[yt-dlp.git] / yt_dlp / extractor / cybrary.py

from .common import InfoExtractor

from ..utils import (
    ExtractorError,
    smuggle_url,
    str_or_none,
    traverse_obj,
    urlencode_postdata
)


class CybraryBaseIE(InfoExtractor):
    _API_KEY = 'AIzaSyCX9ru6j70PX2My1Eq6Q1zoMAhuTdXlzSw'
    _ENDPOINTS = {
        'course': 'https://app.cybrary.it/courses/api/catalog/browse/course/{}',
        'course_enrollment': 'https://app.cybrary.it/courses/api/catalog/{}/enrollment',
        'enrollment': 'https://app.cybrary.it/courses/api/enrollment/{}',
        'launch': 'https://app.cybrary.it/courses/api/catalog/{}/launch',
        'vimeo_oembed': 'https://vimeo.com/api/oembed.json?url=https://vimeo.com/{}',
    }
    _NETRC_MACHINE = 'cybrary'
    _TOKEN = None

    def _perform_login(self, username, password):
        CybraryBaseIE._TOKEN = self._download_json(
            f'https://identitytoolkit.googleapis.com/v1/accounts:signInWithPassword?key={self._API_KEY}',
            None, data=urlencode_postdata({'email': username, 'password': password, 'returnSecureToken': True}),
            note='Logging in')['idToken']

    def _real_initialize(self):
        if not self._TOKEN:
            self.raise_login_required(method='password')

    def _call_api(self, endpoint, item_id):
        return self._download_json(
            self._ENDPOINTS[endpoint].format(item_id), item_id,
            note=f'Downloading {endpoint} JSON metadata',
            headers={'Authorization': f'Bearer {self._TOKEN}'})

    def _get_vimeo_id(self, activity_id):
        launch_api = self._call_api('launch', activity_id)

        if launch_api.get('url'):
            return self._search_regex(r'https?://player\.vimeo\.com/video/(?P<vimeo_id>[0-9]+)', launch_api['url'], 'vimeo_id')
        return traverse_obj(launch_api, ('vendor_data', 'content', ..., 'videoId'), get_all=False)


class CybraryIE(CybraryBaseIE):
    _VALID_URL = r'https?://app.cybrary.it/immersive/(?P<enrollment>[0-9]+)/activity/(?P<id>[0-9]+)'
    _TESTS = [{
        'url': 'https://app.cybrary.it/immersive/12487950/activity/63102',
        'md5': '9ae12d37e555cb2ed554223a71a701d0',
        'info_dict': {
            'id': '646609770',
            'ext': 'mp4',
            'title': 'Getting Started',
            'thumbnail': 'https://i.vimeocdn.com/video/1301817996-76a268f0c56cff18a5cecbbdc44131eb9dda0c80eb0b3a036_1280',
            'series_id': '63111',
            'uploader_url': 'https://vimeo.com/user30867300',
            'duration': 88,
            'uploader_id': 'user30867300',
            'series': 'Cybrary Orientation',
            'uploader': 'Cybrary',
            'chapter': 'Cybrary Orientation Series',
            'chapter_id': '63110'
        },
        'expected_warnings': ['No authenticators for vimeo']
    }, {
        'url': 'https://app.cybrary.it/immersive/12747143/activity/52686',
        'md5': '62f26547dccc59c44363e2a13d4ad08d',
        'info_dict': {
            'id': '445638073',
            'ext': 'mp4',
            'title': 'Azure Virtual Network IP Addressing',
            'thumbnail': 'https://i.vimeocdn.com/video/936667051-1647ace66c627d4a2382185e0dae8deb830309bfddd53f8b2367b2f91e92ed0e-d_1280',
            'series_id': '52733',
            'uploader_url': 'https://vimeo.com/user30867300',
            'duration': 426,
            'uploader_id': 'user30867300',
            'series': 'AZ-500: Microsoft Azure Security Technologies',
            'uploader': 'Cybrary',
            'chapter': 'Implement Network Security',
            'chapter_id': '52693'
        },
        'expected_warnings': ['No authenticators for vimeo']
    }]

    def _real_extract(self, url):
        activity_id, enrollment_id = self._match_valid_url(url).group('id', 'enrollment')
        course = self._call_api('enrollment', enrollment_id)['content']
        activity = traverse_obj(course, ('learning_modules', ..., 'activities', lambda _, v: int(activity_id) == v['id']), get_all=False)

        if activity.get('type') not in ['Video Activity', 'Lesson Activity']:
            raise ExtractorError('The activity is not a video', expected=True)

        module = next((m for m in course.get('learning_modules') or []
                      if int(activity_id) in traverse_obj(m, ('activities', ..., 'id') or [])), None)

