]>
Commit | Line | Data |
---|---|---|
6b03e1e2 AT |
1 | import json |
2 | ||
3 | from .common import InfoExtractor | |
e3755a62 | 4 | from ..compat import compat_HTTPError |
6b03e1e2 | 5 | from ..utils import ( |
e3755a62 S |
6 | clean_html, |
7 | ExtractorError, | |
8 | int_or_none, | |
9 | parse_age_limit, | |
6b03e1e2 | 10 | sanitized_Request, |
e3755a62 | 11 | try_get, |
6b03e1e2 AT |
12 | ) |
13 | ||
14 | ||
e3755a62 S |
15 | class HRTiBaseIE(InfoExtractor): |
16 | """ | |
17 | Base Information Extractor for Croatian Radiotelevision | |
18 | video on demand site https://hrti.hrt.hr | |
19 | Reverse engineered from the JavaScript app in app.min.js | |
20 | """ | |
6b03e1e2 AT |
21 | _NETRC_MACHINE = 'hrti' |
22 | ||
e3755a62 S |
23 | _APP_LANGUAGE = 'hr' |
24 | _APP_VERSION = '1.1' | |
25 | _APP_PUBLICATION_ID = 'all_in_one' | |
26 | _API_URL = 'http://clientapi.hrt.hr/client_api.php/config/identify/format/json' | |
52efa4b3 | 27 | _token = None |
6b03e1e2 | 28 | |
52efa4b3 | 29 | def _initialize_pre_login(self): |
e3755a62 S |
30 | init_data = { |
31 | 'application_publication_id': self._APP_PUBLICATION_ID | |
32 | } | |
33 | ||
34 | uuid = self._download_json( | |
35 | self._API_URL, None, note='Downloading uuid', | |
36 | errnote='Unable to download uuid', | |
37 | data=json.dumps(init_data).encode('utf-8'))['uuid'] | |
38 | ||
39 | app_data = { | |
40 | 'uuid': uuid, | |
41 | 'application_publication_id': self._APP_PUBLICATION_ID, | |
42 | 'application_version': self._APP_VERSION | |
43 | } | |
44 | ||
45 | req = sanitized_Request(self._API_URL, data=json.dumps(app_data).encode('utf-8')) | |
6b03e1e2 AT |
46 | req.get_method = lambda: 'PUT' |
47 | ||
48 | resources = self._download_json( | |
e3755a62 S |
49 | req, None, note='Downloading session information', |
50 | errnote='Unable to download session information') | |
51 | ||
52 | self._session_id = resources['session_id'] | |
6b03e1e2 | 53 | |
6b03e1e2 AT |
54 | modules = resources['modules'] |
55 | ||
e3755a62 S |
56 | self._search_url = modules['vod_catalog']['resources']['search']['uri'].format( |
57 | language=self._APP_LANGUAGE, | |
58 | application_id=self._APP_PUBLICATION_ID) | |
6b03e1e2 | 59 | |
3089bc74 S |
60 | self._login_url = (modules['user']['resources']['login']['uri'] |
61 | + '/format/json').format(session_id=self._session_id) | |
6b03e1e2 | 62 | |
e3755a62 | 63 | self._logout_url = modules['user']['resources']['logout']['uri'] |
6b03e1e2 | 64 | |
52efa4b3 | 65 | def _perform_login(self, username, password): |
e3755a62 | 66 | auth_data = { |
6b03e1e2 AT |
67 | 'username': username, |
68 | 'password': password, | |
e3755a62 S |
69 | } |
70 | ||
6b03e1e2 AT |
71 | try: |
72 | auth_info = self._download_json( | |
e3755a62 S |
73 | self._login_url, None, note='Logging in', errnote='Unable to log in', |
74 | data=json.dumps(auth_data).encode('utf-8')) | |
75 | except ExtractorError as e: | |
76 | if isinstance(e.cause, compat_HTTPError) and e.cause.code == 406: | |
77 | auth_info = self._parse_json(e.cause.read().encode('utf-8'), None) | |
78 | else: | |
79 | raise | |
80 | ||
81 | error_message = auth_info.get('error', {}).get('message') | |
82 | if error_message: | |
83 | raise ExtractorError( | |
84 | '%s said: %s' % (self.IE_NAME, error_message), | |
85 | expected=True) | |
86 | ||
87 | self._token = auth_info['secure_streaming_token'] | |
6b03e1e2 AT |
88 | |
89 | def _real_initialize(self): | |
52efa4b3 | 90 | if not self._token: |
91 | # TODO: figure out authentication with cookies | |
92 | self.raise_login_required(method='password') | |
6b03e1e2 | 93 | |
e3755a62 S |
94 | |
95 | class HRTiIE(HRTiBaseIE): | |
96 | _VALID_URL = r'''(?x) | |
97 | (?: | |
98 | hrti:(?P<short_id>[0-9]+)| | |
99 | https?:// | |
b0dde668 | 100 | hrti\.hrt\.hr/(?:\#/)?video/show/(?P<id>[0-9]+)/(?P<display_id>[^/]+)? |
e3755a62 S |
101 | ) |
102 | ''' | |
103 | _TESTS = [{ | |
104 | 'url': 'https://hrti.hrt.hr/#/video/show/2181385/republika-dokumentarna-serija-16-hd', | |
105 | 'info_dict': { | |
106 | 'id': '2181385', | |
