From 11b15abf7e70f4b106a800cb39f47eb72ab47376 Mon Sep 17 00:00:00 2001 From: 1100101 <1100101+automatic@gmail.com> Date: Fri, 16 May 2025 12:08:16 +0200 Subject: [PATCH 1/5] Fix playlist support for arte.tv --- yt_dlp/extractor/arte.py | 42 ++++++++++++++++++++++++++++++---------- 1 file changed, 32 insertions(+), 10 deletions(-) diff --git a/yt_dlp/extractor/arte.py b/yt_dlp/extractor/arte.py index 3f17da463d..e0ca52109e 100644 --- a/yt_dlp/extractor/arte.py +++ b/yt_dlp/extractor/arte.py @@ -293,33 +293,55 @@ class ArteTVPlaylistIE(ArteTVBaseIE): 'only_matching': True, }, { 'url': 'https://www.arte.tv/pl/videos/RC-014123/arte-reportage/', - 'playlist_mincount': 100, + 'playlist_mincount': 20, 'info_dict': { 'description': 'md5:84e7bf1feda248bc325ebfac818c476e', 'id': 'RC-014123', 'title': 'ARTE Reportage - najlepsze reportaże', }, + }, { + 'url': 'https://www.arte.tv/de/videos/RC-025470/ramy/', + 'playlist_mincount': 30, + 'info_dict': { + 'description': 'md5:8766d73504ddccd12dbd1395a1d56815', + 'id': 'RC-025470', + 'title': 'Ramy', + }, }] def _real_extract(self, url): lang, playlist_id = self._match_valid_url(url).group('lang', 'id') - playlist = self._download_json( - f'{self._API_BASE}/playlist/{lang}/{playlist_id}', playlist_id)['data']['attributes'] + webpage = self._download_webpage(url, playlist_id) + + unescape_func = lambda jstring: jstring.replace('\\"', '"').replace('\\\\', '\\') + playlist_data = self._search_json(r'\$L23.+?', webpage, 'series data', + playlist_id, + end_pattern=r'\],\[\[', + transform_source=unescape_func) + + playlist_item_filter = lambda _, v: re.match(rf'collection_(?:videos|subcollection)_{playlist_id}', v['code']) + collections = traverse_obj(playlist_data, + ('data', + 'zones', + playlist_item_filter, + 'content', + 'data', + ...)) entries = [{ '_type': 'url_transparent', - 'url': video['config']['url'], + 'url': f'https://www.arte.tv{video['url']}', 'ie_key': ArteTVIE.ie_key(), - 'id': video.get('providerId'), + 'id': video.get('id'), 'title': video.get('title'), 'alt_title': video.get('subtitle'), - 'thumbnail': url_or_none(traverse_obj(video, ('mainImage', 'url'))), - 'duration': int_or_none(traverse_obj(video, ('duration', 'seconds'))), - } for video in traverse_obj(playlist, ('items', lambda _, v: v['config']['url']))] + 'duration': int_or_none(traverse_obj(video, ('duration'))), + 'age_limit': int_or_none(traverse_obj(video, 'ageRating')), + } for video in collections] return self.playlist_result(entries, playlist_id, - traverse_obj(playlist, ('metadata', 'title')), - traverse_obj(playlist, ('metadata', 'description'))) + traverse_obj(playlist_data, ('data', 'metadata', 'title')), + traverse_obj(playlist_data, ('data', 'metadata', 'description'))) class ArteTVCategoryIE(ArteTVBaseIE): From 57ae7d34c31523b43b99f7d88176082263ae0211 Mon Sep 17 00:00:00 2001 From: 1100101 <1100101+automatic@gmail.com> Date: Fri, 16 May 2025 13:04:42 +0200 Subject: [PATCH 2/5] Refactored to work with Python 3.9 --- yt_dlp/extractor/arte.py | 51 +++++++++++++++++++++------------------- 1 file changed, 27 insertions(+), 24 deletions(-) diff --git a/yt_dlp/extractor/arte.py b/yt_dlp/extractor/arte.py index e0ca52109e..1a4670b778 100644 --- a/yt_dlp/extractor/arte.py +++ b/yt_dlp/extractor/arte.py @@ -309,16 +309,7 @@ class ArteTVPlaylistIE(ArteTVBaseIE): }, }] - def _real_extract(self, url): - lang, playlist_id = self._match_valid_url(url).group('lang', 'id') - webpage = self._download_webpage(url, playlist_id) - - unescape_func = lambda jstring: jstring.replace('\\"', '"').replace('\\\\', '\\') - playlist_data = self._search_json(r'\$L23.