        vimeo_id = self._get_vimeo_id(activity_id)

        return {
            '_type': 'url_transparent',
            'series': traverse_obj(course, ('content_description', 'title')),
            'series_id': str_or_none(traverse_obj(course, ('content_description', 'id'))),
            'id': vimeo_id,
            'chapter': module.get('title'),
            'chapter_id': str_or_none(module.get('id')),
            'title': activity.get('title'),
            'url': smuggle_url(f'https://player.vimeo.com/video/{vimeo_id}', {'http_headers': {'Referer': 'https://api.cybrary.it'}})
        }


class CybraryCourseIE(CybraryBaseIE):
    _VALID_URL = r'https://app.cybrary.it/browse/course/(?P<id>[\w-]+)/?(?:$|[#?])'
    _TESTS = [{
        'url': 'https://app.cybrary.it/browse/course/az-500-microsoft-azure-security-technologies',
        'info_dict': {
            'id': 898,
            'title': 'AZ-500: Microsoft Azure Security Technologies',
            'description': 'md5:69549d379c0fc1dec92926d4e8b6fbd4'
        },
        'playlist_count': 59
    }, {
        'url': 'https://app.cybrary.it/browse/course/cybrary-orientation',
        'info_dict': {
            'id': 1245,
            'title': 'Cybrary Orientation',
            'description': 'md5:9e69ff66b32fe78744e0ad4babe2e88e'
        },
        'playlist_count': 4
    }]

    def _real_extract(self, url):
        course_id = self._match_id(url)
        course = self._call_api('course', course_id)
        enrollment_info = self._call_api('course_enrollment', course['id'])

        entries = [self.url_result(
            f'https://app.cybrary.it/immersive/{enrollment_info["id"]}/activity/{activity["id"]}')
            for activity in traverse_obj(course, ('content_item', 'learning_modules', ..., 'activities', ...))]

