[HRTi] Implement extractor for Croatian Radiotelevision
This commit is contained in:
		 Aleksandar Topuzovic
					Aleksandar Topuzovic
				
			
				
					committed by
					
						 Sergey M․
						Sergey M․
					
				
			
			
				
	
			
			
			 Sergey M․
						Sergey M․
					
				
			
						parent
						
							712b0b5b70
						
					
				
				
					commit
					6b03e1e25d
				
			| @@ -326,6 +326,7 @@ from .hotnewhiphop import HotNewHipHopIE | ||||
| from .hotstar import HotStarIE | ||||
| from .howcast import HowcastIE | ||||
| from .howstuffworks import HowStuffWorksIE | ||||
| from .hrti import HRTiIE | ||||
| from .huffpost import HuffPostIE | ||||
| from .hypem import HypemIE | ||||
| from .iconosquare import IconosquareIE | ||||
|   | ||||
							
								
								
									
										149
									
								
								youtube_dl/extractor/hrti.py
									
									
									
									
									
										Normal file
									
								
							
							
						
						
									
										149
									
								
								youtube_dl/extractor/hrti.py
									
									
									
									
									
										Normal file
									
								
							| @@ -0,0 +1,149 @@ | ||||
| # coding: utf-8 | ||||
| from __future__ import unicode_literals | ||||
|  | ||||
| import json | ||||
|  | ||||
| from .common import InfoExtractor | ||||
| from ..compat import ( | ||||
|     compat_HTTPError | ||||
| ) | ||||
| from ..utils import ( | ||||
|     sanitized_Request, | ||||
|     ExtractorError | ||||
| ) | ||||
|  | ||||
|  | ||||
| class HRTiIE(InfoExtractor): | ||||
|     ''' | ||||
|     Information Extractor for Croatian Radiotelevision video on demand site | ||||
|     https://hrti.hrt.hr | ||||
|     Reverse engineered from the JavaScript app in app.min.js | ||||
|     ''' | ||||
|     _NETRC_MACHINE = 'hrti' | ||||
|  | ||||
|     APP_LANGUAGE = 'hr' | ||||
|     APP_VERSION = '1.1' | ||||
|     APP_PUBLICATION_ID = 'all_in_one' | ||||
|  | ||||
|     _VALID_URL = r'https?://hrti.hrt.hr/#/video/show/(?P<id>[0-9]+)/(?P<name>(\w|-)+)?' | ||||
|     _TEST = { | ||||
|         'url': 'https://hrti.hrt.hr/#/video/show/2181385/republika-dokumentarna-serija-16-hd', | ||||
|         'info_dict': { | ||||
|             'id': '2181385', | ||||
|             'ext': 'mp4', | ||||
|             'name': 'REPUBLIKA, dokumentarna serija (4_6)-2251938', | ||||
|         }, | ||||
|         'skip': 'Requires login' | ||||
|     } | ||||
|  | ||||
|     def _initialize_api(self): | ||||
|         '''Initializes the API and obtains the required urls''' | ||||
|         api_url = 'http://clientapi.hrt.hr/client_api.php/config/identify/format/json' | ||||
|         app_data = json.dumps({ | ||||
|             'application_publication_id': HRTiIE.APP_PUBLICATION_ID | ||||
|         }) | ||||
|         self.uuid = self._download_json(api_url, None, note='Getting UUID', | ||||
|                                         errnote='Unable to obtain an UUID', | ||||
|                                         data=app_data)['uuid'] | ||||
|  | ||||
|         app_data = json.dumps({ | ||||
|             'uuid': self.uuid, | ||||
|             'application_publication_id': HRTiIE.APP_PUBLICATION_ID, | ||||
|             'screen_height': 1080, | ||||
|             'screen_width': 1920, | ||||
|             'os': 'Windows', | ||||
|             'os_version': 'NT 4.0', | ||||
|             'device_model_string_id': 'chrome 42.0.2311.135', | ||||
|             'application_version': HRTiIE.APP_VERSION | ||||
|         }) | ||||
|  | ||||
|         req = sanitized_Request(api_url, data=app_data) | ||||
|         req.get_method = lambda: 'PUT' | ||||
|  | ||||
|         resources = self._download_json( | ||||
|             req, None, note='Getting API endpoint and session information', | ||||
|             errnote='Unable to get endpoint and session information', | ||||
|             headers={'Content-type': 'application/json'}) | ||||
|  | ||||
|         self.session_id = resources['session_id'] | ||||
|         modules = resources['modules'] | ||||
|  | ||||