+?', webpage, 'series data', - playlist_id, - end_pattern=r'\],\[\[', - transform_source=unescape_func) - + def _entries(self, playlist_data, playlist_id): playlist_item_filter = lambda _, v: re.match(rf'collection_(?:videos|subcollection)_{playlist_id}', v['code']) collections = traverse_obj(playlist_data, ('data', @@ -328,20 +319,32 @@ class ArteTVPlaylistIE(ArteTVBaseIE): 'data', ...)) - entries = [{ - '_type': 'url_transparent', - 'url': f'https://www.arte.tv{video['url']}', - 'ie_key': ArteTVIE.ie_key(), - 'id': video.get('id'), - 'title': video.get('title'), - 'alt_title': video.get('subtitle'), - 'duration': int_or_none(traverse_obj(video, ('duration'))), - 'age_limit': int_or_none(traverse_obj(video, 'ageRating')), - } for video in collections] - - return self.playlist_result(entries, playlist_id, - traverse_obj(playlist_data, ('data', 'metadata', 'title')), - traverse_obj(playlist_data, ('data', 'metadata', 'description'))) + for video in collections: + yield { + '_type': 'url_transparent', + 'url': 'https://www.arte.tv' + video['url'], + 'ie_key': ArteTVIE.ie_key(), + 'id': video['id'], + 'title': video.get('title'), + 'alt_title': video.get('subtitle'), + 'duration': int_or_none(traverse_obj(video, ('duration'))), + 'age_limit': int_or_none(traverse_obj(video, 'ageRating')), + } + + def _real_extract(self, url): + lang, playlist_id = self._match_valid_url(url).group('lang', 'id') + webpage = self._download_webpage(url, playlist_id) + + unescape_func = lambda jstring: jstring.replace('\\"', '"').replace('\\\\', '\\') + json_data = self._search_json(r'\$L23.+?', webpage, 'series data', + playlist_id, + end_pattern=r'\],\[\[', + transform_source=unescape_func) + + return self.playlist_result(self._entries(json_data, playlist_id), + playlist_id, + traverse_obj(json_data, ('data', 'metadata', 'title')), + traverse_obj(json_data, ('data', 'metadata', 'description'))) class ArteTVCategoryIE(ArteTVBaseIE): From 1257a8bd66f66f25b6d2cbd00ca8ded378c95638 Mon Sep 17 00:00:00 2001 From: 1100101 <1100101+automatic@gmail.com> Date: Wed, 24 Sep 2025 21:08:03 +0200 Subject: [PATCH 3/5] Refactor playlist support for arte.tv to use API endpoint, instead of parsing HTML --- yt_dlp/extractor/arte.py | 58 +++++++++++++++++++--------------------- 1 file changed, 28 insertions(+), 30 deletions(-) diff --git a/yt_dlp/extractor/arte.py b/yt_dlp/extractor/arte.py index 1a4670b778..3b2934ed9c 100644 --- a/yt_dlp/extractor/arte.py +++ b/yt_dlp/extractor/arte.py @@ -309,42 +309,40 @@ class ArteTVPlaylistIE(ArteTVBaseIE): }, }] - def _entries(self, playlist_data, playlist_id): - playlist_item_filter = lambda _, v: re.match(rf'collection_(?:videos|subcollection)_{playlist_id}', v['code']) - collections = traverse_obj(playlist_data, - ('data', - 'zones', - playlist_item_filter, - 'content', - 'data', - ...)) - - for video in collections: - yield { - '_type': 'url_transparent', - 'url': 'https://www.arte.tv' + video['url'], - 'ie_key': ArteTVIE.ie_key(), - 'id': video['id'], - 'title': video.get('title'), - 'alt_title': video.get('subtitle'), - 'duration': int_or_none(traverse_obj(video, ('duration'))), - 'age_limit': int_or_none(traverse_obj(video, 'ageRating')), - } + def _entries(self, season_ids, lang, playlist_id): + for season_id in season_ids: + season_data = self._download_json(f'{self._API_BASE}/playlist/{lang}/{season_id}', season_id, headers={ + 'x-validated-age': '18', + }) + + collection = traverse_obj(season_data, ('data', 'attributes', 'items')) + + for video in collection: + yield { + '_type': 'url_transparent', + 'url': traverse_obj(video, ('link', 'url')), + 'ie_key': ArteTVIE.ie_key(), + 'id': video['providerId'], + 'title': video.get('title'), + 'alt_title': video.get('subtitle'), + 'duration': int_or_none(traverse_obj(video, ('duration', 'seconds'))), + 'age_limit': int_or_none(traverse_obj(video, 'ageRating')), + } def _real_extract(self, url): + _API_TOKEN = 'Nzc1Yjc1ZjJkYjk1NWFhN2I2MWEwMmRlMzAzNjI5NmU3NWU3ODg4ODJjOWMxNTMxYzEzZGRjYjg2ZGE4MmIwOA' lang, playlist_id = self._match_valid_url(url).group('lang', 'id') - webpage = self._download_webpage(url, playlist_id) - unescape_func = lambda jstring: jstring.replace('\\"', '"').replace('\\\\', '\\') - json_data = self._search_json(r'\$L23.+?', webpage, 'series data', - playlist_id, - end_pattern=r'\],\[\[', - transform_source=unescape_func) + playlist_info = self._download_json(f'https://api.arte.tv/api/opa/v3/programs/{lang}/{playlist_id}', playlist_id, + headers={ + 'Authorization': f'Bearer {_API_TOKEN}', + }) - return self.playlist_result(self._entries(json_data, playlist_id), + season_ids = traverse_obj(playlist_info, ('programs', ..., 'children', (lambda _, v: v['catalogType'] == 'SEASON'), 'programId')) + return self.playlist_result(self._entries(season_ids, lang, playlist_id), playlist_id, - traverse_obj(json_data, ('data', 'metadata', 'title')), - traverse_obj(json_data, ('data', 'metadata', 'description'))) + traverse_obj(playlist_info, ('programs', ..., 'title')), + traverse_obj(playlist_info, ('programs', ..., 'shortDescription'))) class ArteTVCategoryIE(ArteTVBaseIE): From d6de27342579f5c7b5a58b8d36b8dc4257dbb3de Mon Sep 17 00:00:00 2001 From: 1100101 <1100101+automatic@gmail.com> Date: Wed, 24 Sep 2025 21:13:50 +0200 Subject: [PATCH 4/5] Fix linter issue --- yt_dlp/extractor/arte.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/yt_dlp/extractor/arte.py b/yt_dlp/extractor/arte.py index 3b2934ed9c..25b2f2b1e4 100644 --- a/yt_dlp/extractor/arte.py +++ b/yt_dlp/extractor/arte.py @@ -335,8 +335,8 @@ class ArteTVPlaylistIE(ArteTVBaseIE): playlist_info = self._download_json(f'https://api.arte.tv/api/opa/v3/programs/{lang}/{playlist_id}', playlist_id, headers={ - 'Authorization': f'Bearer {_API_TOKEN}', - }) + 'Authorization': f'Bearer {_API_TOKEN}', + }) season_ids = traverse_obj(playlist_info, ('programs', ..., 'children', (lambda _, v: v['catalogType'] == 'SEASON'), 'programId')) return self.playlist_result(self._entries(season_ids, lang, playlist_id), From 0523ca268f7357a095fb3bc3a22ac11f99a555ff Mon Sep 17 00:00:00 2001 From: Frank Aurich <1100101@gmail.com> Date: Wed, 24 Sep 2025 21:19:38 +0200 Subject: [PATCH 5/5] Fix linter issue --- yt_dlp/extractor/arte.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/yt_dlp/extractor/arte.py b/yt_dlp/extractor/arte.py index 25b2f2b1e4..e9eabad6ef 100644 --- a/yt_dlp/extractor/arte.py +++ b/yt_dlp/extractor/arte.py @@ -336,7 +336,7 @@ class ArteTVPlaylistIE(ArteTVBaseIE): playlist_info = self._download_json(f'https://api.arte.tv/api/opa/v3/programs/{lang}/{playlist_id}', playlist_id, headers={ 'Authorization': f'Bearer {_API_TOKEN}', - }) + }) season_ids = traverse_obj(playlist_info, ('programs', ..., 'children', (lambda _, v: v['catalogType'] == 'SEASON'), 'programId')) return self.playlist_result(self._entries(season_ids, lang, playlist_id),