InfoExtractor):
17 IE_DESC = 'vier.be and vijf.be'
18 _VALID_URL = r'https?://(?:www\.)?(?P<site>vier|vijf)\.be/(?:[^/]+/videos/(?P<display_id>[^/]+)(?:/(?P<id>\d+))?|video/v3/embed/(?P<embed_id>\d+))'
19 _NETRC_MACHINE = 'vier'
21 'url': 'http://www.vier.be/planb/videos/het-wordt-warm-de-moestuin/16129',
22 'md5': 'e4ae2054a6b040ef1e289e20d111b46e',
25 'display_id': 'het-wordt-warm-de-moestuin',
27 'title': 'Het wordt warm in De Moestuin',
28 'description': 'De vele uren werk eisen hun tol. Wim droomt van assistentie...',
29 'upload_date': '20121025',
32 'url': 'http://www.vijf.be/temptationisland/videos/zo-grappig-temptation-island-hosts-moeten-kiezen-tussen-onmogelijke-dilemmas/2561614',
35 'display_id': 'zo-grappig-temptation-island-hosts-moeten-kiezen-tussen-onmogelijke-dilemmas',
37 'title': 'md5:84f45fe48b8c1fa296a7f6d208d080a7',
38 'description': 'md5:0356d4981e58b8cbee19355cbd51a8fe',
39 'upload_date': '20170228',
42 'skip_download': True,
45 'url': 'http://www.vier.be/janigaat/videos/jani-gaat-naar-tokio-aflevering-4/2674839',
48 'display_id': 'jani-gaat-naar-tokio-aflevering-4',
50 'title': 'Jani gaat naar Tokio - Aflevering 4',
51 'description': 'md5:aa8d611541db6ae9e863125704511f88',
52 'upload_date': '20170501',
56 'skip_download': True,
58 'skip': 'Requires account credentials',
60 # Requires account credentials but bypassed extraction via v3/embed page
62 'url': 'http://www.vier.be/janigaat/videos/jani-gaat-naar-tokio-aflevering-4/2674839',
65 'display_id': 'jani-gaat-naar-tokio-aflevering-4',
67 'title': 'jani-gaat-naar-tokio-aflevering-4',
70 'skip_download': True,
72 'expected_warnings': ['Log in to extract metadata'],
74 # Without video id in URL
75 'url': 'http://www.vier.be/planb/videos/dit-najaar-plan-b',
76 'only_matching': True,
78 'url': 'http://www.vier.be/video/v3/embed/16129',
79 'only_matching': True,
82 def _real_initialize(self):
83 self._logged_in = False
85 def _login(self, site):
86 username, password = self._get_login_info()
87 if username is None or password is None:
90 login_page = self._download_webpage(
91 'http://www.%s.be/user/login' % site,
92 None, note='Logging in', errnote='Unable to log in',
93 data=urlencode_postdata({
94 'form_id': 'user_login',
98 headers={'Content-Type': 'application/x-www-form-urlencoded'})
100 login_error = self._html_search_regex(
101 r'(?s)<div class="messages error">\s*<div>\s*<h2.+?</h2>(.+?)<',
102 login_page, 'login error', default=None)
104 self.report_warning('Unable to log in: %s' % login_error)
106 self._logged_in = True
108 def _real_extract(self, url):
109 mobj = re.match(self._VALID_URL, url)
110 embed_id = mobj.group('embed_id')
111 display_id = mobj.group('display_id') or embed_id
112 video_id = mobj.group('id') or embed_id
113 site = mobj.group('site')
115 if not self._logged_in:
118 webpage = self._download_webpage(url, display_id)
120 if r'id="user-login"' in webpage:
122 'Log in to extract metadata', video_id=display_id)
123 webpage = self._download_webpage(
124 'http://www.%s.be/video/v3/embed/%s' % (site, video_id),
127 video_id = self._search_regex(
128 [r'data-nid="(\d+)"', r'"nid"\s*:\s*"(\d+)"'],
129 webpage, 'video id', default=video_id or display_id)
130 application = self._search_regex(
131 [r'data-application="([^"]+)"', r'"application"\s*:\s*"([^"]+)"'],
132 webpage, 'application', default=site + '_vod')
133 filename = self._search_regex(
134 [r'data-filename="([^"]+)"', r'"filename"\s*:\s*"([^"]+)"'],
137 playlist_url = 'http://vod.streamcloud.be/%s/_definst_/mp4:%s.mp4/playlist.m3u8' % (application, filename)
138 formats = self._extract_wowza_formats(playlist_url, display_id, skip_protocols=['dash'])
139 self._sort_formats(formats)
141 title = self._og_search_title(webpage, default=display_id)
142 thumbnail = self._og_search_thumbnail(webpage, default=None)
143 description = self._html_search_regex(
144 r'''(?x)<div\ class="[^"]*field-type-text-with-summary[^"]*">\s*
145 (?:<div\ class="[^"]+">\s*)*
146 <p>\s*(?:<span>)?(.+?)</''',
147 webpage, 'description', default=None)
148 episode_number = int_or_none(self._search_regex(
149 r'(?i)aflevering (\d+)', title, 'episode_number', default=None,
151 upload_date = unified_strdate(self._html_search_regex(
152 r'''(?x)<div\ class="[^"]*field-name-post-date[^"]*">\s*
153 (?:<div\ class="[^"]+">\s*)*
154 (\d{2}/\d{2}/\d{4})''',
155 webpage, 'upload_date', default=None))
159 'display_id': display_id,
161 'description': description,
162 'episode_number': episode_number,
163 'upload_date': upload_date,
164 'thumbnail': thumbnail,
169 class VierVideosIE(InfoExtractor):
170 IE_NAME = 'vier:videos'
171 _VALID_URL = r'https?://(?:www\.)?(?P<site>vier|vijf)\.be/(?P<program>[^/]+)/videos(?:\?.*\bpage=(?P<page>\d+)|$)'
173 'url': 'http://www.vier.be/demoestuin/videos',
177 'playlist_mincount': 153,
179 'url': 'http://www.vijf.be/temptationisland/videos',
181 'id': 'temptationisland',
183 'playlist_mincount': 159,
185 'url': 'http://www.vier.be/demoestuin/videos?page=6',
187 'id': 'demoestuin-page6',
189 'playlist_mincount': 20,
191 'url': 'http://www.vier.be/demoestuin/videos?page=7',
193 'id': 'demoestuin-page7',
195 'playlist_mincount': 13,
198 def _real_extract(self, url):
199 mobj = re.match(self._VALID_URL, url)
200 program = mobj.group('program')
201 site = mobj.group('site')
203 page_id = mobj.group('page')
205 page_id = int(page_id)
207 playlist_id = '%s-page%d' % (program, page_id)
210 playlist_id = program
213 for current_page_id in itertools.count(start_page):
214 current_page = self._download_webpage(
215 'http://www.%s.be/%s/videos?page=%d' % (site, program, current_page_id),
217 'Downloading page %d' % (current_page_id + 1))
219 self.url_result('http://www.' + site + '.be' + video_url, 'Vier')
220 for video_url in re.findall(
221 r'<h[23]><a href="(/[^/]+/videos/[^/]+(?:/\d+)?)">', current_page)]
222 entries.extend(page_entries)
223 if page_id or '>Meer<' not in current_page:
226 return self.playlist_result(entries, playlist_id)