1 from __future__
import unicode_literals
5 from .common
import InfoExtractor
6 from .wistia
import WistiaIE
18 class TeachableBaseIE(InfoExtractor
):
19 _NETRC_MACHINE
= 'teachable'
20 _URL_PREFIX
= 'teachable:'
23 # Only notable ones here
24 'v1.upskillcourses.com': 'upskill',
25 'gns3.teachable.com': 'gns3',
26 'academyhacker.com': 'academyhacker',
27 'stackskills.com': 'stackskills',
28 'market.saleshacker.com': 'saleshacker',
29 'learnability.org': 'learnability',
30 'edurila.com': 'edurila',
31 'courses.workitdaily.com': 'workitdaily',
34 _VALID_URL_SUB_TUPLE
= (_URL_PREFIX
, '|'.join(re
.escape(site
) for site
in _SITES
.keys()))
36 def _real_initialize(self
):
37 self
._logged
_in
= False
39 def _login(self
, site
):
43 username
, password
= self
._get
_login
_info
(netrc_machine
=self
._SITES
.get(site
, site
))
47 login_page
, urlh
= self
._download
_webpage
_handle
(
48 'https://%s/sign_in' % site
, None,
49 'Downloading %s login page' % site
)
51 def is_logged(webpage
):
52 return any(re
.search(p
, webpage
) for p
in (
53 r
'class=["\']user
-signout
',
54 r'<a
[^
>]+\bhref
=["\']/sign_out',
57 if is_logged(login_page):
58 self._logged_in = True
61 login_url = urlh.geturl()
63 login_form = self._hidden_inputs(login_page)
66 'user[email]': username,
67 'user[password]': password,
70 post_url = self._search_regex(
71 r'<form[^>]+action=(["\'])(?P
<url
>(?
:(?
!\
1).)+)\
1', login_page,
72 'post url
', default=login_url, group='url
')
74 if not post_url.startswith('http
'):
75 post_url = urljoin(login_url, post_url)
77 response = self._download_webpage(
78 post_url, None, 'Logging
in to
%s' % site,
79 data=urlencode_postdata(login_form),
81 'Content
-Type
': 'application
/x
-www
-form
-urlencoded
',
85 if '>I accept the new Privacy Policy
<' in response:
87 'Unable to login
: %s asks you to accept new Privacy Policy
. '
88 'Go to https
://%s/ and accept
.' % (site, site), expected=True)
91 if is_logged(response):
92 self._logged_in = True
95 message = get_element_by_class('alert
', response)
96 if message is not None:
98 'Unable to login
: %s' % clean_html(message), expected=True)
100 raise ExtractorError('Unable to log
in')
103 class TeachableIE(TeachableBaseIE):
104 _VALID_URL = r'''(?x)
106 %shttps?://(?P<site_t>[^/]+)|
107 https?://(?:www\.)?(?P<site>%s)
109 /courses/[^/]+/lectures/(?P<id>\d+)
110 ''' % TeachableBaseIE._VALID_URL_SUB_TUPLE
113 'url
': 'https
://gns3
.teachable
.com
/courses
/gns3
-certified
-associate
/lectures
/6842364',
118 'description
': 'md5
:071463ff08b86c208811130ea1c2464c
',
120 'timestamp
': 1542315762,
121 'upload_date
': '20181115',
122 'chapter
': 'Welcome
',
126 'skip_download
': True,
129 'url
': 'http
://v1
.upskillcourses
.com
/courses
/119763/lectures
/1747100',
130 'only_matching
': True,
132 'url
': 'https
://gns3
.teachable
.com
/courses
/423415/lectures
/6885939',
133 'only_matching
': True,
135 'url
': 'teachable
:https
://v1
.upskillcourses
.com
/courses
/essential
-web
-developer
-course
/lectures
/1747100',
136 'only_matching
': True,
140 def _is_teachable(webpage):
141 return 'teachableTracker
.linker
:autoLink
' in webpage and re.search(
142 r'<link
[^
>]+href
=["\']https?://(?:process\.fs|assets)\.teachablecdn\.com',
146 def _extract_url(webpage, source_url):
147 if not TeachableIE._is_teachable(webpage):
149 if re.match(r'https?://[^/]+/(?:courses|p)', source_url):
150 return '%s%s' % (TeachableBaseIE._URL_PREFIX, source_url)
152 def _real_extract(self, url):
153 mobj = self._match_valid_url(url)
154 site = mobj.group('site') or mobj.group('site_t')
155 video_id = mobj.group('id')
159 prefixed = url.startswith(self._URL_PREFIX)
161 url = url[len(self._URL_PREFIX):]
163 webpage = self._download_webpage(url, video_id)
165 wistia_urls = WistiaIE._extract_urls(webpage)
167 if any(re.search(p, webpage) for p in (
168 r'class=["\']lecture
-contents
-locked
',
169 r'>\s
*Lecture contents locked
',
170 r'id=["\']lecture-locked',
171 # https://academy.tailoredtutors.co.uk/courses/108779/lectures/1955313
172 r'class=["\'](?