|         self.search_url = modules['vod_catalog']['resources']['search']['uri'] | ||||
|         self.search_url = self.search_url.format( | ||||
|             language=HRTiIE.APP_LANGUAGE, | ||||
|             application_id=HRTiIE.APP_PUBLICATION_ID) | ||||
|  | ||||
|         self.login_url = modules['user']['resources']['login']['uri'] | ||||
|         self.login_url = self.login_url.format(session_id=self.session_id) | ||||
|         self.login_url += '/format/json' | ||||
|  | ||||
|         self.logout_url = modules['user']['resources']['logout']['uri'] | ||||
|  | ||||
|     def _login(self): | ||||
|         '''Performs a login to the webservice''' | ||||
|         (username, password) = self._get_login_info() | ||||
|  | ||||
|         if username is None or password is None: | ||||
|             self.raise_login_required() | ||||
|  | ||||
|         auth_data = json.dumps({ | ||||
|             'username': username, | ||||
|             'password': password, | ||||
|         }) | ||||
|         try: | ||||
|             auth_info = self._download_json( | ||||
|                 self.login_url, None, note='Authenticating', | ||||
|                 errnote='Unable to log in', data=auth_data) | ||||
|         except ExtractorError as ee: | ||||
|             if isinstance(ee.cause, compat_HTTPError) and ee.cause.code == 406: | ||||
|                 raise ExtractorError('Unable to login, ' + | ||||
|                                      'incorrect username and/or password') | ||||
|             raise | ||||
|  | ||||
|         self.token = auth_info['secure_streaming_token'] | ||||
|         self.access_token = auth_info['session_token'] | ||||
|  | ||||
|         self.logout_url = self.logout_url.format(session_id=self.session_id, | ||||
|                                                  access_token=self.access_token) | ||||
|         self.logout_url += '/format/json' | ||||
|  | ||||
|     def _real_initialize(self): | ||||
|         '''Performs necessary operations so that the information extractor is | ||||
|         ready for operation''' | ||||
|         self._initialize_api() | ||||
|         self._login() | ||||
|  | ||||
|     def _logout(self): | ||||
|         '''Performs logout from the webservice''' | ||||
|         self._download_json(self.logout_url, None, note='Logout', | ||||
|                             errnote='Unable to log out', fatal=False) | ||||
|  | ||||
|     def _real_extract(self, url): | ||||
|         '''Extract the data necessary to download the video''' | ||||
|         video_id = self._match_id(url) | ||||
|  | ||||
|         metadata_url = self.search_url + \ | ||||
|             '/video_id/{video_id}/format/json'.format(video_id=video_id) | ||||
|  | ||||
|         metadata = self._download_json(metadata_url, video_id, | ||||
|                                        note='Getting video metadata') | ||||
|         video = metadata['video'][0] | ||||
|         title_info = video.get('title', {}) | ||||
|         title = title_info.get('title_long') | ||||
|         description = title_info.get('summary_long') | ||||
|  | ||||
|         movie = video['video_assets']['movie'][0] | ||||
|         url = movie['url'].format(TOKEN=self.token) | ||||
|  | ||||
|         formats = self._extract_m3u8_formats(url, video_id, 'mp4') | ||||
|  | ||||
|         self._sort_formats(formats) | ||||
|  | ||||
|         self._logout() | ||||
|  | ||||
|         return { | ||||
|             'id': video_id, | ||||
|             'title': title, | ||||
|             'description': description, | ||||
|             'formats': formats, | ||||
|         } | ||||
| @@ -1926,6 +1926,8 @@ def update_Request(req, url=None, data=None, headers={}, query={}): | ||||
|         origin_req_host=req.origin_req_host, unverifiable=req.unverifiable) | ||||
|     if hasattr(req, 'timeout'): | ||||
|         new_req.timeout = req.timeout | ||||
|     if req.get_method() == 'PUT': | ||||
|         new_req.get_method = lambda : 'PUT' | ||||
|     return new_req | ||||
|  | ||||
|  | ||||
|   | ||||
		Reference in New Issue
	
	Block a user