New upstream version 2020.03.24

[youtubedl] / youtube_dl / extractor / teachable.py

diff --git a/youtube_dl/extractor/teachable.py b/youtube_dl/extractor/teachable.py
index 47ac95ee8789ff0ba697cc56625e3a93f2cdded8..a75369dbe8a3582595ae339d58887eaefd220536 100644 (file)
--- a/youtube_dl/extractor/teachable.py
+++ b/youtube_dl/extractor/teachable.py
@@ -4,11 +4,12 @@ import re
 
 from .common import InfoExtractor
 from .wistia import WistiaIE
 
 from .common import InfoExtractor
 from .wistia import WistiaIE

-from ..compat import compat_str

 from ..utils import (
     clean_html,
     ExtractorError,
 from ..utils import (
     clean_html,
     ExtractorError,

+    int_or_none,

     get_element_by_class,
     get_element_by_class,

+    strip_or_none,

     urlencode_postdata,
     urljoin,
 )
     urlencode_postdata,
     urljoin,
 )


@@ -20,13 +21,14 @@ class TeachableBaseIE(InfoExtractor):
 
     _SITES = {
         # Only notable ones here
 
     _SITES = {
         # Only notable ones here

-        'upskillcourses.com': 'upskill',
-        'academy.gns3.com': 'gns3',
+        'v1.upskillcourses.com': 'upskill',
+        'gns3.teachable.com': 'gns3',

         'academyhacker.com': 'academyhacker',
         'stackskills.com': 'stackskills',
         'market.saleshacker.com': 'saleshacker',
         'learnability.org': 'learnability',
         'edurila.com': 'edurila',
         'academyhacker.com': 'academyhacker',
         'stackskills.com': 'stackskills',
         'market.saleshacker.com': 'saleshacker',
         'learnability.org': 'learnability',
         'edurila.com': 'edurila',

+        'courses.workitdaily.com': 'workitdaily',

     }
 
     _VALID_URL_SUB_TUPLE = (_URL_PREFIX, '|'.join(re.escape(site) for site in _SITES.keys()))
     }
 
     _VALID_URL_SUB_TUPLE = (_URL_PREFIX, '|'.join(re.escape(site) for site in _SITES.keys()))


@@ -47,7 +49,17 @@ class TeachableBaseIE(InfoExtractor):
             'https://%s/sign_in' % site, None,
             'Downloading %s login page' % site)
 
             'https://%s/sign_in' % site, None,
             'Downloading %s login page' % site)
 

-        login_url = compat_str(urlh.geturl())
+        def is_logged(webpage):
+            return any(re.search(p, webpage) for p in (
+                r'class=["\']user-signout',
+                r'<a[^>]+\bhref=["\']/sign_out',
+                r'Log\s+[Oo]ut\s*<'))
+
+        if is_logged(login_page):
+            self._logged_in = True
+            return
+
+        login_url = urlh.geturl()

 
         login_form = self._hidden_inputs(login_page)
 
 
         login_form = self._hidden_inputs(login_page)
 


@@ -77,10 +89,7 @@ class TeachableBaseIE(InfoExtractor):
                 'Go to https://%s/ and accept.' % (site, site), expected=True)
 
         # Successful login
                 'Go to https://%s/ and accept.' % (site, site), expected=True)
 
         # Successful login

-        if any(re.search(p, response) for p in (
-                r'class=["\']user-signout',
-                r'<a[^>]+\bhref=["\']/sign_out',
-                r'>\s*Log out\s*<')):
+        if is_logged(response):

             self._logged_in = True
             return
 
             self._logged_in = True
             return
 


@@ -102,27 +111,29 @@ class TeachableIE(TeachableBaseIE):
                     ''' % TeachableBaseIE._VALID_URL_SUB_TUPLE
 
     _TESTS = [{
                     ''' % TeachableBaseIE._VALID_URL_SUB_TUPLE
 
     _TESTS = [{

-        'url': 'http://upskillcourses.com/courses/essential-web-developer-course/lectures/1747100',
+        'url': 'https://gns3.teachable.com/courses/gns3-certified-associate/lectures/6842364',

         'info_dict': {
         'info_dict': {

-            'id': 'uzw6zw58or',
-            'ext': 'mp4',
-            'title': 'Welcome to the Course!',
-            'description': 'md5:65edb0affa582974de4625b9cdea1107',
-            'duration': 138.763,
-            'timestamp': 1479846621,
-            'upload_date': '20161122',
+            'id': 'untlgzk1v7',
+            'ext': 'bin',
+            'title': 'Overview',
+            'description': 'md5:071463ff08b86c208811130ea1c2464c',
+            'duration': 736.4,
+            'timestamp': 1542315762,
+            'upload_date': '20181115',
+            'chapter': 'Welcome',
+            'chapter_number': 1,

         },
         'params': {
             'skip_download': True,
         },
     }, {
         },
         'params': {
             'skip_download': True,
         },
     }, {

-        'url': 'http://upskillcourses.com/courses/119763/lectures/1747100',
+        'url': 'http://v1.upskillcourses.com/courses/119763/lectures/1747100',