:inner
-)?lesson
-locked
',
173 r'>LESSON LOCKED
<')):
174 self.raise_login_required('Lecture contents locked
')
175 raise ExtractorError('Unable to find video URL
')
177 title = self._og_search_title(webpage, default=None)
180 chapter_number = None
181 section_item = self._search_regex(
182 r'(?s
)(?P
<li
><li
[^
>]+\bdata
-lecture
-id=["\']%s[^>]+>.+?</li>)' % video_id,
183 webpage, 'section item', default=None, group='li')
185 chapter_number = int_or_none(self._search_regex(
186 r'data-ss-position=["\'](\d
+)', section_item, 'section
id',
188 if chapter_number is not None:
191 r'(?s
)<div
[^
>]+\bclass
=["\']section-title[^>]+>(.+?)</div>', webpage):
192 section = strip_or_none(clean_html(s))
196 sections.append(section)
197 if chapter_number <= len(sections):
198 chapter = sections[chapter_number - 1]
201 '_type': 'url_transparent',
203 'ie_key': WistiaIE.ie_key(),
206 'chapter_number': chapter_number,
207 } for wistia_url in wistia_urls]
209 return self.playlist_result(entries, video_id, title)
212 class TeachableCourseIE(TeachableBaseIE):
213 _VALID_URL = r'''(?x)
215 %shttps?://(?P<site_t>[^/]+)|
216 https?://(?:www\.)?(?P<site>%s)
218 /(?:courses|p)/(?:enrolled/)?(?P<id>[^/?#&]+)
219 ''' % TeachableBaseIE._VALID_URL_SUB_TUPLE
221 'url': 'http://v1.upskillcourses.com/courses/essential-web-developer-course/',
223 'id': 'essential-web-developer-course',
224 'title': 'The Essential Web Developer Course (Free)',
226 'playlist_count': 192,
228 'url': 'http://v1.upskillcourses.com/courses/119763/',
229 'only_matching': True,
231 'url': 'http://v1.upskillcourses.com/courses/enrolled/119763',
232 'only_matching': True,
234 'url': 'https://gns3.teachable.com/courses/enrolled/423415',
235 'only_matching': True,
237 'url': 'teachable:https://learn.vrdev.school/p/gear-vr-developer-mini',
238 'only_matching': True,
240 'url': 'teachable:https://filmsimplified.com/p/davinci-resolve-15-crash-course',
241 'only_matching': True,
245 def suitable(cls, url):
246 return False if TeachableIE.suitable(url) else super(
247 TeachableCourseIE, cls).suitable(url)
249 def _real_extract(self, url):
250 mobj = self._match_valid_url(url)
251 site = mobj.group('site') or mobj.group('site_t')
252 course_id = mobj.group('id')
256 prefixed = url.startswith(self._URL_PREFIX)
258 prefix = self._URL_PREFIX
259 url = url[len(prefix):]
261 webpage = self._download_webpage(url, course_id)
263 url_base = 'https://%s/' % site
267 for mobj in re.finditer(
268 r'(?s)(?P<li><li[^>]+class=(["\'])(?
:(?
!\
2).)*?section
-item
[^
>]+>.+?
</li
>)',
270 li = mobj.group('li
')
271 if 'fa
-youtube
-play
' not in li and not re.search(r'\d{1,2}
:\d{2}
', li):
273 lecture_url = self._search_regex(
274 r'<a
[^
>]+href
=(["\'])(?P<url>(?:(?!\1).)+)\1', li,
275 'lecture url', default=None, group='url')
278 lecture_id = self._search_regex(
279 r'/lectures/(\d+)', lecture_url, 'lecture id', default=None)
280 title = self._html_search_regex(
281 r'<span[^>]+class=["\']lecture
-name
[^
>]+>([^
<]+)', li,
282 'title
', default=None)
283 entry_url = urljoin(url_base, lecture_url)
285 entry_url = self._URL_PREFIX + entry_url
289 ie=TeachableIE.ie_key(), video_id=lecture_id,
290 video_title=clean_html(title)))
292 course_title = self._html_search_regex(
293 (r'(?s
)<img
[^
>]+class=["\']course-image[^>]+>\s*<h\d>(.+?)</h',
294 r'(?s)<h\d[^>]+class=["\']course
-title
[^
>]+>(.+?
)</h
'),
295 webpage, 'course title
', fatal=False)
297 return self.playlist_result(entries, course_id, course_title)