2 from __future__ import unicode_literals
6 from .common import InfoExtractor
17 class SafariBaseIE(InfoExtractor):
18 _LOGIN_URL = 'https://www.safaribooksonline.com/accounts/login/'
19 _SUCCESSFUL_LOGIN_REGEX = r'<a href="/accounts/logout/"[^>]*>Sign Out</a>'
20 _NETRC_MACHINE = 'safari'
22 _API_BASE = 'https://www.safaribooksonline.com/api/v1'
27 def _real_initialize(self):
31 # We only need to log in once for courses or individual videos
35 (username, password) = self._get_login_info()
39 headers = std_headers.copy()
40 if 'Referer' not in headers:
41 headers['Referer'] = self._LOGIN_URL
42 login_page_request = sanitized_Request(self._LOGIN_URL, headers=headers)
44 login_page = self._download_webpage(
45 login_page_request, None,
46 'Downloading login form')
48 csrf = self._html_search_regex(
49 r"name='csrfmiddlewaretoken'\s+value='([^']+)'",
50 login_page, 'csrf token')
53 'csrfmiddlewaretoken': csrf,
55 'password1': password,
60 request = sanitized_Request(
61 self._LOGIN_URL, urlencode_postdata(login_form), headers=headers)
62 login_page = self._download_webpage(
63 request, None, 'Logging in as %s' % username)
65 if re.search(self._SUCCESSFUL_LOGIN_REGEX, login_page) is None:
67 'Login failed; make sure your credentials are correct and try again.',
70 SafariBaseIE.LOGGED_IN = True
72 self.to_screen('Login successful')
75 class SafariIE(SafariBaseIE):
77 IE_DESC = 'safaribooksonline.com online video'
78 _VALID_URL = r'''(?x)https?://
79 (?:www\.)?safaribooksonline\.com/
85 (?:chapter(?:-content)?/)?
86 (?P<part>part\d+)\.html
90 'url': 'https://www.safaribooksonline.com/library/view/hadoop-fundamentals-livelessons/9780133392838/part00.html',
91 'md5': 'dcc5a425e79f2564148652616af1f2a3',
95 'title': 'Introduction to Hadoop Fundamentals LiveLessons',
96 'timestamp': 1437758058,
97 'upload_date': '20150724',
98 'uploader_id': 'stork',
101 'url': 'https://www.safaribooksonline.com/api/v1/book/9780133392838/chapter/part00.html',
102 'only_matching': True,
104 # non-digits in course id
105 'url': 'https://www.safaribooksonline.com/library/view/create-a-nodejs/100000006A0210/part00.html',
106 'only_matching': True,
109 def _real_extract(self, url):
110 mobj = re.match(self._VALID_URL, url)
111 course_id = mobj.group('course_id')
112 part = mobj.group('part')
114 webpage = self._download_webpage(url, '%s/%s' % (course_id, part))
115 reference_id = self._search_regex(r'data-reference-id="([^"]+)"', webpage, 'kaltura reference id')
116 partner_id = self._search_regex(r'data-partner-id="([^"]+)"', webpage, 'kaltura widget id')
117 ui_id = self._search_regex(r'data-ui-id="([^"]+)"', webpage, 'kaltura uiconf id')
120 'wid': '_%s' % partner_id,
122 'flashvars[referenceId]': reference_id,
126 kaltura_session = self._download_json(
127 '%s/player/kaltura_session/?reference_id=%s' % (self._API_BASE, reference_id),
128 course_id, 'Downloading kaltura session JSON',
129 'Unable to download kaltura session JSON', fatal=False)
131 session = kaltura_session.get('session')
133 query['flashvars[ks]'] = session
135 return self.url_result(update_url_query(
136 'https://cdnapisec.kaltura.com/html5/html5lib/v2.37.1/mwEmbedFrame.php', query),
140 class SafariCourseIE(SafariBaseIE):
141 IE_NAME = 'safari:course'
142 IE_DESC = 'safaribooksonline.com online courses'
144 _VALID_URL = r'https?://(?:www\.)?safaribooksonline\.com/(?:library/view/[^/]+|api/v1/book)/(?P<id>[^/]+)/?(?:[#?]|$)'
147 'url': 'https://www.safaribooksonline.com/library/view/hadoop-fundamentals-livelessons/9780133392838/',
149 'id': '9780133392838',
150 'title': 'Hadoop Fundamentals LiveLessons',
152 'playlist_count': 22,
153 'skip': 'Requires safaribooksonline account credentials',
155 'url': 'https://www.safaribooksonline.com/api/v1/book/9781449396459/?override_format=json',
156 'only_matching': True,
159 def _real_extract(self, url):
160 course_id = self._match_id(url)
162 course_json = self._download_json(
163 '%s/book/%s/?override_format=%s' % (self._API_BASE, course_id, self._API_FORMAT),
164 course_id, 'Downloading course JSON')
166 if 'chapters' not in course_json:
167 raise ExtractorError(
168 'No chapters found for course %s' % course_id, expected=True)
171 self.url_result(chapter, 'Safari')
172 for chapter in course_json['chapters']]
174 course_title = course_json['title']
176 return self.playlist_result(entries, course_id, course_title)