        return self.playlist_result(
            entries,
            traverse_obj(course, ('content_item', 'id'), expected_type=str_or_none),
            course.get('title'), course.get('short_description'))
Commit	Line	Data
86e5f3ed	1	from .common import InfoExtractor
12e022d0 TS	2
	3	from ..utils import (
	4	ExtractorError,
	5	smuggle_url,
	6	str_or_none,
	7	traverse_obj,
	8	urlencode_postdata
	9	)
	10
	11
	12	class CybraryBaseIE(InfoExtractor):
	13	_API_KEY = 'AIzaSyCX9ru6j70PX2My1Eq6Q1zoMAhuTdXlzSw'
	14	_ENDPOINTS = {
	15	'course': 'https://app.cybrary.it/courses/api/catalog/browse/course/{}',
	16	'course_enrollment': 'https://app.cybrary.it/courses/api/catalog/{}/enrollment',
	17	'enrollment': 'https://app.cybrary.it/courses/api/enrollment/{}',
	18	'launch': 'https://app.cybrary.it/courses/api/catalog/{}/launch',
	19	'vimeo_oembed': 'https://vimeo.com/api/oembed.json?url=https://vimeo.com/{}',
	20	}
	21	_NETRC_MACHINE = 'cybrary'
	22	_TOKEN = None
	23
	24	def _perform_login(self, username, password):
	25	CybraryBaseIE._TOKEN = self._download_json(
	26	f'https://identitytoolkit.googleapis.com/v1/accounts:signInWithPassword?key={self._API_KEY}',
	27	None, data=urlencode_postdata({'email': username, 'password': password, 'returnSecureToken': True}),
	28	note='Logging in')['idToken']
	29
	30	def _real_initialize(self):
	31	if not self._TOKEN:
	32	self.raise_login_required(method='password')
	33
	34	def _call_api(self, endpoint, item_id):
	35	return self._download_json(
	36	self._ENDPOINTS[endpoint].format(item_id), item_id,
	37	note=f'Downloading {endpoint} JSON metadata',
	38	headers={'Authorization': f'Bearer {self._TOKEN}'})
	39
	40	def _get_vimeo_id(self, activity_id):
	41	launch_api = self._call_api('launch', activity_id)
	42
	43	if launch_api.get('url'):
	44	return self._search_regex(r'https?://player\.vimeo\.com/video/(?P<vimeo_id>[0-9]+)', launch_api['url'], 'vimeo_id')
	45	return traverse_obj(launch_api, ('vendor_data', 'content', ..., 'videoId'), get_all=False)
	46
	47
	48	class CybraryIE(CybraryBaseIE):
	49	_VALID_URL = r'https?://app.cybrary.it/immersive/(?P<enrollment>[0-9]+)/activity/(?P<id>[0-9]+)'
	50	_TESTS = [{
	51	'url': 'https://app.cybrary.it/immersive/12487950/activity/63102',
	52	'md5': '9ae12d37e555cb2ed554223a71a701d0',
	53	'info_dict': {
	54	'id': '646609770',
	55	'ext': 'mp4',
	56	'title': 'Getting Started',
	57	'thumbnail': 'https://i.vimeocdn.com/video/1301817996-76a268f0c56cff18a5cecbbdc44131eb9dda0c80eb0b3a036_1280',
	58	'series_id': '63111',
	59	'uploader_url': 'https://vimeo.com/user30867300',
	60	'duration': 88,
	61	'uploader_id': 'user30867300',
	62	'series': 'Cybrary Orientation',
	63	'uploader': 'Cybrary',
	64	'chapter': 'Cybrary Orientation Series',
	65	'chapter_id': '63110'
66	},
67	'expected_warnings': ['No authenticators for vimeo']
68	}, {
69	'url': 'https://app.cybrary.it/immersive/12747143/activity/52686',
70	'md5': '62f26547dccc59c44363e2a13d4ad08d',
71	'info_dict': {
72	'id': '445638073',
73	'ext': 'mp4',
74	'title': 'Azure Virtual Network IP Addressing',
75	'thumbnail': 'https://i.vimeocdn.com/video/936667051-1647ace66c627d4a2382185e0dae8deb830309bfddd53f8b2367b2f91e92ed0e-d_1280',
76	'series_id': '52733',
77	'uploader_url': 'https://vimeo.com/user30867300',
78	'duration': 426,
79	'uploader_id': 'user30867300',
80	'series': 'AZ-500: Microsoft Azure Security Technologies',
81	'uploader': 'Cybrary',
82	'chapter': 'Implement Network Security',
83	'chapter_id': '52693'
84	},
85	'expected_warnings': ['No authenticators for vimeo']
86	}]
87
88	def _real_extract(self, url):
89	activity_id, enrollment_id = self._match_valid_url(url).group('id', 'enrollment')
90	course = self._call_api('enrollment', enrollment_id)['content']
91	activity = traverse_obj(course, ('learning_modules', ..., 'activities', lambda _, v: int(activity_id) == v['id']), get_all=False)
92
93	if activity.get('type') not in ['Video Activity', 'Lesson Activity']:
94	raise ExtractorError('The activity is not a video', expected=True)
95
96	module = next((m for m in course.get('learning_modules') or []
97	if int(activity_id) in traverse_obj(m, ('activities', ..., 'id') or [])), None)
98
99	vimeo_id = self._get_vimeo_id(activity_id)
100
101	return {
102	'_type': 'url_transparent',
103	'series': traverse_obj(course, ('content_description', 'title')),
104	'series_id': str_or_none(traverse_obj(course, ('content_description', 'id'))),
105	'id': vimeo_id,
106	'chapter': module.get('title'),
107	'chapter_id': str_or_none(module.get('id')),
108	'title': activity.get('title'),
109	'url': smuggle_url(f'https://player.vimeo.com/video/{vimeo_id}', {'http_headers': {'Referer': 'https://api.cybrary.it'}})
110	}
111
112
113	class CybraryCourseIE(CybraryBaseIE):
114	_VALID_URL = r'https://app.cybrary.it/browse/course/(?P<id>[\w-]+)/?(?:$\|[#?])'
115	_TESTS = [{
116	'url': 'https://app.cybrary.it/browse/course/az-500-microsoft-azure-security-technologies',
117	'info_dict': {
118	'id': 898,
119	'title': 'AZ-500: Microsoft Azure Security Technologies',
120	'description': 'md5:69549d379c0fc1dec92926d4e8b6fbd4'
121	},
122	'playlist_count': 59
123	}, {
124	'url': 'https://app.cybrary.it/browse/course/cybrary-orientation',
125	'info_dict': {
126	'id': 1245,
127	'title': 'Cybrary Orientation',
128	'description': 'md5:9e69ff66b32fe78744e0ad4babe2e88e'
129	},
130	'playlist_count': 4
131	}]
132
133	def _real_extract(self, url):
134	course_id = self._match_id(url)
135	course = self._call_api('course', course_id)
136	enrollment_info = self._call_api('course_enrollment', course['id'])
137
138	entries = [self.url_result(
139	f'https://app.cybrary.it/immersive/{enrollment_info["id"]}/activity/{activity["id"]}')
140	for activity in traverse_obj(course, ('content_item', 'learning_modules', ..., 'activities', ...))]
141
142	return self.playlist_result(
143	entries,
144	traverse_obj(course, ('content_item', 'id'), expected_type=str_or_none),
145	course.get('title'), course.get('short_description'))