yt_dlp/extractor/radiocomercial.py

@@ -17,16 +17,15 @@ from ..utils.traversal import traverse_obj
 
 
 class RadioComercialIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?radiocomercial\.pt/podcasts/[^/]+/\D*(?P<season>\d+)/(?P<id>[\w-]+)'
+    _VALID_URL = r'https?://(?:www\.)?radiocomercial\.pt/podcasts/[^/]+/\w(?P<season>\d+)/(?P<id>[-\w]+)/*$'
-    _TESTS = [
+    _TESTS = [{
-        {
         'url': 'https://radiocomercial.pt/podcasts/o-homem-que-mordeu-o-cao/t6/taylor-swift-entranhando-se-que-nem-uma-espada-no-ventre-dos-fas',
         'md5': '5f4fe8e485b29d2e8fd495605bc2c7e4',
         'info_dict': {
             'id': 'taylor-swift-entranhando-se-que-nem-uma-espada-no-ventre-dos-fas',
             'ext': 'mp3',
             'title': 'Taylor Swift entranhando-se que nem uma espada no ventre dos fãs.',
-                'description': None,
+            'description': '',
             'release_date': '20231025',
             'thumbnail': r're:https://radiocomercial.pt/upload/[^.]+.jpg',
             'season': 6
@@ -39,7 +38,7 @@ class RadioComercialIE(InfoExtractor):
             'id': 'convenca-me-num-minuto-que-os-lobisomens-existem',
             'ext': 'mp3',
             'title': 'Convença-me num minuto que os lobisomens existem',
-                'description': None,
+            'description': '',
             'release_date': '20231026',
             'thumbnail': r're:https://radiocomercial.pt/upload/[^.]+.jpg',
             'season': 3
@@ -56,23 +55,6 @@ class RadioComercialIE(InfoExtractor):
             'release_date': '20231101',
             'thumbnail': r're:https://radiocomercial.pt/upload/[^.]+.jpg',
             'season': 2
-            },
-            'params': {
-                # inconsistant md5
-                'skip_download': True,
-            },
-        },
-        {
-            'url': 'https://radiocomercial.pt/podcasts/tnt-todos-no-top/2023/t-n-t-29-de-outubro',
-            'md5': '91d32d4d4b1407272068b102730fc9fa',
-            'info_dict': {
-                'id': 't-n-t-29-de-outubro',
-                'ext': 'mp3',
-                'title': 'T.N.T 29 de outubro',
-                'description': None,
-                'release_date': '20231029',
-                'thumbnail': r're:https://radiocomercial.pt/upload/[^.]+.jpg',
-                'season': 2023
         }
     },
     ]
@@ -83,7 +65,7 @@ class RadioComercialIE(InfoExtractor):
         return {
             'id': video_id,
             'title': self._html_extract_title(webpage),
-            'description': self._og_search_description(webpage, default=None),
+            'description': self._og_search_description(webpage, default=''),
             'release_date': unified_strdate(
                 get_element_by_class('date', get_element_html_by_class('descriptions', webpage))),
             'thumbnail': self._og_search_thumbnail(webpage),
@@ -93,52 +75,49 @@ class RadioComercialIE(InfoExtractor):
 
 
 class RadioComercialPlaylistIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?radiocomercial\.pt/podcasts/(?P<id>[\w-]+)(?:\D*(?P<season>\d+))?'
+    _VALID_URL = r'https?://(?:www\.)?radiocomercial\.pt/podcasts/(?P<id>[-\w]+)[/\w\d+]*$'
     _PAGE_SIZE = 19
-    _TESTS = [
+    _TESTS = [{
-        {
         'url': 'https://radiocomercial.pt/podcasts/convenca-me-num-minuto/t3',
         'info_dict': {
             'id': 'convenca-me-num-minuto',
             'title': 'Convença-me num Minuto - Temporada 3',
         },
         'playlist_mincount': 32
-        },
+    }, {
-        {
         'url': 'https://radiocomercial.pt/podcasts/o-homem-que-mordeu-o-cao',
         'info_dict': {
             'id': 'o-homem-que-mordeu-o-cao',
             'title': 'O Homem Que Mordeu o Cão',
         },
         'playlist_mincount': 19
-        },
+    }, {
-        {
         'url': 'https://radiocomercial.pt/podcasts/as-minhas-coisas-favoritas',
         'info_dict': {
             'id': 'as-minhas-coisas-favoritas',
             'title': 'As Minhas Coisas Favoritas',
         },
         'playlist_mincount': 131
-        }
+    },
     ]
 
-    def _fetch_page(self, podcast, season, page):
+    def _fetch_page(self, url, season, page):
         page += 1
-        url = f'https://radiocomercial.pt/podcasts/{podcast}' + (f'/t{season}' if season else '') + f'/{page}'
+        next_page = f'{url}{"/" + str(page) if page > 1 else ""}'
-        playlist_id = join_nonempty(podcast, season, delim='_')
+        webpage = self._download_webpage(next_page, season, headers={'X-Requested-With': 'XMLHttpRequest'},
-        webpage = self._download_webpage(url, playlist_id, note=f'Downloading page: {page}')
+                                         note=f'Downloading page: {next_page}')
-
         episodes = set(traverse_obj(get_elements_html_by_class('tm-ouvir-podcast', webpage),
                                     (..., {extract_attributes}, 'href')))
         for entry in episodes:
             yield self.url_result(f'https://radiocomercial.pt{entry}', RadioComercialIE)
 
     def _real_extract(self, url):
-        podcast, season = self._match_valid_url(url).group('id', 'season')
+        podcast = self._match_id(url)
         webpage = self._download_webpage(url, podcast)
 
         name = try_call(lambda: get_element_text_and_html_by_tag('h1', webpage)[0])
-        title = name if name == season else join_nonempty(name, season, delim=' - Temporada ')
+        season = self._html_extract_title(webpage)
+        title = name if name == season else join_nonempty(name, season, delim=' - ')
 
-        return self.playlist_result(OnDemandPagedList(functools.partial(self._fetch_page, podcast, season),
+        return self.playlist_result(OnDemandPagedList(functools.partial(self._fetch_page, url, season),
                                                       self._PAGE_SIZE), podcast, title)