107 | 'display_id': 'republika-dokumentarna-serija-16-hd', | |
108 | 'ext': 'mp4', | |
109 | 'title': 'REPUBLIKA, dokumentarna serija (1/6) (HD)', | |
110 | 'description': 'md5:48af85f620e8e0e1df4096270568544f', | |
111 | 'duration': 2922, | |
112 | 'view_count': int, | |
113 | 'average_rating': int, | |
114 | 'episode_number': int, | |
115 | 'season_number': int, | |
116 | 'age_limit': 12, | |
117 | }, | |
118 | 'skip': 'Requires account credentials', | |
119 | }, { | |
120 | 'url': 'https://hrti.hrt.hr/#/video/show/2181385/', | |
121 | 'only_matching': True, | |
122 | }, { | |
123 | 'url': 'hrti:2181385', | |
124 | 'only_matching': True, | |
b0dde668 AT |
125 | }, { |
126 | 'url': 'https://hrti.hrt.hr/video/show/3873068/cuvar-dvorca-dramska-serija-14', | |
127 | 'only_matching': True, | |
e3755a62 | 128 | }] |
6b03e1e2 AT |
129 | |
130 | def _real_extract(self, url): | |
5ad28e7f | 131 | mobj = self._match_valid_url(url) |
e3755a62 S |
132 | video_id = mobj.group('short_id') or mobj.group('id') |
133 | display_id = mobj.group('display_id') or video_id | |
6b03e1e2 | 134 | |
e3755a62 S |
135 | video = self._download_json( |
136 | '%s/video_id/%s/format/json' % (self._search_url, video_id), | |
137 | display_id, 'Downloading video metadata JSON')['video'][0] | |
6b03e1e2 | 138 | |
e3755a62 S |
139 | title_info = video['title'] |
140 | title = title_info['title_long'] | |
6b03e1e2 AT |
141 | |
142 | movie = video['video_assets']['movie'][0] | |
e3755a62 S |
143 | m3u8_url = movie['url'].format(TOKEN=self._token) |
144 | formats = self._extract_m3u8_formats( | |
145 | m3u8_url, display_id, 'mp4', entry_protocol='m3u8_native', | |
146 | m3u8_id='hls') | |
6b03e1e2 AT |
147 | self._sort_formats(formats) |
148 | ||
e3755a62 S |
149 | description = clean_html(title_info.get('summary_long')) |
150 | age_limit = parse_age_limit(video.get('parental_control', {}).get('rating')) | |
151 | view_count = int_or_none(video.get('views')) | |
152 | average_rating = int_or_none(video.get('user_rating')) | |
153 | duration = int_or_none(movie.get('duration')) | |
6b03e1e2 AT |
154 | |
155 | return { | |
156 | 'id': video_id, | |
e3755a62 | 157 | 'display_id': display_id, |
6b03e1e2 AT |
158 | 'title': title, |
159 | 'description': description, | |
e3755a62 S |
160 | 'duration': duration, |
161 | 'view_count': view_count, | |
162 | 'average_rating': average_rating, | |
163 | 'age_limit': age_limit, | |
6b03e1e2 AT |
164 | 'formats': formats, |
165 | } | |
e3755a62 S |
166 | |
167 | ||
168 | class HRTiPlaylistIE(HRTiBaseIE): | |
ae5af890 | 169 | _VALID_URL = r'https?://hrti\.hrt\.hr/(?:#/)?video/list/category/(?P<id>[0-9]+)/(?P<display_id>[^/]+)?' |
e3755a62 S |
170 | _TESTS = [{ |
171 | 'url': 'https://hrti.hrt.hr/#/video/list/category/212/ekumena', | |
172 | 'info_dict': { | |
173 | 'id': '212', | |
174 | 'title': 'ekumena', | |
175 | }, | |
176 | 'playlist_mincount': 8, | |
177 | 'skip': 'Requires account credentials', | |
178 | }, { | |
179 | 'url': 'https://hrti.hrt.hr/#/video/list/category/212/', | |
180 | 'only_matching': True, | |
ae5af890 S |
181 | }, { |
182 | 'url': 'https://hrti.hrt.hr/video/list/category/212/ekumena', | |
183 | 'only_matching': True, | |
e3755a62 S |
184 | }] |
185 | ||
186 | def _real_extract(self, url): | |
5ad28e7f | 187 | mobj = self._match_valid_url(url) |
e3755a62 S |
188 | category_id = mobj.group('id') |
189 | display_id = mobj.group('display_id') or category_id | |
190 | ||
191 | response = self._download_json( | |
192 | '%s/category_id/%s/format/json' % (self._search_url, category_id), | |
193 | display_id, 'Downloading video metadata JSON') | |
194 | ||
195 | video_ids = try_get( | |
196 | response, lambda x: x['video_listings'][0]['alternatives'][0]['list'], | |
197 | list) or [video['id'] for video in response.get('videos', []) if video.get('id')] | |
198 | ||
4cb13d0d | 199 | entries = [self.url_result('hrti:%s' % video_id) for video_id in video_ids] |
e3755a62 S |
200 | |
201 | return self.playlist_result(entries, category_id, display_id) |