]>
Commit | Line | Data |
---|---|---|
59040888 | 1 | # coding: utf-8 |
8c820776 S |
2 | from __future__ import unicode_literals |
3 | ||
d7df3089 RA |
4 | import re |
5 | ||
59040888 | 6 | from .common import InfoExtractor |
98b7506e RA |
7 | from ..utils import ( |
8 | int_or_none, | |
9 | js_to_json, | |
98b7506e RA |
10 | urlencode_postdata, |
11 | extract_attributes, | |
12 | smuggle_url, | |
13 | ) | |
59040888 PH |
14 | |
15 | ||
16 | class TouTvIE(InfoExtractor): | |
98b7506e | 17 | _NETRC_MACHINE = 'toutv' |
8c820776 | 18 | IE_NAME = 'tou.tv' |
e6b8803d | 19 | _VALID_URL = r'https?://ici\.tou\.tv/(?P<id>[a-zA-Z0-9_-]+(?:/S[0-9]+[EC][0-9]+)?)' |
98b7506e RA |
20 | _access_token = None |
21 | _claims = None | |
59040888 | 22 | |
de328af3 | 23 | _TESTS = [{ |
882af14d | 24 | 'url': 'http://ici.tou.tv/garfield-tout-court/S2015E17', |
8c820776 | 25 | 'info_dict': { |
882af14d | 26 | 'id': '122017', |
057c0609 | 27 | 'ext': 'mp4', |
882af14d | 28 | 'title': 'Saison 2015 Épisode 17', |
29 | 'description': 'La photo de famille 2', | |
30 | 'upload_date': '20100717', | |
59040888 | 31 | }, |
8c820776 | 32 | 'params': { |
882af14d | 33 | # m3u8 download |
34 | 'skip_download': True, | |
59040888 | 35 | }, |
98b7506e | 36 | 'skip': '404 Not Found', |
de328af3 S |
37 | }, { |
38 | 'url': 'http://ici.tou.tv/hackers', | |
39 | 'only_matching': True, | |
e6b8803d RA |
40 | }, { |
41 | 'url': 'https://ici.tou.tv/l-age-adulte/S01C501', | |
42 | 'only_matching': True, | |
de328af3 | 43 | }] |
59040888 | 44 | |
98b7506e RA |
45 | def _real_initialize(self): |
46 | email, password = self._get_login_info() | |
47 | if email is None: | |
48 | return | |
d7df3089 | 49 | state = 'http://ici.tou.tv/' |
98b7506e RA |
50 | webpage = self._download_webpage(state, None, 'Downloading homepage') |
51 | toutvlogin = self._parse_json(self._search_regex( | |
52 | r'(?s)toutvlogin\s*=\s*({.+?});', webpage, 'toutvlogin'), None, js_to_json) | |
53 | authorize_url = toutvlogin['host'] + '/auth/oauth/v2/authorize' | |
54 | login_webpage = self._download_webpage( | |
55 | authorize_url, None, 'Downloading login page', query={ | |
56 | 'client_id': toutvlogin['clientId'], | |
57 | 'redirect_uri': 'https://ici.tou.tv/login/loginCallback', | |
58 | 'response_type': 'token', | |
59 | 'scope': 'media-drmt openid profile email id.write media-validation.read.privileged', | |
60 | 'state': state, | |
61 | }) | |
d7df3089 RA |
62 | |
63 | def extract_form_url_and_data(wp, default_form_url, form_spec_re=''): | |
64 | form, form_elem = re.search( | |
65 | r'(?s)((<form[^>]+?%s[^>]*?>).+?</form>)' % form_spec_re, wp).groups() | |
66 | form_data = self._hidden_inputs(form) | |
67 | form_url = extract_attributes(form_elem).get('action') or default_form_url | |
68 | return form_url, form_data | |
69 | ||
70 | post_url, form_data = extract_form_url_and_data( | |
71 | login_webpage, | |
72 | 'https://services.radio-canada.ca/auth/oauth/v2/authorize/login', | |
73 | r'(?:id|name)="Form-login"') | |
98b7506e RA |
74 | form_data.update({ |
75 | 'login-email': email, | |
76 | 'login-password': password, | |
77 | }) | |
d7df3089 | 78 | consent_webpage = self._download_webpage( |
98b7506e | 79 | post_url, None, 'Logging in', data=urlencode_postdata(form_data)) |
d7df3089 RA |
80 | post_url, form_data = extract_form_url_and_data( |
81 | consent_webpage, | |
82 | 'https://services.radio-canada.ca/auth/oauth/v2/authorize/consent') | |
83 | _, urlh = self._download_webpage_handle( | |
84 | post_url, None, 'Following Redirection', | |
85 | data=urlencode_postdata(form_data)) | |
98b7506e RA |
86 | self._access_token = self._search_regex( |
87 | r'access_token=([\da-f]{8}-[\da-f]{4}-[\da-f]{4}-[\da-f]{4}-[\da-f]{12})', | |
88 | urlh.geturl(), 'access token') | |
89 | self._claims = self._download_json( | |
90 | 'https://services.radio-canada.ca/media/validation/v2/getClaims', | |
91 | None, 'Extracting Claims', query={ | |
92 | 'token': self._access_token, | |
93 | 'access_token': self._access_token, | |
94 | })['claims'] | |
95 | ||
59040888 | 96 | def _real_extract(self, url): |
882af14d | 97 | path = self._match_id(url) |
98 | metadata = self._download_json('http://ici.tou.tv/presentation/%s' % path, path) | |
8d7a24af S |
99 | # IsDrm does not necessarily mean the video is DRM protected (see |
100 | # https://github.com/rg3/youtube-dl/issues/13994). | |
98b7506e | 101 | if metadata.get('IsDrm'): |
8d7a24af | 102 | self.report_warning('This video is probably DRM protected.', path) |
882af14d | 103 | video_id = metadata['IdMedia'] |
104 | details = metadata['Details'] | |
105 | title = details['OriginalTitle'] | |
98b7506e RA |
106 | video_url = 'radiocanada:%s:%s' % (metadata.get('AppCode', 'toutv'), video_id) |
107 | if self._access_token and self._claims: | |
108 | video_url = smuggle_url(video_url, { | |
109 | 'access_token': self._access_token, | |
110 | 'claims': self._claims, | |
111 | }) | |
59040888 PH |
112 | |
113 | return { | |
882af14d | 114 | '_type': 'url_transparent', |
98b7506e | 115 | 'url': video_url, |
59040888 | 116 | 'id': video_id, |
882af14d | 117 | 'title': title, |
118 | 'thumbnail': details.get('ImageUrl'), | |
119 | 'duration': int_or_none(details.get('LengthInSeconds')), | |
59040888 | 120 | } |