         'only_matching': True,
     }, {
         'only_matching': True,
     }, {

-        'url': 'https://academy.gns3.com/courses/423415/lectures/6885939',
+        'url': 'https://gns3.teachable.com/courses/423415/lectures/6885939',

         'only_matching': True,
     }, {
         'only_matching': True,
     }, {

-        'url': 'teachable:https://upskillcourses.com/courses/essential-web-developer-course/lectures/1747100',
+        'url': 'teachable:https://v1.upskillcourses.com/courses/essential-web-developer-course/lectures/1747100',

         'only_matching': True,
     }]
 
         'only_matching': True,
     }]
 


@@ -152,22 +163,51 @@ class TeachableIE(TeachableBaseIE):
 
         webpage = self._download_webpage(url, video_id)
 
 
         webpage = self._download_webpage(url, video_id)
 

-        wistia_url = WistiaIE._extract_url(webpage)
-        if not wistia_url:
+        wistia_urls = WistiaIE._extract_urls(webpage)
+        if not wistia_urls:

             if any(re.search(p, webpage) for p in (
                     r'class=["\']lecture-contents-locked',
                     r'>\s*Lecture contents locked',
             if any(re.search(p, webpage) for p in (
                     r'class=["\']lecture-contents-locked',
                     r'>\s*Lecture contents locked',

-                    r'id=["\']lecture-locked')):
+                    r'id=["\']lecture-locked',
+                    # https://academy.tailoredtutors.co.uk/courses/108779/lectures/1955313
+                    r'class=["\'](?:inner-)?lesson-locked',
+                    r'>LESSON LOCKED<')):

                 self.raise_login_required('Lecture contents locked')
                 self.raise_login_required('Lecture contents locked')

+            raise ExtractorError('Unable to find video URL')

 
         title = self._og_search_title(webpage, default=None)
 
 
         title = self._og_search_title(webpage, default=None)
 

-        return {
+        chapter = None
+        chapter_number = None
+        section_item = self._search_regex(
+            r'(?s)(?P<li><li[^>]+\bdata-lecture-id=["\']%s[^>]+>.+?</li>)' % video_id,
+            webpage, 'section item', default=None, group='li')
+        if section_item:
+            chapter_number = int_or_none(self._search_regex(
+                r'data-ss-position=["\'](\d+)', section_item, 'section id',
+                default=None))
+            if chapter_number is not None:
+                sections = []
+                for s in re.findall(
+                        r'(?s)<div[^>]+\bclass=["\']section-title[^>]+>(.+?)</div>', webpage):
+                    section = strip_or_none(clean_html(s))
+                    if not section:
+                        sections = []
+                        break
+                    sections.append(section)
+                if chapter_number <= len(sections):
+                    chapter = sections[chapter_number - 1]
+
+        entries = [{

             '_type': 'url_transparent',
             'url': wistia_url,
             'ie_key': WistiaIE.ie_key(),
             'title': title,
             '_type': 'url_transparent',
             'url': wistia_url,
             'ie_key': WistiaIE.ie_key(),
             'title': title,

-        }
+            'chapter': chapter,
+            'chapter_number': chapter_number,
+        } for wistia_url in wistia_urls]
+
+        return self.playlist_result(entries, video_id, title)

 
 
 class TeachableCourseIE(TeachableBaseIE):
 
 
 class TeachableCourseIE(TeachableBaseIE):


@@ -179,20 +219,20 @@ class TeachableCourseIE(TeachableBaseIE):
                         /(?:courses|p)/(?:enrolled/)?(?P<id>[^/?#&]+)
                     ''' % TeachableBaseIE._VALID_URL_SUB_TUPLE
     _TESTS = [{
                         /(?:courses|p)/(?:enrolled/)?(?P<id>[^/?#&]+)
                     ''' % TeachableBaseIE._VALID_URL_SUB_TUPLE
     _TESTS = [{

-        'url': 'http://upskillcourses.com/courses/essential-web-developer-course/',
+        'url': 'http://v1.upskillcourses.com/courses/essential-web-developer-course/',

         'info_dict': {
             'id': 'essential-web-developer-course',
             'title': 'The Essential Web Developer Course (Free)',
         },
         'playlist_count': 192,
     }, {
         'info_dict': {
             'id': 'essential-web-developer-course',
             'title': 'The Essential Web Developer Course (Free)',
         },
         'playlist_count': 192,
     }, {

-        'url': 'http://upskillcourses.com/courses/119763/',
+        'url': 'http://v1.upskillcourses.com/courses/119763/',

         'only_matching': True,
     }, {
         'only_matching': True,
     }, {

-        'url': 'http://upskillcourses.com/courses/enrolled/119763',
+        'url': 'http://v1.upskillcourses.com/courses/enrolled/119763',

         'only_matching': True,
     }, {
         'only_matching': True,
     }, {

-        'url': 'https://academy.gns3.com/courses/enrolled/423415',
+        'url': 'https://gns3.teachable.com/courses/enrolled/423415',

         'only_matching': True,
     }, {
         'url': 'teachable:https://learn.vrdev.school/p/gear-vr-developer-mini',
         'only_matching': True,
     }, {
         'url': 'teachable:https://learn.vrdev.school/p/gear-vr-developer-mini',