[telequebec] Add support for emissions and refactor (closes #14649, closes #14655)

youtube_dl/extractor/extractors.py

@@ -1051,6 +1051,7 @@ from .telegraaf import TelegraafIE
 from .telemb import TeleMBIE
 from .telequebec import (
     TeleQuebecIE,
+    TeleQuebecEmissionIE,
     TeleQuebecLiveIE,
 )
 from .teletask import TeleTaskIE

youtube_dl/extractor/telequebec.py

@@ -10,19 +10,33 @@ from ..utils import (
 )
 
 
-class TeleQuebecIE(InfoExtractor):
+class TeleQuebecBaseIE(InfoExtractor):
+    @staticmethod
+    def _limelight_result(media_id):
+        return {
+            '_type': 'url_transparent',
+            'url': smuggle_url(
+                'limelight:media:' + media_id, {'geo_countries': ['CA']}),
+            'ie_key': 'LimelightMedia',
+        }
+
+
+class TeleQuebecIE(TeleQuebecBaseIE):
     _VALID_URL = r'https?://zonevideo\.telequebec\.tv/media/(?P<id>\d+)'
     _TESTS = [{
-        'url': 'http://zonevideo.telequebec.tv/media/20984/le-couronnement-de-new-york/couronnement-de-new-york',
-        'md5': 'fe95a0957e5707b1b01f5013e725c90f',
+        # available till 01.01.2023
+        'url': 'http://zonevideo.telequebec.tv/media/37578/un-petit-choc-et-puis-repart/un-chef-a-la-cabane',
         'info_dict': {
-            'id': '20984',
+            'id': '577116881b4b439084e6b1cf4ef8b1b3',
             'ext': 'mp4',
-            'title': 'Le couronnement de New York',
-            'description': 'md5:f5b3d27a689ec6c1486132b2d687d432',
-            'upload_date': '20170201',
-            'timestamp': 1485972222,
-        }
+            'title': 'Un petit choc et puis repart!',
+            'description': 'md5:b04a7e6b3f74e32d7b294cffe8658374',
+            'upload_date': '20180222',
+            'timestamp': 1519326631,
+        },
+        'params': {
+            'skip_download': True,
+        },
     }, {
         # no description
         'url': 'http://zonevideo.telequebec.tv/media/30261',
@@ -31,22 +45,57 @@ class TeleQuebecIE(InfoExtractor):
 
     def _real_extract(self, url):
         media_id = self._match_id(url)
+
         media_data = self._download_json(
             'https://mnmedias.api.telequebec.tv/api/v2/media/' + media_id,
             media_id)['media']
-        return {
-            '_type': 'url_transparent',
-            'id': media_id,
-            'url': smuggle_url(
-                'limelight:media:' + media_data['streamInfo']['sourceId'],
-                {'geo_countries': ['CA']}),
-            'title': media_data['title'],
+
+        info = self._limelight_result(media_data['streamInfo']['sourceId'])
+        info.update({
+            'title': media_data.get('title'),
             'description': try_get(
                 media_data, lambda x: x['descriptions'][0]['text'], compat_str),
             'duration': int_or_none(
                 media_data.get('durationInMilliseconds'), 1000),
-            'ie_key': 'LimelightMedia',
-        }
+        })
+        return info
+
+
+class TeleQuebecEmissionIE(TeleQuebecBaseIE):
+    _VALID_URL = r'https?://[^/]+\.telequebec\.tv/emissions/(?P<id>[^?#&]+)'
+    _TESTS = [{
+        'url': 'http://lindicemcsween.telequebec.tv/emissions/100430013/des-soins-esthetiques-a-377-d-interets-annuels-ca-vous-tente',
+        'info_dict': {
+            'id': '66648a6aef914fe3badda25e81a4d50a',
+            'ext': 'mp4',
+            'title': "Des soins esthétiques à 377 % d'intérêts annuels, ça vous tente?",
+            'description': 'md5:369e0d55d0083f1fc9b71ffb640ea014',
+            'upload_date': '20171024',
+            'timestamp': 1508862118,
+        },
+        'params': {
+            'skip_download': True,
+        },
+    }, {
+        'url': 'http://bancpublic.telequebec.tv/emissions/emission-49/31986/jeunes-meres-sous-pression',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        display_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, display_id)
+
+        media_id = self._search_regex(
+            r'mediaUID\s*:\s*["\'][Ll]imelight_(?P<id>[a-z0-9]{32})', webpage,
+            'limelight id')
+
+        info = self._limelight_result(media_id)
+        info.update({
+            'title': self._og_search_title(webpage, default=None),
+            'description': self._og_search_description(webpage, default=None),
+        })
+        return info
 
 
 class TeleQuebecLiveIE(InfoExtractor):